Annotation of gforth/doc/gforth.ds, revision 1.28

1.1       anton       1: \input texinfo   @c -*-texinfo-*-
                      2: @comment The source is gforth.ds, from which gforth.texi is generated
1.28    ! crook       3: 
1.21      crook       4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
1.28    ! crook       5: @comment 1. x-ref all ambiguous or implementation-defined features?
        !             6: @comment 2. Describe the use of Auser Avariable AConstant A, etc.
        !             7: @comment 3. words in miscellaneous section need a home.
        !             8: @comment 4. search for TODO for other minor and major works required.
        !             9: @comment 5. [rats] change all @var to @i in Forth source so that info
        !            10: @comment    file looks decent.
        !            11: 
1.1       anton      12: @comment %**start of header (This is for running Texinfo on a region.)
                     13: @setfilename gforth.info
                     14: @settitle Gforth Manual
                     15: @dircategory GNU programming tools
                     16: @direntry
                     17: * Gforth: (gforth).             A fast interpreter for the Forth language.
                     18: @end direntry
                     19: @comment @setchapternewpage odd
1.12      anton      20: @macro progstyle {}
                     21: Programming style note:
1.3       anton      22: @end macro
1.1       anton      23: @comment %**end of header (This is for running Texinfo on a region.)
                     24: 
1.10      anton      25: @include version.texi
                     26: 
1.1       anton      27: @ifinfo
1.11      anton      28: This file documents Gforth @value{VERSION}
1.1       anton      29: 
1.26      crook      30: Copyright @copyright{} 1995-1999 Free Software Foundation, Inc.
1.1       anton      31: 
                     32:      Permission is granted to make and distribute verbatim copies of
                     33:      this manual provided the copyright notice and this permission notice
                     34:      are preserved on all copies.
                     35:      
                     36: @ignore
                     37:      Permission is granted to process this file through TeX and print the
                     38:      results, provided the printed document carries a copying permission
                     39:      notice identical to this one except for the removal of this paragraph
                     40:      (this paragraph not being relevant to the printed manual).
                     41:      
                     42: @end ignore
                     43:      Permission is granted to copy and distribute modified versions of this
                     44:      manual under the conditions for verbatim copying, provided also that the
                     45:      sections entitled "Distribution" and "General Public License" are
                     46:      included exactly as in the original, and provided that the entire
                     47:      resulting derived work is distributed under the terms of a permission
                     48:      notice identical to this one.
                     49:      
                     50:      Permission is granted to copy and distribute translations of this manual
                     51:      into another language, under the above conditions for modified versions,
                     52:      except that the sections entitled "Distribution" and "General Public
                     53:      License" may be included in a translation approved by the author instead
                     54:      of in the original English.
                     55: @end ifinfo
                     56: 
                     57: @finalout
                     58: @titlepage
                     59: @sp 10
                     60: @center @titlefont{Gforth Manual}
                     61: @sp 2
1.11      anton      62: @center for version @value{VERSION}
1.1       anton      63: @sp 2
                     64: @center Anton Ertl
1.6       pazsan     65: @center Bernd Paysan
1.5       anton      66: @center Jens Wilke
1.23      crook      67: @center Neal Crook
1.1       anton      68: @sp 3
1.28    ! crook      69: @center This manual is permanently under construction and was last updated on 16-Apr-1999
1.1       anton      70: 
                     71: @comment  The following two commands start the copyright page.
                     72: @page
                     73: @vskip 0pt plus 1filll
1.13      pazsan     74: Copyright @copyright{} 1995--1998 Free Software Foundation, Inc.
1.1       anton      75: 
                     76: @comment !! Published by ... or You can get a copy of this manual ...
                     77: 
                     78:      Permission is granted to make and distribute verbatim copies of
                     79:      this manual provided the copyright notice and this permission notice
                     80:      are preserved on all copies.
                     81:      
                     82:      Permission is granted to copy and distribute modified versions of this
                     83:      manual under the conditions for verbatim copying, provided also that the
                     84:      sections entitled "Distribution" and "General Public License" are
                     85:      included exactly as in the original, and provided that the entire
                     86:      resulting derived work is distributed under the terms of a permission
                     87:      notice identical to this one.
                     88:      
                     89:      Permission is granted to copy and distribute translations of this manual
                     90:      into another language, under the above conditions for modified versions,
                     91:      except that the sections entitled "Distribution" and "General Public
                     92:      License" may be included in a translation approved by the author instead
                     93:      of in the original English.
                     94: @end titlepage
                     95: 
                     96: 
                     97: @node Top, License, (dir), (dir)
                     98: @ifinfo
                     99: Gforth is a free implementation of ANS Forth available on many
1.11      anton     100: personal machines. This manual corresponds to version @value{VERSION}.
1.1       anton     101: @end ifinfo
                    102: 
                    103: @menu
1.21      crook     104: * License::                     The GPL
1.26      crook     105: * Goals::                       About the Gforth Project
1.21      crook     106: * Introduction::                An introduction to ANS Forth
1.28    ! crook     107: * Gforth Environment::          Starting (and exiting) Gforth
1.1       anton     108: * Words::                       Forth words available in Gforth
1.24      anton     109: * Error messages::              How to interpret them
1.1       anton     110: * Tools::                       Programming tools
                    111: * ANS conformance::             Implementation-defined options etc.
                    112: * Model::                       The abstract machine of Gforth
                    113: * Integrating Gforth::          Forth as scripting language for applications
                    114: * Emacs and Gforth::            The Gforth Mode
                    115: * Image Files::                 @code{.fi} files contain compiled code
                    116: * Engine::                      The inner interpreter and the primitives
1.24      anton     117: * Binding to System Library::   
1.13      pazsan    118: * Cross Compiler::              The Cross Compiler
1.1       anton     119: * Bugs::                        How to report them
                    120: * Origin::                      Authors and ancestors of Gforth
1.21      crook     121: * Forth-related information::   Books and places to look on the WWW
1.1       anton     122: * Word Index::                  An item for each Forth word
                    123: * Concept Index::               A menu covering many topics
1.12      anton     124: 
1.24      anton     125: @detailmenu --- The Detailed Node Listing ---
1.12      anton     126: 
1.26      crook     127: Goals of Gforth
                    128: 
                    129: * Gforth Extensions Sinful?::
                    130: 
1.24      anton     131: An Introduction to ANS Forth
                    132: 
                    133: * Introducing the Text Interpreter::
                    134: * Stacks and Postfix notation::
                    135: * Your first definition::
                    136: * How does that work?::
                    137: * Forth is written in Forth::
                    138: * Review - elements of a Forth system::
                    139: * Exercises::
                    140: 
1.28    ! crook     141: 
        !           142: Gforth Environment
        !           143: 
        !           144: * Invoking Gforth::
        !           145: * Leaving Gforth::
        !           146: * Command-line editing::
        !           147: * Upper and lower case::
        !           148: * Environment variables::
        !           149: * Gforth Files::
        !           150: 
        !           151: 
1.12      anton     152: Forth Words
                    153: 
                    154: * Notation::                    
1.21      crook     155: * Comments::
                    156: * Boolean Flags::
1.12      anton     157: * Arithmetic::                  
                    158: * Stack Manipulation::          
                    159: * Memory::                      
                    160: * Control Structures::          
                    161: * Defining Words::              
1.21      crook     162: * The Text Interpreter::
1.12      anton     163: * Tokens for Words::            
1.21      crook     164: * Word Lists::                   
                    165: * Environmental Queries::
1.12      anton     166: * Files::                       
                    167: * Blocks::                      
                    168: * Other I/O::                   
                    169: * Programming Tools::           
                    170: * Assembler and Code Words::    
                    171: * Threading Words::             
1.26      crook     172: * Locals::                      
                    173: * Structures::                  
                    174: * Object-oriented Forth::       
1.21      crook     175: * Passing Commands to the OS::
                    176: * Miscellaneous Words::
1.12      anton     177: 
                    178: Arithmetic
                    179: 
                    180: * Single precision::            
                    181: * Bitwise operations::          
1.21      crook     182: * Double precision::            Double-cell integer arithmetic
                    183: * Numeric comparison::
1.12      anton     184: * Mixed precision::             operations with single and double-cell integers
                    185: * Floating Point::              
                    186: 
                    187: Stack Manipulation
                    188: 
                    189: * Data stack::                  
                    190: * Floating point stack::        
                    191: * Return stack::                
                    192: * Locals stack::                
                    193: * Stack pointer manipulation::  
                    194: 
                    195: Memory
                    196: 
1.27      crook     197: * Reserving Data Space::
1.12      anton     198: * Memory Access::      
1.27      crook     199: * Address Arithmetic::          
                    200: * Memory Blocks::
                    201: * Dynamic Allocation::        
1.12      anton     202: 
                    203: Control Structures
                    204: 
                    205: * Selection::                   
                    206: * Simple Loops::                
                    207: * Counted Loops::               
                    208: * Arbitrary control structures::  
                    209: * Calls and returns::           
                    210: * Exception Handling::          
                    211: 
                    212: Defining Words
                    213: 
                    214: * Simple Defining Words::       
                    215: * Colon Definitions::           
                    216: * User-defined Defining Words::  
                    217: * Supplying names::             
                    218: * Interpretation and Compilation Semantics::  
                    219: 
1.21      crook     220: The Text Interpreter
                    221: 
                    222: * Number Conversion::
                    223: * Interpret/Compile states::
                    224: * Literals::
                    225: * Interpreter Directives::
1.27      crook     226: * Input Sources::
1.21      crook     227: 
1.26      crook     228: Word Lists
                    229: 
                    230: * Why use word lists?::
                    231: * Word list examples::
                    232: 
                    233: Files
                    234: 
                    235: * Forth source files::
                    236: * General files::         
                    237: * Search Paths::                 
                    238: * Forth Search Paths::    
                    239: * General Search Paths::        
                    240: 
                    241: Other I/O
                    242: 
                    243: * Simple numeric output::
                    244: * Formatted numeric output::
                    245: * String Formats::
                    246: * Displaying characters and strings::
                    247: * Input::
                    248: 
                    249: Programming Tools
                    250: 
                    251: * Debugging::                   Simple and quick.
                    252: * Assertions::                  Making your programs self-checking.
                    253: * Singlestep Debugger::                Executing your program word by word.
                    254: 
                    255: Locals
                    256: 
                    257: * Gforth locals::               
                    258: * ANS Forth locals::            
                    259: 
                    260: Gforth locals
                    261: 
                    262: * Where are locals visible by name?::  
                    263: * How long do locals live?::    
                    264: * Programming Style::           
                    265: * Implementation::              
                    266: 
1.12      anton     267: Structures
                    268: 
                    269: * Why explicit structure support?::  
                    270: * Structure Usage::             
                    271: * Structure Naming Convention::  
                    272: * Structure Implementation::    
                    273: * Structure Glossary::          
                    274: 
                    275: Object-oriented Forth
                    276: 
1.24      anton     277: * Why object-oriented programming?::
                    278: * Object-Oriented Terminology::
                    279: * Objects::
                    280: * OOF::
                    281: * Mini-OOF::
1.23      crook     282: * Comparison with other object models::  
1.12      anton     283: 
1.24      anton     284: The @file{objects.fs} model
1.12      anton     285: 
                    286: * Properties of the Objects model::  
                    287: * Basic Objects Usage::         
1.23      crook     288: * The Objects base class::            
1.12      anton     289: * Creating objects::            
                    290: * Object-Oriented Programming Style::  
                    291: * Class Binding::               
                    292: * Method conveniences::         
                    293: * Classes and Scoping::         
                    294: * Object Interfaces::           
                    295: * Objects Implementation::      
                    296: * Objects Glossary::            
                    297: 
1.24      anton     298: The @file{oof.fs} model
1.12      anton     299: 
                    300: * Properties of the OOF model::
                    301: * Basic OOF Usage::
1.23      crook     302: * The OOF base class::
1.12      anton     303: * Class Declaration::
                    304: * Class Implementation::
                    305: 
1.24      anton     306: The @file{mini-oof.fs} model
1.23      crook     307: 
                    308: * Basic Mini-OOF Usage::
                    309: * Mini-OOF Example::
                    310: * Mini-OOF Implementation::
                    311: 
1.12      anton     312: Tools
                    313: 
                    314: * ANS Report::                  Report the words used, sorted by wordset.
                    315: 
                    316: ANS conformance
                    317: 
                    318: * The Core Words::              
                    319: * The optional Block word set::  
                    320: * The optional Double Number word set::  
                    321: * The optional Exception word set::  
                    322: * The optional Facility word set::  
                    323: * The optional File-Access word set::  
                    324: * The optional Floating-Point word set::  
                    325: * The optional Locals word set::  
                    326: * The optional Memory-Allocation word set::  
                    327: * The optional Programming-Tools word set::  
                    328: * The optional Search-Order word set::  
                    329: 
                    330: The Core Words
                    331: 
                    332: * core-idef::                   Implementation Defined Options                   
                    333: * core-ambcond::                Ambiguous Conditions                
                    334: * core-other::                  Other System Documentation                  
                    335: 
                    336: The optional Block word set
                    337: 
                    338: * block-idef::                  Implementation Defined Options
                    339: * block-ambcond::               Ambiguous Conditions               
                    340: * block-other::                 Other System Documentation                 
                    341: 
                    342: The optional Double Number word set
                    343: 
                    344: * double-ambcond::              Ambiguous Conditions              
                    345: 
                    346: The optional Exception word set
                    347: 
                    348: * exception-idef::              Implementation Defined Options              
                    349: 
                    350: The optional Facility word set
                    351: 
                    352: * facility-idef::               Implementation Defined Options               
                    353: * facility-ambcond::            Ambiguous Conditions            
                    354: 
                    355: The optional File-Access word set
                    356: 
                    357: * file-idef::                   Implementation Defined Options
                    358: * file-ambcond::                Ambiguous Conditions                
                    359: 
                    360: The optional Floating-Point word set
                    361: 
                    362: * floating-idef::               Implementation Defined Options
                    363: * floating-ambcond::            Ambiguous Conditions            
                    364: 
                    365: The optional Locals word set
                    366: 
                    367: * locals-idef::                 Implementation Defined Options                 
                    368: * locals-ambcond::              Ambiguous Conditions              
                    369: 
                    370: The optional Memory-Allocation word set
                    371: 
                    372: * memory-idef::                 Implementation Defined Options                 
                    373: 
                    374: The optional Programming-Tools word set
                    375: 
                    376: * programming-idef::            Implementation Defined Options            
                    377: * programming-ambcond::         Ambiguous Conditions         
                    378: 
                    379: The optional Search-Order word set
                    380: 
                    381: * search-idef::                 Implementation Defined Options                 
                    382: * search-ambcond::              Ambiguous Conditions              
                    383: 
                    384: Image Files
                    385: 
1.24      anton     386: * Image Licensing Issues::      Distribution terms for images.
                    387: * Image File Background::       Why have image files?
                    388: * Non-Relocatable Image Files::  don't always work.
                    389: * Data-Relocatable Image Files::  are better.
1.12      anton     390: * Fully Relocatable Image Files::  better yet.
1.24      anton     391: * Stack and Dictionary Sizes::  Setting the default sizes for an image.
                    392: * Running Image Files::         @code{gforth -i @var{file}} or @var{file}.
                    393: * Modifying the Startup Sequence::  and turnkey applications.
1.12      anton     394: 
                    395: Fully Relocatable Image Files
                    396: 
1.27      crook     397: * gforthmi::                    The normal way
1.12      anton     398: * cross.fs::                    The hard way
                    399: 
                    400: Engine
                    401: 
                    402: * Portability::                 
                    403: * Threading::                   
                    404: * Primitives::                  
                    405: * Performance::                 
                    406: 
                    407: Threading
                    408: 
                    409: * Scheduling::                  
                    410: * Direct or Indirect Threaded?::  
                    411: * DOES>::                       
                    412: 
                    413: Primitives
                    414: 
                    415: * Automatic Generation::        
                    416: * TOS Optimization::            
                    417: * Produced code::               
1.13      pazsan    418: 
                    419: Cross Compiler
                    420: 
                    421: * Using the Cross Compiler::
                    422: * How the Cross Compiler Works::
                    423: 
1.24      anton     424: Other Forth-related information
1.21      crook     425: 
                    426: * Internet resources::
                    427: * Books::
                    428: * The Forth Interest Group::
                    429: * Conferences::
                    430: 
1.24      anton     431: @end detailmenu
1.1       anton     432: @end menu
                    433: 
1.26      crook     434: @node License, Goals, Top, Top
1.1       anton     435: @unnumbered GNU GENERAL PUBLIC LICENSE
                    436: @center Version 2, June 1991
                    437: 
                    438: @display
                    439: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
                    440: 675 Mass Ave, Cambridge, MA 02139, USA
                    441: 
                    442: Everyone is permitted to copy and distribute verbatim copies
                    443: of this license document, but changing it is not allowed.
                    444: @end display
                    445: 
                    446: @unnumberedsec Preamble
                    447: 
                    448:   The licenses for most software are designed to take away your
                    449: freedom to share and change it.  By contrast, the GNU General Public
                    450: License is intended to guarantee your freedom to share and change free
                    451: software---to make sure the software is free for all its users.  This
                    452: General Public License applies to most of the Free Software
                    453: Foundation's software and to any other program whose authors commit to
                    454: using it.  (Some other Free Software Foundation software is covered by
                    455: the GNU Library General Public License instead.)  You can apply it to
                    456: your programs, too.
                    457: 
                    458:   When we speak of free software, we are referring to freedom, not
                    459: price.  Our General Public Licenses are designed to make sure that you
                    460: have the freedom to distribute copies of free software (and charge for
                    461: this service if you wish), that you receive source code or can get it
                    462: if you want it, that you can change the software or use pieces of it
                    463: in new free programs; and that you know you can do these things.
                    464: 
                    465:   To protect your rights, we need to make restrictions that forbid
                    466: anyone to deny you these rights or to ask you to surrender the rights.
                    467: These restrictions translate to certain responsibilities for you if you
                    468: distribute copies of the software, or if you modify it.
                    469: 
                    470:   For example, if you distribute copies of such a program, whether
                    471: gratis or for a fee, you must give the recipients all the rights that
                    472: you have.  You must make sure that they, too, receive or can get the
                    473: source code.  And you must show them these terms so they know their
                    474: rights.
                    475: 
                    476:   We protect your rights with two steps: (1) copyright the software, and
                    477: (2) offer you this license which gives you legal permission to copy,
                    478: distribute and/or modify the software.
                    479: 
                    480:   Also, for each author's protection and ours, we want to make certain
                    481: that everyone understands that there is no warranty for this free
                    482: software.  If the software is modified by someone else and passed on, we
                    483: want its recipients to know that what they have is not the original, so
                    484: that any problems introduced by others will not reflect on the original
                    485: authors' reputations.
                    486: 
                    487:   Finally, any free program is threatened constantly by software
                    488: patents.  We wish to avoid the danger that redistributors of a free
                    489: program will individually obtain patent licenses, in effect making the
                    490: program proprietary.  To prevent this, we have made it clear that any
                    491: patent must be licensed for everyone's free use or not licensed at all.
                    492: 
                    493:   The precise terms and conditions for copying, distribution and
                    494: modification follow.
                    495: 
                    496: @iftex
                    497: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
                    498: @end iftex
                    499: @ifinfo
                    500: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
                    501: @end ifinfo
                    502: 
                    503: @enumerate 0
                    504: @item
                    505: This License applies to any program or other work which contains
                    506: a notice placed by the copyright holder saying it may be distributed
                    507: under the terms of this General Public License.  The ``Program'', below,
                    508: refers to any such program or work, and a ``work based on the Program''
                    509: means either the Program or any derivative work under copyright law:
                    510: that is to say, a work containing the Program or a portion of it,
                    511: either verbatim or with modifications and/or translated into another
                    512: language.  (Hereinafter, translation is included without limitation in
                    513: the term ``modification''.)  Each licensee is addressed as ``you''.
                    514: 
                    515: Activities other than copying, distribution and modification are not
                    516: covered by this License; they are outside its scope.  The act of
                    517: running the Program is not restricted, and the output from the Program
                    518: is covered only if its contents constitute a work based on the
                    519: Program (independent of having been made by running the Program).
                    520: Whether that is true depends on what the Program does.
                    521: 
                    522: @item
                    523: You may copy and distribute verbatim copies of the Program's
                    524: source code as you receive it, in any medium, provided that you
                    525: conspicuously and appropriately publish on each copy an appropriate
                    526: copyright notice and disclaimer of warranty; keep intact all the
                    527: notices that refer to this License and to the absence of any warranty;
                    528: and give any other recipients of the Program a copy of this License
                    529: along with the Program.
                    530: 
                    531: You may charge a fee for the physical act of transferring a copy, and
                    532: you may at your option offer warranty protection in exchange for a fee.
                    533: 
                    534: @item
                    535: You may modify your copy or copies of the Program or any portion
                    536: of it, thus forming a work based on the Program, and copy and
                    537: distribute such modifications or work under the terms of Section 1
                    538: above, provided that you also meet all of these conditions:
                    539: 
                    540: @enumerate a
                    541: @item
                    542: You must cause the modified files to carry prominent notices
                    543: stating that you changed the files and the date of any change.
                    544: 
                    545: @item
                    546: You must cause any work that you distribute or publish, that in
                    547: whole or in part contains or is derived from the Program or any
                    548: part thereof, to be licensed as a whole at no charge to all third
                    549: parties under the terms of this License.
                    550: 
                    551: @item
                    552: If the modified program normally reads commands interactively
                    553: when run, you must cause it, when started running for such
                    554: interactive use in the most ordinary way, to print or display an
                    555: announcement including an appropriate copyright notice and a
                    556: notice that there is no warranty (or else, saying that you provide
                    557: a warranty) and that users may redistribute the program under
                    558: these conditions, and telling the user how to view a copy of this
                    559: License.  (Exception: if the Program itself is interactive but
                    560: does not normally print such an announcement, your work based on
                    561: the Program is not required to print an announcement.)
                    562: @end enumerate
                    563: 
                    564: These requirements apply to the modified work as a whole.  If
                    565: identifiable sections of that work are not derived from the Program,
                    566: and can be reasonably considered independent and separate works in
                    567: themselves, then this License, and its terms, do not apply to those
                    568: sections when you distribute them as separate works.  But when you
                    569: distribute the same sections as part of a whole which is a work based
                    570: on the Program, the distribution of the whole must be on the terms of
                    571: this License, whose permissions for other licensees extend to the
                    572: entire whole, and thus to each and every part regardless of who wrote it.
                    573: 
                    574: Thus, it is not the intent of this section to claim rights or contest
                    575: your rights to work written entirely by you; rather, the intent is to
                    576: exercise the right to control the distribution of derivative or
                    577: collective works based on the Program.
                    578: 
                    579: In addition, mere aggregation of another work not based on the Program
                    580: with the Program (or with a work based on the Program) on a volume of
                    581: a storage or distribution medium does not bring the other work under
                    582: the scope of this License.
                    583: 
                    584: @item
                    585: You may copy and distribute the Program (or a work based on it,
                    586: under Section 2) in object code or executable form under the terms of
                    587: Sections 1 and 2 above provided that you also do one of the following:
                    588: 
                    589: @enumerate a
                    590: @item
                    591: Accompany it with the complete corresponding machine-readable
                    592: source code, which must be distributed under the terms of Sections
                    593: 1 and 2 above on a medium customarily used for software interchange; or,
                    594: 
                    595: @item
                    596: Accompany it with a written offer, valid for at least three
                    597: years, to give any third party, for a charge no more than your
                    598: cost of physically performing source distribution, a complete
                    599: machine-readable copy of the corresponding source code, to be
                    600: distributed under the terms of Sections 1 and 2 above on a medium
                    601: customarily used for software interchange; or,
                    602: 
                    603: @item
                    604: Accompany it with the information you received as to the offer
                    605: to distribute corresponding source code.  (This alternative is
                    606: allowed only for noncommercial distribution and only if you
                    607: received the program in object code or executable form with such
                    608: an offer, in accord with Subsection b above.)
                    609: @end enumerate
                    610: 
                    611: The source code for a work means the preferred form of the work for
                    612: making modifications to it.  For an executable work, complete source
                    613: code means all the source code for all modules it contains, plus any
                    614: associated interface definition files, plus the scripts used to
                    615: control compilation and installation of the executable.  However, as a
                    616: special exception, the source code distributed need not include
                    617: anything that is normally distributed (in either source or binary
                    618: form) with the major components (compiler, kernel, and so on) of the
                    619: operating system on which the executable runs, unless that component
                    620: itself accompanies the executable.
                    621: 
                    622: If distribution of executable or object code is made by offering
                    623: access to copy from a designated place, then offering equivalent
                    624: access to copy the source code from the same place counts as
                    625: distribution of the source code, even though third parties are not
                    626: compelled to copy the source along with the object code.
                    627: 
                    628: @item
                    629: You may not copy, modify, sublicense, or distribute the Program
                    630: except as expressly provided under this License.  Any attempt
                    631: otherwise to copy, modify, sublicense or distribute the Program is
                    632: void, and will automatically terminate your rights under this License.
                    633: However, parties who have received copies, or rights, from you under
                    634: this License will not have their licenses terminated so long as such
                    635: parties remain in full compliance.
                    636: 
                    637: @item
                    638: You are not required to accept this License, since you have not
                    639: signed it.  However, nothing else grants you permission to modify or
                    640: distribute the Program or its derivative works.  These actions are
                    641: prohibited by law if you do not accept this License.  Therefore, by
                    642: modifying or distributing the Program (or any work based on the
                    643: Program), you indicate your acceptance of this License to do so, and
                    644: all its terms and conditions for copying, distributing or modifying
                    645: the Program or works based on it.
                    646: 
                    647: @item
                    648: Each time you redistribute the Program (or any work based on the
                    649: Program), the recipient automatically receives a license from the
                    650: original licensor to copy, distribute or modify the Program subject to
                    651: these terms and conditions.  You may not impose any further
                    652: restrictions on the recipients' exercise of the rights granted herein.
                    653: You are not responsible for enforcing compliance by third parties to
                    654: this License.
                    655: 
                    656: @item
                    657: If, as a consequence of a court judgment or allegation of patent
                    658: infringement or for any other reason (not limited to patent issues),
                    659: conditions are imposed on you (whether by court order, agreement or
                    660: otherwise) that contradict the conditions of this License, they do not
                    661: excuse you from the conditions of this License.  If you cannot
                    662: distribute so as to satisfy simultaneously your obligations under this
                    663: License and any other pertinent obligations, then as a consequence you
                    664: may not distribute the Program at all.  For example, if a patent
                    665: license would not permit royalty-free redistribution of the Program by
                    666: all those who receive copies directly or indirectly through you, then
                    667: the only way you could satisfy both it and this License would be to
                    668: refrain entirely from distribution of the Program.
                    669: 
                    670: If any portion of this section is held invalid or unenforceable under
                    671: any particular circumstance, the balance of the section is intended to
                    672: apply and the section as a whole is intended to apply in other
                    673: circumstances.
                    674: 
                    675: It is not the purpose of this section to induce you to infringe any
                    676: patents or other property right claims or to contest validity of any
                    677: such claims; this section has the sole purpose of protecting the
                    678: integrity of the free software distribution system, which is
                    679: implemented by public license practices.  Many people have made
                    680: generous contributions to the wide range of software distributed
                    681: through that system in reliance on consistent application of that
                    682: system; it is up to the author/donor to decide if he or she is willing
                    683: to distribute software through any other system and a licensee cannot
                    684: impose that choice.
                    685: 
                    686: This section is intended to make thoroughly clear what is believed to
                    687: be a consequence of the rest of this License.
                    688: 
                    689: @item
                    690: If the distribution and/or use of the Program is restricted in
                    691: certain countries either by patents or by copyrighted interfaces, the
                    692: original copyright holder who places the Program under this License
                    693: may add an explicit geographical distribution limitation excluding
                    694: those countries, so that distribution is permitted only in or among
                    695: countries not thus excluded.  In such case, this License incorporates
                    696: the limitation as if written in the body of this License.
                    697: 
                    698: @item
                    699: The Free Software Foundation may publish revised and/or new versions
                    700: of the General Public License from time to time.  Such new versions will
                    701: be similar in spirit to the present version, but may differ in detail to
                    702: address new problems or concerns.
                    703: 
                    704: Each version is given a distinguishing version number.  If the Program
                    705: specifies a version number of this License which applies to it and ``any
                    706: later version'', you have the option of following the terms and conditions
                    707: either of that version or of any later version published by the Free
                    708: Software Foundation.  If the Program does not specify a version number of
                    709: this License, you may choose any version ever published by the Free Software
                    710: Foundation.
                    711: 
                    712: @item
                    713: If you wish to incorporate parts of the Program into other free
                    714: programs whose distribution conditions are different, write to the author
                    715: to ask for permission.  For software which is copyrighted by the Free
                    716: Software Foundation, write to the Free Software Foundation; we sometimes
                    717: make exceptions for this.  Our decision will be guided by the two goals
                    718: of preserving the free status of all derivatives of our free software and
                    719: of promoting the sharing and reuse of software generally.
                    720: 
                    721: @iftex
                    722: @heading NO WARRANTY
                    723: @end iftex
                    724: @ifinfo
                    725: @center NO WARRANTY
                    726: @end ifinfo
                    727: 
                    728: @item
                    729: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
                    730: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
                    731: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
                    732: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
                    733: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
                    734: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
                    735: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
                    736: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
                    737: REPAIR OR CORRECTION.
                    738: 
                    739: @item
                    740: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
                    741: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
                    742: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
                    743: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
                    744: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
                    745: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
                    746: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
                    747: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
                    748: POSSIBILITY OF SUCH DAMAGES.
                    749: @end enumerate
                    750: 
                    751: @iftex
                    752: @heading END OF TERMS AND CONDITIONS
                    753: @end iftex
                    754: @ifinfo
                    755: @center END OF TERMS AND CONDITIONS
                    756: @end ifinfo
                    757: 
                    758: @page
                    759: @unnumberedsec How to Apply These Terms to Your New Programs
                    760: 
                    761:   If you develop a new program, and you want it to be of the greatest
                    762: possible use to the public, the best way to achieve this is to make it
                    763: free software which everyone can redistribute and change under these terms.
                    764: 
                    765:   To do so, attach the following notices to the program.  It is safest
                    766: to attach them to the start of each source file to most effectively
                    767: convey the exclusion of warranty; and each file should have at least
                    768: the ``copyright'' line and a pointer to where the full notice is found.
                    769: 
                    770: @smallexample
                    771: @var{one line to give the program's name and a brief idea of what it does.}
                    772: Copyright (C) 19@var{yy}  @var{name of author}
                    773: 
                    774: This program is free software; you can redistribute it and/or modify 
                    775: it under the terms of the GNU General Public License as published by 
                    776: the Free Software Foundation; either version 2 of the License, or 
                    777: (at your option) any later version.
                    778: 
                    779: This program is distributed in the hope that it will be useful,
                    780: but WITHOUT ANY WARRANTY; without even the implied warranty of
                    781: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
                    782: GNU General Public License for more details.
                    783: 
                    784: You should have received a copy of the GNU General Public License
                    785: along with this program; if not, write to the Free Software
                    786: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
                    787: @end smallexample
                    788: 
                    789: Also add information on how to contact you by electronic and paper mail.
                    790: 
                    791: If the program is interactive, make it output a short notice like this
                    792: when it starts in an interactive mode:
                    793: 
                    794: @smallexample
                    795: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
                    796: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
                    797: type `show w'.  
                    798: This is free software, and you are welcome to redistribute it 
                    799: under certain conditions; type `show c' for details.
                    800: @end smallexample
                    801: 
                    802: The hypothetical commands @samp{show w} and @samp{show c} should show
                    803: the appropriate parts of the General Public License.  Of course, the
                    804: commands you use may be called something other than @samp{show w} and
                    805: @samp{show c}; they could even be mouse-clicks or menu items---whatever
                    806: suits your program.
                    807: 
                    808: You should also get your employer (if you work as a programmer) or your
                    809: school, if any, to sign a ``copyright disclaimer'' for the program, if
                    810: necessary.  Here is a sample; alter the names:
                    811: 
                    812: @smallexample
                    813: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
                    814: `Gnomovision' (which makes passes at compilers) written by James Hacker.
                    815: 
                    816: @var{signature of Ty Coon}, 1 April 1989
                    817: Ty Coon, President of Vice
                    818: @end smallexample
                    819: 
                    820: This General Public License does not permit incorporating your program into
                    821: proprietary programs.  If your program is a subroutine library, you may
                    822: consider it more useful to permit linking proprietary applications with the
                    823: library.  If this is what you want to do, use the GNU Library General
                    824: Public License instead of this License.
                    825: 
                    826: @iftex
                    827: @unnumbered Preface
                    828: @cindex Preface
1.21      crook     829: This manual documents Gforth. Some introductory material is provided for
                    830: readers who are unfamiliar with Forth or who are migrating to Gforth
                    831: from other Forth compilers. However, this manual is primarily a
                    832: reference manual.
1.1       anton     833: @end iftex
                    834: 
1.28    ! crook     835: @comment TODO much more blurb here.
1.26      crook     836: 
                    837: @c ******************************************************************
                    838: @node Goals, Introduction, License, Top
                    839: @comment node-name,     next,           previous, up
                    840: @chapter Goals of Gforth
                    841: @cindex goals of the Gforth project
                    842: The goal of the Gforth Project is to develop a standard model for
                    843: ANS Forth. This can be split into several subgoals:
                    844: 
                    845: @itemize @bullet
                    846: @item
                    847: Gforth should conform to the ANS Forth Standard.
                    848: @item
                    849: It should be a model, i.e. it should define all the
                    850: implementation-dependent things.
                    851: @item
                    852: It should become standard, i.e. widely accepted and used. This goal
                    853: is the most difficult one.
                    854: @end itemize
                    855: 
                    856: To achieve these goals Gforth should be
                    857: @itemize @bullet
                    858: @item
                    859: Similar to previous models (fig-Forth, F83)
                    860: @item
                    861: Powerful. It should provide for all the things that are considered
                    862: necessary today and even some that are not yet considered necessary.
                    863: @item
                    864: Efficient. It should not get the reputation of being exceptionally
                    865: slow.
                    866: @item
                    867: Free.
                    868: @item
                    869: Available on many machines/easy to port.
                    870: @end itemize
                    871: 
                    872: Have we achieved these goals? Gforth conforms to the ANS Forth
                    873: standard. It may be considered a model, but we have not yet documented
                    874: which parts of the model are stable and which parts we are likely to
                    875: change. It certainly has not yet become a de facto standard, but it
                    876: appears to be quite popular. It has some similarities to and some
                    877: differences from previous models. It has some powerful features, but not
                    878: yet everything that we envisioned. We certainly have achieved our
                    879: execution speed goals (@pxref{Performance}).  It is free and available
                    880: on many machines.
                    881: 
                    882: @menu
                    883: * Gforth Extensions Sinful?::
                    884: @end menu
                    885: 
                    886: @node Gforth Extensions Sinful?, , Goals, Goals
                    887: @comment node-name,     next,           previous, up
                    888: @section Is it a Sin to use Gforth Extensions?
                    889: @cindex Gforth extensions
                    890: 
                    891: If you've been paying attention, you will have realised that there is an
                    892: ANS (American National Standard) for Forth. As you read through the rest
                    893: of this manual, you will see documentation for @var{Standard} words, and
                    894: documentation for some appealing Gforth @var{extensions}. You might ask
                    895: yourself the question: @var{``Given that there is a standard, would I be
                    896: committing a sin to use (non-Standard) Gforth extensions?''}
                    897: 
                    898: The answer to that question is somewhat pragmatic and somewhat
                    899: philosophical. Consider these points:
                    900: 
                    901: @itemize @bullet
                    902: @item
                    903: A number of the Gforth extensions can be implemented in ANS Forth using
                    904: files provided in the @file{compat/} directory. These are mentioned in
                    905: the text in passing.
                    906: @item
                    907: Forth has a rich historical precedent for programmers taking advantage
                    908: of implementation-dependent features of their tools (for example,
                    909: relying on a knowledge of the dictionary structure). Sometimes these
                    910: techniques are necessary to extract every last bit of performance from
                    911: the hardware, sometimes they are just a programming shorthand.
                    912: @item
                    913: The best way to break the rules is to know what the rules are. To learn
                    914: the rules, there is no substitute for studying the text of the Standard
                    915: itself. In particular, Appendix A of the Standard (@var{Rationale})
                    916: provides a valuable insight into the thought processes of the technical
                    917: committee.
                    918: @item
                    919: The best reason to break a rule is because you have to; because it's
                    920: more productive to do that, because it makes your code run fast enough
                    921: or because you can see no Standard way to achieve what you want to
                    922: achieve.
                    923: @end itemize
                    924: 
                    925: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
                    926: analyse your program and determine what non-Standard definitions it
                    927: relies upon.
                    928: 
                    929: @c ******************************************************************
1.28    ! crook     930: @node    Introduction, Gforth Environment, Goals, Top
1.21      crook     931: @comment node-name,     next,           previous, up
                    932: @chapter An Introduction to ANS Forth
                    933: @cindex Forth - an introduction
                    934: 
                    935: The primary purpose of this manual is to document Gforth. However, since
                    936: Forth is not a widely-known language and there is a lack of up-to-date
                    937: teaching material, it seems worthwhile to provide some introductory
                    938: material. @xref{Forth-related information} for other sources of Forth-related
                    939: information.
                    940: 
1.26      crook     941: The examples in this section should work on any ANS Forth; the
                    942: output shown was produced using Gforth. Each example attempts to
1.21      crook     943: reproduce the exact output that Gforth produces. If you try out the
                    944: examples (and you should), what you should type is shown @kbd{like this}
                    945: and Gforth's response is shown @code{like this}. The single exception is
                    946: that, where the example shows @kbd{<return>} it means that you should
1.26      crook     947: press the ``carriage return'' key. Unfortunately, some output formats for
1.21      crook     948: this manual cannot show the difference between @kbd{this} and
                    949: @code{this} which will make trying out the examples harder (but not
                    950: impossible).
                    951: 
                    952: Forth is an unusual language. It provides an interactive development
                    953: environment which includes both an interpreter and compiler. Forth
                    954: programming style encourages you to break a problem down into many
                    955: @cindex factoring
                    956: small fragments (@var{factoring}), and then to develop and test each
                    957: fragment interactively. Forth advocates assert that breaking the
                    958: edit-compile-test cycle used by conventional programming languages can
                    959: lead to great productivity improvements.
                    960: 
                    961: @menu
                    962: * Introducing the Text Interpreter::
                    963: * Stacks and Postfix notation::
                    964: * Your first definition::
                    965: * How does that work?::
                    966: * Forth is written in Forth::
                    967: * Review - elements of a Forth system::
                    968: * Exercises::
                    969: @end menu
                    970: 
                    971: @comment ----------------------------------------------
                    972: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
                    973: @section Introducing the Text Interpreter
                    974: @cindex text interpreter
                    975: @cindex outer interpreter
                    976: 
                    977: When you invoke the Forth image, you will see a startup banner printed
                    978: and nothing else (if you have Gforth installed on your system, try
                    979: invoking it now, by typing @kbd{gforth<return>}). Forth is now running
                    980: its command line interpreter, which is called the @var{Text Interpreter}
1.26      crook     981: (also known as the @var{Outer Interpreter}).  (You will learn a lot
                    982: about the text interpreter as you read through this chapter,
                    983: but @pxref{The Text Interpreter} for more detail).
1.21      crook     984: 
1.26      crook     985: Although it's not obvious, Forth is actually waiting for your
1.21      crook     986: input. Type a number and press the <return> key:
                    987: 
                    988: @example
                    989: @kbd{45<return>}  ok
                    990: @end example
                    991: 
                    992: Rather than give you a prompt to invite you to input something, the text
                    993: interpreter prints a status message @var{after} it has processed a line
1.26      crook     994: of input. The status message in this case (``@code{ ok}'' followed by
1.21      crook     995: carriage-return) indicates that the text interpreter was able to process
                    996: all of your input successfully. Now type something illegal:
                    997: 
                    998: @example
                    999: @kbd{qwer341<return>}
1.26      crook    1000: :1: Undefined word
                   1001: qwer341
1.21      crook    1002: ^^^^^^^
1.26      crook    1003: $400D2BA8 Bounce
                   1004: $400DBDA8 no.extensions
1.21      crook    1005: @end example
                   1006: 
1.26      crook    1007: The exact text, other than the ``Undefined word'' may differ slightly on
                   1008: your system, but the effect is the same; when the text interpreter
                   1009: detects an error, it discards any remaining text on a line, resets
                   1010: certain internal state and prints an error message.
                   1011: 
1.27      crook    1012: The text interpreter waits for you to press carriage-return, and then
1.26      crook    1013: processes your input line. Starting at the beginning of the line, it
                   1014: breaks the line into groups of characters separated by spaces. For each
                   1015: group of characters in turn, it makes two attempts to do something:
1.21      crook    1016: 
                   1017: @itemize @bullet
                   1018: @item
                   1019: It tries to treat it as a command. It does this by searching a @var{name
                   1020: dictionary}. If the group of characters matches an entry in the name
                   1021: dictionary, the name dictionary provides the text interpreter with
                   1022: information that allows the text interpreter perform some actions. In
                   1023: Forth jargon, we say that the group
                   1024: @cindex word
                   1025: @cindex definition
                   1026: @cindex execution token
                   1027: @cindex xt
                   1028: of characters names a @var{word}, that the dictionary search returns an
                   1029: @var{execution token (xt)} corresponding to the @var{definition} of the
                   1030: word, and that the text interpreter executes the xt. Often, the terms
                   1031: @var{word} and @var{definition} are used interchangeably.
                   1032: @item
                   1033: If the text interpreter fails to find a match in the name dictionary, it
                   1034: tries to treat the group of characters as a number in the current number
                   1035: base (when you start up Forth, the current number base is base 10). If
                   1036: the group of characters legitimately represents a number, the text
                   1037: interpreter pushes the number onto a stack (we'll learn more about that
                   1038: in the next section).
                   1039: @end itemize
                   1040: 
                   1041: If the text interpreter is unable to do either of these things with any
1.26      crook    1042: group of characters, it discards the group of characters and the rest of
                   1043: the line, then prints an error message. If the text interpreter reaches
                   1044: the end of the line without error, it prints the status message ``@code{ ok}''
                   1045: followed by carriage-return.
1.21      crook    1046: 
                   1047: This is the simplest command we can give to the text interpreter:
                   1048: 
                   1049: @example
                   1050: @kbd{<return>}  ok
                   1051: @end example
                   1052: 
                   1053: The text interpreter did everything we asked it to do (nothing) without
1.26      crook    1054: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1.21      crook    1055: command:
                   1056: 
                   1057: @example
                   1058: @kbd{12 dup fred dup<return>}
1.26      crook    1059: :1: Undefined word
                   1060: 12 dup fred dup
1.21      crook    1061:        ^^^^
1.26      crook    1062: $400D2BA8 Bounce
                   1063: $400DBDA8 no.extensions
1.21      crook    1064: @end example
                   1065: 
1.26      crook    1066: When you press the carriage-return key, the text interpreter starts to
                   1067: work its way along the line:
1.21      crook    1068: 
                   1069: @itemize @bullet
                   1070: @item
                   1071: When it gets to the space after the @code{2}, it takes the group of
                   1072: characters @code{12} and looks them up in the name
                   1073: dictionary@footnote{We can't tell if it found them or not, but assume
                   1074: for now that it did not}. There is no match for this group of characters
                   1075: in the name dictionary, so it tries to treat them as a number. It is
1.26      crook    1076: able to do this successfully, so it puts the number, 12, ``on the stack''
1.21      crook    1077: (whatever that means).
                   1078: @item
                   1079: The text interpreter resumes scanning the line and gets the next group
1.26      crook    1080: of characters, @code{dup}. It looks it up in the name dictionary and
                   1081: (you'll have to take my word for this) finds it, and executes the word
1.21      crook    1082: @code{dup} (whatever that means).
                   1083: @item
                   1084: Once again, the text interpreter resumes scanning the line and gets the
                   1085: group of characters @code{fred}. It looks them up in the name
                   1086: dictionary, but can't find them. It tries to treat them as a number, but
                   1087: they don't represent any legal number.
                   1088: @end itemize
                   1089: 
                   1090: At this point, the text interpreter gives up and prints an error
                   1091: message. The error message shows exactly how far the text interpreter
                   1092: got in processing the line. In particular, it shows that the text
                   1093: interpreter made no attempt to do anything with the final character
                   1094: group, @code{dup}, even though we have good reason to believe that the
                   1095: text interpreter would have had no problems with looking that word up
                   1096: and executing it a second time.
                   1097: 
                   1098: 
                   1099: @comment ----------------------------------------------
                   1100: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
                   1101: @section Stacks, postfix notation and parameter passing
                   1102: @cindex text interpreter
                   1103: @cindex outer interpreter
                   1104: 
                   1105: In procedural programming languages (like C and Pascal), the
1.26      crook    1106: building-block of programs is the @var{function} or @var{procedure}. These
                   1107: functions or procedures are called with @var{explicit parameters}. For
1.21      crook    1108: example, in C we might write:
                   1109: 
                   1110: @example
                   1111: total = total + new_volume(length,height,depth);
                   1112: @end example
                   1113: 
1.26      crook    1114: @noindent
                   1115: where new_volume is a function-call to another piece of code, and total,
                   1116: length, height and depth are all variables. length, height and depth are
                   1117: parameters to the function-call.
1.21      crook    1118: 
1.26      crook    1119: In Forth, the equivalent of the function or procedure is the
1.21      crook    1120: @var{definition} and parameters are implicitly passed between
                   1121: definitions using a shared stack that is visible to the
                   1122: programmer. Although Forth does support variables, the existence of the
                   1123: stack means that they are used far less often than in most other
                   1124: programming languages. When the text interpreter encounters a number, it
                   1125: will place (@var{push}) it on the stack. There are several stacks (the
                   1126: actual number is implementation-dependent ..) and the particular stack
                   1127: used for any operation is implied unambiguously by the operation being
                   1128: performed. The stack used for all integer operations is called the @var{data
                   1129: stack} and, since this is the stack used most commonly, references to
1.26      crook    1130: ``the data stack'' are often abbreviated to ``the stack''.
1.21      crook    1131: 
                   1132: The stacks have a last-in, first-out (LIFO) organisation. If you type:
                   1133: 
                   1134: @example
                   1135: @kbd{1 2 3<return>}  ok
                   1136: @end example
                   1137: 
1.26      crook    1138: Then this instructs the text interpreter to placed three numbers on the
                   1139: (data) stack. An analogy for the behaviour of the stack is to take a
                   1140: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
                   1141: the table. The 3 was the last card onto the pile (``last-in'') and if
                   1142: you take a card off the pile then, unless you're prepared to fiddle a
                   1143: bit, the card that you take off will be the 3 (``first-out''). The
                   1144: number that will be first-out of the stack is called the @var{top of
                   1145: stack}, which
                   1146: @cindex TOS definition
1.21      crook    1147: is often abbreviated to @var{TOS}.
                   1148: 
1.26      crook    1149: To understand how parameters are passed in Forth, consider the
                   1150: behaviour of the definition @code{+} (pronounced ``plus''). You will not
                   1151: be surprised to learn that this definition performs addition. More
1.21      crook    1152: precisely, it adds two number together and produces a result. Where does
1.26      crook    1153: it get the two numbers from? It takes the top two numbers off the
1.21      crook    1154: stack. Where does it place the result? On the stack. You can act-out the
                   1155: behaviour of @code{+} with your playing cards like this:
                   1156: 
                   1157: @itemize @bullet
                   1158: @item
1.26      crook    1159: Pick up two cards from the stack on the table
1.21      crook    1160: @item
1.26      crook    1161: Stare at them intently and ask yourself ``what @var{is} the sum of these two
                   1162: numbers''
1.21      crook    1163: @item
                   1164: Decide that the answer is 5
                   1165: @item
                   1166: Shuffle the two cards back into the pack and find a 5
                   1167: @item
                   1168: Put a 5 on the remaining ace that's on the table.
                   1169: @end itemize
                   1170: 
                   1171: If you don't have a pack of cards handy but you do have Forth running,
1.26      crook    1172: you can use the definition @code{.s} to show the current state of the stack,
1.21      crook    1173: without affecting the stack. Type:
                   1174: 
                   1175: @example
                   1176: @kbd{clearstack 1 2 3<return>} ok
1.26      crook    1177: @kbd{.s<return>} <3> 1 2 3  ok
1.21      crook    1178: @end example
                   1179: 
                   1180: The text interpreter looks up the word @code{clearstack} and executes
                   1181: it; it tidies up the stack and removes any entries that may have been
                   1182: left on it by earlier examples. The text interpreter pushes each of the
                   1183: three numbers in turn onto the stack. Finally, the text interpreter
                   1184: looks up the word @code{.s} and executes it. The effect of executing
1.26      crook    1185: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
                   1186: followed by a list of all the items on the stack; the item on the far
                   1187: right-hand side is the TOS.
1.21      crook    1188: 
                   1189: You can now type:
                   1190: 
1.26      crook    1191: @example
                   1192: @kbd{+ .s<return>} <2> 1 5  ok
                   1193: @end example
1.21      crook    1194: 
1.26      crook    1195: @noindent
1.21      crook    1196: which is correct; there are now 2 items on the stack and the result of
                   1197: the addition is 5.
                   1198: 
1.26      crook    1199: If you're playing with cards, try doing a second addition: pick up the
1.21      crook    1200: two cards, work out that their sum is 6, shuffle them into the pack,
1.26      crook    1201: look for a 6 and place that on the table. You now have just one item on
                   1202: the stack. What happens if you try to do a third addition? Pick up the
                   1203: first card, pick up the second card -- ah! There is no second card. This
                   1204: is called a @var{stack underflow} and consitutes an error. If you try to
                   1205: do the same thing with Forth it will report an error (probably a Stack
                   1206: Underflow or an Invalid Memory Address error).
                   1207: 
                   1208: The opposite situation to a stack underflow is a @var{stack overflow},
                   1209: which simply accepts that there is a finite amount of storage space
                   1210: reserved for the stack. To stretch the playing card analogy, if you had
                   1211: enough packs of cards and you piled the cards up on the table, you would
                   1212: eventually be unable to add another card; you'd hit the ceiling. Gforth
                   1213: allows you to set the maximum size of the stacks. In general, the only
                   1214: time that you will get a stack overflow is because a definition has a
                   1215: bug in it and is generating data on the stack uncontrollably.
1.21      crook    1216: 
                   1217: There's one final use for the playing card analogy. If you model your
                   1218: stack using a pack of playing cards, the maximum number of items on
                   1219: your stack will be 52 (I assume you didn't use the Joker). The maximum
1.26      crook    1220: @var{value} of any item on the stack is 13 (the King). In fact, the only
1.21      crook    1221: possible numbers are positive integer numbers 1 through 13; you can't
                   1222: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
                   1223: think about some of the cards, you can accommodate different
                   1224: numbers. For example, you could think of the Jack as representing 0,
                   1225: the Queen as representing -1 and the King as representing -2. Your
                   1226: *range* remains unchanged (you can still only represent a total of 13
                   1227: numbers) but the numbers that you can represent are -2 through 10.
                   1228: 
                   1229: In that analogy, the limit was the amount of information that a single
                   1230: stack entry could hold, and Forth has a similar limit. In Forth, the
1.26      crook    1231: size of a stack entry is called a @var{cell}. The actual size of a cell is
1.21      crook    1232: implementation dependent and affects the maximum value that a stack
                   1233: entry can hold. A Standard Forth provides a cell size of at least
                   1234: 16-bits, and most desktop systems use a cell size of 32-bits.
                   1235: 
                   1236: Forth does not do any type checking for you, so you are free to
                   1237: manipulate and combine stack items in any way you wish. A convenient
                   1238: ways of treating stack items is as 2's complement signed integers, and
1.26      crook    1239: that is what Standard words like ``+'' do. Therefore you can type:
1.21      crook    1240: 
1.26      crook    1241: @example
                   1242: @kbd{-5 12 + .s<return>} <1> 7  ok
                   1243: @end example
1.21      crook    1244: 
1.26      crook    1245: If you use numbers and definitions like ``+'' in order to turn Forth
1.21      crook    1246: into a great big pocket calculator, you will realise that it's rather
                   1247: different from a normal calculator. Rather than typing 2 + 3 = you had
1.26      crook    1248: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1.21      crook    1249: result). The terminology used to describe this difference is to say
1.26      crook    1250: that your calculator uses @var{Infix Notation} (parameters and operators
                   1251: are mixed) whilst Forth uses @var{Postfix Notation} (parameters and
                   1252: operators are separate), also called @var{Reverse Polish Notation}.
1.21      crook    1253: 
                   1254: Whilst postfix notation might look confusing to begin with, it has
                   1255: several important advantages:
                   1256: 
1.26      crook    1257: @itemize @bullet
                   1258: @item
                   1259: it is unambiguous
                   1260: @item
                   1261: it is more concise
                   1262: @item
                   1263: it fits naturally with a stack-based system
                   1264: @end itemize
1.21      crook    1265: 
                   1266: To examine these claims in more detail, consider these sums:
                   1267: 
1.26      crook    1268: @example
1.21      crook    1269: 6 + 5 * 4 =
                   1270: 4 * 5 + 6 =
1.26      crook    1271: @end example
1.21      crook    1272: 
                   1273: If you're just learning maths or your maths is very rusty, you will
                   1274: probably come up with the answer 44 for the first and 26 for the
                   1275: second. If you are a bit of a whizz at maths you will remember the
1.26      crook    1276: @var{convention} that multiplication takes precendence over addition, and
1.21      crook    1277: you'd come up with the answer 26 both times. To explain the answer 26
                   1278: to someone who got the answer 44, you'd probably rewrite the first sum
                   1279: like this:
                   1280: 
1.26      crook    1281: @example
1.21      crook    1282: 6 + (5 * 4) =
1.26      crook    1283: @end example
1.21      crook    1284: 
                   1285: If what you really wanted was to perform the addition before the
                   1286: multiplication, you would have to use parentheses to force it.
                   1287: 
                   1288: If you did the first two sums on a pocket calculator you would probably
                   1289: get the right answers, unless you were very cautious and entered them using
                   1290: these keystroke sequences:
                   1291: 
                   1292: 6 + 5 = * 4 =
                   1293: 4 * 5 = + 6 =
                   1294: 
                   1295: Postfix notation is unambiguous because the order that the operators
                   1296: are applied is always explicit; that also means that parentheses are
1.26      crook    1297: never required. The operators are @var{active} (the act of quoting the
                   1298: operator makes the operation occur) which removes the need for ``=''.
1.21      crook    1299: 
                   1300: The sum 6 + 5 * 4 can be written (in postfix notation) in two
                   1301: equivalent ways:
                   1302: 
1.26      crook    1303: @example
1.21      crook    1304: 6 5 4 * +      or:
                   1305: 5 4 * 6 +
1.26      crook    1306: @end example
1.21      crook    1307: 
1.23      crook    1308: An important thing that you should notice about this notation is that
                   1309: the @var{order} of the numbers does not change; if you want to subtract
                   1310: 2 from 10 you type @code{10 2 -}.
                   1311: 
1.26      crook    1312: The reason that Forth uses postfix notation is very simple to explain: it
1.23      crook    1313: makes the implementation extremely simple, and it follows naturally from
                   1314: using the stack as a mechanism for passing parameters. Another way of
                   1315: thinking about this is to realise that all Forth definitions are
                   1316: @var{active}; they execute as they are encountered by the text
1.26      crook    1317: interpreter. The result of this is that the syntax of Forth is trivially
                   1318: simple.
1.23      crook    1319: 
                   1320: 
                   1321: 
                   1322: @comment ----------------------------------------------
                   1323: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
                   1324: @section Your first Forth definition
                   1325: @cindex first definition
1.21      crook    1326: 
1.23      crook    1327: Until now, the examples we've seen have been trivial; we've just been
                   1328: using Forth an a bigger-than-pocket calculator. Also, each calculation
1.26      crook    1329: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1.23      crook    1330: again@footnote{That's not quite true. If you press the up-arrow key on
                   1331: your keyboard you should be able to scroll back to any earlier command,
                   1332: edit it and re-enter it.} In this section we'll see how to add new
                   1333: word to Forth's vocabulary.
                   1334: 
1.26      crook    1335: The easiest way to create a new word is to use a @var{colon
                   1336: definition}. We'll define a few and try them out before we worry too
1.23      crook    1337: much about how they work. Try typing in these examples; be careful to
                   1338: copy the spaces accurately:
                   1339: 
                   1340: @example
                   1341: : add-two 2 + . ;
                   1342: : greet ." Hello and welcome" ;
                   1343: : demo 5 add-two ;
                   1344: @end example
1.21      crook    1345: 
1.23      crook    1346: @noindent
                   1347: Now try them out:
1.21      crook    1348: 
1.23      crook    1349: @example
                   1350: @kbd{greet<return>} Hello and welcome  ok
                   1351: @kbd{greet greet<return>} Hello and welcomeHello and welcome  ok
                   1352: @kbd{4 add-two<return>} 6  ok
                   1353: @kbd{demo<return>} 7  ok
                   1354: @kbd{9 greet demo add-two<return>} Hello and welcome7 11  ok
                   1355: @end example
1.21      crook    1356: 
1.23      crook    1357: The first new thing that we've introduced here is the pair of words
                   1358: @code{:} and @code{;}. These are used to start and terminate a new
                   1359: definition, respectively. The first word after the @code{:} is the name
                   1360: for the new definition.
1.21      crook    1361: 
1.23      crook    1362: As you can see from the examples, a definition is built up of words that
                   1363: have already been defined; Forth makes no distinction between
                   1364: definitions that existed when you started the system up, and those that
                   1365: you define yourself.
1.21      crook    1366: 
1.23      crook    1367: The examples also introduce the words @code{.} (dot), @code{."} (dot-quote)
                   1368: and @code{dup} (dewp). Dot takes the value from the top of the stack and
                   1369: displays it. It's like @code{.s} except that it only displays the top
                   1370: item of the stack and it is destructive; after it has executed the
                   1371: number is no longer on the top of the stack. There is always one space
                   1372: printed after the number, and no spaces before it. Dot-quote defines a
                   1373: string (a sequence of characters) that will be printed when the word is
                   1374: executed. The string can contain any printable characters except
                   1375: @code{"}. A @code{"} has a special function; it is not itself a Forth
                   1376: word but it acts as a delimiter. The way that it works is described in
                   1377: the next section. Finally, @code{dup} duplicates the value at the top of
                   1378: the stack. Try typing @code{5 dup .s} to see what it does.
1.21      crook    1379: 
1.23      crook    1380: We already know that the text interpreter searches through the
                   1381: dictionary to locate names. If you've followed the examples earlier, you
                   1382: will already have a definition called @code{add-two}. Lets try modifying
                   1383: it by typing in a new definition:
1.21      crook    1384: 
1.23      crook    1385: @example
                   1386: @kbd{: add-two dup . ." + 2 =" 2 + . ;<return>} redefined add-two  ok
                   1387: @end example
1.21      crook    1388: 
1.23      crook    1389: Forth recognised that we were defining a word that already exists, and
                   1390: printed a message to warn us of that fact. Let's try out the new
                   1391: definition:
1.21      crook    1392: 
1.23      crook    1393: @example
                   1394: @kbd{9 add-two<return>} 9 + 2 =11  ok
                   1395: @end example
1.21      crook    1396: 
1.23      crook    1397: @noindent
                   1398: All that we've actually done here, though, is to create a new
                   1399: definition, with a particular name. The fact that there was already a
                   1400: definition with the same name did not make any difference to the way
                   1401: that the new definition was created (except that Forth printed a warning
                   1402: message). The old definition of add-two still exists (try @code{demo}
                   1403: again to see that this is true). Any new definition will use the new
                   1404: definition of @code{add-two}, but old definitions continue to use the
                   1405: version that already existed at the time that they were @code{compiled}.
1.21      crook    1406: 
1.23      crook    1407: Before you go on to the next section, try defining and redefining some
                   1408: words of your own.
1.21      crook    1409: 
                   1410: @comment ----------------------------------------------
                   1411: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
                   1412: @section How does that work?
                   1413: @cindex parsing words
                   1414: 
1.23      crook    1415: Now we're going to take another look at the definition of @code{add-two}
                   1416: from the previous section. From our knowledge of the way that the text
                   1417: interpreter works, we would have expected this result when we tried to
                   1418: define @code{add-two}:
1.21      crook    1419: 
1.23      crook    1420: @example
                   1421: @kbd{: add-two 2 + . " ;<return>}
                   1422:   ^^^^^^^
                   1423: Error: Undefined word
                   1424: @end example
1.21      crook    1425: 
1.23      crook    1426: The reason that this didn't happen is bound up in the way that @code{:}
                   1427: works. The word @code{:} does two special things. The first special
                   1428: thing that it does prevents the text interpreter from ever seeing the
                   1429: characters @code{add-two}. The text interpreter uses a variable called
                   1430: @cindex modifying >IN
                   1431: @code{>IN} (pronounced ''to-in'') to keep track of where it is in the
                   1432: input line. When it encounters the word @code{:} it behaves in exactly
                   1433: the same way as it does for any other word; it looks it up in the name
                   1434: dictionary, finds its xt and executes it. When @code{:} executes, it
                   1435: looks at the input buffer, finds the word @code{add-two} and advances the
                   1436: value of @code{>IN} to point past it. It then does some other stuff
                   1437: associated with creating the new definition (including creating an entry
                   1438: for @code{add-two} in the name dictionary). When the execution of @code{:}
                   1439: completes, control returns to the text interpreter, which is oblivious
                   1440: to the fact that it has been tricked into ignoring part of the input
                   1441: line.
1.21      crook    1442: 
1.23      crook    1443: @cindex parsing words
                   1444: Words like @code{:} -- words that advance the value of @code{>IN} and so
                   1445: prevent the text interpreter from acting on the whole of the input line
                   1446: -- are called @var{parsing words}.
                   1447: 
1.28    ! crook    1448: @cindex @code{state} - effect on the text interpreter
1.23      crook    1449: @cindex text interpreter - effect of state
                   1450: The second special thing that @code{:} does is to change the value of a
                   1451: variable called @code{state}, which affects the way that the text
                   1452: interpreter behaves. When Gforth starts up, @code{state} has the value
                   1453: 0, and the text interpreter is said to be in @var{interpret}
                   1454: mode. During a colon definition (started with @code{:}), @code{state} is
                   1455: set to -1 and the text interpreter is said to be in @var{compile}
                   1456: mode. The word @code{;} ends the definition -- one of the things that it
                   1457: does is to change the value of @code{state} back to 0.
                   1458: 
                   1459: When the text interpreter is in @var{interpret} mode, we already know
                   1460: how it behaves; it looks for each character sequence in the dictionary,
                   1461: finds its xt and executes it, or it converts it to a number and pushes
                   1462: it onto the stack, or it fails to do either and generates an error.
                   1463: 
                   1464: When the text interpreter is in @var{compile} mode, its behaviour is
                   1465: slightly different; it still looks for each character sequence in the
                   1466: dictionary and finds its xt, or converts it to a number, or fails to do
                   1467: either and generates an error. However, instead of executing the xt or
                   1468: pushing the number onto the stack it lays down (@var{compiles}) some
                   1469: magic to make that xt or number get executed or pushed at a later time;
                   1470: at the time that @code{add-two} is @var{executed}. Therefore, when you
                   1471: execute @code{add-two} its @var{run-time effect} is exactly the same as
                   1472: if you had typed @code{2 + .} outside of a definition, and pressed
1.26      crook    1473: carriage-return.
1.21      crook    1474: 
1.23      crook    1475: In Forth, every word or number can be described in terms of three
                   1476: properties:
1.21      crook    1477: 
                   1478: @itemize @bullet
                   1479: @item
1.23      crook    1480: Its behaviour at @var{compile} time
1.21      crook    1481: @item
1.23      crook    1482: Its behaviour at @var{interpret} time
1.21      crook    1483: @item
1.23      crook    1484: Its behaviour at @var{execution} time.
1.21      crook    1485: @end itemize
                   1486: 
1.23      crook    1487: These behaviours are called the @var{semantics} of the word or
                   1488: number. The value of @var{state} determines whether the text
                   1489: interpreter will use the compile or interpret semantics of a word or
                   1490: number that it encounters.
1.21      crook    1491: 
                   1492: @itemize @bullet
                   1493: @item
1.23      crook    1494: @cindex interpretation semantics
                   1495: When the text interpreter encounters a word or number in @var{interpret}
                   1496: state, it performs the @var{interpretation semantics} of the word or
                   1497: number.
1.21      crook    1498: @item
1.23      crook    1499: @cindex compilation semantics
                   1500: When the text interpreter encounters a word or number in @var{compile}
                   1501: state, it performs the @var{compilation semantics} of the word or
                   1502: number.
1.21      crook    1503: @end itemize
                   1504: 
1.23      crook    1505: The behaviour of numbers is always the same:
1.21      crook    1506: 
                   1507: @itemize @bullet
                   1508: @item
1.23      crook    1509: When the number is @var{compiled}, it is appended to the current
                   1510: definition so that its run-time behaviour is to execute. (In other
                   1511: words, the compilation semantics of a number are to postpone its
                   1512: execution semantics until the run-time of the definition that it is
                   1513: being compiled into.)
                   1514: @item
                   1515: When the number is @var{interpreted}, its behaviour is to execute. (In
                   1516: other words, the interpretation semantics of a number are to perform its
                   1517: execution semantics.)
1.21      crook    1518: @item
1.23      crook    1519: @cindex execution semantics
                   1520: When the number is @var{executed}, its behaviour is to push its value
                   1521: onto the stack. (In other words, the execution semantics of a number are
                   1522: to push its value onto the stack.)
1.21      crook    1523: @end itemize
                   1524: 
1.23      crook    1525: The behaviour of a word is not so regular, but the vast majority behave
                   1526: like this:
1.21      crook    1527: 
                   1528: @itemize @bullet
                   1529: @item
1.23      crook    1530: The @var{compilation semantics} of the word are to append its
                   1531: @var{execution semantics} to the current definition (so that its
                   1532: run-time behaviour is to execute).
1.21      crook    1533: @item
1.23      crook    1534: The @var{interpretation semantics} of the word are to execute.
                   1535: @item
                   1536: The @var{execution semantics} of the word are to do something useful.
1.21      crook    1537: @end itemize
                   1538: 
                   1539: 
1.23      crook    1540: The actual behaviour of any particular word depends upon the way in
                   1541: which it was defined. In all cases, the text interpreter decides what to
                   1542: do with the word; when it searches the name dictionary for a definition,
                   1543: it not only retrieves the xt for the word, it also retrieves a flag
                   1544: called the @var{immediate flag}. If the flag is set, the text
                   1545: interpreter will @var{execute} the word rather than @var{compiling}
                   1546: @cindex immediate words
                   1547: it. In other words, these so-called @var{immediate} words behave like
                   1548: this:
1.21      crook    1549: 
                   1550: @itemize @bullet
                   1551: @item
1.23      crook    1552: The @var{compilation semantics} of the word are to perform its
                   1553: @var{execution semantics} (so that its compile-time behaviour is to
                   1554: execute).
1.21      crook    1555: @item
1.23      crook    1556: The @var{interpretation semantics} of the word are to execute.
                   1557: @item
                   1558: The @var{execution semantics} of the word are to do something useful.
1.21      crook    1559: @end itemize
                   1560: 
1.23      crook    1561: This example shows the difference between an immediate and a
                   1562: non-immediate word:
1.21      crook    1563: 
                   1564: @example
1.23      crook    1565: : show-state state @ . ;
                   1566: : show-state-now show-state ; immediate
                   1567: : word1 show-state ;
                   1568: : word2 show-state-now ;
                   1569: @end example
                   1570: 
                   1571: The word @code{immediate} after the definition of @code{show-state-now}
                   1572: makes that word an immediate word. These definitions introduce a new
1.27      crook    1573: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
1.23      crook    1574: variable, and leaves it on the stack. Therefore, the behaviour of
                   1575: @code{show-state} is to print a number that represents the current value
                   1576: of @code{state}.
                   1577: 
                   1578: When you execute @code{word1}, it prints the number 0, indicating
                   1579: that the system is in interpret state. When the text interpreter
                   1580: compiled the definition of @code{word1}, it encountered
                   1581: @code{show-state} whose compilation semantics are to append its
                   1582: execution semantics to the current definition. When you execute
                   1583: @code{word1}, it performs the execution semantics of @code{show-state}.
                   1584: At the time that @code{word1} (and therefore @code{show-state}) are
                   1585: executed, the system is in interpret state.
                   1586: 
                   1587: When you pressed <return> after entering the definition of @code{word2},
                   1588: you should have seen the number -1 printed, followed by @code{ ok}. When
                   1589: the text interpreter compiled the definition of @code{word2}, it
                   1590: encountered @code{show-state-now}, an immediate word, whose compilation
                   1591: semantics are therefore to perform its execution semantics. It is
                   1592: executed straight away (even before the text interpreter has moved on
                   1593: to process another group of characters; the @code{;} in this
                   1594: example). The effect of executing it are to display the value of
                   1595: @code{state} @var{at the time that the definition of} @code{word2}
                   1596: @var{is being defined}. Printing -1 demonstrates that the system is in
                   1597: compilation state at this time. If you execute @code{word2} it does
                   1598: nothing at all.
                   1599: 
1.26      crook    1600: @cindex @code{."}, how it works
1.23      crook    1601: Before leaving the subject of immediate words, consider the behaviour of
                   1602: @code{."} in the definition of @code{greet}, in the previous
                   1603: section. This word is both a parsing word and an immediate word. Notice
                   1604: that there is a space between @code{."} and the start of the text
                   1605: @code{Hello and welcome}, but that there is no space between the last
                   1606: letter of @code{welcome} and the @code{"} character. The reason for this
                   1607: is that @code{."} is a Forth word; it must have a space after it so that
                   1608: the text interpreter can identify it. The @code{"} is not a Forth word;
                   1609: it is a @var{delimiter}. The examples earlier show that, when the string
                   1610: is displayed, there is neither a space before the @code{H} nor after the
                   1611: @code{e}. Since @code{."} is an immediate word, it executes at the time
1.26      crook    1612: that @code{greet} is defined. When it executes, it searches forward in
1.23      crook    1613: the input line looking for the delimiter. When it finds the delimiter,
                   1614: it updates @code{>in} to point past the delimiter. It also compiles some
                   1615: magic code into the definition of @code{greet}; the xt of a run-time
                   1616: routine that prints a text string. It compiles the string @code{Hello
                   1617: and welcome} into memory so that it is available to be printed
                   1618: later. When the text interpreter gains control, the next word it finds
                   1619: in the input stream is @code{;} and so it terminates the definition of
                   1620: @code{greet}.
1.21      crook    1621: 
                   1622: 
1.23      crook    1623: @comment ----------------------------------------------
                   1624: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
                   1625: @section Forth is written in Forth
                   1626: @cindex structure of Forth programs
1.21      crook    1627: 
1.23      crook    1628: When you start up a Forth compiler, a large number of definitions
                   1629: already exist. In Forth, you develop a new application using bottom-up
                   1630: programming techniques to create new definitions that are defined in
                   1631: terms of existing definitions. As you create each definition you can
                   1632: test and debug it interactively.
                   1633: 
                   1634: If you have tried out the examples in this section, you will probably
                   1635: have typed them in by hand; when you leave Gforth, your definitions will
                   1636: be deleted. You can avoid this by using a text editor to enter Forth
                   1637: source code into a file, and then load all of the code from the file
1.26      crook    1638: using @code{include} (@xref{Forth source files}). A Forth source
1.23      crook    1639: file is processed by the text interpreter, just as though you had typed
                   1640: it in by hand@footnote{Actually, there are some subtle differences, like
                   1641: the fact that it doesn't print @code{ ok} at the end of each line}. 
                   1642: 
                   1643: Gforth also supports the traditional Forth alternative to using text
                   1644: files for program entry (@xref{Blocks}).
                   1645: 
                   1646: In common with many, if not most, Forth compilers, most of Gforth is
1.28    ! crook    1647: actually written in Forth. All of the @file{.fs} files in the
        !          1648: installation directory@footnote{For example,
        !          1649: @file{/usr/local/share/gforth..}} are Forth source files, which you can
        !          1650: study to see examples of Forth programming.
        !          1651: 
        !          1652: Gforth maintains a history file that records every line that you type to
        !          1653: the text interpreter. This file is preserved between sessions, and is
        !          1654: used to provide a command-line recall facility. If you enter long
        !          1655: definitions by hand, you can use a text editor to paste them out of the
        !          1656: history file into a Forth source file for reuse at a later time
        !          1657: (@pxref{Command-line editing} for more information).
1.21      crook    1658: 
                   1659: 
1.23      crook    1660: @comment ----------------------------------------------
                   1661: @node Review - elements of a Forth system, Exercises, Forth is written in Forth, Introduction
                   1662: @section Review - elements of a Forth system
                   1663: @cindex elements of a Forth system
1.21      crook    1664: 
1.23      crook    1665: To summarise this chapter:
1.21      crook    1666: 
                   1667: 
1.23      crook    1668: @itemize @bullet
                   1669: @item
                   1670: Forth programs use @var{factoring} to break a problem down into small
                   1671: fragments called @var{words} or @var{definitions}.
                   1672: @item
                   1673: Forth program development is an interactive process.
                   1674: @item
                   1675: The main command loop that accepts input, and controls both
                   1676: interpretation and compilation, is called the @var{text interpreter}
1.26      crook    1677: (also known as the @var{outer interpreter}).
1.23      crook    1678: @item
                   1679: Forth has a very simple syntax, consisting of words and numbers
                   1680: separated by spaces or carriage-return characters. Any additional syntax
                   1681: is imposed by @var{parsing words}.
                   1682: @item
                   1683: Forth uses a stack to pass parameters between words. As a result, it
                   1684: uses postfix notation.
                   1685: @item
                   1686: To use a word that has previously been defined, the text interpreter
                   1687: searches for the word in the @var{name dictionary}.
                   1688: @item
                   1689: Words have @var{interpretation semantics}, @var{compilation semantics}
                   1690: and @var{execution semantics}.
                   1691: @item
                   1692: The text interpreter uses the value of @code{state} to select between
                   1693: the use of the @var{interpretation semantics} and the  @var{compilation
                   1694: semantics} of a word that it encounters.
                   1695: @item
                   1696: The relationship between the @var{interpretation semantics}, @var{compilation semantics}
                   1697: and @var{execution semantics} for a word depend upon the way in which
1.26      crook    1698: the word was defined (for example, whether it is an @var{immediate} word).
1.23      crook    1699: @item
                   1700: Forth definitions can be implemented in Forth (called @var{high-level
                   1701: definitions}) or in some other way (usually a lower-level language and
                   1702: as a result often called @var{low-level definitions}, @var{code
                   1703: definitions} or @var{primitives}).
                   1704: @item
                   1705: Many Forth systems are implemented mainly in Forth.
                   1706: @item
                   1707: You now know enough to read and understand the rest of this manual and
1.26      crook    1708: the ANS Forth document.
1.23      crook    1709: @end itemize
1.21      crook    1710: 
                   1711: 
1.23      crook    1712: @comment TODO - other defining words
                   1713: @comment other parsing words
                   1714: @comment Your first loop
                   1715: @comment syntax and semantics
                   1716: @comment DOES>
                   1717: @comment taste of other elements of Forth
1.21      crook    1718: 
                   1719: 
                   1720: 
                   1721: @comment ----------------------------------------------
1.23      crook    1722: @node Exercises, ,Review - elements of a Forth system, Introduction
                   1723: @section Exercises
1.21      crook    1724: @cindex elements of a Forth system
                   1725: 
1.23      crook    1726: Amazing as it may seem, if you have read (and understood) this far, you
                   1727: know almost all the fundamentals about the inner workings of a Forth
                   1728: system. You certainly know enough to be able to read and understand the
                   1729: rest of this manual, to learn more about the facilities that Gforth
                   1730: provides. Even scarier, you know almost enough to implement your own Forth
                   1731: system. However, that's not a good idea just yet.. better to try writing
                   1732: some programs in Gforth.
                   1733: 
1.26      crook    1734: The large number of Forth words available in ANS Forth and
1.23      crook    1735: Gforth make learning Forth somewhat daunting. To make the problem
                   1736: easier, use the index of this manual to learn more about these words:
1.21      crook    1737: 
1.23      crook    1738: ..levels of Forth words.
1.21      crook    1739: 
                   1740: 
                   1741: Ideally, provide a set of programming excercises linked into the stuff
                   1742: done already and into other sections of the manual. Provide solutions to
                   1743: all the exercises in a .fs file in the distribution. Get some
                   1744: inspiration from Starting Forth and Kelly&Spies.
                   1745: 
                   1746: 
1.28    ! crook    1747: @c excercises:
        !          1748: @c 1. take inches and convert to feet and inches.
        !          1749: @c 2. take temperature and convert from fahrenheight to celcius;
        !          1750: @c    may need to care about symmetric vs floored??
        !          1751: @c 3. take input line and do character substitution
        !          1752: @c    to encipher or decipher
        !          1753: @c 4. as above but work on a file for in and out
        !          1754: @c 5. take input line and convert to pig-latin 
        !          1755: @c
        !          1756: @c thing of sets of things to exercise then come up with
        !          1757: @c problems that need those things.
        !          1758: 
1.26      crook    1759: @c ******************************************************************
1.28    ! crook    1760: @node Gforth Environment, Words, Introduction, Top
        !          1761: @chapter Gforth Environment
        !          1762: @cindex Gforth environment
        !          1763: 
        !          1764: Note: ultimately, the gforth man page will be auto-geenrated from the
        !          1765: material in this chapter.
        !          1766: 
        !          1767: @menu
        !          1768: * Invoking Gforth::
        !          1769: * Leaving Gforth::
        !          1770: * Command-line editing::
        !          1771: * Upper and lower case::
        !          1772: * Environment variables::
        !          1773: * Gforth Files::
        !          1774: @end menu
        !          1775: 
        !          1776: 
        !          1777: @comment ----------------------------------------------
        !          1778: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
        !          1779: @section Invoking Gforth
1.26      crook    1780: @cindex invoking Gforth
                   1781: @cindex running Gforth
                   1782: @cindex command-line options
                   1783: @cindex options on the command line
                   1784: @cindex flags on the command line
                   1785: 
                   1786: You will usually just say @code{gforth}. In many other cases the default
                   1787: Gforth image will be invoked like this:
                   1788: @example
                   1789: gforth [files] [-e forth-code]
                   1790: @end example
                   1791: This interprets the contents of the files and the Forth code in the order they
                   1792: are given.
1.23      crook    1793: 
1.26      crook    1794: In general, the command line looks like this:
1.1       anton    1795: 
1.26      crook    1796: @example
                   1797: gforth [initialization options] [image-specific options]
                   1798: @end example
1.1       anton    1799: 
1.26      crook    1800: The initialization options must come before the rest of the command
                   1801: line. They are:
1.1       anton    1802: 
1.26      crook    1803: @table @code
                   1804: @cindex -i, command-line option
                   1805: @cindex --image-file, command-line option
                   1806: @item --image-file @var{file}
                   1807: @itemx -i @var{file}
                   1808: Loads the Forth image @var{file} instead of the default
                   1809: @file{gforth.fi} (@pxref{Image Files}).
1.1       anton    1810: 
1.26      crook    1811: @cindex --path, command-line option
1.1       anton    1812: @cindex -p, command-line option
                   1813: @item --path @var{path}
                   1814: @itemx -p @var{path}
                   1815: Uses @var{path} for searching the image file and Forth source code files
                   1816: instead of the default in the environment variable @code{GFORTHPATH} or
                   1817: the path specified at installation time (e.g.,
                   1818: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
                   1819: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
                   1820: 
                   1821: @cindex --dictionary-size, command-line option
                   1822: @cindex -m, command-line option
                   1823: @cindex @var{size} parameters for command-line options
                   1824: @cindex size of the dictionary and the stacks
                   1825: @item --dictionary-size @var{size}
                   1826: @itemx -m @var{size}
                   1827: Allocate @var{size} space for the Forth dictionary space instead of
                   1828: using the default specified in the image (typically 256K). The
1.21      crook    1829: @var{size} specification for this and subsequent options consists of
                   1830: an integer and a unit (e.g.,
1.1       anton    1831: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1.12      anton    1832: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
                   1833: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
                   1834: @code{e} is used.
1.1       anton    1835: 
                   1836: @cindex --data-stack-size, command-line option
                   1837: @cindex -d, command-line option
                   1838: @item --data-stack-size @var{size}
                   1839: @itemx -d @var{size}
                   1840: Allocate @var{size} space for the data stack instead of using the
                   1841: default specified in the image (typically 16K).
                   1842: 
                   1843: @cindex --return-stack-size, command-line option
                   1844: @cindex -r, command-line option
                   1845: @item --return-stack-size @var{size}
                   1846: @itemx -r @var{size}
                   1847: Allocate @var{size} space for the return stack instead of using the
                   1848: default specified in the image (typically 15K).
                   1849: 
                   1850: @cindex --fp-stack-size, command-line option
                   1851: @cindex -f, command-line option
                   1852: @item --fp-stack-size @var{size}
                   1853: @itemx -f @var{size}
                   1854: Allocate @var{size} space for the floating point stack instead of
                   1855: using the default specified in the image (typically 15.5K). In this case
                   1856: the unit specifier @code{e} refers to floating point numbers.
                   1857: 
                   1858: @cindex --locals-stack-size, command-line option
                   1859: @cindex -l, command-line option
                   1860: @item --locals-stack-size @var{size}
                   1861: @itemx -l @var{size}
                   1862: Allocate @var{size} space for the locals stack instead of using the
                   1863: default specified in the image (typically 14.5K).
                   1864: 
                   1865: @cindex -h, command-line option
                   1866: @cindex --help, command-line option
                   1867: @item --help
                   1868: @itemx -h
                   1869: Print a message about the command-line options
                   1870: 
                   1871: @cindex -v, command-line option
                   1872: @cindex --version, command-line option
                   1873: @item --version
                   1874: @itemx -v
                   1875: Print version and exit
                   1876: 
                   1877: @cindex --debug, command-line option
                   1878: @item --debug
                   1879: Print some information useful for debugging on startup.
                   1880: 
                   1881: @cindex --offset-image, command-line option
                   1882: @item --offset-image
                   1883: Start the dictionary at a slightly different position than would be used
                   1884: otherwise (useful for creating data-relocatable images,
                   1885: @pxref{Data-Relocatable Image Files}).
                   1886: 
1.5       anton    1887: @cindex --no-offset-im, command-line option
                   1888: @item --no-offset-im
                   1889: Start the dictionary at the normal position.
                   1890: 
1.1       anton    1891: @cindex --clear-dictionary, command-line option
                   1892: @item --clear-dictionary
                   1893: Initialize all bytes in the dictionary to 0 before loading the image
                   1894: (@pxref{Data-Relocatable Image Files}).
1.5       anton    1895: 
                   1896: @cindex --die-on-signal, command-line-option
                   1897: @item --die-on-signal
                   1898: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
                   1899: or the segmentation violation SIGSEGV) by translating it into a Forth
                   1900: @code{THROW}. With this option, Gforth exits if it receives such a
                   1901: signal. This option is useful when the engine and/or the image might be
                   1902: severely broken (such that it causes another signal before recovering
                   1903: from the first); this option avoids endless loops in such cases.
1.1       anton    1904: @end table
                   1905: 
                   1906: @cindex loading files at startup
                   1907: @cindex executing code on startup
                   1908: @cindex batch processing with Gforth
                   1909: As explained above, the image-specific command-line arguments for the
                   1910: default image @file{gforth.fi} consist of a sequence of filenames and
                   1911: @code{-e @var{forth-code}} options that are interpreted in the sequence
                   1912: in which they are given. The @code{-e @var{forth-code}} or
1.21      crook    1913: @code{--evaluate @var{forth-code}} option evaluates the Forth
1.1       anton    1914: code. This option takes only one argument; if you want to evaluate more
1.26      crook    1915: Forth words, you have to quote them or use @code{-e} several times. To exit
1.1       anton    1916: after processing the command line (instead of entering interactive mode)
                   1917: append @code{-e bye} to the command line.
                   1918: 
                   1919: @cindex versions, invoking other versions of Gforth
                   1920: If you have several versions of Gforth installed, @code{gforth} will
                   1921: invoke the version that was installed last. @code{gforth-@var{version}}
                   1922: invokes a specific version. You may want to use the option
                   1923: @code{--path}, if your environment contains the variable
                   1924: @code{GFORTHPATH}.
                   1925: 
                   1926: Not yet implemented:
                   1927: On startup the system first executes the system initialization file
                   1928: (unless the option @code{--no-init-file} is given; note that the system
                   1929: resulting from using this option may not be ANS Forth conformant). Then
                   1930: the user initialization file @file{.gforth.fs} is executed, unless the
                   1931: option @code{--no-rc} is given; this file is first searched in @file{.},
                   1932: then in @file{~}, then in the normal path (see above).
                   1933: 
1.21      crook    1934: 
1.28    ! crook    1935: 
        !          1936: @comment ----------------------------------------------
        !          1937: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
        !          1938: @section Leaving Gforth
1.21      crook    1939: @cindex Gforth - leaving
                   1940: @cindex leaving Gforth
                   1941: 
1.28    ! crook    1942: You can leave Gforth by typing @code{bye} or Ctrl-D or (if you invoked
        !          1943: Gforth with the @code{--die-on-signal} option) Ctrl-C. When you leave
        !          1944: Gforth, all of your definitions and data are discarded. @xref{Image
        !          1945: Files} for ways of saving the state of the system before leaving Gforth.
1.21      crook    1946: 
                   1947: doc-bye
                   1948: 
                   1949: 
1.28    ! crook    1950: @comment ----------------------------------------------
        !          1951: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
        !          1952: @section Command-line editing
        !          1953: @cindex command-line editing
        !          1954: 
        !          1955: Gforth maintains a history file that records every line that you type to
        !          1956: the text interpreter. This file is preserved between sessions, and is
        !          1957: used to provide a command-line recall facility; if you type ctrl-P
        !          1958: repeatedly you can recall successively older command from this (or
        !          1959: previous) session(s). The full list of command-line editing facilities is:
        !          1960: 
        !          1961: @itemize @bullet
        !          1962: @item
        !          1963: ctrl-P (``previous'') (or up-arrow) to recall successively older
        !          1964: commands from the history buffer.
        !          1965: @item
        !          1966: ctrl-N (``next'') (or down-arrow) to recall successively newer commands
        !          1967: from the history buffer.
        !          1968: @item
        !          1969: ctrl-F (or right-arrow) to move the cursor right, non-destructively.
        !          1970: @item
        !          1971: ctrl-B (or left-arrow) to move the cursor left, non-destructively.
        !          1972: @item
        !          1973: ctrl-H (backspace) to delete the character to the left of the cursor,
        !          1974: closing up the line.
        !          1975: @item
        !          1976: ctrl-K to delete (``kill'') from the cursor to the end of the line.
        !          1977: @item
        !          1978: ctrl-A to move the cursor to the start of the line.
        !          1979: @item
        !          1980: ctrl-E to move the cursor to the end of the line.
        !          1981: @item
        !          1982: carriage-return or line-feed (ctrl-J, ctrl-M) to submit the current
        !          1983: line.
        !          1984: @item
        !          1985: tab to step through all possible full-word completions of the word
        !          1986: currently being typed.
        !          1987: @item
        !          1988: ctrl-D to terminate Gforth (gracefully, using @code{bye}).
        !          1989: @end itemize
        !          1990: 
        !          1991: When editing, displayable characters are inserted to the left of the
        !          1992: cursor position; the line is always in ``insert'' (as opposed to
        !          1993: ``overstrike'') mode.
        !          1994: 
        !          1995: @cindex history file
        !          1996: @cindex @file{.gforth-history}
        !          1997: On Unix systems, the history file is @file{~/.gforth-history} by
        !          1998: default@footnote{i.e. it is stored in the user's home directory.}. You
        !          1999: can find out the name and location of your history file using:
        !          2000: 
        !          2001: @example 
        !          2002: history-file type \ Unix-class systems
        !          2003: 
        !          2004: history-file type \ Other systems
        !          2005: history-dir  type
        !          2006: @end example
1.23      crook    2007: 
1.28    ! crook    2008: If you enter long definitions by hand, you can use a text editor to
        !          2009: paste them out of the history file into a Forth source file for reuse at
        !          2010: a later time.
        !          2011: 
        !          2012: Gforth never trims the size of the history file, so you should do this
        !          2013: periodically, if necessary.
        !          2014: 
        !          2015: @comment this is all defined in history.fs
        !          2016: @comment TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
        !          2017: @comment chosen?
        !          2018: 
        !          2019: 
        !          2020: 
        !          2021: @comment ----------------------------------------------
        !          2022: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
        !          2023: @section Upper and lower case
        !          2024: @cindex case-sensitivity
        !          2025: @cindex upper and lower case
        !          2026: 
        !          2027: Gforth is case-insensitive, so you can enter definitions and invoke
        !          2028: Standard words using upper, lower or mixed case (however,
        !          2029: @pxref{core-idef, Implementation-defined options, Implementation-defined
        !          2030: options}).
        !          2031: 
        !          2032: ANS Forth only @i{requires} implementations to recognise Standard words when
        !          2033: they are typed entirely in upper case. Therefore, a Standard program
        !          2034: must use upper case for all Standard words@footnote{You can use whatever
        !          2035: case you like for words that you define.}.
        !          2036: 
        !          2037: 
        !          2038: @comment ----------------------------------------------
        !          2039: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
        !          2040: @section Environment variables
        !          2041: @cindex environment variables
        !          2042: 
        !          2043: Gforth uses these environment variables:
        !          2044: 
        !          2045: @itemize @bullet
        !          2046: @item
        !          2047: @cindex GFORTHHIST - environment variable
        !          2048: GFORTHHIST - (Unix systems only) specifies the directory in which to
        !          2049: open/create the history file, @file{.gforth-history}. Default:
        !          2050: @code{$HOME}.
        !          2051: 
        !          2052: @item
        !          2053: @cindex GFORTHPATH - environment variable
        !          2054: GFORTHPATH - specifies the path used when searching for the gforth image file and
        !          2055: for Forth source-code files.
        !          2056: 
        !          2057: @item
        !          2058: @cindex GFORTH - environment variable
        !          2059: GFORTH - used by @file{gforthmi} @xref{gforthmi}.
        !          2060: 
        !          2061: @item
        !          2062: @cindex GFORTHD - environment variable
        !          2063: GFORTHD - used by @file{gforthmi} @xref{gforthmi}.
        !          2064: 
        !          2065: @item
        !          2066: @cindex TMP, TEMP - environment variable
        !          2067: TMP, TEMP - (non-Unix systems only) used as a potential location for the
        !          2068: history file.
        !          2069: @end itemize
        !          2070: 
        !          2071: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
        !          2072: @comment mentioning these.
        !          2073: 
        !          2074: All the Gforth environment variables default to sensible values if they
        !          2075: are not set.
        !          2076: 
        !          2077: 
        !          2078: @comment ----------------------------------------------
        !          2079: @node Gforth Files, ,Environment variables,Gforth Environment
        !          2080: @section Gforth files
        !          2081: @cindex Gforth files
        !          2082: 
        !          2083: When Gforth is installed on a Unix system it installs files in these
        !          2084: locations:
        !          2085: 
        !          2086: @itemize @bullet
        !          2087: @item
        !          2088: @file{/usr/local/bin/gforth}
        !          2089: @item
        !          2090: @file{/usr/local/bin/gforthmi}
        !          2091: @item
        !          2092: @file{/usr/local/man/man1/gforth.1} - man page.
        !          2093: @item
        !          2094: @file{/usr/local/info} - the Info version of this manual.
        !          2095: @item
        !          2096: @file{/usr/local/lib/gforth/<version>/..} - Gforth @file{.fi} files.
        !          2097: @item
        !          2098: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
        !          2099: @item
        !          2100: @file{/usr/local/share/gforth/<version>/..} - Gforth source files.
        !          2101: @item
        !          2102: @file{../emacs/site-lisp/gforth.el} - Emacs gforth mode.
        !          2103: @end itemize
1.23      crook    2104: 
1.26      crook    2105: @c ******************************************************************
1.28    ! crook    2106: @node Words, Error messages, Gforth Environment, Top
1.1       anton    2107: @chapter Forth Words
1.26      crook    2108: @cindex words
1.1       anton    2109: 
                   2110: @menu
                   2111: * Notation::                    
1.21      crook    2112: * Comments::
                   2113: * Boolean Flags::
1.1       anton    2114: * Arithmetic::                  
                   2115: * Stack Manipulation::          
1.5       anton    2116: * Memory::                      
1.1       anton    2117: * Control Structures::          
                   2118: * Defining Words::              
1.21      crook    2119: * The Text Interpreter::
1.12      anton    2120: * Tokens for Words::            
1.21      crook    2121: * Word Lists::                   
                   2122: * Environmental Queries::
1.12      anton    2123: * Files::                       
                   2124: * Blocks::                      
                   2125: * Other I/O::                   
                   2126: * Programming Tools::           
                   2127: * Assembler and Code Words::    
                   2128: * Threading Words::             
1.26      crook    2129: * Locals::                      
                   2130: * Structures::                  
                   2131: * Object-oriented Forth::       
1.21      crook    2132: * Passing Commands to the OS::
                   2133: * Miscellaneous Words::
1.1       anton    2134: @end menu
                   2135: 
1.21      crook    2136: @node Notation, Comments, Words, Words
1.1       anton    2137: @section Notation
                   2138: @cindex notation of glossary entries
                   2139: @cindex format of glossary entries
                   2140: @cindex glossary notation format
                   2141: @cindex word glossary entry format
                   2142: 
                   2143: The Forth words are described in this section in the glossary notation
                   2144: that has become a de-facto standard for Forth texts, i.e.,
                   2145: 
                   2146: @format
                   2147: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
                   2148: @end format
                   2149: @var{Description}
                   2150: 
                   2151: @table @var
                   2152: @item word
1.28    ! crook    2153: The name of the word.
1.1       anton    2154: 
                   2155: @item Stack effect
                   2156: @cindex stack effect
                   2157: The stack effect is written in the notation @code{@var{before} --
                   2158: @var{after}}, where @var{before} and @var{after} describe the top of
                   2159: stack entries before and after the execution of the word. The rest of
                   2160: the stack is not touched by the word. The top of stack is rightmost,
                   2161: i.e., a stack sequence is written as it is typed in. Note that Gforth
                   2162: uses a separate floating point stack, but a unified stack
                   2163: notation. Also, return stack effects are not shown in @var{stack
                   2164: effect}, but in @var{Description}. The name of a stack item describes
                   2165: the type and/or the function of the item. See below for a discussion of
                   2166: the types.
                   2167: 
                   2168: All words have two stack effects: A compile-time stack effect and a
                   2169: run-time stack effect. The compile-time stack-effect of most words is
                   2170: @var{ -- }. If the compile-time stack-effect of a word deviates from
                   2171: this standard behaviour, or the word does other unusual things at
                   2172: compile time, both stack effects are shown; otherwise only the run-time
                   2173: stack effect is shown.
                   2174: 
                   2175: @cindex pronounciation of words
                   2176: @item pronunciation
                   2177: How the word is pronounced.
                   2178: 
                   2179: @cindex wordset
                   2180: @item wordset
1.21      crook    2181: The ANS Forth standard is divided into several word sets. A standard
                   2182: system need not support all of them. Therefore, in theory, the fewer
                   2183: word sets your program uses the more portable it will be. However, we
                   2184: suspect that most ANS Forth systems on personal machines will feature
1.26      crook    2185: all word sets. Words that are not defined in ANS Forth have
1.21      crook    2186: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
1.1       anton    2187: describes words that will work in future releases of Gforth;
                   2188: @code{gforth-internal} words are more volatile. Environmental query
                   2189: strings are also displayed like words; you can recognize them by the
1.21      crook    2190: @code{environment} in the word set field.
1.1       anton    2191: 
                   2192: @item Description
                   2193: A description of the behaviour of the word.
                   2194: @end table
                   2195: 
                   2196: @cindex types of stack items
                   2197: @cindex stack item types
                   2198: The type of a stack item is specified by the character(s) the name
                   2199: starts with:
                   2200: 
                   2201: @table @code
                   2202: @item f
                   2203: @cindex @code{f}, stack item type
                   2204: Boolean flags, i.e. @code{false} or @code{true}.
                   2205: @item c
                   2206: @cindex @code{c}, stack item type
                   2207: Char
                   2208: @item w
                   2209: @cindex @code{w}, stack item type
                   2210: Cell, can contain an integer or an address
                   2211: @item n
                   2212: @cindex @code{n}, stack item type
                   2213: signed integer
                   2214: @item u
                   2215: @cindex @code{u}, stack item type
                   2216: unsigned integer
                   2217: @item d
                   2218: @cindex @code{d}, stack item type
                   2219: double sized signed integer
                   2220: @item ud
                   2221: @cindex @code{ud}, stack item type
                   2222: double sized unsigned integer
                   2223: @item r
                   2224: @cindex @code{r}, stack item type
                   2225: Float (on the FP stack)
1.21      crook    2226: @item a-
1.1       anton    2227: @cindex @code{a_}, stack item type
                   2228: Cell-aligned address
1.21      crook    2229: @item c-
1.1       anton    2230: @cindex @code{c_}, stack item type
                   2231: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.21      crook    2232: @item f-
1.1       anton    2233: @cindex @code{f_}, stack item type
                   2234: Float-aligned address
1.21      crook    2235: @item df-
1.1       anton    2236: @cindex @code{df_}, stack item type
                   2237: Address aligned for IEEE double precision float
1.21      crook    2238: @item sf-
1.1       anton    2239: @cindex @code{sf_}, stack item type
                   2240: Address aligned for IEEE single precision float
                   2241: @item xt
                   2242: @cindex @code{xt}, stack item type
                   2243: Execution token, same size as Cell
                   2244: @item wid
                   2245: @cindex @code{wid}, stack item type
1.21      crook    2246: Word list ID, same size as Cell
1.1       anton    2247: @item f83name
                   2248: @cindex @code{f83name}, stack item type
                   2249: Pointer to a name structure
                   2250: @item "
                   2251: @cindex @code{"}, stack item type
1.12      anton    2252: string in the input stream (not on the stack). The terminating character
                   2253: is a blank by default. If it is not a blank, it is shown in @code{<>}
1.1       anton    2254: quotes.
                   2255: @end table
                   2256: 
1.21      crook    2257: @node Comments, Boolean Flags, Notation, Words
                   2258: @section Comments
1.26      crook    2259: @cindex comments
1.21      crook    2260: 
1.26      crook    2261: Forth supports two styles of comment; the traditional @var{in-line} comment,
                   2262: @code{(} and its modern cousin, the @var{comment to end of line}; @code{\}.
1.21      crook    2263: 
1.23      crook    2264: doc-(
1.21      crook    2265: doc-\
1.23      crook    2266: doc-\G
1.21      crook    2267: 
                   2268: @node Boolean Flags, Arithmetic, Comments, Words
                   2269: @section Boolean Flags
1.26      crook    2270: @cindex Boolean flags
1.21      crook    2271: 
                   2272: A Boolean flag is cell-sized. A cell with all bits clear represents the
                   2273: flag @code{false} and a flag with all bits set represents the flag
1.26      crook    2274: @code{true}. Words that check a flag (for example, @code{IF}) will treat
1.21      crook    2275: a cell that has @var{any} bit set as @code{true}.
                   2276: 
                   2277: doc-true
                   2278: doc-false
                   2279: 
                   2280: 
                   2281: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
1.1       anton    2282: @section Arithmetic
                   2283: @cindex arithmetic words
                   2284: 
                   2285: @cindex division with potentially negative operands
                   2286: Forth arithmetic is not checked, i.e., you will not hear about integer
                   2287: overflow on addition or multiplication, you may hear about division by
                   2288: zero if you are lucky. The operator is written after the operands, but
                   2289: the operands are still in the original order. I.e., the infix @code{2-1}
                   2290: corresponds to @code{2 1 -}. Forth offers a variety of division
                   2291: operators. If you perform division with potentially negative operands,
                   2292: you do not want to use @code{/} or @code{/mod} with its undefined
                   2293: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
                   2294: former, @pxref{Mixed precision}).
1.26      crook    2295: @comment TODO discuss the different division forms and the std approach
1.1       anton    2296: 
                   2297: @menu
                   2298: * Single precision::            
                   2299: * Bitwise operations::          
1.21      crook    2300: * Double precision::            Double-cell integer arithmetic
                   2301: * Numeric comparison::
1.1       anton    2302: * Mixed precision::             operations with single and double-cell integers
                   2303: * Floating Point::              
                   2304: @end menu
                   2305: 
                   2306: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
                   2307: @subsection Single precision
                   2308: @cindex single precision arithmetic words
                   2309: 
1.21      crook    2310: By default, numbers in Forth are single-precision integers that are 1
1.26      crook    2311: cell in size. They can be signed or unsigned, depending upon how you
1.21      crook    2312: treat them. @xref{Number Conversion} for the rules used by the text
                   2313: interpreter for recognising single-precision integers.
                   2314: 
1.1       anton    2315: doc-+
1.21      crook    2316: doc-1+
1.1       anton    2317: doc--
1.21      crook    2318: doc-1-
1.1       anton    2319: doc-*
                   2320: doc-/
                   2321: doc-mod
                   2322: doc-/mod
                   2323: doc-negate
                   2324: doc-abs
                   2325: doc-min
                   2326: doc-max
1.21      crook    2327: doc-d>s
1.27      crook    2328: doc-floored
1.1       anton    2329: 
1.21      crook    2330: @node Bitwise operations, Double precision, Single precision, Arithmetic
1.1       anton    2331: @subsection Bitwise operations
                   2332: @cindex bitwise operation words
                   2333: 
                   2334: doc-and
                   2335: doc-or
                   2336: doc-xor
                   2337: doc-invert
1.21      crook    2338: doc-lshift
                   2339: doc-rshift
1.1       anton    2340: doc-2*
1.21      crook    2341: doc-d2*
1.1       anton    2342: doc-2/
1.21      crook    2343: doc-d2/
                   2344: 
                   2345: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
                   2346: @subsection Double precision
                   2347: @cindex double precision arithmetic words
                   2348: 
                   2349: @xref{Number Conversion} for the rules used by the text interpreter for
                   2350: recognising double-precision integers.
                   2351: 
                   2352: A double precision number is represented by a cell pair, with the most
1.26      crook    2353: significant digit at the TOS. It is trivial to convert an unsigned
                   2354: single to an (unsigned) double; simply push a @code{0} onto the
                   2355: TOS. Since numbers are represented by Gforth using 2's complement
                   2356: arithmetic, converting a signed single to a (signed) double requires
                   2357: sign-extension across the most significant digit. This can be achieved
                   2358: using @code{s>d}. The moral of the story is that you cannot convert a
                   2359: number without knowing whether it represents an unsigned or a
                   2360: signed number.
1.21      crook    2361: 
                   2362: doc-s>d
                   2363: doc-d+
                   2364: doc-d-
                   2365: doc-dnegate
                   2366: doc-dabs
                   2367: doc-dmin
                   2368: doc-dmax
                   2369: 
                   2370: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
                   2371: @subsection Numeric comparison
                   2372: @cindex numeric comparison words
                   2373: 
1.28    ! crook    2374: doc-<
        !          2375: doc-<=
        !          2376: doc-<>
        !          2377: doc-=
        !          2378: doc->
        !          2379: doc->=
        !          2380: 
1.21      crook    2381: doc-0<
1.23      crook    2382: doc-0<=
1.21      crook    2383: doc-0<>
                   2384: doc-0=
1.23      crook    2385: doc-0>
                   2386: doc-0>=
1.28    ! crook    2387: 
        !          2388: doc-u<
        !          2389: doc-u<=
        !          2390: @comment TODO why u<> and u= .. they are the same as <> and =
        !          2391: doc-u<>
        !          2392: doc-u=
        !          2393: doc-u>
        !          2394: doc-u>=
        !          2395: 
        !          2396: doc-within
        !          2397: 
        !          2398: doc-d<
        !          2399: doc-d<=
        !          2400: doc-d<>
        !          2401: doc-d=
        !          2402: doc-d>
        !          2403: doc-d>=
1.23      crook    2404: 
1.21      crook    2405: doc-d0<
1.23      crook    2406: doc-d0<=
                   2407: doc-d0<>
1.21      crook    2408: doc-d0=
1.23      crook    2409: doc-d0>
                   2410: doc-d0>=
                   2411: 
1.21      crook    2412: doc-du<
1.28    ! crook    2413: doc-du<=
        !          2414: doc-du<>
        !          2415: doc-du=
        !          2416: doc-du>
        !          2417: doc-du>=
1.1       anton    2418: 
1.21      crook    2419: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
1.1       anton    2420: @subsection Mixed precision
                   2421: @cindex mixed precision arithmetic words
                   2422: 
                   2423: doc-m+
                   2424: doc-*/
                   2425: doc-*/mod
                   2426: doc-m*
                   2427: doc-um*
                   2428: doc-m*/
                   2429: doc-um/mod
                   2430: doc-fm/mod
                   2431: doc-sm/rem
                   2432: 
1.21      crook    2433: @node Floating Point,  , Mixed precision, Arithmetic
1.1       anton    2434: @subsection Floating Point
                   2435: @cindex floating point arithmetic words
                   2436: 
1.21      crook    2437: @xref{Number Conversion} for the rules used by the text interpreter for
                   2438: recognising floating-point numbers.
1.1       anton    2439: 
                   2440: @cindex angles in trigonometric operations
                   2441: @cindex trigonometric operations
                   2442: Angles in floating point operations are given in radians (a full circle
1.26      crook    2443: has 2 pi radians). Gforth has a separate floating point
                   2444: stack, but the documentation uses the unified notation.
1.1       anton    2445: 
                   2446: @cindex floating-point arithmetic, pitfalls
                   2447: Floating point numbers have a number of unpleasant surprises for the
                   2448: unwary (e.g., floating point addition is not associative) and even a few
                   2449: for the wary. You should not use them unless you know what you are doing
                   2450: or you don't care that the results you get are totally bogus. If you
                   2451: want to learn about the problems of floating point numbers (and how to
                   2452: avoid them), you might start with @cite{David Goldberg, What Every
                   2453: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
1.17      anton    2454: Computing Surveys 23(1):5@minus{}48, March 1991}
                   2455: (@url{http://www.validgh.com/goldberg/paper.ps}).
1.1       anton    2456: 
1.21      crook    2457: doc-d>f
                   2458: doc-f>d
1.1       anton    2459: doc-f+
                   2460: doc-f-
                   2461: doc-f*
                   2462: doc-f/
                   2463: doc-fnegate
                   2464: doc-fabs
                   2465: doc-fmax
                   2466: doc-fmin
                   2467: doc-floor
                   2468: doc-fround
                   2469: doc-f**
                   2470: doc-fsqrt
                   2471: doc-fexp
                   2472: doc-fexpm1
                   2473: doc-fln
                   2474: doc-flnp1
                   2475: doc-flog
                   2476: doc-falog
                   2477: doc-fsin
                   2478: doc-fcos
                   2479: doc-fsincos
                   2480: doc-ftan
                   2481: doc-fasin
                   2482: doc-facos
                   2483: doc-fatan
                   2484: doc-fatan2
                   2485: doc-fsinh
                   2486: doc-fcosh
                   2487: doc-ftanh
                   2488: doc-fasinh
                   2489: doc-facosh
                   2490: doc-fatanh
1.21      crook    2491: doc-pi
1.28    ! crook    2492: 
1.21      crook    2493: doc-f0<
1.28    ! crook    2494: doc-f0<=
        !          2495: doc-f0<>
1.21      crook    2496: doc-f0=
1.28    ! crook    2497: doc-f0>
        !          2498: doc-f0>=
        !          2499: 
1.21      crook    2500: doc-f<
                   2501: doc-f<=
                   2502: doc-f<>
                   2503: doc-f=
                   2504: doc-f>
                   2505: doc-f>=
1.28    ! crook    2506: 
1.21      crook    2507: doc-f2*
                   2508: doc-f2/
                   2509: doc-1/f
                   2510: doc-f~
                   2511: doc-precision
                   2512: doc-set-precision
1.1       anton    2513: 
                   2514: @node Stack Manipulation, Memory, Arithmetic, Words
                   2515: @section Stack Manipulation
                   2516: @cindex stack manipulation words
                   2517: 
                   2518: @cindex floating-point stack in the standard
1.21      crook    2519: Gforth maintains a number of separate stacks:
                   2520: 
                   2521: @itemize @bullet
                   2522: @item
                   2523: A data stack (aka parameter stack) -- for characters, cells,
                   2524: addresses, and double cells.
                   2525: 
                   2526: @item
                   2527: A floating point stack -- for floating point numbers.
                   2528: 
                   2529: @item
                   2530: A return stack -- for storing the return addresses of colon
                   2531: definitions and other data.
                   2532: 
                   2533: @item
                   2534: A locals stack for storing local variables.
                   2535: @end itemize
                   2536: 
                   2537: Whilst every sane Forth has a separate floating-point stack, it is not
                   2538: strictly required; an ANS Forth system could theoretically keep
                   2539: floating-point numbers on the data stack. As an additional difficulty,
                   2540: you don't know how many cells a floating-point number takes. It is
                   2541: reportedly possible to write words in a way that they work also for a
                   2542: unified stack model, but we do not recommend trying it. Instead, just
                   2543: say that your program has an environmental dependency on a separate
                   2544: floating-point stack.
                   2545: 
                   2546: doc-floating-stack
1.1       anton    2547: 
                   2548: @cindex return stack and locals
                   2549: @cindex locals and return stack
1.21      crook    2550: A Forth system is allowed to keep local variables on the
1.1       anton    2551: return stack. This is reasonable, as local variables usually eliminate
                   2552: the need to use the return stack explicitly. So, if you want to produce
1.21      crook    2553: a standard compliant program and you are using local variables in a
                   2554: word, forget about return stack manipulations in that word (refer to the
1.1       anton    2555: standard document for the exact rules).
                   2556: 
                   2557: @menu
                   2558: * Data stack::                  
                   2559: * Floating point stack::        
                   2560: * Return stack::                
                   2561: * Locals stack::                
                   2562: * Stack pointer manipulation::  
                   2563: @end menu
                   2564: 
                   2565: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
                   2566: @subsection Data stack
                   2567: @cindex data stack manipulation words
                   2568: @cindex stack manipulations words, data stack
                   2569: 
                   2570: doc-drop
                   2571: doc-nip
                   2572: doc-dup
                   2573: doc-over
                   2574: doc-tuck
                   2575: doc-swap
1.21      crook    2576: doc-pick
1.1       anton    2577: doc-rot
                   2578: doc--rot
                   2579: doc-?dup
                   2580: doc-roll
                   2581: doc-2drop
                   2582: doc-2nip
                   2583: doc-2dup
                   2584: doc-2over
                   2585: doc-2tuck
                   2586: doc-2swap
                   2587: doc-2rot
                   2588: 
                   2589: @node Floating point stack, Return stack, Data stack, Stack Manipulation
                   2590: @subsection Floating point stack
                   2591: @cindex floating-point stack manipulation words
                   2592: @cindex stack manipulation words, floating-point stack
                   2593: 
                   2594: doc-fdrop
                   2595: doc-fnip
                   2596: doc-fdup
                   2597: doc-fover
                   2598: doc-ftuck
                   2599: doc-fswap
1.21      crook    2600: doc-fpick
1.1       anton    2601: doc-frot
                   2602: 
                   2603: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
                   2604: @subsection Return stack
                   2605: @cindex return stack manipulation words
                   2606: @cindex stack manipulation words, return stack
                   2607: 
                   2608: doc->r
                   2609: doc-r>
                   2610: doc-r@
                   2611: doc-rdrop
                   2612: doc-2>r
                   2613: doc-2r>
                   2614: doc-2r@
                   2615: doc-2rdrop
                   2616: 
                   2617: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
                   2618: @subsection Locals stack
                   2619: 
1.26      crook    2620: @comment TODO
1.21      crook    2621: 
1.1       anton    2622: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
                   2623: @subsection Stack pointer manipulation
                   2624: @cindex stack pointer manipulation words
                   2625: 
1.21      crook    2626: doc-sp0
                   2627: doc-s0
1.1       anton    2628: doc-sp@
                   2629: doc-sp!
1.21      crook    2630: doc-fp0
1.1       anton    2631: doc-fp@
                   2632: doc-fp!
1.21      crook    2633: doc-rp0
                   2634: doc-r0
1.1       anton    2635: doc-rp@
                   2636: doc-rp!
1.21      crook    2637: doc-lp0
                   2638: doc-l0
1.1       anton    2639: doc-lp@
                   2640: doc-lp!
                   2641: 
                   2642: @node Memory, Control Structures, Stack Manipulation, Words
                   2643: @section Memory
1.26      crook    2644: @cindex memory words
1.1       anton    2645: 
1.27      crook    2646: @cindex dictionary
                   2647: Forth definitions are organised in memory structures that are
                   2648: collectively called the @var{dictionary}. The dictionary can be
                   2649: considered as three logical memory regions:
                   2650: 
                   2651: @itemize @bullet
                   2652: @item
                   2653: @cindex code space
                   2654: @cindex code dictionary
                   2655: Code space, also known as the @var{code dictionary}.
                   2656: @item
                   2657: @cindex name space
                   2658: @cindex name dictionary
                   2659: Name space, also known as the @var{name dictionary}@footnote{Sometimes,
                   2660: people use the term @var{dictionary} to simply refer to the name
                   2661: dictionary, because it is the one region that is used for looking up
                   2662: names, just as you would in a conventional dictionary.}.
                   2663: @item
                   2664: @cindex data space
                   2665: Data space
                   2666: @end itemize
                   2667: 
                   2668: When you create a colon definition, the text interpreter compiles
                   2669: the definition itself into the code dictionary and compiles the name
                   2670: of the definition into the name dictionary, together with other
                   2671: information about the definition (such as its execution token).
                   2672: 
                   2673: When you create a variable, the execution of @code{variable} will
                   2674: compile some code, assign once cell in data space, and compile the name
                   2675: of the variable into the name dictionary.
                   2676: 
                   2677: @cindex memory regions - relationship between them
                   2678: ANS Forth does not specify the relationship between the three memory
                   2679: regions, and specifies that a Standard program must not access code or
                   2680: data space directly -- it may only access data space directly. In
                   2681: addition, the Standard defines what relationships you may and may not
                   2682: rely on when allocating regions in data space. These constraints are
                   2683: simply a reflection of the many diverse techniques that are used to
                   2684: implement Forth systems; understanding and following the requirements of
                   2685: the Standard allows you to write portable programs -- programs that run
                   2686: in the same way on any of these diverse systems. Another way of looking
                   2687: at this is to say that ANS Forth was designed to permit compliant Forth
                   2688: systems to be implemented in many diverse ways.
                   2689: 
                   2690: @cindex memory regions - how they are assigned
                   2691: Here are some examples of the way in which name, code and data spaces
                   2692: are assigned:
                   2693: 
                   2694: @itemize @bullet
                   2695: @item
                   2696: For a Forth system that runs from RAM under a general-purpose operating
                   2697: system, it can be convenient to interleave name, code and data spaces in
                   2698: a single contiguous memory region. This organisation can be
                   2699: memory-efficient (for example, because the relationship between the name
                   2700: dictionary entry and the associated code dictionary entry can be
                   2701: implicit, rather than requiring an explicit memory pointer to reference
                   2702: from the name dictionary and the code dictionary). This is the
                   2703: organisation used by Gforth, as this example@footnote{The addresses
                   2704: in the example have been truncated to fit it onto the page, and the
                   2705: addresses and data shown will not match the output from your system} shows:
                   2706: @example
                   2707: hex
                   2708: variable fred 123456 fred !
                   2709: variable jim abcd jim !
                   2710: : foo + / - ;
                   2711: ' fred 10 - 50 dump 
                   2712: ..80: 5C 46 0E 40  84 66 72 65 - 64 20 20 20  20 20 20 20  \F.@.fred       
                   2713: ..90: D0 9B 04 08  00 00 00 00 - 56 34 12 00  80 46 0E 40  ........V4...F.@
                   2714: ..A0: 83 6A 69 6D  20 20 20 20 - D0 9B 04 08  00 00 00 00  .jim    ........
                   2715: ..B0: CD AB 00 00  9C 46 0E 40 - 83 66 6F 6F  20 20 20 20  .....F.@.foo    
                   2716: ..C0: 80 9B 04 08  00 00 00 00 - E4 2E 05 08  0C 2F 05 08  ............./..
                   2717: @end example
                   2718: 
                   2719: @item
                   2720: For a high-performance system running on a modern RISC processor with a
                   2721: modified Harvard architecture (one that has a unified main memory but
                   2722: separate instruction and data caches), it is desirable to separate
                   2723: processor instructions from processor data. This encourages a high cache
                   2724: density and therefore a high cache hit rate. The Forth code dictionary
                   2725: is not necessarily made up entirely of processor instructions; its
                   2726: nature is dependent upon the Forth implementation. 
                   2727: 
                   2728: @item
                   2729: A Forth compiler that runs on a segmented 8086 processor could be
                   2730: designed to interleave the name, code and data spaces within a single
                   2731: 64Kbyte segment. A more common implementation choice is to use a
                   2732: separate 64Kbyte segment for each region, which provides more memory
                   2733: overall but provides an address map in which only the data space is
                   2734: accessible.
                   2735: 
                   2736: @item
                   2737: Microprocessors exist that run Forth (or many of the primitives required
                   2738: to implement the Forth virtual machine efficiently) directly. On these
                   2739: processors, the relationship between name, code and data spaces may be
                   2740: imposed as a side-effect of the microarchitecture of the processor.
                   2741: 
                   2742: @item
                   2743: A Forth compiler that executes from ROM on an embedded system needs its
                   2744: data space separated from the name and code spaces so that the data
                   2745: space can be mapped to a RAM area.
                   2746: 
                   2747: @item 
                   2748: A Forth compiler that runs on an embedded system may have a requirement
                   2749: for a small memory footprint. On such a system it can be useful to
                   2750: separate the name space from the data and code spaces; once the
                   2751: application has been compiled, the name dictionary is no longer
                   2752: required@footnote{more strictly speaking, most applications can be
                   2753: designed so that this is the case}. The name dictionary can be deleted
                   2754: entirely, or could be stored in memory on a remote @var{host} system for
                   2755: debug and development purposes. In the latter case, the compiler running
                   2756: on the @var{target} system could implement a protocol across a
                   2757: communication link that would allow it to interrogate the name dictionary.
                   2758: @end itemize
                   2759: 
1.1       anton    2760: @menu
1.27      crook    2761: * Reserving Data Space::
                   2762: * Memory Access::
                   2763: * Address Arithmetic::
                   2764: * Memory Blocks::
                   2765: * Dynamic Allocation::
1.1       anton    2766: @end menu
                   2767: 
1.27      crook    2768: 
                   2769: @node Reserving Data Space, Memory Access, Memory, Memory
                   2770: @subsection Reserving Data Space
                   2771: @cindex reserving data space
                   2772: @cindex data space - reserving some
                   2773: 
                   2774: @cindex data space pointer - alignment
                   2775: These factors affect the alignment of @code{here}, the data
                   2776: space pointer:
                   2777: 
                   2778: @itemize @bullet
                   2779: @item
                   2780: If the data-space pointer is aligned@footnote{In ANS Forth-speak,
                   2781: @var{aligned} implictly means @code{CELL}-aligned} before an
                   2782: @code{allot}, and a whole number of characters are reserved or released, it
                   2783: will remain aligned after the @code{allot}.
                   2784: 
                   2785: @item
                   2786: If the data-space pointer is character-aligned before an @code{allot},
                   2787: and a whole number of cells are reserved or released, it will remain
                   2788: character-aligned after the @code{allot}.
                   2789: 
                   2790: @item
                   2791: The initial contents of data space reserved using @code{allot} is
                   2792: undefined.
                   2793: 
                   2794: @item
                   2795: Definitions created by @code{create}, @code{variable}, @code{2variable}
                   2796: return aligned addresses.
                   2797: 
                   2798: @item
                   2799: After a definition is compiled or @code{align} is executed, the data
                   2800: space pointer is guaranteed to be aligned.
                   2801: @end itemize
                   2802: 
                   2803: @cindex data space pointer - contiguous regions
                   2804: Contiguous regions may be created in data space under these conditions:
                   2805: @itemize @bullet
                   2806: @item
                   2807: The value of the data-space pointer, @code{here}, always defines the
                   2808: beginning of a contiguous region of data space.
                   2809: 
                   2810: @item
                   2811: @code{CREATE} establishes the beginning of a contiguous region of data
                   2812: space (the @code{CREATE}d definition returns the initial address of the
                   2813: region).
                   2814: 
                   2815: @item
                   2816: @code{variable} does @var{not} establish the beginning of a contiguous
                   2817: region in data space; @code{variable} followed by @code{allot} is not
                   2818: guaranteed to allocate data space region that is contiguous with the
                   2819: storage allocated by @code{variable}. Instead, use @code{create} --
                   2820: @xref{Simple Defining Words} for examples.
                   2821: 
                   2822: @item
                   2823: Successive calls to @code{allot}, @code{,} (comma), @code{2,} (2-comma),
                   2824: @code{c,} (c-comma) and @code{align} reserve a single contiguous region
                   2825: in data space. The contiguity of the region is interrupted by compiling
                   2826: (or removing) definitions from the dictionary.
                   2827: 
                   2828: @item
                   2829: The most recently reserved contiguous region may be released by calling
                   2830: @code{allot} with a negative argument, provided that the region has not
                   2831: been interrupted by compiling (or removing) definitions from the
                   2832: dictionary.
                   2833: @end itemize
                   2834: 
                   2835: doc-here
                   2836: doc-unused
                   2837: doc-allot
                   2838: doc-c,
                   2839: doc-,
                   2840: doc-2,
                   2841: 
1.28    ! crook    2842: @comment TODO may want to add description of similar user-space words,
        !          2843: @comment but only if its accompanied by clear description of what user
        !          2844: @comment space is and when it is useful. Words are udp uallot
1.27      crook    2845: 
                   2846: @node Memory Access, Address Arithmetic, Reserving Data Space, Memory
1.1       anton    2847: @subsection Memory Access
                   2848: @cindex memory access words
                   2849: 
                   2850: doc-@
                   2851: doc-!
                   2852: doc-+!
                   2853: doc-c@
                   2854: doc-c!
                   2855: doc-2@
                   2856: doc-2!
                   2857: doc-f@
                   2858: doc-f!
                   2859: doc-sf@
                   2860: doc-sf!
                   2861: doc-df@
                   2862: doc-df!
                   2863: 
1.27      crook    2864: @node Address Arithmetic, Memory Blocks, Memory Access, Memory
                   2865: @subsection Address Arithmetic
1.1       anton    2866: @cindex address arithmetic words
                   2867: 
                   2868: ANS Forth does not specify the sizes of the data types. Instead, it
                   2869: offers a number of words for computing sizes and doing address
                   2870: arithmetic. Basically, address arithmetic is performed in terms of
                   2871: address units (aus); on most systems the address unit is one byte. Note
                   2872: that a character may have more than one au, so @code{chars} is no noop
                   2873: (on systems where it is a noop, it compiles to nothing).
                   2874: 
                   2875: @cindex alignment of addresses for types
                   2876: ANS Forth also defines words for aligning addresses for specific
                   2877: types. Many computers require that accesses to specific data types
                   2878: must only occur at specific addresses; e.g., that cells may only be
                   2879: accessed at addresses divisible by 4. Even if a machine allows unaligned
                   2880: accesses, it can usually perform aligned accesses faster. 
                   2881: 
                   2882: For the performance-conscious: alignment operations are usually only
                   2883: necessary during the definition of a data structure, not during the
                   2884: (more frequent) accesses to it.
                   2885: 
                   2886: ANS Forth defines no words for character-aligning addresses. This is not
                   2887: an oversight, but reflects the fact that addresses that are not
                   2888: char-aligned have no use in the standard and therefore will not be
                   2889: created.
                   2890: 
                   2891: @cindex @code{CREATE} and alignment
1.26      crook    2892: AND Forth guarantees that addresses returned by @code{CREATE}d words
1.1       anton    2893: are cell-aligned; in addition, Gforth guarantees that these addresses
                   2894: are aligned for all purposes.
                   2895: 
1.26      crook    2896: Note that the ANS Forth word @code{char} has nothing to do with address
                   2897: arithmetic.
1.1       anton    2898: 
                   2899: doc-chars
                   2900: doc-char+
                   2901: doc-cells
                   2902: doc-cell+
                   2903: doc-cell
                   2904: doc-align
                   2905: doc-aligned
                   2906: doc-floats
                   2907: doc-float+
                   2908: doc-float
                   2909: doc-falign
                   2910: doc-faligned
                   2911: doc-sfloats
                   2912: doc-sfloat+
                   2913: doc-sfalign
                   2914: doc-sfaligned
                   2915: doc-dfloats
                   2916: doc-dfloat+
                   2917: doc-dfalign
                   2918: doc-dfaligned
                   2919: doc-maxalign
                   2920: doc-maxaligned
                   2921: doc-cfalign
                   2922: doc-cfaligned
                   2923: doc-address-unit-bits
                   2924: 
1.27      crook    2925: @node Memory Blocks, Dynamic Allocation, Address Arithmetic, Memory
1.1       anton    2926: @subsection Memory Blocks
                   2927: @cindex memory block words
1.27      crook    2928: @cindex character strings - moving and copying
                   2929: 
                   2930: Memory blocks often represent character strings; @xref{String Formats}
                   2931: for ways of storing character strings in memory. @xref{Displaying
                   2932: characters and strings} for other string-processing words.
1.1       anton    2933: 
1.21      crook    2934: Some of these words work on address units (increments of @code{CELL}),
                   2935: and expect a @code{CELL}-aligned address. Others work on character units
                   2936: (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
                   2937: address. Choose the correct operation depending upon your data type. If
                   2938: you are moving a block of memory (for example, a region reserved by
                   2939: @code{allot}) it is safe to use @code{move}, and it should be faster
                   2940: than using @code{cmove}. If you are moving (for example) a string
                   2941: compiled using @code{S"}, it is not portable to use @code{move}; the
                   2942: alignment of the string in memory could change, and the relationship
                   2943: between @code{CELL} and @code{CHAR} could change.
                   2944: 
                   2945: When copying characters between overlapping memory regions, choose
                   2946: carefully between @code{cmove} and @code{cmove>}.
                   2947: 
                   2948: You can only use any of these words @var{portably} to access data space.
                   2949: 
1.27      crook    2950: @comment TODO - think the naming of the arguments is wrong for move
1.1       anton    2951: doc-move
                   2952: doc-erase
1.27      crook    2953: @comment TODO - think the naming of the arguments is wrong for cmove
1.1       anton    2954: doc-cmove
1.27      crook    2955: @comment TODO - think the naming of the arguments is wrong for cmove>
1.1       anton    2956: doc-cmove>
                   2957: doc-fill
                   2958: doc-blank
1.21      crook    2959: doc-compare
                   2960: doc-search
1.27      crook    2961: doc--trailing
                   2962: doc-/string
                   2963: 
                   2964: @comment TODO examples
                   2965: 
                   2966: @node Dynamic Allocation, ,Memory Blocks, Memory
                   2967: @subsection Dynamic Allocation of Memory
                   2968: @cindex dynamic allocation of memory
                   2969: @cindex memory-allocation word set
                   2970: 
                   2971: The ANS Forth memory-allocation word set allows memory regions to be
                   2972: dynamically assigned, resized and released without affecting the data
                   2973: space pointer. In Gforth, these words are implemented using
                   2974: the standard C library calls malloc(), free() and resize().
                   2975: 
                   2976: doc-allocate
                   2977: doc-free
                   2978: doc-resize
                   2979: 
1.1       anton    2980: 
1.26      crook    2981: @node Control Structures, Defining Words, Memory, Words
1.1       anton    2982: @section Control Structures
                   2983: @cindex control structures
                   2984: 
                   2985: Control structures in Forth cannot be used in interpret state, only in
                   2986: compile state@footnote{More precisely, they have no interpretation
                   2987: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
                   2988: a colon definition. We do not like this limitation, but have not seen a
                   2989: satisfying way around it yet, although many schemes have been proposed.
                   2990: 
                   2991: @menu
                   2992: * Selection::                   
                   2993: * Simple Loops::                
                   2994: * Counted Loops::               
                   2995: * Arbitrary control structures::  
                   2996: * Calls and returns::           
                   2997: * Exception Handling::          
                   2998: @end menu
                   2999: 
                   3000: @node Selection, Simple Loops, Control Structures, Control Structures
                   3001: @subsection Selection
                   3002: @cindex selection control structures
                   3003: @cindex control structures for selection
                   3004: 
                   3005: @cindex @code{IF} control structure
                   3006: @example
                   3007: @var{flag}
                   3008: IF
                   3009:   @var{code}
                   3010: ENDIF
                   3011: @end example
1.21      crook    3012: @noindent
1.1       anton    3013: or
                   3014: @example
                   3015: @var{flag}
                   3016: IF
                   3017:   @var{code1}
                   3018: ELSE
                   3019:   @var{code2}
                   3020: ENDIF
                   3021: @end example
                   3022: 
                   3023: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
                   3024: standard, and @code{ENDIF} is not, although it is quite popular. We
                   3025: recommend using @code{ENDIF}, because it is less confusing for people
                   3026: who also know other languages (and is not prone to reinforcing negative
                   3027: prejudices against Forth in these people). Adding @code{ENDIF} to a
                   3028: system that only supplies @code{THEN} is simple:
                   3029: @example
1.21      crook    3030: : ENDIF   POSTPONE THEN ; immediate
1.1       anton    3031: @end example
                   3032: 
                   3033: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
                   3034: (adv.)}  has the following meanings:
                   3035: @quotation
                   3036: ... 2b: following next after in order ... 3d: as a necessary consequence
                   3037: (if you were there, then you saw them).
                   3038: @end quotation
                   3039: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
                   3040: and many other programming languages has the meaning 3d.]
                   3041: 
1.21      crook    3042: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
1.1       anton    3043: you can avoid using @code{?dup}. Using these alternatives is also more
1.26      crook    3044: efficient than using @code{?dup}. Definitions in ANS Forth
1.1       anton    3045: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
                   3046: @file{compat/control.fs}.
                   3047: 
                   3048: @cindex @code{CASE} control structure
                   3049: @example
                   3050: @var{n}
                   3051: CASE
                   3052:   @var{n1} OF @var{code1} ENDOF
                   3053:   @var{n2} OF @var{code2} ENDOF
                   3054:   @dots{}
                   3055: ENDCASE
                   3056: @end example
                   3057: 
                   3058: Executes the first @var{codei}, where the @var{ni} is equal to
                   3059: @var{n}. A default case can be added by simply writing the code after
                   3060: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
                   3061: but must not consume it.
                   3062: 
                   3063: @node Simple Loops, Counted Loops, Selection, Control Structures
                   3064: @subsection Simple Loops
                   3065: @cindex simple loops
                   3066: @cindex loops without count 
                   3067: 
                   3068: @cindex @code{WHILE} loop
                   3069: @example
                   3070: BEGIN
                   3071:   @var{code1}
                   3072:   @var{flag}
                   3073: WHILE
                   3074:   @var{code2}
                   3075: REPEAT
                   3076: @end example
                   3077: 
                   3078: @var{code1} is executed and @var{flag} is computed. If it is true,
                   3079: @var{code2} is executed and the loop is restarted; If @var{flag} is
                   3080: false, execution continues after the @code{REPEAT}.
                   3081: 
                   3082: @cindex @code{UNTIL} loop
                   3083: @example
                   3084: BEGIN
                   3085:   @var{code}
                   3086:   @var{flag}
                   3087: UNTIL
                   3088: @end example
                   3089: 
                   3090: @var{code} is executed. The loop is restarted if @code{flag} is false.
                   3091: 
                   3092: @cindex endless loop
                   3093: @cindex loops, endless
                   3094: @example
                   3095: BEGIN
                   3096:   @var{code}
                   3097: AGAIN
                   3098: @end example
                   3099: 
                   3100: This is an endless loop.
                   3101: 
                   3102: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
                   3103: @subsection Counted Loops
                   3104: @cindex counted loops
                   3105: @cindex loops, counted
                   3106: @cindex @code{DO} loops
                   3107: 
                   3108: The basic counted loop is:
                   3109: @example
                   3110: @var{limit} @var{start}
                   3111: ?DO
                   3112:   @var{body}
                   3113: LOOP
                   3114: @end example
                   3115: 
                   3116: This performs one iteration for every integer, starting from @var{start}
1.21      crook    3117: and up to, but excluding @var{limit}. The counter, or @var{index}, can be
                   3118: accessed with @code{i}. For example, the loop:
1.1       anton    3119: @example
                   3120: 10 0 ?DO
                   3121:   i .
                   3122: LOOP
                   3123: @end example
1.21      crook    3124: @noindent
                   3125: prints @code{0 1 2 3 4 5 6 7 8 9}
                   3126: 
1.1       anton    3127: The index of the innermost loop can be accessed with @code{i}, the index
                   3128: of the next loop with @code{j}, and the index of the third loop with
                   3129: @code{k}.
                   3130: 
                   3131: doc-i
                   3132: doc-j
                   3133: doc-k
                   3134: 
                   3135: The loop control data are kept on the return stack, so there are some
1.21      crook    3136: restrictions on mixing return stack accesses and counted loop words. In
                   3137: particuler, if you put values on the return stack outside the loop, you
                   3138: cannot read them inside the loop@footnote{well, not in a way that is
                   3139: portable.}. If you put values on the return stack within a loop, you
                   3140: have to remove them before the end of the loop and before accessing the
                   3141: index of the loop.
1.1       anton    3142: 
                   3143: There are several variations on the counted loop:
                   3144: 
1.21      crook    3145: @itemize @bullet
                   3146: @item
                   3147: @code{LEAVE} leaves the innermost counted loop immediately; execution
                   3148: continues after the associated @code{LOOP} or @code{NEXT}. For example:
                   3149: 
                   3150: @example
                   3151: 10 0 ?DO  i DUP . 3 = IF LEAVE THEN LOOP
                   3152: @end example
                   3153: prints @code{0 1 2 3}
                   3154: 
1.1       anton    3155: 
1.21      crook    3156: @item
                   3157: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
                   3158: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
                   3159: return stack so @code{EXIT} can get to its return address. For example:
                   3160: 
                   3161: @example
                   3162: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
                   3163: @end example
                   3164: prints @code{0 1 2 3}
                   3165: 
                   3166: 
                   3167: @item
1.1       anton    3168: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
                   3169: (and @code{LOOP} iterates until they become equal by wrap-around
                   3170: arithmetic). This behaviour is usually not what you want. Therefore,
                   3171: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
                   3172: @code{?DO}), which do not enter the loop if @var{start} is greater than
                   3173: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
                   3174: unsigned loop parameters.
                   3175: 
1.21      crook    3176: @item
                   3177: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
                   3178: the loop, independent of the loop parameters. Do not use @code{DO}, even
                   3179: if you know that the loop is entered in any case. Such knowledge tends
                   3180: to become invalid during maintenance of a program, and then the
                   3181: @code{DO} will make trouble.
                   3182: 
                   3183: @item
1.1       anton    3184: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
                   3185: index by @var{n} instead of by 1. The loop is terminated when the border
                   3186: between @var{limit-1} and @var{limit} is crossed. E.g.:
                   3187: 
1.21      crook    3188: @example
                   3189: 4 0 +DO  i .  2 +LOOP
                   3190: @end example
                   3191: @noindent
                   3192: prints @code{0 2}
                   3193: 
                   3194: @example
                   3195: 4 1 +DO  i .  2 +LOOP
                   3196: @end example
                   3197: @noindent
                   3198: prints @code{1 3}
1.1       anton    3199: 
                   3200: 
                   3201: @cindex negative increment for counted loops
                   3202: @cindex counted loops with negative increment
                   3203: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
                   3204: 
1.21      crook    3205: @example
                   3206: -1 0 ?DO  i .  -1 +LOOP
                   3207: @end example
                   3208: @noindent
                   3209: prints @code{0 -1}
1.1       anton    3210: 
1.21      crook    3211: @example
                   3212: 0 0 ?DO  i .  -1 +LOOP
                   3213: @end example
                   3214: prints nothing.
1.1       anton    3215: 
                   3216: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
                   3217: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
                   3218: index by @var{u} each iteration. The loop is terminated when the border
                   3219: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
                   3220: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
                   3221: 
1.21      crook    3222: @example
                   3223: -2 0 -DO  i .  1 -LOOP
                   3224: @end example
                   3225: @noindent
                   3226: prints @code{0 -1}
1.1       anton    3227: 
1.21      crook    3228: @example
                   3229: -1 0 -DO  i .  1 -LOOP
                   3230: @end example
                   3231: @noindent
                   3232: prints @code{0}
                   3233: 
                   3234: @example
                   3235: 0 0 -DO  i .  1 -LOOP
                   3236: @end example
                   3237: @noindent
                   3238: prints nothing.
1.1       anton    3239: 
1.21      crook    3240: @end itemize
1.1       anton    3241: 
                   3242: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1.26      crook    3243: @code{-LOOP} are not defined in ANS Forth. However, an implementation
                   3244: for these words that uses only standard words is provided in
                   3245: @file{compat/loops.fs}.
1.1       anton    3246: 
                   3247: 
                   3248: @cindex @code{FOR} loops
1.26      crook    3249: Another counted loop is:
1.1       anton    3250: @example
                   3251: @var{n}
                   3252: FOR
                   3253:   @var{body}
                   3254: NEXT
                   3255: @end example
                   3256: This is the preferred loop of native code compiler writers who are too
1.26      crook    3257: lazy to optimize @code{?DO} loops properly. This loop structure is not
                   3258: defined in ANS Forth. In Gforth, this loop iterates @var{n+1} times;
                   3259: @code{i} produces values starting with @var{n} and ending with 0. Other
                   3260: Forth systems may behave differently, even if they support @code{FOR}
                   3261: loops. To avoid problems, don't use @code{FOR} loops.
1.1       anton    3262: 
                   3263: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
                   3264: @subsection Arbitrary control structures
                   3265: @cindex control structures, user-defined
                   3266: 
                   3267: @cindex control-flow stack
                   3268: ANS Forth permits and supports using control structures in a non-nested
                   3269: way. Information about incomplete control structures is stored on the
                   3270: control-flow stack. This stack may be implemented on the Forth data
                   3271: stack, and this is what we have done in Gforth.
                   3272: 
                   3273: @cindex @code{orig}, control-flow stack item
                   3274: @cindex @code{dest}, control-flow stack item
                   3275: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
                   3276: entry represents a backward branch target. A few words are the basis for
                   3277: building any control structure possible (except control structures that
                   3278: need storage, like calls, coroutines, and backtracking).
                   3279: 
                   3280: doc-if
                   3281: doc-ahead
                   3282: doc-then
                   3283: doc-begin
                   3284: doc-until
                   3285: doc-again
                   3286: doc-cs-pick
                   3287: doc-cs-roll
                   3288: 
1.21      crook    3289: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
                   3290: manipulate the control-flow stack in a portable way. Without them, you
                   3291: would need to know how many stack items are occupied by a control-flow
                   3292: entry (many systems use one cell. In Gforth they currently take three,
                   3293: but this may change in the future).
                   3294: 
1.1       anton    3295: Some standard control structure words are built from these words:
                   3296: 
                   3297: doc-else
                   3298: doc-while
                   3299: doc-repeat
                   3300: 
                   3301: Gforth adds some more control-structure words:
                   3302: 
                   3303: doc-endif
                   3304: doc-?dup-if
                   3305: doc-?dup-0=-if
                   3306: 
                   3307: Counted loop words constitute a separate group of words:
                   3308: 
                   3309: doc-?do
                   3310: doc-+do
                   3311: doc-u+do
                   3312: doc--do
                   3313: doc-u-do
                   3314: doc-do
                   3315: doc-for
                   3316: doc-loop
                   3317: doc-+loop
                   3318: doc--loop
                   3319: doc-next
                   3320: doc-leave
                   3321: doc-?leave
                   3322: doc-unloop
                   3323: doc-done
                   3324: 
1.21      crook    3325: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
                   3326: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
1.1       anton    3327: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
                   3328: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
                   3329: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
                   3330: resolved (by using one of the loop-ending words or @code{DONE}).
                   3331: 
1.26      crook    3332: Another group of control structure words are:
1.1       anton    3333: 
                   3334: doc-case
                   3335: doc-endcase
                   3336: doc-of
                   3337: doc-endof
                   3338: 
1.21      crook    3339: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
                   3340: @code{CS-ROLL}.
1.1       anton    3341: 
                   3342: @subsubsection Programming Style
                   3343: 
                   3344: In order to ensure readability we recommend that you do not create
                   3345: arbitrary control structures directly, but define new control structure
                   3346: words for the control structure you want and use these words in your
1.26      crook    3347: program. For example, instead of writing:
1.1       anton    3348: 
                   3349: @example
1.26      crook    3350: BEGIN
1.1       anton    3351:   ...
1.26      crook    3352: IF [ 1 CS-ROLL ]
1.1       anton    3353:   ...
1.26      crook    3354: AGAIN THEN
1.1       anton    3355: @end example
                   3356: 
1.21      crook    3357: @noindent
1.1       anton    3358: we recommend defining control structure words, e.g.,
                   3359: 
                   3360: @example
1.26      crook    3361: : WHILE ( DEST -- ORIG DEST )
                   3362:  POSTPONE IF
                   3363:  1 CS-ROLL ; immediate
                   3364: 
                   3365: : REPEAT ( orig dest -- )
                   3366:  POSTPONE AGAIN
                   3367:  POSTPONE THEN ; immediate
1.1       anton    3368: @end example
                   3369: 
1.21      crook    3370: @noindent
1.1       anton    3371: and then using these to create the control structure:
                   3372: 
                   3373: @example
1.26      crook    3374: BEGIN
1.1       anton    3375:   ...
1.26      crook    3376: WHILE
1.1       anton    3377:   ...
1.26      crook    3378: REPEAT
1.1       anton    3379: @end example
                   3380: 
                   3381: That's much easier to read, isn't it? Of course, @code{REPEAT} and
                   3382: @code{WHILE} are predefined, so in this example it would not be
                   3383: necessary to define them.
                   3384: 
                   3385: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
                   3386: @subsection Calls and returns
                   3387: @cindex calling a definition
                   3388: @cindex returning from a definition
                   3389: 
1.3       anton    3390: @cindex recursive definitions
                   3391: A definition can be called simply be writing the name of the definition
1.26      crook    3392: to be called. Normally a definition is invisible during its own
1.3       anton    3393: definition. If you want to write a directly recursive definition, you
1.26      crook    3394: can use @code{recursive} to make the current definition visible, or
                   3395: @code{recurse} to call the current definition directly.
1.3       anton    3396: 
                   3397: doc-recursive
                   3398: doc-recurse
                   3399: 
1.21      crook    3400: @comment TODO add example of the two recursion methods
1.12      anton    3401: @quotation
                   3402: @progstyle
                   3403: I prefer using @code{recursive} to @code{recurse}, because calling the
                   3404: definition by name is more descriptive (if the name is well-chosen) than
                   3405: the somewhat cryptic @code{recurse}.  E.g., in a quicksort
                   3406: implementation, it is much better to read (and think) ``now sort the
                   3407: partitions'' than to read ``now do a recursive call''.
                   3408: @end quotation
1.3       anton    3409: 
1.21      crook    3410: @comment TODO maybe move deferred words to Defining Words section and x-ref
                   3411: @comment from here.. that is where these two are glossed.
                   3412: 
1.3       anton    3413: For mutual recursion, use @code{defer}red words, like this:
                   3414: 
                   3415: @example
1.28    ! crook    3416: Defer foo
1.3       anton    3417: 
                   3418: : bar ( ... -- ... )
                   3419:  ... foo ... ;
                   3420: 
                   3421: :noname ( ... -- ... )
                   3422:  ... bar ... ;
                   3423: IS foo
                   3424: @end example
                   3425: 
1.26      crook    3426: The current definition returns control to the calling definition when
                   3427: the end of the definition is reached or @code{EXIT} is encountered.
1.1       anton    3428: 
                   3429: doc-exit
                   3430: doc-;s
                   3431: 
                   3432: @node Exception Handling,  , Calls and returns, Control Structures
                   3433: @subsection Exception Handling
1.26      crook    3434: @cindex exceptions
1.1       anton    3435: 
1.26      crook    3436: If your program detects a fatal error condition, the simplest action
                   3437: that it can take is to @code{quit}. This resets the return stack and
                   3438: restarts the text interpreter, but does not print any error message.
1.21      crook    3439: 
1.26      crook    3440: The next stage in severity is to execute @code{abort}, which has the
                   3441: same effect as @code{quit}, with the addition that it resets the data
                   3442: stack.
1.1       anton    3443: 
1.26      crook    3444: A slightly more sophisticated approach is use use @code{abort"}, which
                   3445: compiles a string to be used as an error message and does a conditional
                   3446: @code{abort} at run-time. For example:
1.1       anton    3447: 
1.26      crook    3448: @example
                   3449: @kbd{: checker abort" That flag was true" ." A false flag" ;<return>}  ok
                   3450: @kbd{0 checker<return>} A false flag ok
                   3451: @kbd{1 checker<return>}
                   3452: :1: That flag was true
                   3453: 1 checker
                   3454:   ^^^^^^^
                   3455: $400D1648 throw 
                   3456: $400E4660
                   3457: @end example
1.1       anton    3458: 
1.26      crook    3459: These simple techniques allow a program to react to a fatal error
                   3460: condition, but they are not exactly user-friendly. The ANS Forth
                   3461: Exception word set provides the pair of words @code{throw} and
                   3462: @code{catch}, which can be used to provide sophisticated error-handling.
1.1       anton    3463: 
1.26      crook    3464: @code{catch} has a similar behaviour to @code{execute}, in that it takes
                   3465: an @var{xt} as a parameter and starts execution of the xt. However,
                   3466: before passing control to the xt, @code{catch} pushes an
                   3467: @var{exception frame} onto the @var{exception stack}. This exception
                   3468: frame is used to restore the system to a known state if a detected error
                   3469: occurs during the execution of the xt. A typical way to use @code{catch}
                   3470: would be:
1.1       anton    3471: 
1.26      crook    3472: @example
                   3473: ... ['] foo catch IF ...
                   3474: @end example
1.1       anton    3475: 
1.26      crook    3476: Whilst @code{foo} executes, it can call other words to any level of
                   3477: nesting, as usual.  If @code{foo} (and all the words that it calls)
                   3478: execute successfully, control will ultimately passes to the word following
                   3479: the @code{catch}, and there will be a @code{true} flag (0) at
                   3480: TOS. However, if any word detects an error, it can terminate the
                   3481: execution of @code{foo} by pushing an error code onto the stack and then
                   3482: performing a @code{throw}. The execution of @code{throw} will pass 
                   3483: control to the word following the @code{catch}, but this time the TOS
                   3484: will hold the error code. Therefore, the @code{IF} in the example
                   3485: can be used to determine whether @code{foo} executed successfully.
1.1       anton    3486: 
1.26      crook    3487: This simple example shows how you can use @code{throw} and @code{catch}
                   3488: to ``take over'' exception handling from the system:
1.1       anton    3489: @example
1.26      crook    3490: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
1.1       anton    3491: @end example
                   3492: 
1.26      crook    3493: The next example is more sophisticated and shows a multi-level
                   3494: @code{throw} and @code{catch}. To understand this example, start at the
                   3495: definition of @code{top-level} and work backwards:
                   3496: 
1.1       anton    3497: @example
1.26      crook    3498: : lowest-level ( -- c )
                   3499:     key dup 27 = if
                   3500:        1 throw \ ESCAPE key pressed
                   3501:     else
                   3502:        ." lowest-level successfull" CR
                   3503:     then
                   3504: ;
                   3505: 
                   3506: : lower-level ( -- c )
                   3507:     lowest-level
                   3508:     \ at this level consider a CTRL-U to be a fatal error
                   3509:     dup 21 = if \ CTRL-U
                   3510:        2 throw
                   3511:     else
                   3512:        ." lower-level successfull" CR
                   3513:     then
                   3514: ;
                   3515: 
                   3516: : low-level ( -- c )
                   3517:     ['] lower-level catch
                   3518:     ?dup if
                   3519:        \ error occurred - do we recognise it?
                   3520:        dup 1 = if
                   3521:            \ ESCAPE key pressed.. pretend it was an E
                   3522:            [char] E
                   3523:        else throw \ propogate the error upwards
                   3524:        then
                   3525:     then
                   3526:     ." low-level successfull" CR
                   3527: ;
                   3528: 
                   3529: : top-level ( -- )
                   3530:     CR ['] low-level catch \ CATCH is used like EXECUTE
                   3531:     ?dup if \ error occurred..
                   3532:        ." Error " . ." occurred - contact your supplier"
                   3533:     else
                   3534:        ." The '" emit ." ' key was pressed" CR
                   3535:     then
                   3536: ;
1.1       anton    3537: @end example
                   3538: 
1.26      crook    3539: The ANS Forth document assigns @code{throw} codes thus:
1.1       anton    3540: 
1.26      crook    3541: @itemize @bullet
                   3542: @item
                   3543: codes in the range -1 -- -255 are reserved to be assigned by the
                   3544: Standard. Assignments for codes in the range -1 -- -58 are currently
                   3545: documented in the Standard. In particular, @code{-1 throw} is equivalent
                   3546: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
                   3547: @item
                   3548: codes in the range -256 -- -4095 are reserved to be assigned by the system.
                   3549: @item
                   3550: all other codes may be assigned by programs.
                   3551: @end itemize
1.1       anton    3552: 
1.26      crook    3553: Gforth provides the word @code{exception} as a mechanism for assigning
                   3554: system throw codes to applications. This allows multiple applications to
                   3555: co-exist in memory without any clash of @code{throw} codes. A definition
                   3556: of @code{exception} in ANS Forth is provided in
                   3557: @file{compat/exception.fs}.
1.1       anton    3558: 
1.26      crook    3559: doc-quit
                   3560: doc-abort
                   3561: doc-abort"
1.1       anton    3562: 
1.26      crook    3563: doc-catch
                   3564: doc-throw
                   3565: doc---exception-exception
1.1       anton    3566: 
                   3567: 
1.26      crook    3568: @c -------------------------------------------------------------
                   3569: @node Defining Words, The Text Interpreter, Control Structures, Words
                   3570: @section Defining Words
                   3571: @cindex defining words
1.1       anton    3572: 
1.26      crook    3573: @comment TODO much more intro material here. 3 classes: colon defn, variables/constants
                   3574: @comment values, user-defined defining words.
1.1       anton    3575: 
                   3576: @menu
1.27      crook    3577: * Simple Defining Words::
                   3578: * Colon Definitions::
                   3579: * User-defined Defining Words::
                   3580: * Supplying names::
                   3581: * Interpretation and Compilation Semantics::
1.1       anton    3582: @end menu
                   3583: 
1.26      crook    3584: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
                   3585: @subsection Simple Defining Words
                   3586: @cindex simple defining words
                   3587: @cindex defining words, simple
                   3588: 
1.27      crook    3589: @comment TODO include examples of reserving data space for buffers
                   3590: @comment etc. using variable, allot, create and build up to the point
                   3591: @comment where it is appropriate to x-ref to the "structures" section.
                   3592: 
1.26      crook    3593: doc-constant
                   3594: doc-2constant
                   3595: doc-fconstant
                   3596: doc-variable
                   3597: doc-2variable
                   3598: doc-fvariable
                   3599: doc-create
                   3600: doc-user
                   3601: doc-value
                   3602: doc-to
                   3603: doc-defer
                   3604: doc-is
1.28    ! crook    3605: doc-defers
        !          3606: doc-alias
1.26      crook    3607: 
                   3608: Definitions in ANS Forth for @code{defer}, @code{<is>} and
                   3609: @code{[is]} are provided in @file{compat/defer.fs}.
                   3610: @comment TODO - what do the two "is" words do?
1.1       anton    3611: 
1.26      crook    3612: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
                   3613: @subsection Colon Definitions
                   3614: @cindex colon definitions
1.1       anton    3615: 
1.26      crook    3616: @example
                   3617: : name ( ... -- ... )
                   3618:     word1 word2 word3 ;
                   3619: @end example
1.1       anton    3620: 
1.26      crook    3621: creates a word called @code{name}, that, upon execution, executes
                   3622: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
1.1       anton    3623: 
1.26      crook    3624: The explanation above is somewhat superficial. @xref{Interpretation and
                   3625: Compilation Semantics} for an in-depth discussion of some of the issues
                   3626: involved.
                   3627: 
                   3628: doc-:
                   3629: doc-;
1.1       anton    3630: 
1.26      crook    3631: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
                   3632: @subsection User-defined Defining Words
                   3633: @cindex user-defined defining words
                   3634: @cindex defining words, user-defined
1.1       anton    3635: 
1.26      crook    3636: You can create new defining words simply by wrapping defining-time code
                   3637: around existing defining words and putting the sequence in a colon
                   3638: definition.
1.1       anton    3639: 
1.26      crook    3640: @comment TODO example
1.1       anton    3641: 
1.26      crook    3642: @cindex @code{CREATE} ... @code{DOES>}
                   3643: If you want the words defined with your defining words to behave
                   3644: differently from words defined with standard defining words, you can
                   3645: write your defining word like this:
1.1       anton    3646: 
                   3647: @example
1.26      crook    3648: : def-word ( "name" -- )
                   3649:     Create @var{code1}
                   3650: DOES> ( ... -- ... )
                   3651:     @var{code2} ;
                   3652: 
                   3653: def-word name
1.1       anton    3654: @end example
                   3655: 
1.26      crook    3656: Technically, this fragment defines a defining word @code{def-word}, and
                   3657: a word @code{name}; when you execute @code{name}, the address of the
                   3658: body of @code{name} is put on the data stack and @var{code2} is executed
                   3659: (the address of the body of @code{name} is the address @code{HERE}
                   3660: returns immediately after the @code{CREATE}). The word @code{name} is
                   3661: sometimes called a @var{child} of @code{def-word}.
1.1       anton    3662: 
1.26      crook    3663: In other words, if you make the following definitions:
1.1       anton    3664: 
                   3665: @example
1.26      crook    3666: : def-word1 ( "name" -- )
                   3667:     Create @var{code1} ;
                   3668: 
                   3669: : action1 ( ... -- ... )
                   3670:     @var{code2} ;
                   3671: 
                   3672: def-word name1
1.1       anton    3673: @end example
                   3674: 
1.26      crook    3675: Using @code{name1 action1} is equivalent to using @code{name}.
                   3676: 
                   3677: The classic example is that you can define @code{Constant} in this way:
                   3678: 
1.1       anton    3679: @example
1.26      crook    3680: : constant ( w "name" -- )
                   3681:     create ,
                   3682: DOES> ( -- w )
                   3683:     @@ ;
1.1       anton    3684: @end example
                   3685: 
1.26      crook    3686: @comment that is the classic example.. maybe it should be earlier. There
                   3687: @comment is a beautiful description of how this works and what it does in
                   3688: @comment the Forthwrite 100th edition.
                   3689: 
                   3690: When you create a constant with @code{5 constant five}, first a new word
                   3691: @code{five} is created, then the value 5 is laid down in the body of
                   3692: @code{five} with @code{,}. When @code{five} is invoked, the address of
                   3693: the body is put on the stack, and @code{@@} retrieves the value 5.
                   3694: 
                   3695: @cindex stack effect of @code{DOES>}-parts
                   3696: @cindex @code{DOES>}-parts, stack effect
                   3697: In the example above the stack comment after the @code{DOES>} specifies
                   3698: the stack effect of the defined words, not the stack effect of the
                   3699: following code (the following code expects the address of the body on
                   3700: the top of stack, which is not reflected in the stack comment). This is
                   3701: the convention that I use and recommend (it clashes a bit with using
                   3702: locals declarations for stack effect specification, though).
1.1       anton    3703: 
1.26      crook    3704: @subsubsection Applications of @code{CREATE..DOES>}
                   3705: @cindex @code{CREATE} ... @code{DOES>}, applications
1.1       anton    3706: 
1.26      crook    3707: You may wonder how to use this feature. Here are some usage patterns:
1.1       anton    3708: 
1.26      crook    3709: @cindex factoring similar colon definitions
                   3710: When you see a sequence of code occurring several times, and you can
                   3711: identify a meaning, you will factor it out as a colon definition. When
                   3712: you see similar colon definitions, you can factor them using
                   3713: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
                   3714: that look very similar:
1.1       anton    3715: @example
1.26      crook    3716: : ori, ( reg-target reg-source n -- )
                   3717:     0 asm-reg-reg-imm ;
                   3718: : andi, ( reg-target reg-source n -- )
                   3719:     1 asm-reg-reg-imm ;
1.1       anton    3720: @end example
                   3721: 
1.26      crook    3722: @noindent
                   3723: This could be factored with:
                   3724: @example
                   3725: : reg-reg-imm ( op-code -- )
                   3726:     CREATE ,
                   3727: DOES> ( reg-target reg-source n -- )
                   3728:     @@ asm-reg-reg-imm ;
                   3729: 
                   3730: 0 reg-reg-imm ori,
                   3731: 1 reg-reg-imm andi,
                   3732: @end example
1.1       anton    3733: 
1.26      crook    3734: @cindex currying
                   3735: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
                   3736: supply a part of the parameters for a word (known as @dfn{currying} in
                   3737: the functional language community). E.g., @code{+} needs two
                   3738: parameters. Creating versions of @code{+} with one parameter fixed can
                   3739: be done like this:
1.1       anton    3740: @example
1.26      crook    3741: : curry+ ( n1 -- )
                   3742:     CREATE ,
                   3743: DOES> ( n2 -- n1+n2 )
                   3744:     @@ + ;
                   3745: 
                   3746:  3 curry+ 3+
                   3747: -2 curry+ 2-
1.1       anton    3748: @end example
                   3749: 
1.26      crook    3750: @subsubsection The gory details of @code{CREATE..DOES>}
                   3751: @cindex @code{CREATE} ... @code{DOES>}, details
1.1       anton    3752: 
1.26      crook    3753: doc-does>
1.1       anton    3754: 
1.26      crook    3755: @cindex @code{DOES>} in a separate definition
                   3756: This means that you need not use @code{CREATE} and @code{DOES>} in the
                   3757: same definition; you can put the @code{DOES>}-part in a separate
                   3758: definition. This allows us to, e.g., select among different DOES>-parts:
                   3759: @example
                   3760: : does1 
                   3761: DOES> ( ... -- ... )
                   3762:     ... ;
1.1       anton    3763: 
1.26      crook    3764: : does2
                   3765: DOES> ( ... -- ... )
                   3766:     ... ;
1.1       anton    3767: 
1.26      crook    3768: : def-word ( ... -- ... )
                   3769:     create ...
                   3770:     IF
                   3771:        does1
                   3772:     ELSE
                   3773:        does2
                   3774:     ENDIF ;
                   3775: @end example
1.1       anton    3776: 
1.26      crook    3777: In this example, the selection of whether to use @code{does1} or
                   3778: @code{does2} is made at compile-time; at the time that the child word is
                   3779: @code{Create}d.
1.1       anton    3780: 
1.26      crook    3781: @cindex @code{DOES>} in interpretation state
                   3782: In a standard program you can apply a @code{DOES>}-part only if the last
                   3783: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
                   3784: will override the behaviour of the last word defined in any case. In a
                   3785: standard program, you can use @code{DOES>} only in a colon
                   3786: definition. In Gforth, you can also use it in interpretation state, in a
                   3787: kind of one-shot mode; for example:
1.1       anton    3788: @example
1.26      crook    3789: CREATE name ( ... -- ... )
                   3790:   @var{initialization}
                   3791: DOES>
                   3792:   @var{code} ;
1.1       anton    3793: @end example
                   3794: 
1.26      crook    3795: @noindent
                   3796: is equivalent to the standard:
1.1       anton    3797: @example
1.26      crook    3798: :noname
                   3799: DOES>
                   3800:     @var{code} ;
                   3801: CREATE name EXECUTE ( ... -- ... )
                   3802:     @var{initialization}
1.1       anton    3803: @end example
                   3804: 
1.26      crook    3805: You can get the address of the body of a word with:
                   3806: 
                   3807: doc->body
1.1       anton    3808: 
1.26      crook    3809: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
                   3810: @subsection Supplying names for the defined words
                   3811: @cindex names for defined words
                   3812: @cindex defining words, name parameter
1.1       anton    3813: 
1.26      crook    3814: @cindex defining words, name given in a string
                   3815: By default, defining words take the names for the defined words from the
                   3816: input stream. Sometimes you want to supply the name from a string. You
                   3817: can do this with:
1.1       anton    3818: 
1.26      crook    3819: doc-nextname
1.1       anton    3820: 
1.26      crook    3821: For example:
1.1       anton    3822: 
1.26      crook    3823: @example
                   3824: s" foo" nextname create
                   3825: @end example
                   3826: @noindent
                   3827: is equivalent to:
                   3828: @example
                   3829: create foo
                   3830: @end example
1.1       anton    3831: 
1.26      crook    3832: @cindex defining words without name
                   3833: Sometimes you want to define an @var{anonymous word}; a word without a
                   3834: name. You can do this with:
1.1       anton    3835: 
1.26      crook    3836: doc-:noname
1.1       anton    3837: 
1.26      crook    3838: This leaves the execution token for the word on the stack after the
                   3839: closing @code{;}. Here's an example in which a deferred word is
                   3840: initialised with an @code{xt} from an anonymous colon definition:
                   3841: @example
                   3842: Defer deferred
                   3843: :noname ( ... -- ... )
                   3844:   ... ;
                   3845: IS deferred
                   3846: @end example
1.1       anton    3847: 
1.26      crook    3848: Gforth provides an alternative way of doing this, using two separate
                   3849: words:
1.1       anton    3850: 
1.26      crook    3851: doc-noname
                   3852: @cindex execution token of last defined word
                   3853: doc-lastxt
1.1       anton    3854: 
1.26      crook    3855: The previous example can be rewritten using @code{noname} and
                   3856: @code{lastxt}:
1.1       anton    3857: 
1.26      crook    3858: @example
                   3859: Defer deferred
                   3860: noname : ( ... -- ... )
                   3861:   ... ;
                   3862: lastxt IS deferred
                   3863: @end example
1.1       anton    3864: 
1.26      crook    3865: @code{lastxt} also works when the last word was not defined as
                   3866: @code{noname}. 
1.1       anton    3867: 
                   3868: 
1.26      crook    3869: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
                   3870: @subsection Interpretation and Compilation Semantics
                   3871: @cindex semantics, interpretation and compilation
1.1       anton    3872: 
1.26      crook    3873: @cindex interpretation semantics
                   3874: The @dfn{interpretation semantics} of a word are what the text
                   3875: interpreter does when it encounters the word in interpret state. It also
                   3876: appears in some other contexts, e.g., the execution token returned by
                   3877: @code{' @var{word}} identifies the interpretation semantics of
                   3878: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
                   3879: interpret-state text interpretation of @code{@var{word}}).
1.1       anton    3880: 
1.26      crook    3881: @cindex compilation semantics
                   3882: The @dfn{compilation semantics} of a word are what the text interpreter
                   3883: does when it encounters the word in compile state. It also appears in
                   3884: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
                   3885: standard terminology, ``appends to the current definition''.} the
                   3886: compilation semantics of @var{word}.
1.1       anton    3887: 
1.26      crook    3888: @cindex execution semantics
                   3889: The standard also talks about @dfn{execution semantics}. They are used
                   3890: only for defining the interpretation and compilation semantics of many
                   3891: words. By default, the interpretation semantics of a word are to
                   3892: @code{execute} its execution semantics, and the compilation semantics of
                   3893: a word are to @code{compile,} its execution semantics.@footnote{In
                   3894: standard terminology: The default interpretation semantics are its
                   3895: execution semantics; the default compilation semantics are to append its
                   3896: execution semantics to the execution semantics of the current
                   3897: definition.}
                   3898: 
                   3899: @comment TODO expand, make it co-operate with new sections on text interpreter.
                   3900: 
                   3901: @cindex immediate words
                   3902: @cindex compile-only words
                   3903: You can change the semantics of the most-recently defined word:
                   3904: 
                   3905: doc-immediate
                   3906: doc-compile-only
                   3907: doc-restrict
                   3908: 
                   3909: Note that ticking (@code{'}) a compile-only word gives an error
                   3910: (``Interpreting a compile-only word'').
1.1       anton    3911: 
1.26      crook    3912: Gforth also allows you to define words with arbitrary combinations of
                   3913: interpretation and compilation semantics.
1.1       anton    3914: 
1.26      crook    3915: doc-interpret/compile:
1.1       anton    3916: 
1.26      crook    3917: This feature was introduced for implementing @code{TO} and @code{S"}. I
                   3918: recommend that you do not define such words, as cute as they may be:
                   3919: they make it hard to get at both parts of the word in some contexts.
                   3920: E.g., assume you want to get an execution token for the compilation
                   3921: part. Instead, define two words, one that embodies the interpretation
                   3922: part, and one that embodies the compilation part.  Once you have done
                   3923: that, you can define a combined word with @code{interpret/compile:} for
                   3924: the convenience of your users.
1.1       anton    3925: 
1.26      crook    3926: You might try to use this feature to provide an optimizing
                   3927: implementation of the default compilation semantics of a word. For
                   3928: example, by defining:
1.1       anton    3929: @example
1.26      crook    3930: :noname
                   3931:    foo bar ;
                   3932: :noname
                   3933:    POSTPONE foo POSTPONE bar ;
                   3934: interpret/compile: foobar
1.1       anton    3935: @end example
1.26      crook    3936: 
1.23      crook    3937: @noindent
1.26      crook    3938: as an optimizing version of:
                   3939: 
1.1       anton    3940: @example
1.26      crook    3941: : foobar
                   3942:     foo bar ;
1.1       anton    3943: @end example
                   3944: 
1.26      crook    3945: Unfortunately, this does not work correctly with @code{[compile]},
                   3946: because @code{[compile]} assumes that the compilation semantics of all
                   3947: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
                   3948: foobar} would compile the compilation semantics for the optimizing
                   3949: @code{foobar}, whereas it would compile the interpretation semantics for
                   3950: the non-optimizing @code{foobar}.
1.1       anton    3951: 
1.26      crook    3952: @cindex state-smart words (are a bad idea)
                   3953: Some people try to use @var{state-smart} words to emulate the feature provided
                   3954: by @code{interpret/compile:} (words are state-smart if they check
                   3955: @code{STATE} during execution). E.g., they would try to code
                   3956: @code{foobar} like this:
1.1       anton    3957: 
1.26      crook    3958: @example
                   3959: : foobar
                   3960:   STATE @@
                   3961:   IF ( compilation state )
                   3962:     POSTPONE foo POSTPONE bar
                   3963:   ELSE
                   3964:     foo bar
                   3965:   ENDIF ; immediate
                   3966: @end example
1.1       anton    3967: 
1.26      crook    3968: Although this works if @code{foobar} is only processed by the text
                   3969: interpreter, it does not work in other contexts (like @code{'} or
                   3970: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
                   3971: for a state-smart word, not for the interpretation semantics of the
                   3972: original @code{foobar}; when you execute this execution token (directly
                   3973: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
                   3974: state, the result will not be what you expected (i.e., it will not
                   3975: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
                   3976: write them@footnote{For a more detailed discussion of this topic, see
                   3977: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
                   3978: Ertl; presented at EuroForth '98 and available from
                   3979: @url{http://www.complang.tuwien.ac.at/papers/}}!
1.1       anton    3980: 
1.26      crook    3981: @cindex defining words with arbitrary semantics combinations
                   3982: It is also possible to write defining words that define words with
                   3983: arbitrary combinations of interpretation and compilation semantics. In
                   3984: general, they look like this:
1.1       anton    3985: 
1.26      crook    3986: @example
                   3987: : def-word
                   3988:     create-interpret/compile
                   3989:     @var{code1}
                   3990: interpretation>
                   3991:     @var{code2}
                   3992: <interpretation
                   3993: compilation>
                   3994:     @var{code3}
                   3995: <compilation ;
                   3996: @end example
1.1       anton    3997: 
1.26      crook    3998: For a @var{word} defined with @code{def-word}, the interpretation
                   3999: semantics are to push the address of the body of @var{word} and perform
                   4000: @var{code2}, and the compilation semantics are to push the address of
                   4001: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
                   4002: can also be defined like this (except that the defined constants don't
                   4003: behave correctly when @code{[compile]}d):
1.1       anton    4004: 
1.26      crook    4005: @example
                   4006: : constant ( n "name" -- )
                   4007:     create-interpret/compile
                   4008:     ,
                   4009: interpretation> ( -- n )
                   4010:     @@
                   4011: <interpretation
                   4012: compilation> ( compilation. -- ; run-time. -- n )
                   4013:     @@ postpone literal
                   4014: <compilation ;
                   4015: @end example
1.1       anton    4016: 
1.26      crook    4017: doc-create-interpret/compile
                   4018: doc-interpretation>
                   4019: doc-<interpretation
                   4020: doc-compilation>
                   4021: doc-<compilation
1.1       anton    4022: 
1.26      crook    4023: Note that words defined with @code{interpret/compile:} and
                   4024: @code{create-interpret/compile} have an extended header structure that
                   4025: differs from other words; however, unless you try to access them with
                   4026: plain address arithmetic, you should not notice this. Words for
                   4027: accessing the header structure usually know how to deal with this; e.g.,
                   4028: @code{' word >body} also gives you the body of a word created with
                   4029: @code{create-interpret/compile}.
1.1       anton    4030: 
1.27      crook    4031: doc-postpone
                   4032: 
                   4033: 
                   4034: 
1.26      crook    4035: @c ----------------------------------------------------------
                   4036: @node The Text Interpreter, Tokens for Words, Defining Words, Words
                   4037: @section  The Text Interpreter
                   4038: @cindex interpreter - outer
                   4039: @cindex text interpreter
                   4040: @cindex outer interpreter
1.1       anton    4041: 
1.27      crook    4042: @comment index..
1.1       anton    4043: 
1.27      crook    4044: When a Forth system starts up, the final stages of initialisation are to
                   4045: set @code{state} to 0 (interperetation state) and execute @code{quit},
                   4046: to start the text interpreter.
                   4047: 
                   4048: The text interpreter is an endless loop that accepts input from various
                   4049: devices (by default the user input device -- the keyboard). A popular
                   4050: implementation technique for Forth is to implement a @var{forth virtual
                   4051: machine} using a loop called the @var{inner interpreter}. Because of
                   4052: this naming, the text interpreter is also known as the @var{outer
                   4053: interpreter}.
                   4054: 
                   4055: The text interpreter works on input one line at a time. Starting at the
                   4056: beginning of the line, it skips leading spaces (called @var{delimiters})
                   4057: then parses a string (a sequence of non-space characters) until it
                   4058: either reaches a space character or it reaches the end of the
                   4059: line. Having parsed a string, it then makes two attempts to do something
                   4060: with it:
                   4061: 
                   4062: @itemize @bullet
                   4063: @item
                   4064: It looks the string up in a dictionary of definitions. If the string is
                   4065: found in the dictionary, the string names a @var{definition} (also known
                   4066: as a @var{word}) and the dictionary search will return an @var{execution
                   4067: token} (xt) for the definition and some flags that show when the
                   4068: definition can be used legally. If the definition can be legally
                   4069: executed in @var{interpret} mode then the text interpreter will use the
                   4070: xt to execute it, otherwise it will issue an error message. The
                   4071: dictionary is described in more detail in <TODO>.
                   4072: @item
                   4073: If the string is not found in the dictionary, the text interpreter
                   4074: attempts to treat it as a number in the current radix (base 10 after
                   4075: initial startup). If the string represents a legal number in the current
                   4076: radix, the number is pushed onto the appropriate parameter stack.
                   4077: See @ref{Number Conversion} for details.
                   4078: @end itemize
                   4079: If both of these attempts fail, the remainder of the input line is
                   4080: discarded and the text interpreter isses an error message. If one of
                   4081: these attempts succeeds, the text interpreter repeats the parsing
                   4082: process until the end of the line has been reached. At this point, 
                   4083: it prints the status message ``  ok'' and waits for more input.
                   4084: 
                   4085: There are two important things to note about the behaviour of the text
                   4086: interpreter:
                   4087: 
                   4088: @itemize @bullet
                   4089: @item
                   4090: It processes each input string to completion before parsing additional
                   4091: characters from the input line.
                   4092: @item
                   4093: It keeps track of its position in the input line using a variable
                   4094: (called @code{>IN}, pronounced ``to-in''). The value of @code{>IN} can
                   4095: be modified by the execution of definitions in the input line. This
                   4096: means that definitions can ``trick'' the text interpreter either into
                   4097: skipping sections of the input line or into parsing a section of the
                   4098: input line more than once.
                   4099: @end itemize
1.21      crook    4100: 
1.26      crook    4101: doc->in
1.27      crook    4102: doc-source
                   4103: 
1.26      crook    4104: doc-tib
                   4105: doc-#tib
1.1       anton    4106: 
1.26      crook    4107: @menu
                   4108: * Number Conversion::
                   4109: * Interpret/Compile states::
                   4110: * Literals::
                   4111: * Interpreter Directives::
1.27      crook    4112: * Input Sources::
1.26      crook    4113: @end menu
1.1       anton    4114: 
                   4115: 
1.26      crook    4116: @node Number Conversion, Interpret/Compile states, The Text Interpreter, The Text Interpreter
                   4117: @subsection Number Conversion
                   4118: @cindex number conversion
                   4119: @cindex double-cell numbers, input format
                   4120: @cindex input format for double-cell numbers
                   4121: @cindex single-cell numbers, input format
                   4122: @cindex input format for single-cell numbers
                   4123: @cindex floating-point numbers, input format
                   4124: @cindex input format for floating-point numbers
1.1       anton    4125: 
1.26      crook    4126: If the text interpreter fails to find a particular string in the name
                   4127: dictionary, it attempts to convert it to a number using a set of rules.
1.1       anton    4128: 
1.26      crook    4129: Let <digit> represent any character that is a legal digit in the current
                   4130: number base (for example, 0-9 when the number base is decimal or 0-9, A-F
                   4131: when the number base is hexadecimal).
1.1       anton    4132: 
1.26      crook    4133: Let <decimal digit> represent any character in the range 0-9.
1.1       anton    4134: 
1.26      crook    4135: @comment TODO need to extend the next defn to support fp format
                   4136: Let @{+ | -@} represent the optional presence of either a @code{+} or
                   4137: @code{-} character.
1.1       anton    4138: 
1.26      crook    4139: Let * represent any number of instances of the previous character
                   4140: (including none).
1.1       anton    4141: 
1.26      crook    4142: Let any other character represent itself.
1.1       anton    4143: 
1.26      crook    4144: Now, the conversion rules are:
1.21      crook    4145: 
1.26      crook    4146: @itemize @bullet
                   4147: @item
                   4148: A string of the form <digit><digit>* is treated as a single-precision
                   4149: (CELL-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
                   4150: @item
                   4151: A string of the form -<digit><digit>* is treated as a single-precision
                   4152: (CELL-sized) negative integer, and is represented using 2's-complement
                   4153: arithmetic. Examples are -45 -5681 -0
                   4154: @item
                   4155: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
                   4156: (double-CELL-sized) positive integer. Examples are 3465. 3.465 34.65
                   4157: (and note that these all represent the same number).
                   4158: @item
                   4159: A string of the form -<digit><digit>*.<digit>* is treated as a
                   4160: double-precision (double-CELL-sized) negative integer, and is
                   4161: represented using 2's-complement arithmetic. Examples are -3465. -3.465
                   4162: -34.65 (and note that these all represent the same number).
                   4163: @item
                   4164: A string of the form @{+ | -@}<decimal digit>@{.@}<decimal digit>*@{e | E@}@{+
                   4165: | -@}<decimal digit><decimal digit>* is treated as floating-point
                   4166: number. Examples are 1e0 1.e 1.e0 +1e+0 (which all represent the same
                   4167: number) +12.E-4 
                   4168: @end itemize
1.1       anton    4169: 
1.26      crook    4170: By default, the number base used for integer number conversion is given
                   4171: by the contents of a variable named @code{BASE}. Base 10 (decimal) is
                   4172: always used for floating-point number conversion.
1.1       anton    4173: 
1.26      crook    4174: doc-base
                   4175: doc-hex
                   4176: doc-decimal
1.1       anton    4177: 
1.26      crook    4178: @cindex '-prefix for character strings
                   4179: @cindex &-prefix for decimal numbers
                   4180: @cindex %-prefix for binary numbers
                   4181: @cindex $-prefix for hexadecimal numbers
                   4182: Gforth allows you to override the value of @code{BASE} by using a prefix
                   4183: before the first digit of an (integer) number. Four prefixes are
                   4184: supported:
1.1       anton    4185: 
1.26      crook    4186: @itemize @bullet
                   4187: @item
                   4188: @code{&} -- decimal number
                   4189: @item
                   4190: @code{%} -- binary number
                   4191: @item
                   4192: @code{$} -- hexadecimal number
                   4193: @item
                   4194: @code{'} -- base 256 number
                   4195: @end itemize
1.1       anton    4196: 
1.26      crook    4197: Here are some examples, with the equivalent decimal number shown after
                   4198: in braces:
1.1       anton    4199: 
1.26      crook    4200: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
                   4201: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
                   4202: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
                   4203: &905 (905), $abc (2478), $ABC (2478).
1.1       anton    4204: 
1.26      crook    4205: @cindex number conversion - traps for the unwary
                   4206: Number conversion has a number of traps for the unwary:
1.1       anton    4207: 
1.26      crook    4208: @itemize @bullet
                   4209: @item
                   4210: You cannot determine the current number base using the code sequence
                   4211: @code{BASE @@ .} -- the number base is always 10 in the current number
                   4212: base. Instead, use something like @code{BASE @@ DECIMAL DUP . BASE !}
                   4213: @item
                   4214: If the number base is set to a value greater than 14 (for example,
                   4215: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
                   4216: it to be intepreted as either a single-precision integer or a
                   4217: floating-point number (Gforth treats it as an integer). The ambiguity
                   4218: can be resolved by explicitly stating the sign of the mantissa and/or
                   4219: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
                   4220: ambiguity arises; either representation will be treated as a
                   4221: floating-point number.
                   4222: @item
                   4223: There is a word @code{bin} but it does @var{not} set the number base!
                   4224: It is used to specify file types.
                   4225: @item
                   4226: ANS Forth requires the @code{.} of a double-precision number to
                   4227: be the final character in the string. Allowing the @code{.} to be
                   4228: anywhere after the first digit is a Gforth extension.
                   4229: @item
                   4230: The number conversion process does not check for overflow.
                   4231: @item
                   4232: In Gforth, number conversion to floating-point numbers always use base
                   4233: 10, irrespective of the value of @code{BASE}. In ANS Forth,
                   4234: conversion to floating-point numbers whilst the value of
                   4235: @code{BASE} is not 10 is an ambiguous condition.
                   4236: @end itemize
1.1       anton    4237: 
                   4238: 
1.26      crook    4239: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
                   4240: @subsection Interpret/Compile states
                   4241: @cindex Interpret/Compile states
1.1       anton    4242: 
1.26      crook    4243: @comment TODO Intro blah.
1.1       anton    4244: 
1.26      crook    4245: doc-state
                   4246: doc-[
                   4247: doc-]
1.1       anton    4248: 
                   4249: 
1.26      crook    4250: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
                   4251: @subsection Literals
                   4252: @cindex Literals
1.21      crook    4253: 
1.26      crook    4254: @comment TODO Intro blah.
1.23      crook    4255: 
1.26      crook    4256: doc-literal
                   4257: doc-]L
                   4258: doc-2literal
                   4259: doc-fliteral
1.1       anton    4260: 
1.27      crook    4261: @node Interpreter Directives, Input Sources, Literals, The Text Interpreter
1.26      crook    4262: @subsection Interpreter Directives
                   4263: @cindex interpreter directives
1.1       anton    4264: 
1.26      crook    4265: These words are usually used outside of definitions; for example, to
                   4266: control which parts of a source file are processed by the text
                   4267: interpreter. There are only a few ANS Forth Standard words, but Gforth
                   4268: supplements these with a rich set of immediate control structure words
                   4269: to compensate for the fact that the non-immediate versions can only be
                   4270: used in compile state (@pxref{Control Structures}).
1.1       anton    4271: 
1.26      crook    4272: doc-[IF]
                   4273: doc-[ELSE]
                   4274: doc-[THEN]
                   4275: doc-[ENDIF]
1.1       anton    4276: 
1.26      crook    4277: doc-[IFDEF]
                   4278: doc-[IFUNDEF]
1.1       anton    4279: 
1.26      crook    4280: doc-[?DO]
                   4281: doc-[DO]
                   4282: doc-[FOR]
                   4283: doc-[LOOP]
                   4284: doc-[+LOOP]
                   4285: doc-[NEXT]
1.1       anton    4286: 
1.26      crook    4287: doc-[BEGIN]
                   4288: doc-[UNTIL]
                   4289: doc-[AGAIN]
                   4290: doc-[WHILE]
                   4291: doc-[REPEAT]
1.1       anton    4292: 
1.27      crook    4293: 
                   4294: @node Input Sources, , Interpreter Directives, The Text Interpreter
                   4295: @subsection Input Sources
                   4296: @cindex input sources
                   4297: @cindex text interpreter - input sources
                   4298: 
                   4299: The text interpreter can process input from these sources:
                   4300: 
                   4301: @itemize @bullet
                   4302: @item
                   4303: The user input device -- the keyboard. This is the default input for the
                   4304: text interpreter when Forth is started up.
                   4305: @item
                   4306: A file, using the words described in @ref{Forth source files}.
                   4307: @item
                   4308: A block, using the words described in @ref{Blocks}.
                   4309: @item
                   4310: A text string, using @code{evaluate}.
                   4311: @end itemize
                   4312: 
                   4313: A program can determine the current input device by checking the values
                   4314: of @code{source-id} and @code{blk}.
                   4315: 
                   4316: doc-source-id
                   4317: doc-blk
                   4318: 
                   4319: doc-save-input
                   4320: doc-restore-input
                   4321: 
                   4322: doc-evaluate
                   4323: 
                   4324: 
1.26      crook    4325: @c -------------------------------------------------------------
                   4326: @node Tokens for Words, Word Lists, The Text Interpreter, Words
                   4327: @section Tokens for Words
                   4328: @cindex tokens for words
1.1       anton    4329: 
1.28    ! crook    4330: This section describes the creation and use of tokens that represent
1.26      crook    4331: words on the stack (and in data space).
1.21      crook    4332: 
1.26      crook    4333: Named words have interpretation and compilation semantics. Unnamed words
                   4334: just have execution semantics.
1.21      crook    4335: 
1.26      crook    4336: @comment TODO ?normally interpretation semantics are the execution semantics.
                   4337: @comment this should all be covered in earlier ss
1.21      crook    4338: 
1.26      crook    4339: @cindex execution token
                   4340: An @dfn{execution token} represents the execution semantics of an
                   4341: unnamed word. An execution token occupies one cell. As explained in
                   4342: @ref{Supplying names}, the execution token of the last word
                   4343: defined can be produced with @code{lastxt}.
1.1       anton    4344: 
1.26      crook    4345: doc-execute
                   4346: doc-compile,
1.1       anton    4347: 
1.26      crook    4348: @cindex code field address
                   4349: @cindex CFA
                   4350: In Gforth, the abstract data type @emph{execution token} is implemented
                   4351: as a code field address (CFA).
                   4352: @comment TODO note that the standard does not say what it represents..
                   4353: @comment and you cannot necessarily compile it in all Forths (eg native
                   4354: @comment compilers?).
1.1       anton    4355: 
1.26      crook    4356: The interpretation semantics of a named word are also represented by an
                   4357: execution token. You can get it with:
1.1       anton    4358: 
1.26      crook    4359: doc-[']
                   4360: doc-'
1.1       anton    4361: 
1.26      crook    4362: For literals, you use @code{'} in interpreted code and @code{[']} in
                   4363: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusually
                   4364: by complaining about compile-only words. To get an execution token for a
                   4365: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
                   4366: @var{X} drop}.
1.1       anton    4367: 
1.26      crook    4368: @cindex compilation token
                   4369: The compilation semantics are represented by a @dfn{compilation token}
                   4370: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
                   4371: execution token. The compilation semantics represented by the
                   4372: compilation token can be performed with @code{execute}, which consumes
                   4373: the whole compilation token, with an additional stack effect determined
                   4374: by the represented compilation semantics.
1.1       anton    4375: 
1.26      crook    4376: doc-[comp']
                   4377: doc-comp'
1.1       anton    4378: 
1.26      crook    4379: You can compile the compilation semantics with @code{postpone,}. I.e.,
                   4380: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
                   4381: @var{word}}.
1.1       anton    4382: 
1.26      crook    4383: doc-postpone,
1.1       anton    4384: 
1.26      crook    4385: At present, the @var{w} part of a compilation token is an execution
                   4386: token, and the @var{xt} part represents either @code{execute} or
                   4387: @code{compile,}. However, don't rely on that knowledge, unless necessary;
                   4388: we may introduce unusual compilation tokens in the future (e.g.,
                   4389: compilation tokens representing the compilation semantics of literals).
1.21      crook    4390: 
1.26      crook    4391: @cindex name token
                   4392: @cindex name field address
                   4393: @cindex NFA
                   4394: Named words are also represented by the @dfn{name token}, (@var{nt}). The abstract
                   4395: data type @emph{name token} is implemented as a name field address (NFA).
1.1       anton    4396: 
1.26      crook    4397: doc-find-name
                   4398: doc-name>int
                   4399: doc-name?int
                   4400: doc-name>comp
                   4401: doc-name>string
1.1       anton    4402: 
1.26      crook    4403: @c -------------------------------------------------------------
                   4404: @node Word Lists, Environmental Queries, Tokens for Words, Words
                   4405: @section Word Lists
                   4406: @cindex word lists
                   4407: @cindex name dictionary
1.1       anton    4408: 
1.26      crook    4409: @cindex wid
                   4410: All definitions other than those created by @code{:noname} have an entry
                   4411: in the name dictionary. The name dictionary is fragmented into a number
                   4412: of parts, called @var{word lists}. A word list is identified by a
                   4413: cell-sized word list identifier (@var{wid}) in much the same way as a
                   4414: file is identified by a file handle. The numerical value of the wid has
                   4415: no (portable) meaning, and might change from session to session.
1.1       anton    4416: 
1.26      crook    4417: @cindex compilation word list
                   4418: At any one time, a single word list is defined as the word list to which
                   4419: all new definitions will be added -- this is called the @var{compilation
                   4420: word list}. When Gforth is started, the compilation word list is the
                   4421: word list called @code{FORTH-WORDLIST}.
1.1       anton    4422: 
1.26      crook    4423: @cindex search order stack
                   4424: Forth maintains a stack of word lists, representing the @var{search
                   4425: order}.  When the name dictionary is searched (for example, when
                   4426: attempting to find a word's execution token during compilation), only
                   4427: those word lists that are currently in the search order are
                   4428: searched. The most recently-defined word in the word list at the top of
                   4429: the word list stack is searched first, and the search proceeds until
                   4430: either the word is located or the oldest definition in the word list at
                   4431: the bottom of the stack is reached. Definitions of the word may exist in
                   4432: more than one word lists; the search order determines which version will
                   4433: be found.
1.1       anton    4434: 
1.26      crook    4435: The ANS Forth Standard ``Search order'' word set is intended to provide a
                   4436: set of low-level tools that allow various different schemes to be
                   4437: implemented. Gforth provides @code{vocabulary}, a traditional Forth
                   4438: word.  @file{compat/vocabulary.fs} provides an implementation in ANS
                   4439: Standard Forth.
1.1       anton    4440: 
1.27      crook    4441: @comment TODO: locals section refers to here, saying that every word list (aka
                   4442: @comment vocabulary) has its own methods for searching etc. Need to document that.
1.1       anton    4443: 
1.27      crook    4444: @comment the thisone- prefix is used to pick out the true definition of a
                   4445: @comment word from the source files, rather than some alias.
1.26      crook    4446: doc-forth-wordlist
                   4447: doc-definitions
                   4448: doc-get-current
                   4449: doc-set-current
                   4450: doc-get-order
1.27      crook    4451: doc---thisone-set-order
1.26      crook    4452: doc-wordlist
                   4453: doc-also
1.27      crook    4454: doc---thisone-forth
1.26      crook    4455: doc-only
1.27      crook    4456: doc---thisone-order
1.26      crook    4457: doc-previous
1.15      anton    4458: 
1.26      crook    4459: doc-find
                   4460: doc-search-wordlist
1.15      anton    4461: 
1.26      crook    4462: doc-words
                   4463: doc-vlist
1.1       anton    4464: 
1.26      crook    4465: doc-mappedwordlist
                   4466: doc-root
                   4467: doc-vocabulary
                   4468: doc-seal
                   4469: doc-vocs
                   4470: doc-current
                   4471: doc-context
1.1       anton    4472: 
1.26      crook    4473: @menu
                   4474: * Why use word lists?::
                   4475: * Word list examples::
                   4476: @end menu
                   4477: 
                   4478: @node Why use word lists?, Word list examples, Word Lists, Word Lists
                   4479: @subsection Why use word lists?
                   4480: @cindex word lists - why use them?
                   4481: 
                   4482: There are several reasons for using multiple word lists:
                   4483: 
                   4484: @itemize @bullet
                   4485: @item
                   4486: To improve compilation speed by reducing the number of name dictionary
                   4487: entries that must be searched. This is achieved by creating a new
                   4488: word list that contains all of the definitions that are used in the
                   4489: definition of a Forth system but which would not usually be used by
                   4490: programs running on that system. That word list would be on the search
                   4491: list when the Forth system was compiled but would be removed from the
                   4492: search list for normal operation. This can be a useful technique for
                   4493: low-performance systems (for example, 8-bit processors in embedded
                   4494: systems) but is unlikely to be necessary in high-performance desktop
                   4495: systems.
                   4496: @item
                   4497: To prevent a set of words from being used outside the context in which
                   4498: they are valid. Two classic examples of this are an integrated editor
                   4499: (all of the edit commands are defined in a separate word list; the
                   4500: search order is set to the editor word list when the editor is invoked;
                   4501: the old search order is restored when the editor is terminated) and an
                   4502: integrated assembler (the op-codes for the machine are defined in a
                   4503: separate word list which is used when a @code{CODE} word is defined).
                   4504: @item
                   4505: To prevent a name-space clash between multiple definitions with the same
                   4506: name. For example, when building a cross-compiler you might have a word
                   4507: @code{IF} that generates conditional code for your target system. By
                   4508: placing this definition in a different word list you can control whether
                   4509: the host system's @code{IF} or the target system's @code{IF} get used in
                   4510: any particular context by controlling the order of the word lists on the
                   4511: search order stack.
                   4512: @end itemize
1.1       anton    4513: 
1.26      crook    4514: @node Word list examples, ,Why use word lists?, Word Lists
                   4515: @subsection Word list examples
                   4516: @cindex word lists - examples
1.1       anton    4517: 
1.26      crook    4518: Here is an example of creating and using a new wordlist using ANS
                   4519: Forth Standard words:
1.1       anton    4520: 
                   4521: @example
1.26      crook    4522: wordlist constant my-new-words-wordlist
                   4523: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
1.21      crook    4524: 
1.26      crook    4525: \ add it to the search order
                   4526: also my-new-words
1.21      crook    4527: 
1.26      crook    4528: \ alternatively, add it to the search order and make it
                   4529: \ the compilation word list
                   4530: also my-new-words definitions
                   4531: \ type "order" to see the problem
1.21      crook    4532: @end example
                   4533: 
1.26      crook    4534: The problem with this example is that @code{order} has no way to
                   4535: associate the name @code{my-new-words} with the wid of the word list (in
                   4536: Gforth, @code{order} and @code{vocs} will display @code{???}  for a wid
                   4537: that has no associated name). There is no Standard way of associating a
                   4538: name with a wid.
                   4539: 
                   4540: In Gforth, this example can be re-coded using @code{vocabulary}, which
                   4541: associates a name with a wid:
1.21      crook    4542: 
1.26      crook    4543: @example
                   4544: vocabulary my-new-words
1.21      crook    4545: 
1.26      crook    4546: \ add it to the search order
                   4547: my-new-words
1.21      crook    4548: 
1.26      crook    4549: \ alternatively, add it to the search order and make it
                   4550: \ the compilation word list
                   4551: my-new-words definitions
                   4552: \ type "order" to see that the problem is solved
                   4553: @end example
1.23      crook    4554: 
1.26      crook    4555: @c -------------------------------------------------------------
                   4556: @node Environmental Queries, Files, Word Lists, Words
                   4557: @section Environmental Queries
                   4558: @cindex environmental queries
1.21      crook    4559: 
1.26      crook    4560: ANS Forth introduced the idea of ``environmental queries'' as a way
                   4561: for a program running on a system to determine certain characteristics of the system.
                   4562: The Standard specifies a number of strings that might be recognised by a system.
1.21      crook    4563: 
1.26      crook    4564: The Standard requires that the name space used for environmental queries
                   4565: be distinct from the name space used for definitions.
1.21      crook    4566: 
1.26      crook    4567: Typically, environmental queries are supported by creating a set of
                   4568: definitions in a word list that is @var{only} used during environmental
                   4569: queries; that is what Gforth does. There is no Standard way of adding
                   4570: definitions to the set of recognised environmental queries, but any
                   4571: implementation that supports the loading of optional word sets must have
                   4572: some mechanism for doing this (after loading the word set, the
                   4573: associated environmental query string must return @code{true}). In
                   4574: Gforth, the word list used to honour environmental queries can be
                   4575: manipulated just like any other word list.
1.21      crook    4576: 
1.26      crook    4577: doc-environment?
                   4578: doc-environment-wordlist
1.21      crook    4579: 
1.26      crook    4580: doc-gforth
                   4581: doc-os-class
1.21      crook    4582: 
1.26      crook    4583: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
                   4584: returning two items on the stack, querying it using @code{environment?}
                   4585: will return an additional item; the @code{true} flag that shows that the
                   4586: string was recognised.
1.21      crook    4587: 
1.26      crook    4588: @comment TODO Document the standard strings or note where they are documented herein
1.21      crook    4589: 
1.26      crook    4590: Here are some examples of using environmental queries:
1.21      crook    4591: 
1.26      crook    4592: @example
                   4593: s" address-unit-bits" environment? 0=
                   4594: [IF]
                   4595:      cr .( environmental attribute address-units-bits unknown... ) cr
                   4596: [THEN]
1.21      crook    4597: 
1.26      crook    4598: s" block" environment? [IF] DROP include block.fs [THEN]
1.21      crook    4599: 
1.26      crook    4600: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
1.21      crook    4601: 
1.26      crook    4602: s" gforth" environment? [IF] .( Gforth version ) TYPE
                   4603:                         [ELSE] .( Not Gforth..) [THEN]
                   4604: @end example
1.21      crook    4605: 
                   4606: 
1.26      crook    4607: Here is an example of adding a definition to the environment word list:
1.21      crook    4608: 
1.26      crook    4609: @example
                   4610: get-current environment-wordlist set-current
                   4611: true constant block
                   4612: true constant block-ext
                   4613: set-current
                   4614: @end example
1.21      crook    4615: 
1.26      crook    4616: You can see what definitions are in the environment word list like this:
1.21      crook    4617: 
1.26      crook    4618: @example
                   4619: get-order 1+ environment-wordlist swap set-order words previous
                   4620: @end example
1.21      crook    4621: 
                   4622: 
1.26      crook    4623: @c -------------------------------------------------------------
                   4624: @node Files, Blocks, Environmental Queries, Words
                   4625: @section Files
1.28    ! crook    4626: @cindex files
        !          4627: @cindex I/O - file-handling
1.21      crook    4628: 
1.26      crook    4629: Gforth provides facilities for accessing files that are stored in the
                   4630: host operating system's file-system. Files that are processed by Gforth
                   4631: can be divided into two categories:
1.21      crook    4632: 
1.23      crook    4633: @itemize @bullet
                   4634: @item
1.26      crook    4635: Files that are processed by the Text Interpreter (@var{Forth source files}).
1.23      crook    4636: @item
1.26      crook    4637: Files that are processed by some other program (@var{general files}).
                   4638: @end itemize
                   4639: 
                   4640: @menu
                   4641: * Forth source files::
                   4642: * General files::         
                   4643: * Search Paths::                 
                   4644: * Forth Search Paths::    
                   4645: * General Search Paths::        
                   4646: @end menu
                   4647: 
1.21      crook    4648: 
1.26      crook    4649: @c -------------------------------------------------------------
                   4650: @node Forth source files, General files, Files, Files
                   4651: @subsection Forth source files
                   4652: @cindex including files
                   4653: @cindex Forth source files
1.21      crook    4654: 
1.26      crook    4655: The simplest way to interpret the contents of a file is to use one of
                   4656: these two formats:
1.21      crook    4657: 
1.26      crook    4658: @example
                   4659: include mysource.fs
                   4660: s" mysource.fs" included
                   4661: @end example
1.21      crook    4662: 
1.26      crook    4663: Sometimes you want to include a file only if it is not included already
                   4664: (by, say, another source file). In that case, you can use one of these
                   4665: fomats:
1.21      crook    4666: 
1.26      crook    4667: @example
                   4668: require mysource.fs
                   4669: needs mysource.fs
                   4670: s" mysource.fs" required
                   4671: @end example
1.21      crook    4672: 
1.26      crook    4673: @cindex stack effect of included files
                   4674: @cindex including files, stack effect
                   4675: I recommend that you write your source files such that interpreting them
                   4676: does not change the stack. This allows using these files with
                   4677: @code{required} and friends without complications. For example:
1.21      crook    4678: 
1.26      crook    4679: @example
                   4680: 1 require foo.fs drop
                   4681: @end example
1.21      crook    4682: 
1.26      crook    4683: doc-include-file
                   4684: doc-included
1.28    ! crook    4685: doc-included?
1.26      crook    4686: doc-include
                   4687: @comment TODO describe what happens on error. Describes how the require
                   4688: @comment stuff works and describe how to clear/reset the history (eg
1.28    ! crook    4689: @comment for debug). Add examples. Describe the scope of the file
        !          4690: @comment history.
1.26      crook    4691: doc-required
                   4692: doc-require
                   4693: doc-needs
1.28    ! crook    4694: doc-init-included-files
1.21      crook    4695: 
1.26      crook    4696: A definition in ANS Forth for @code{required} is provided in
                   4697: @file{compat/required.fs}.
1.21      crook    4698: 
1.26      crook    4699: @c -------------------------------------------------------------
                   4700: @node General files, Search Paths, Forth source files, Files
                   4701: @subsection General files
                   4702: @cindex general files
                   4703: @cindex file-handling
1.21      crook    4704: 
1.26      crook    4705: Files are opened/created by name and type. The following types are
                   4706: recognised:
1.1       anton    4707: 
1.26      crook    4708: doc-r/o
                   4709: doc-r/w
                   4710: doc-w/o
                   4711: doc-bin
1.1       anton    4712: 
1.26      crook    4713: When a file is opened/created, it returns a file identifier,
                   4714: @var{wfileid} that is used for all other file commands. All file
                   4715: commands also return a status value, @var{wior}, that is 0 for a
                   4716: successful operation and an implementation-defined non-zero value in the
                   4717: case of an error.
1.21      crook    4718: 
1.26      crook    4719: doc-open-file
                   4720: doc-create-file
1.21      crook    4721: 
1.26      crook    4722: doc-close-file
                   4723: doc-delete-file
                   4724: doc-rename-file
                   4725: doc-read-file
                   4726: doc-read-line
                   4727: doc-write-file
                   4728: doc-write-line
                   4729: doc-emit-file
                   4730: doc-flush-file
1.21      crook    4731: 
1.26      crook    4732: doc-file-status
                   4733: doc-file-position
                   4734: doc-reposition-file
                   4735: doc-file-size
                   4736: doc-resize-file
1.21      crook    4737: 
1.26      crook    4738: @c ---------------------------------------------------------
                   4739: @node Search Paths, Forth Search Paths, General files, Files
                   4740: @subsection Search Paths
                   4741: @cindex path for @code{included}
                   4742: @cindex file search path
                   4743: @cindex @code{include} search path
                   4744: @cindex search path for files
1.21      crook    4745: 
1.28    ! crook    4746: @comment TODO what uses these search paths.. just include and friends?
1.26      crook    4747: If you specify an absolute filename (i.e., a filename starting with
                   4748: @file{/} or @file{~}, or with @file{:} in the second position (as in
                   4749: @samp{C:...})) for @code{included} and friends, that file is included
                   4750: just as you would expect.
1.21      crook    4751: 
1.26      crook    4752: For relative filenames, Gforth uses a search path similar to Forth's
                   4753: search order (@pxref{Word Lists}). It tries to find the given filename
                   4754: in the directories present in the path, and includes the first one it
                   4755: finds. There are separate search paths for Forth source files and
                   4756: general files.
1.21      crook    4757: 
1.26      crook    4758: If the search path contains the directory @file{.} (as it should), this
                   4759: refers to the directory that the present file was @code{included}
                   4760: from. This allows files to include other files relative to their own
                   4761: position (irrespective of the current working directory or the absolute
                   4762: position).  This feature is essential for libraries consisting of
                   4763: several files, where a file may include other files from the library.
                   4764: It corresponds to @code{#include "..."} in C. If the current input
                   4765: source is not a file, @file{.} refers to the directory of the innermost
                   4766: file being included, or, if there is no file being included, to the
                   4767: current working directory.
1.21      crook    4768: 
1.26      crook    4769: Use @file{~+} to refer to the current working directory (as in the
                   4770: @code{bash}).
1.1       anton    4771: 
1.26      crook    4772: If the filename starts with @file{./}, the search path is not searched
                   4773: (just as with absolute filenames), and the @file{.} has the same meaning
                   4774: as described above.
1.1       anton    4775: 
1.26      crook    4776: @c ---------------------------------------------------------
                   4777: @node Forth Search Paths, General Search Paths, Search Paths, Files
                   4778: @subsubsection Forth Search Paths
1.28    ! crook    4779: @cindex search path control - Forth
1.5       anton    4780: 
1.26      crook    4781: The search path is initialized when you start Gforth (@pxref{Invoking
                   4782: Gforth}). You can display it and change it using these words:
1.5       anton    4783: 
1.26      crook    4784: doc-.fpath
                   4785: doc-fpath+
                   4786: doc-fpath=
                   4787: doc-open-fpath-file
1.5       anton    4788: 
1.26      crook    4789: Here is an example of using @code{fpath} and @code{require}:
1.5       anton    4790: 
1.26      crook    4791: @example
                   4792: fpath= /usr/lib/forth/|./
                   4793: require timer.fs
                   4794: @end example
1.5       anton    4795: 
1.26      crook    4796: @c ---------------------------------------------------------
                   4797: @node General Search Paths,  , Forth Search Paths, Files
                   4798: @subsubsection General Search Paths
                   4799: @cindex search path control - for user applications
1.5       anton    4800: 
1.26      crook    4801: Your application may need to search files in several directories, like
                   4802: @code{included} does. To facilitate this, Gforth allows you to define
                   4803: and use your own search paths, by providing generic equivalents of the
                   4804: Forth search path words:
1.5       anton    4805: 
1.26      crook    4806: doc-.path
                   4807: doc-path+
                   4808: doc-path=
                   4809: doc-open-path-file
1.5       anton    4810: 
1.26      crook    4811: Here's an example of creating a search path:
1.5       anton    4812: 
1.26      crook    4813: @example
                   4814: \ Make a buffer for the path:
                   4815: create mypath   100 chars ,     \ maximum length (is checked)
                   4816:                 0 ,             \ real len
                   4817:                 100 chars allot \ space for path
                   4818: @end example
1.5       anton    4819: 
1.26      crook    4820: @c -------------------------------------------------------------
                   4821: @node Blocks, Other I/O, Files, Words
                   4822: @section Blocks
1.28    ! crook    4823: @cindex I/O - blocks
        !          4824: @cindex blocks
        !          4825: 
        !          4826: @comment TODO finish the TODOs below and add more index entries
        !          4827: 
        !          4828: When you run Gforth on a modern desk-top computer, it runs under the
        !          4829: control of an operating system which provides certain services.  One of
        !          4830: these services is @var{file services}, which allows Forth source code
        !          4831: and data to be stored in files and read into Gforth (@pxref{Files}).
        !          4832: 
        !          4833: Traditionally, Forth has been an important programming language on
        !          4834: systems where it has interfaced directly to the underlying hardware with
        !          4835: no intervening operating system. Forth provides a mechanism, called
        !          4836: @var{blocks}, for accessing mass storage on such systems.
        !          4837: 
        !          4838: A block is a 1024-byte data area, which can be used to hold data or
        !          4839: Forth source code. No structure is imposed on the contents of the
        !          4840: block. A block is identified by its number; blocks are numbered
        !          4841: contiguously from 1 to an implementation-defined maximum.
        !          4842: 
        !          4843: A typical system that used blocks but no operating system might use a
        !          4844: single floppy-disk drive for mass storage, with the disks formatted to
        !          4845: provide 256-byte sectors. Blocks would be implemented by assigning the
        !          4846: first four sectors of the disk to block 1, the second four sectors to
        !          4847: block 2 and so on, up to the limit of the capacity of the disk. The disk
        !          4848: would not contain any file system information, just the set of blocks.
        !          4849: 
        !          4850: On systems that do provide file services, blocks are typically
        !          4851: implemented by storing a sequence of blocks within a single @var{blocks
        !          4852: file}.  The size of the blocks file will be an exact multiple of 1024
        !          4853: bytes, corresponding to the number of blocks it contains. This is the
        !          4854: mechanism that Gforth uses.
        !          4855: 
        !          4856: Only 1 blocks file can be open at a time. If you use block words without
        !          4857: having specified a blocks file, Gforth defaults to the blocks file
        !          4858: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
        !          4859: locate a blocks file (@pxref{Forth Search Paths}).
        !          4860: 
        !          4861: When you read and write blocks under program control, Gforth uses a
        !          4862: number of @var{block buffers} as intermediate storage. These buffers are
        !          4863: not used when you use @code{load} to interpret the contents of a block.
        !          4864: 
        !          4865: The behaviour of the block buffers is directly analagous to that of a
        !          4866: cache. Each block buffer has three states:
        !          4867: 
        !          4868: @itemize @bullet
        !          4869: @item
        !          4870: Unassigned
        !          4871: @item
        !          4872: Assigned-clean
        !          4873: @item
        !          4874: Assigned-dirty
        !          4875: @end itemize
        !          4876: 
        !          4877: Initially, all block buffers are @var{unassigned}. In order to access a
        !          4878: block, the block (specified by its block number) must be assigned to a
        !          4879: block buffer.
        !          4880: 
        !          4881: The assignment of a block to a block buffer is performed by @code{block}
        !          4882: or @code{buffer}. Use @code{block} when you wish to modify the existing
        !          4883: contents of a block. Use @code{buffer} when you don't care about the
        !          4884: existing contents of the block@footnote{The ANS Forth definition of
        !          4885: @code{block} is intended not to cause disk I/O; if the data associated
        !          4886: with the particular block is already stored in a block buffer due to an
        !          4887: earlier @code{block} command, @code{buffer} will return that block
        !          4888: buffer and the existing contents of the block will be
        !          4889: available. Otherwise, @code{buffer} will simply assign a new, empty
        !          4890: block buffer for the block}.
        !          4891: 
        !          4892: Once a block has been assigned to a block buffer, the block buffer state
        !          4893: becomes @var{assigned-clean}. Data can now be manipulated within the
        !          4894: block buffer.
        !          4895: 
        !          4896: When the contents of a block buffer is changed it is necessary,
        !          4897: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
        !          4898: either abandon the changes (by doing nothing) or commit the changes,
        !          4899: using @code{update}. Using @code{update} does not change the blocks
        !          4900: file; it simply changes a block buffer's state to @var{assigned-dirty}.
        !          4901: 
        !          4902: The word @code{flush} causes all @var{assigned-dirty} blocks to be
        !          4903: written back to the blocks file on disk. Leaving Gforth using @code{bye}
        !          4904: also causes a @code{flush} to be performed.
        !          4905: 
        !          4906: In Gforth, @code{block} and @code{buffer} use a @var{direct-mapped}
        !          4907: algorithm to assign a block buffer to a block. That means that any
        !          4908: particular block can only be assigned to one specific block buffer,
        !          4909: called (for the particular operation) the @var{victim buffer}. If the
        !          4910: victim buffer is @var{unassigned} or @var{assigned-clean} it can be
        !          4911: allocated to the new block immediately. If it is @var{assigned-dirty}
        !          4912: its current contents must be written out to disk before it can be
        !          4913: allocated to the new block.
        !          4914: 
        !          4915: Although no structure is imposed on the contents of a block, it is
        !          4916: traditional to display the contents as 16 lines each of 64 characters.  A
        !          4917: block provides a single, continuous stream of input (for example, it
        !          4918: acts as a single parse area) -- there are no end-of-line characters
        !          4919: within a block, and no end-of-file character at the end of a
        !          4920: block. There are two consequences of this:
1.26      crook    4921: 
1.28    ! crook    4922: @itemize @bullet
        !          4923: @item
        !          4924: The last character of one line wraps straight into the first character
        !          4925: of the following line
        !          4926: @item
        !          4927: The word @code{\} -- comment to end of line -- requires special
        !          4928: treatment; in the context of a block it causes all characters until the
        !          4929: end of the current 64-character ``line'' to be ignored.
        !          4930: @end itemize
        !          4931: 
        !          4932: In Gforth, when you use @code{block} with a non-existent block number,
        !          4933: the current block file will be extended to the appropriate size and the
        !          4934: block buffer will be initialised with spaces.
        !          4935: 
        !          4936: Gforth doesn't encourage the use of blocks@footnote{See Frank Sergeant's
        !          4937: Pygmy Forth to see just how well blocks can be integrated into a Forth
        !          4938: programming environment}; the mechanism is only provided for backward
        !          4939: compatibility -- ANS Forth requires blocks to be available when files
        !          4940: are.
        !          4941: 
        !          4942: Common techniques that are used when working with blocks include:
        !          4943: 
        !          4944: @itemize @bullet
        !          4945: @item
        !          4946: A screen editor that allows you to edit blocks without leaving the Forth
        !          4947: environment.
        !          4948: @item
        !          4949: Shadow screens; where every code block has an associated block
        !          4950: containing comments (for example: code in odd block numbers, comments in
        !          4951: even block numbers). Typically, the block editor provides a convenient
        !          4952: mechanism to toggle between code and comments.
        !          4953: @item
        !          4954: Load blocks; a single block (typically block 1) contains a number of
        !          4955: @code{thru} commands which @code{load} the whole of the application.
        !          4956: @item
        !          4957: Chaining blocks; a block terminates with a @code{-->} so that a whole
        !          4958: application can be @code{load}ed by @code{load}ing a single block.
        !          4959: @end itemize
1.26      crook    4960: 
                   4961: 
                   4962: @comment TODO what about errors on open-blocks?
                   4963: doc-open-blocks
                   4964: doc-use
                   4965: doc-get-block-fid
                   4966: doc-block-position
1.28    ! crook    4967: 
        !          4968: doc-scr
        !          4969: doc-list
        !          4970: 
        !          4971: doc---block-block
        !          4972: doc-buffer
        !          4973: 
1.26      crook    4974: doc-update
1.28    ! crook    4975: doc-updated?
1.26      crook    4976: doc-save-buffers
                   4977: doc-empty-buffers
                   4978: doc-empty-buffer
                   4979: doc-flush
1.28    ! crook    4980: 
1.26      crook    4981: doc-load
                   4982: doc-thru
                   4983: doc-+load
                   4984: doc-+thru
                   4985: doc---block--->
                   4986: doc-block-included
                   4987: 
                   4988: @c -------------------------------------------------------------
                   4989: @node Other I/O, Programming Tools, Blocks, Words
                   4990: @section Other I/O
1.28    ! crook    4991: @cindex I/O - keyboard and display
1.26      crook    4992: 
                   4993: @menu
                   4994: * Simple numeric output::       Predefined formats
                   4995: * Formatted numeric output::    Formatted (pictured) output
                   4996: * String Formats::              How Forth stores strings in memory
                   4997: * Displaying characters and strings:: Other stuff
                   4998: * Input::                       Input
                   4999: @end menu
                   5000: 
                   5001: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
                   5002: @subsection Simple numeric output
1.28    ! crook    5003: @cindex numeric output - simple/free-format
1.5       anton    5004: 
1.26      crook    5005: The simplest output functions are those that display numbers from the
                   5006: data or floating-point stacks. Floating-point output is always displayed
                   5007: using base 10. Numbers displayed from the data stack use the value stored
                   5008: in @code{base}.
1.5       anton    5009: 
1.26      crook    5010: doc-.
                   5011: doc-dec.
                   5012: doc-hex.
                   5013: doc-u.
                   5014: doc-.r
                   5015: doc-u.r
                   5016: doc-d.
                   5017: doc-ud.
                   5018: doc-d.r
                   5019: doc-ud.r
                   5020: doc-f.
                   5021: doc-fe.
                   5022: doc-fs.
1.5       anton    5023: 
1.26      crook    5024: Examples of printing the number 1234.5678E23 in the different floating-point output
                   5025: formats are shown below:
1.5       anton    5026: 
                   5027: @example
1.26      crook    5028: f. 123456779999999000000000000.
                   5029: fe. 123.456779999999E24
                   5030: fs. 1.23456779999999E26
1.5       anton    5031: @end example
                   5032: 
                   5033: 
1.26      crook    5034: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
                   5035: @subsection Formatted numeric output
1.28    ! crook    5036: @cindex formatted numeric output
1.26      crook    5037: @cindex pictured numeric output
1.28    ! crook    5038: @cindex numeric output - formatted
1.26      crook    5039: 
                   5040: Forth traditionally uses a technique called @var{pictured numeric
                   5041: output} for formatted printing of integers.  In this technique, digits
                   5042: are extracted from the number (using the current output radix defined by
                   5043: @code{base}), converted to ASCII codes and appended to a string that is
                   5044: built in a scratch-pad area of memory (@pxref{core-idef,
                   5045: Implementation-defined options, Implementation-defined
                   5046: options}). Arbitrary characters can be appended to the string during the
                   5047: extraction process. The completed string is specified by an address
                   5048: and length and can be manipulated (@code{TYPE}ed, copied, modified)
                   5049: under program control.
1.5       anton    5050: 
1.26      crook    5051: All of the words described in the previous section for simple numeric
                   5052: output are implemented in Gforth using pictured numeric output.
1.5       anton    5053: 
1.26      crook    5054: Three important things to remember about Pictured Numeric Output:
1.5       anton    5055: 
1.26      crook    5056: @itemize @bullet
                   5057: @item
1.28    ! crook    5058: It always operates on double-precision numbers; to display a
        !          5059: single-precision number, convert it first (@pxref{Double precision} for
        !          5060: ways of doing this).
1.26      crook    5061: @item
1.28    ! crook    5062: It always treats the double-precision number as though it were
        !          5063: unsigned. The examples below show ways of printing signed numbers.
1.26      crook    5064: @item
                   5065: The string is built up from right to left; least significant digit first.
                   5066: @end itemize
1.5       anton    5067: 
1.26      crook    5068: doc-<#
                   5069: doc-#
                   5070: doc-#s
                   5071: doc-hold
                   5072: doc-sign
                   5073: doc-#>
1.5       anton    5074: 
1.26      crook    5075: doc-represent
1.5       anton    5076: 
1.26      crook    5077: Here are some examples of using pictured numeric output:
1.5       anton    5078: 
                   5079: @example
1.26      crook    5080: : my-u. ( u -- )
                   5081:   \ Simplest use of pns.. behaves like Standard u. 
                   5082:   0              \ convert to unsigned double
                   5083:   <#             \ start conversion
                   5084:   #s             \ convert all digits
                   5085:   #>             \ complete conversion
                   5086:   TYPE SPACE ;   \ display, with trailing space
1.5       anton    5087: 
1.26      crook    5088: : cents-only ( u -- )
                   5089:   0              \ convert to unsigned double
                   5090:   <#             \ start conversion
                   5091:   # #            \ convert two least-significant digits
                   5092:   #>             \ complete conversion, discard other digits
                   5093:   TYPE SPACE ;   \ display, with trailing space
1.5       anton    5094: 
1.26      crook    5095: : dollars-and-cents ( u -- )
                   5096:   0              \ convert to unsigned double
                   5097:   <#             \ start conversion
                   5098:   # #            \ convert two least-significant digits
                   5099:   [char] . hold  \ insert decimal point
                   5100:   #s             \ convert remaining digits
                   5101:   [char] $ hold  \ append currency symbol
                   5102:   #>             \ complete conversion
                   5103:   TYPE SPACE ;   \ display, with trailing space
1.5       anton    5104: 
1.26      crook    5105: : my-. ( n -- )
                   5106:   \ handling negatives.. behaves like Standard .
                   5107:   s>d            \ convert to signed double
                   5108:   swap over dabs \ leave sign byte followed by unsigned double
                   5109:   <#             \ start conversion
                   5110:   #s             \ convert all digits
                   5111:   rot sign       \ get at sign byte, append "-" if needed
                   5112:   #>             \ complete conversion
                   5113:   TYPE SPACE ;   \ display, with trailing space
1.5       anton    5114: 
1.26      crook    5115: : account. ( n -- )
                   5116:   \ accountants don't like minus signs, they use braces
                   5117:   \ for negative numbers
                   5118:   s>d            \ convert to signed double
                   5119:   swap over dabs \ leave sign byte followed by unsigned double
                   5120:   <#             \ start conversion
                   5121:   2 pick         \ get copy of sign byte
                   5122:   0< IF [char] ) hold THEN \ right-most character of output
                   5123:   #s             \ convert all digits
                   5124:   rot            \ get at sign byte
                   5125:   0< IF [char] ( hold THEN
                   5126:   #>             \ complete conversion
                   5127:   TYPE SPACE ;   \ display, with trailing space
1.5       anton    5128: @end example
                   5129: 
1.26      crook    5130: Here are some examples of using these words:
1.5       anton    5131: 
                   5132: @example
1.26      crook    5133: 1 my-u. 1
                   5134: hex -1 my-u. decimal FFFFFFFF
                   5135: 1 cents-only 01
                   5136: 1234 cents-only 34
                   5137: 2 dollars-and-cents $0.02
                   5138: 1234 dollars-and-cents $12.34
                   5139: 123 my-. 123
                   5140: -123 my. -123
                   5141: 123 account. 123
                   5142: -456 account. (456)
1.5       anton    5143: @end example
                   5144: 
                   5145: 
1.26      crook    5146: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
                   5147: @subsection String Formats
1.27      crook    5148: @cindex strings - see character strings
                   5149: @cindex character strings - formats
1.28    ! crook    5150: @cindex I/O - see character strings
1.26      crook    5151: 
1.27      crook    5152: Forth commonly uses two different methods for representing character
                   5153: strings:
1.26      crook    5154: 
                   5155: @itemize @bullet
                   5156: @item
                   5157: @cindex address of counted string
                   5158: As a @var{counted string}, represented by a @var{c-addr}. The char
                   5159: addressed by @var{c-addr} contains a character-count, @var{n}, of the
                   5160: string and the string occupies the subsequent @var{n} char addresses in
                   5161: memory.
                   5162: @item
                   5163: As cell pair on the stack; @var{c-addr u}, where @var{u} is the length
                   5164: of the string in characters, and @var{c-addr} is the address of the
                   5165: first byte of the string.
                   5166: @end itemize
                   5167: 
                   5168: ANS Forth encourages the use of the second format when representing
                   5169: strings on the stack, whilst conceeding that the counted string format
                   5170: remains useful as a way of storing strings in memory.
                   5171: 
                   5172: doc-count
                   5173: 
                   5174: @xref{Memory Blocks} for words that move, copy and search
                   5175: for strings. @xref{Displaying characters and strings,} for words that
                   5176: display characters and strings.
                   5177: 
                   5178: 
                   5179: @node Displaying characters and strings, Input, String Formats, Other I/O
                   5180: @subsection Displaying characters and strings
1.27      crook    5181: @cindex characters - compiling and displaying
                   5182: @cindex character strings - compiling and displaying
1.26      crook    5183: 
                   5184: This section starts with a glossary of Forth words and ends with a set
                   5185: of examples.
                   5186: 
                   5187: doc-bl
                   5188: doc-space
                   5189: doc-spaces
                   5190: doc-emit
                   5191: doc-toupper
                   5192: doc-."
                   5193: doc-.(
                   5194: doc-type
                   5195: doc-cr
1.27      crook    5196: @cindex cursor control
1.26      crook    5197: doc-at-xy
                   5198: doc-page
                   5199: doc-s"
                   5200: doc-c"
                   5201: doc-char
                   5202: doc-[char]
                   5203: doc-sliteral
                   5204: 
                   5205: As an example, consider the following text, stored in a file @file{test.fs}:
1.5       anton    5206: 
                   5207: @example
1.26      crook    5208: .( text-1)
                   5209: : my-word
                   5210:   ." text-2" cr
                   5211:   .( text-3)
                   5212: ;
                   5213: 
                   5214: ." text-4"
                   5215: 
                   5216: : my-char
                   5217:   [char] ALPHABET emit
                   5218:   char emit
                   5219: ;
1.5       anton    5220: @end example
                   5221: 
1.26      crook    5222: When you load this code into Gforth, the following output is generated:
1.5       anton    5223: 
1.26      crook    5224: @example
                   5225: @kbd{include test.fs <return>} text-1text-3text-4 ok
                   5226: @end example
1.5       anton    5227: 
1.26      crook    5228: @itemize @bullet
                   5229: @item
                   5230: Messages @code{text-1} and @code{text-3} are displayed because @code{.(} 
                   5231: is an immediate word; it behaves in the same way whether it is used inside
                   5232: or outside a colon definition.
                   5233: @item
                   5234: Message @code{text-4} is displayed because of Gforth's added interpretation
                   5235: semantics for @code{."}.
                   5236: @item
                   5237: Message @code{text-2} is @var{not} displayed, because the text interpreter
                   5238: performs the compilation semantics for @code{."} within the definition of
                   5239: @code{my-word}.
                   5240: @end itemize
1.5       anton    5241: 
1.26      crook    5242: Here are some examples of executing @code{my-word} and @code{my-char}:
1.5       anton    5243: 
1.26      crook    5244: @example
                   5245: @kbd{my-word <return>} text-2
                   5246:  ok
                   5247: @kbd{my-char fred <return>} Af ok
                   5248: @kbd{my-char jim <return>} Aj ok
                   5249: @end example
1.5       anton    5250: 
                   5251: @itemize @bullet
                   5252: @item
1.26      crook    5253: Message @code{text-2} is displayed because of the run-time behaviour of
                   5254: @code{."}.
                   5255: @item
                   5256: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
                   5257: on the stack at run-time. @code{emit} always displays the character
                   5258: when @code{my-char} is executed.
                   5259: @item
                   5260: @code{char} parses a string at run-time and the second @code{emit} displays
                   5261: the first character of the string.
1.5       anton    5262: @item
1.26      crook    5263: If you type @code{see my-char} you can see that @code{[char]} discarded
                   5264: the text ``LPHABET'' and only compiled the display code for ``A'' into the
                   5265: definition of @code{my-char}.
1.5       anton    5266: @end itemize
                   5267: 
                   5268: 
                   5269: 
1.26      crook    5270: @node Input, , Displaying characters and strings, Other I/O
                   5271: @subsection Input
                   5272: @cindex input
1.28    ! crook    5273: @cindex I/O - see input
        !          5274: @cindex parsing a string
        !          5275: @comment TODO more index entries.. particularly wrt parsing
1.5       anton    5276: 
1.27      crook    5277: @xref{String Formats} for ways of storing character strings in memory.
1.5       anton    5278: 
1.27      crook    5279: @comment TODO examples for >number >float accept key key? pad parse word refill
                   5280: 
                   5281: doc-key
                   5282: doc-key?
1.26      crook    5283: doc->number
                   5284: doc->float
                   5285: doc-accept
1.27      crook    5286: doc-pad
                   5287: doc-parse
                   5288: doc-word
                   5289: doc-sword
                   5290: doc-refill
                   5291: @comment obsolescent words..
                   5292: doc-convert
1.26      crook    5293: doc-query
                   5294: doc-expect
1.27      crook    5295: doc-span
1.5       anton    5296: 
                   5297: 
                   5298: 
                   5299: @c -------------------------------------------------------------
1.26      crook    5300: @node Programming Tools, Assembler and Code Words, Other I/O, Words
                   5301: @section Programming Tools
                   5302: @cindex programming tools
1.12      anton    5303: 
                   5304: @menu
1.26      crook    5305: * Debugging::                   Simple and quick.
                   5306: * Assertions::                  Making your programs self-checking.
                   5307: * Singlestep Debugger::                Executing your program word by word.
1.5       anton    5308: @end menu
                   5309: 
1.26      crook    5310: @node Debugging, Assertions, Programming Tools, Programming Tools
                   5311: @subsection Debugging
                   5312: @cindex debugging
1.5       anton    5313: 
1.26      crook    5314: Languages with a slow edit/compile/link/test development loop tend to
                   5315: require sophisticated tracing/stepping debuggers to facilate
                   5316: productive debugging.
1.5       anton    5317: 
1.26      crook    5318: A much better (faster) way in fast-compiling languages is to add
                   5319: printing code at well-selected places, let the program run, look at
                   5320: the output, see where things went wrong, add more printing code, etc.,
                   5321: until the bug is found.
1.5       anton    5322: 
1.26      crook    5323: The simple debugging aids provided in @file{debugs.fs}
                   5324: are meant to support this style of debugging. In addition, there are
                   5325: words for non-destructively inspecting the stack and memory:
1.5       anton    5326: 
1.26      crook    5327: doc-.s
                   5328: doc-f.s
1.5       anton    5329: 
1.26      crook    5330: There is a word @code{.r} but it does @var{not} display the return
                   5331: stack! It is used for formatted numeric output.
1.5       anton    5332: 
1.26      crook    5333: doc-depth
                   5334: doc-fdepth
                   5335: doc-clearstack
                   5336: doc-?
                   5337: doc-dump
1.5       anton    5338: 
1.26      crook    5339: The word @code{~~} prints debugging information (by default the source
                   5340: location and the stack contents). It is easy to insert. If you use Emacs
                   5341: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
                   5342: query-replace them with nothing). The deferred words
                   5343: @code{printdebugdata} and @code{printdebugline} control the output of
                   5344: @code{~~}. The default source location output format works well with
                   5345: Emacs' compilation mode, so you can step through the program at the
                   5346: source level using @kbd{C-x `} (the advantage over a stepping debugger
                   5347: is that you can step in any direction and you know where the crash has
                   5348: happened or where the strange data has occurred).
1.5       anton    5349: 
1.26      crook    5350: The default actions of @code{~~} clobber the contents of the pictured
                   5351: numeric output string, so you should not use @code{~~}, e.g., between
                   5352: @code{<#} and @code{#>}.
1.5       anton    5353: 
1.26      crook    5354: doc-~~
                   5355: doc-printdebugdata
                   5356: doc-printdebugline
1.5       anton    5357: 
1.26      crook    5358: doc-see
                   5359: doc-marker
1.5       anton    5360: 
1.26      crook    5361: Here's an example of using @code{marker} at the start of a source file
                   5362: that you are debugging; it ensures that you only ever have one copy of
                   5363: the file's definitions compiled at any time:
1.5       anton    5364: 
1.26      crook    5365: @example
                   5366: [IFDEF] my-code
                   5367:     my-code
                   5368: [ENDIF]
1.5       anton    5369: 
1.26      crook    5370: marker my-code
1.28    ! crook    5371: init-included-files
1.5       anton    5372: 
1.26      crook    5373: \ .. definitions start here
                   5374: \ .
                   5375: \ .
                   5376: \ end
                   5377: @end example
1.5       anton    5378: 
                   5379: 
                   5380: 
1.26      crook    5381: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
                   5382: @subsection Assertions
                   5383: @cindex assertions
1.5       anton    5384: 
1.26      crook    5385: It is a good idea to make your programs self-checking, especially if you
                   5386: make an assumption that may become invalid during maintenance (for
                   5387: example, that a certain field of a data structure is never zero). Gforth
                   5388: supports @var{assertions} for this purpose. They are used like this:
1.23      crook    5389: 
1.26      crook    5390: @example
                   5391: assert( @var{flag} )
                   5392: @end example
1.23      crook    5393: 
1.26      crook    5394: The code between @code{assert(} and @code{)} should compute a flag, that
                   5395: should be true if everything is alright and false otherwise. It should
                   5396: not change anything else on the stack. The overall stack effect of the
                   5397: assertion is @code{( -- )}. E.g.
1.23      crook    5398: 
1.26      crook    5399: @example
                   5400: assert( 1 1 + 2 = ) \ what we learn in school
                   5401: assert( dup 0<> ) \ assert that the top of stack is not zero
                   5402: assert( false ) \ this code should not be reached
                   5403: @end example
1.23      crook    5404: 
1.26      crook    5405: The need for assertions is different at different times. During
                   5406: debugging, we want more checking, in production we sometimes care more
                   5407: for speed. Therefore, assertions can be turned off, i.e., the assertion
                   5408: becomes a comment. Depending on the importance of an assertion and the
                   5409: time it takes to check it, you may want to turn off some assertions and
                   5410: keep others turned on. Gforth provides several levels of assertions for
                   5411: this purpose:
1.23      crook    5412: 
1.26      crook    5413: doc-assert0(
                   5414: doc-assert1(
                   5415: doc-assert2(
                   5416: doc-assert3(
                   5417: doc-assert(
                   5418: doc-)
1.23      crook    5419: 
1.26      crook    5420: The variable @code{assert-level} specifies the highest assertions that
                   5421: are turned on. I.e., at the default @code{assert-level} of one,
                   5422: @code{assert0(} and @code{assert1(} assertions perform checking, while
                   5423: @code{assert2(} and @code{assert3(} assertions are treated as comments.
                   5424: 
                   5425: The value of @code{assert-level} is evaluated at compile-time, not at
                   5426: run-time. Therefore you cannot turn assertions on or off at run-time;
                   5427: you have to set the @code{assert-level} appropriately before compiling a
                   5428: piece of code. You can compile different pieces of code at different
                   5429: @code{assert-level}s (e.g., a trusted library at level 1 and
                   5430: newly-written code at level 3).
1.23      crook    5431: 
1.26      crook    5432: doc-assert-level
1.23      crook    5433: 
1.26      crook    5434: If an assertion fails, a message compatible with Emacs' compilation mode
                   5435: is produced and the execution is aborted (currently with @code{ABORT"}.
                   5436: If there is interest, we will introduce a special throw code. But if you
                   5437: intend to @code{catch} a specific condition, using @code{throw} is
                   5438: probably more appropriate than an assertion).
1.23      crook    5439: 
1.26      crook    5440: Definitions in ANS Forth for these assertion words are provided
                   5441: in @file{compat/assert.fs}.
1.23      crook    5442: 
                   5443: 
1.26      crook    5444: @node Singlestep Debugger, , Assertions, Programming Tools
                   5445: @subsection Singlestep Debugger
                   5446: @cindex singlestep Debugger
                   5447: @cindex debugging Singlestep
                   5448: @cindex @code{dbg}
                   5449: @cindex @code{BREAK:}
                   5450: @cindex @code{BREAK"}
1.23      crook    5451: 
1.26      crook    5452: When you create a new word there's often the need to check whether it
                   5453: behaves correctly or not. You can do this by typing @code{dbg
                   5454: badword}. A debug session might look like this:
1.23      crook    5455: 
1.26      crook    5456: @example
                   5457: : badword 0 DO i . LOOP ;  ok
                   5458: 2 dbg badword 
                   5459: : badword  
                   5460: Scanning code...
1.23      crook    5461: 
1.26      crook    5462: Nesting debugger ready!
1.23      crook    5463: 
1.26      crook    5464: 400D4738  8049BC4 0              -> [ 2 ] 00002 00000 
                   5465: 400D4740  8049F68 DO             -> [ 0 ] 
                   5466: 400D4744  804A0C8 i              -> [ 1 ] 00000 
                   5467: 400D4748 400C5E60 .              -> 0 [ 0 ] 
                   5468: 400D474C  8049D0C LOOP           -> [ 0 ] 
                   5469: 400D4744  804A0C8 i              -> [ 1 ] 00001 
                   5470: 400D4748 400C5E60 .              -> 1 [ 0 ] 
                   5471: 400D474C  8049D0C LOOP           -> [ 0 ] 
                   5472: 400D4758  804B384 ;              ->  ok
                   5473: @end example
1.23      crook    5474: 
1.26      crook    5475: Each line displayed is one step. You always have to hit return to
                   5476: execute the next word that is displayed. If you don't want to execute
                   5477: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
                   5478: an overview what keys are available:
1.23      crook    5479: 
1.26      crook    5480: @table @i
1.23      crook    5481: 
1.26      crook    5482: @item <return>
                   5483: Next; Execute the next word.
1.23      crook    5484: 
1.26      crook    5485: @item n
                   5486: Nest; Single step through next word.
1.5       anton    5487: 
1.26      crook    5488: @item u
                   5489: Unnest; Stop debugging and execute rest of word. If we got to this word
                   5490: with nest, continue debugging with the calling word.
1.5       anton    5491: 
1.26      crook    5492: @item d
                   5493: Done; Stop debugging and execute rest.
1.5       anton    5494: 
1.26      crook    5495: @item s
                   5496: Stop; Abort immediately.
1.5       anton    5497: 
1.26      crook    5498: @end table
1.5       anton    5499: 
1.26      crook    5500: Debugging large application with this mechanism is very difficult, because
                   5501: you have to nest very deeply into the program before the interesting part
                   5502: begins. This takes a lot of time. 
1.5       anton    5503: 
1.26      crook    5504: To do it more directly put a @code{BREAK:} command into your source code.
                   5505: When program execution reaches @code{BREAK:} the single step debugger is
                   5506: invoked and you have all the features described above.
1.23      crook    5507: 
1.26      crook    5508: If you have more than one part to debug it is useful to know where the
                   5509: program has stopped at the moment. You can do this by the 
                   5510: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
                   5511: string is typed out when the ``breakpoint'' is reached.
                   5512: 
                   5513: doc-dbg
                   5514: doc-BREAK:
                   5515: doc-BREAK"
                   5516: 
                   5517: 
                   5518: @c -------------------------------------------------------------
                   5519: @node Assembler and Code Words, Threading Words, Programming Tools, Words
                   5520: @section Assembler and Code Words
                   5521: @cindex assembler
                   5522: @cindex code words
1.5       anton    5523: 
1.26      crook    5524: Gforth provides some words for defining primitives (words written in
                   5525: machine code), and for defining the the machine-code equivalent of
                   5526: @code{DOES>}-based defining words. However, the machine-independent
                   5527: nature of Gforth poses a few problems: First of all, Gforth runs on
                   5528: several architectures, so it can provide no standard assembler. What's
                   5529: worse is that the register allocation not only depends on the processor,
                   5530: but also on the @code{gcc} version and options used.
1.5       anton    5531: 
1.26      crook    5532: The words that Gforth offers encapsulate some system dependences (e.g., the
                   5533: header structure), so a system-independent assembler may be used in
                   5534: Gforth. If you do not have an assembler, you can compile machine code
                   5535: directly with @code{,} and @code{c,}.
1.5       anton    5536: 
1.26      crook    5537: doc-assembler
                   5538: doc-code
                   5539: doc-end-code
                   5540: doc-;code
                   5541: doc-flush-icache
1.5       anton    5542: 
1.26      crook    5543: If @code{flush-icache} does not work correctly, @code{code} words
                   5544: etc. will not work (reliably), either.
1.5       anton    5545: 
1.26      crook    5546: @code{flush-icache} is always present. The other words are rarely used
                   5547: and reside in @code{code.fs}, which is usually not loaded. You can load
                   5548: it with @code{require code.fs}.
1.5       anton    5549: 
1.26      crook    5550: @cindex registers of the inner interpreter
                   5551: In the assembly code you will want to refer to the inner interpreter's
                   5552: registers (e.g., the data stack pointer) and you may want to use other
                   5553: registers for temporary storage. Unfortunately, the register allocation
                   5554: is installation-dependent.
1.5       anton    5555: 
1.26      crook    5556: The easiest solution is to use explicit register declarations
                   5557: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
                   5558: GNU C Manual}) for all of the inner interpreter's registers: You have to
                   5559: compile Gforth with @code{-DFORCE_REG} (configure option
                   5560: @code{--enable-force-reg}) and the appropriate declarations must be
                   5561: present in the @code{machine.h} file (see @code{mips.h} for an example;
                   5562: you can find a full list of all declarable register symbols with
                   5563: @code{grep register engine.c}). If you give explicit registers to all
                   5564: variables that are declared at the beginning of @code{engine()}, you
                   5565: should be able to use the other caller-saved registers for temporary
                   5566: storage. Alternatively, you can use the @code{gcc} option
                   5567: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
                   5568: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
                   5569: (however, this restriction on register allocation may slow Gforth
                   5570: significantly).
1.5       anton    5571: 
1.26      crook    5572: If this solution is not viable (e.g., because @code{gcc} does not allow
                   5573: you to explicitly declare all the registers you need), you have to find
                   5574: out by looking at the code where the inner interpreter's registers
                   5575: reside and which registers can be used for temporary storage. You can
                   5576: get an assembly listing of the engine's code with @code{make engine.s}.
1.5       anton    5577: 
1.26      crook    5578: In any case, it is good practice to abstract your assembly code from the
                   5579: actual register allocation. E.g., if the data stack pointer resides in
                   5580: register @code{$17}, create an alias for this register called @code{sp},
                   5581: and use that in your assembly code.
1.5       anton    5582: 
1.26      crook    5583: @cindex code words, portable
                   5584: Another option for implementing normal and defining words efficiently
                   5585: is to add the desired functionality to the source of Gforth. For normal
                   5586: words you just have to edit @file{primitives} (@pxref{Automatic
                   5587: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
                   5588: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
                   5589: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.5       anton    5590: 
                   5591: 
1.26      crook    5592: @c -------------------------------------------------------------
                   5593: @node Threading Words, Locals, Assembler and Code Words, Words
                   5594: @section Threading Words
                   5595: @cindex threading words
1.5       anton    5596: 
1.26      crook    5597: @cindex code address
                   5598: These words provide access to code addresses and other threading stuff
                   5599: in Gforth (and, possibly, other interpretive Forths). It more or less
                   5600: abstracts away the differences between direct and indirect threading
                   5601: (and, for direct threading, the machine dependences). However, at
                   5602: present this wordset is still incomplete. It is also pretty low-level;
                   5603: some day it will hopefully be made unnecessary by an internals wordset
                   5604: that abstracts implementation details away completely.
1.5       anton    5605: 
1.26      crook    5606: doc-threading-method
                   5607: doc->code-address
                   5608: doc->does-code
                   5609: doc-code-address!
                   5610: doc-does-code!
                   5611: doc-does-handler!
                   5612: doc-/does-handler
1.5       anton    5613: 
1.26      crook    5614: The code addresses produced by various defining words are produced by
                   5615: the following words:
1.5       anton    5616: 
1.26      crook    5617: doc-docol:
                   5618: doc-docon:
                   5619: doc-dovar:
                   5620: doc-douser:
                   5621: doc-dodefer:
                   5622: doc-dofield:
1.5       anton    5623: 
1.26      crook    5624: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
                   5625: with @code{>does-code}. If the word was defined in that way, the value
                   5626: returned is non-zero and identifies the @code{DOES>} used by the
                   5627: defining word.
                   5628: @comment TODO should that be ``identifies the xt of the DOES> ??''
1.5       anton    5629: 
1.26      crook    5630: @c -------------------------------------------------------------
                   5631: @node Locals, Structures, Threading Words, Words
                   5632: @section Locals
                   5633: @cindex locals
1.5       anton    5634: 
1.26      crook    5635: Local variables can make Forth programming more enjoyable and Forth
                   5636: programs easier to read. Unfortunately, the locals of ANS Forth are
                   5637: laden with restrictions. Therefore, we provide not only the ANS Forth
                   5638: locals wordset, but also our own, more powerful locals wordset (we
                   5639: implemented the ANS Forth locals wordset through our locals wordset).
1.5       anton    5640: 
1.26      crook    5641: The ideas in this section have also been published in the paper
                   5642: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
                   5643: at EuroForth '94; it is available at
                   5644: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1.5       anton    5645: 
1.26      crook    5646: @menu
                   5647: * Gforth locals::               
                   5648: * ANS Forth locals::            
                   5649: @end menu
1.5       anton    5650: 
1.26      crook    5651: @node Gforth locals, ANS Forth locals, Locals, Locals
                   5652: @subsection Gforth locals
                   5653: @cindex Gforth locals
                   5654: @cindex locals, Gforth style
1.5       anton    5655: 
1.26      crook    5656: Locals can be defined with
1.5       anton    5657: 
                   5658: @example
1.26      crook    5659: @{ local1 local2 ... -- comment @}
                   5660: @end example
                   5661: or
                   5662: @example
                   5663: @{ local1 local2 ... @}
1.5       anton    5664: @end example
                   5665: 
1.26      crook    5666: E.g.,
1.5       anton    5667: @example
1.26      crook    5668: : max @{ n1 n2 -- n3 @}
                   5669:  n1 n2 > if
                   5670:    n1
                   5671:  else
                   5672:    n2
                   5673:  endif ;
1.5       anton    5674: @end example
                   5675: 
1.26      crook    5676: The similarity of locals definitions with stack comments is intended. A
                   5677: locals definition often replaces the stack comment of a word. The order
                   5678: of the locals corresponds to the order in a stack comment and everything
                   5679: after the @code{--} is really a comment.
1.5       anton    5680: 
1.26      crook    5681: This similarity has one disadvantage: It is too easy to confuse locals
                   5682: declarations with stack comments, causing bugs and making them hard to
                   5683: find. However, this problem can be avoided by appropriate coding
                   5684: conventions: Do not use both notations in the same program. If you do,
                   5685: they should be distinguished using additional means, e.g. by position.
                   5686: 
                   5687: @cindex types of locals
                   5688: @cindex locals types
                   5689: The name of the local may be preceded by a type specifier, e.g.,
                   5690: @code{F:} for a floating point value:
                   5691: 
                   5692: @example
                   5693: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
                   5694: \ complex multiplication
                   5695:  Ar Br f* Ai Bi f* f-
                   5696:  Ar Bi f* Ai Br f* f+ ;
                   5697: @end example
                   5698: 
                   5699: @cindex flavours of locals
                   5700: @cindex locals flavours
                   5701: @cindex value-flavoured locals
                   5702: @cindex variable-flavoured locals
                   5703: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
                   5704: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
                   5705: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
                   5706: with @code{W:}, @code{D:} etc.) produces its value and can be changed
                   5707: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
                   5708: produces its address (which becomes invalid when the variable's scope is
                   5709: left). E.g., the standard word @code{emit} can be defined in terms of
                   5710: @code{type} like this:
1.5       anton    5711: 
                   5712: @example
1.26      crook    5713: : emit @{ C^ char* -- @}
                   5714:     char* 1 type ;
1.5       anton    5715: @end example
                   5716: 
1.26      crook    5717: @cindex default type of locals
                   5718: @cindex locals, default type
                   5719: A local without type specifier is a @code{W:} local. Both flavours of
                   5720: locals are initialized with values from the data or FP stack.
1.5       anton    5721: 
1.26      crook    5722: Currently there is no way to define locals with user-defined data
                   5723: structures, but we are working on it.
1.5       anton    5724: 
1.26      crook    5725: Gforth allows defining locals everywhere in a colon definition. This
                   5726: poses the following questions:
1.5       anton    5727: 
1.26      crook    5728: @menu
                   5729: * Where are locals visible by name?::  
                   5730: * How long do locals live?::    
                   5731: * Programming Style::           
                   5732: * Implementation::              
                   5733: @end menu
1.5       anton    5734: 
1.26      crook    5735: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
                   5736: @subsubsection Where are locals visible by name?
                   5737: @cindex locals visibility
                   5738: @cindex visibility of locals
                   5739: @cindex scope of locals
1.5       anton    5740: 
1.26      crook    5741: Basically, the answer is that locals are visible where you would expect
                   5742: it in block-structured languages, and sometimes a little longer. If you
                   5743: want to restrict the scope of a local, enclose its definition in
                   5744: @code{SCOPE}...@code{ENDSCOPE}.
1.5       anton    5745: 
1.26      crook    5746: doc-scope
                   5747: doc-endscope
1.5       anton    5748: 
1.26      crook    5749: These words behave like control structure words, so you can use them
                   5750: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
                   5751: arbitrary ways.
1.5       anton    5752: 
1.26      crook    5753: If you want a more exact answer to the visibility question, here's the
                   5754: basic principle: A local is visible in all places that can only be
                   5755: reached through the definition of the local@footnote{In compiler
                   5756: construction terminology, all places dominated by the definition of the
                   5757: local.}. In other words, it is not visible in places that can be reached
                   5758: without going through the definition of the local. E.g., locals defined
                   5759: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
                   5760: defined in @code{BEGIN}...@code{UNTIL} are visible after the
                   5761: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1.5       anton    5762: 
1.26      crook    5763: The reasoning behind this solution is: We want to have the locals
                   5764: visible as long as it is meaningful. The user can always make the
                   5765: visibility shorter by using explicit scoping. In a place that can
                   5766: only be reached through the definition of a local, the meaning of a
                   5767: local name is clear. In other places it is not: How is the local
                   5768: initialized at the control flow path that does not contain the
                   5769: definition? Which local is meant, if the same name is defined twice in
                   5770: two independent control flow paths?
1.5       anton    5771: 
1.26      crook    5772: This should be enough detail for nearly all users, so you can skip the
                   5773: rest of this section. If you really must know all the gory details and
                   5774: options, read on.
1.5       anton    5775: 
1.26      crook    5776: In order to implement this rule, the compiler has to know which places
                   5777: are unreachable. It knows this automatically after @code{AHEAD},
                   5778: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
                   5779: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
                   5780: compiler that the control flow never reaches that place. If
                   5781: @code{UNREACHABLE} is not used where it could, the only consequence is
                   5782: that the visibility of some locals is more limited than the rule above
                   5783: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
                   5784: lie to the compiler), buggy code will be produced.
1.5       anton    5785: 
1.26      crook    5786: doc-unreachable
1.5       anton    5787: 
1.26      crook    5788: Another problem with this rule is that at @code{BEGIN}, the compiler
                   5789: does not know which locals will be visible on the incoming
                   5790: back-edge. All problems discussed in the following are due to this
                   5791: ignorance of the compiler (we discuss the problems using @code{BEGIN}
                   5792: loops as examples; the discussion also applies to @code{?DO} and other
                   5793: loops). Perhaps the most insidious example is:
1.5       anton    5794: @example
1.26      crook    5795: AHEAD
                   5796: BEGIN
                   5797:   x
                   5798: [ 1 CS-ROLL ] THEN
                   5799:   @{ x @}
                   5800:   ...
                   5801: UNTIL
                   5802: @end example
1.5       anton    5803: 
1.26      crook    5804: This should be legal according to the visibility rule. The use of
                   5805: @code{x} can only be reached through the definition; but that appears
                   5806: textually below the use.
1.5       anton    5807: 
1.26      crook    5808: From this example it is clear that the visibility rules cannot be fully
                   5809: implemented without major headaches. Our implementation treats common
                   5810: cases as advertised and the exceptions are treated in a safe way: The
                   5811: compiler makes a reasonable guess about the locals visible after a
                   5812: @code{BEGIN}; if it is too pessimistic, the
                   5813: user will get a spurious error about the local not being defined; if the
                   5814: compiler is too optimistic, it will notice this later and issue a
                   5815: warning. In the case above the compiler would complain about @code{x}
                   5816: being undefined at its use. You can see from the obscure examples in
                   5817: this section that it takes quite unusual control structures to get the
                   5818: compiler into trouble, and even then it will often do fine.
1.5       anton    5819: 
1.26      crook    5820: If the @code{BEGIN} is reachable from above, the most optimistic guess
                   5821: is that all locals visible before the @code{BEGIN} will also be
                   5822: visible after the @code{BEGIN}. This guess is valid for all loops that
                   5823: are entered only through the @code{BEGIN}, in particular, for normal
                   5824: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
                   5825: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
                   5826: compiler. When the branch to the @code{BEGIN} is finally generated by
                   5827: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
                   5828: warns the user if it was too optimistic:
                   5829: @example
                   5830: IF
                   5831:   @{ x @}
                   5832: BEGIN
                   5833:   \ x ? 
                   5834: [ 1 cs-roll ] THEN
                   5835:   ...
                   5836: UNTIL
1.5       anton    5837: @end example
                   5838: 
1.26      crook    5839: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
                   5840: optimistically assumes that it lives until the @code{THEN}. It notices
                   5841: this difference when it compiles the @code{UNTIL} and issues a
                   5842: warning. The user can avoid the warning, and make sure that @code{x}
                   5843: is not used in the wrong area by using explicit scoping:
                   5844: @example
                   5845: IF
                   5846:   SCOPE
                   5847:   @{ x @}
                   5848:   ENDSCOPE
                   5849: BEGIN
                   5850: [ 1 cs-roll ] THEN
                   5851:   ...
                   5852: UNTIL
                   5853: @end example
1.5       anton    5854: 
1.26      crook    5855: Since the guess is optimistic, there will be no spurious error messages
                   5856: about undefined locals.
1.5       anton    5857: 
1.26      crook    5858: If the @code{BEGIN} is not reachable from above (e.g., after
                   5859: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
                   5860: optimistic guess, as the locals visible after the @code{BEGIN} may be
                   5861: defined later. Therefore, the compiler assumes that no locals are
                   5862: visible after the @code{BEGIN}. However, the user can use
                   5863: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
                   5864: visible at the BEGIN as at the point where the top control-flow stack
                   5865: item was created.
1.5       anton    5866: 
1.26      crook    5867: doc-assume-live
1.5       anton    5868: 
1.26      crook    5869: E.g.,
1.5       anton    5870: @example
1.26      crook    5871: @{ x @}
                   5872: AHEAD
                   5873: ASSUME-LIVE
                   5874: BEGIN
                   5875:   x
                   5876: [ 1 CS-ROLL ] THEN
                   5877:   ...
                   5878: UNTIL
1.5       anton    5879: @end example
                   5880: 
1.26      crook    5881: Other cases where the locals are defined before the @code{BEGIN} can be
                   5882: handled by inserting an appropriate @code{CS-ROLL} before the
                   5883: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
                   5884: behind the @code{ASSUME-LIVE}).
1.5       anton    5885: 
1.26      crook    5886: Cases where locals are defined after the @code{BEGIN} (but should be
                   5887: visible immediately after the @code{BEGIN}) can only be handled by
                   5888: rearranging the loop. E.g., the ``most insidious'' example above can be
                   5889: arranged into:
1.5       anton    5890: @example
1.26      crook    5891: BEGIN
                   5892:   @{ x @}
                   5893:   ... 0=
                   5894: WHILE
                   5895:   x
                   5896: REPEAT
1.5       anton    5897: @end example
                   5898: 
1.26      crook    5899: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
                   5900: @subsubsection How long do locals live?
                   5901: @cindex locals lifetime
                   5902: @cindex lifetime of locals
1.5       anton    5903: 
1.26      crook    5904: The right answer for the lifetime question would be: A local lives at
                   5905: least as long as it can be accessed. For a value-flavoured local this
                   5906: means: until the end of its visibility. However, a variable-flavoured
                   5907: local could be accessed through its address far beyond its visibility
                   5908: scope. Ultimately, this would mean that such locals would have to be
                   5909: garbage collected. Since this entails un-Forth-like implementation
                   5910: complexities, I adopted the same cowardly solution as some other
                   5911: languages (e.g., C): The local lives only as long as it is visible;
                   5912: afterwards its address is invalid (and programs that access it
                   5913: afterwards are erroneous).
1.5       anton    5914: 
1.26      crook    5915: @node Programming Style, Implementation, How long do locals live?, Gforth locals
                   5916: @subsubsection Programming Style
                   5917: @cindex locals programming style
                   5918: @cindex programming style, locals
1.5       anton    5919: 
1.26      crook    5920: The freedom to define locals anywhere has the potential to change
                   5921: programming styles dramatically. In particular, the need to use the
                   5922: return stack for intermediate storage vanishes. Moreover, all stack
                   5923: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
                   5924: determined arguments) can be eliminated: If the stack items are in the
                   5925: wrong order, just write a locals definition for all of them; then
                   5926: write the items in the order you want.
1.5       anton    5927: 
1.26      crook    5928: This seems a little far-fetched and eliminating stack manipulations is
                   5929: unlikely to become a conscious programming objective. Still, the number
                   5930: of stack manipulations will be reduced dramatically if local variables
                   5931: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
                   5932: a traditional implementation of @code{max}).
1.5       anton    5933: 
1.26      crook    5934: This shows one potential benefit of locals: making Forth programs more
                   5935: readable. Of course, this benefit will only be realized if the
                   5936: programmers continue to honour the principle of factoring instead of
                   5937: using the added latitude to make the words longer.
1.5       anton    5938: 
1.26      crook    5939: @cindex single-assignment style for locals
                   5940: Using @code{TO} can and should be avoided.  Without @code{TO},
                   5941: every value-flavoured local has only a single assignment and many
                   5942: advantages of functional languages apply to Forth. I.e., programs are
                   5943: easier to analyse, to optimize and to read: It is clear from the
                   5944: definition what the local stands for, it does not turn into something
                   5945: different later.
1.5       anton    5946: 
1.26      crook    5947: E.g., a definition using @code{TO} might look like this:
1.5       anton    5948: @example
1.26      crook    5949: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   5950:  u1 u2 min 0
                   5951:  ?do
                   5952:    addr1 c@@ addr2 c@@ -
                   5953:    ?dup-if
                   5954:      unloop exit
                   5955:    then
                   5956:    addr1 char+ TO addr1
                   5957:    addr2 char+ TO addr2
                   5958:  loop
                   5959:  u1 u2 - ;
1.5       anton    5960: @end example
1.26      crook    5961: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
                   5962: every loop iteration. @code{strcmp} is a typical example of the
                   5963: readability problems of using @code{TO}. When you start reading
                   5964: @code{strcmp}, you think that @code{addr1} refers to the start of the
                   5965: string. Only near the end of the loop you realize that it is something
                   5966: else.
1.5       anton    5967: 
1.26      crook    5968: This can be avoided by defining two locals at the start of the loop that
                   5969: are initialized with the right value for the current iteration.
1.5       anton    5970: @example
1.26      crook    5971: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   5972:  addr1 addr2
                   5973:  u1 u2 min 0 
                   5974:  ?do @{ s1 s2 @}
                   5975:    s1 c@@ s2 c@@ -
                   5976:    ?dup-if
                   5977:      unloop exit
                   5978:    then
                   5979:    s1 char+ s2 char+
                   5980:  loop
                   5981:  2drop
                   5982:  u1 u2 - ;
1.5       anton    5983: @end example
1.26      crook    5984: Here it is clear from the start that @code{s1} has a different value
                   5985: in every loop iteration.
1.5       anton    5986: 
1.26      crook    5987: @node Implementation,  , Programming Style, Gforth locals
                   5988: @subsubsection Implementation
                   5989: @cindex locals implementation
                   5990: @cindex implementation of locals
1.5       anton    5991: 
1.26      crook    5992: @cindex locals stack
                   5993: Gforth uses an extra locals stack. The most compelling reason for
                   5994: this is that the return stack is not float-aligned; using an extra stack
                   5995: also eliminates the problems and restrictions of using the return stack
                   5996: as locals stack. Like the other stacks, the locals stack grows toward
                   5997: lower addresses. A few primitives allow an efficient implementation:
1.5       anton    5998: 
1.26      crook    5999: doc-@local#
                   6000: doc-f@local#
                   6001: doc-laddr#
                   6002: doc-lp+!#
                   6003: doc-lp!
                   6004: doc->l
                   6005: doc-f>l
1.5       anton    6006: 
1.26      crook    6007: In addition to these primitives, some specializations of these
                   6008: primitives for commonly occurring inline arguments are provided for
                   6009: efficiency reasons, e.g., @code{@@local0} as specialization of
                   6010: @code{@@local#} for the inline argument 0. The following compiling words
                   6011: compile the right specialized version, or the general version, as
                   6012: appropriate:
1.6       pazsan   6013: 
1.26      crook    6014: doc-compile-@local
                   6015: doc-compile-f@local
                   6016: doc-compile-lp+!
1.12      anton    6017: 
1.26      crook    6018: Combinations of conditional branches and @code{lp+!#} like
                   6019: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
                   6020: is taken) are provided for efficiency and correctness in loops.
1.6       pazsan   6021: 
1.26      crook    6022: A special area in the dictionary space is reserved for keeping the
                   6023: local variable names. @code{@{} switches the dictionary pointer to this
                   6024: area and @code{@}} switches it back and generates the locals
                   6025: initializing code. @code{W:} etc.@ are normal defining words. This
                   6026: special area is cleared at the start of every colon definition.
1.6       pazsan   6027: 
1.26      crook    6028: @cindex word list for defining locals
                   6029: A special feature of Gforth's dictionary is used to implement the
                   6030: definition of locals without type specifiers: every word list (aka
                   6031: vocabulary) has its own methods for searching
                   6032: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
                   6033: with a special search method: When it is searched for a word, it
                   6034: actually creates that word using @code{W:}. @code{@{} changes the search
                   6035: order to first search the word list containing @code{@}}, @code{W:} etc.,
                   6036: and then the word list for defining locals without type specifiers.
1.12      anton    6037: 
1.26      crook    6038: The lifetime rules support a stack discipline within a colon
                   6039: definition: The lifetime of a local is either nested with other locals
                   6040: lifetimes or it does not overlap them.
1.6       pazsan   6041: 
1.26      crook    6042: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
                   6043: pointer manipulation is generated. Between control structure words
                   6044: locals definitions can push locals onto the locals stack. @code{AGAIN}
                   6045: is the simplest of the other three control flow words. It has to
                   6046: restore the locals stack depth of the corresponding @code{BEGIN}
                   6047: before branching. The code looks like this:
                   6048: @format
                   6049: @code{lp+!#} current-locals-size @minus{} dest-locals-size
                   6050: @code{branch} <begin>
                   6051: @end format
1.6       pazsan   6052: 
1.26      crook    6053: @code{UNTIL} is a little more complicated: If it branches back, it
                   6054: must adjust the stack just like @code{AGAIN}. But if it falls through,
                   6055: the locals stack must not be changed. The compiler generates the
                   6056: following code:
                   6057: @format
                   6058: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
                   6059: @end format
                   6060: The locals stack pointer is only adjusted if the branch is taken.
1.6       pazsan   6061: 
1.26      crook    6062: @code{THEN} can produce somewhat inefficient code:
                   6063: @format
                   6064: @code{lp+!#} current-locals-size @minus{} orig-locals-size
                   6065: <orig target>:
                   6066: @code{lp+!#} orig-locals-size @minus{} new-locals-size
                   6067: @end format
                   6068: The second @code{lp+!#} adjusts the locals stack pointer from the
                   6069: level at the @var{orig} point to the level after the @code{THEN}. The
                   6070: first @code{lp+!#} adjusts the locals stack pointer from the current
                   6071: level to the level at the orig point, so the complete effect is an
                   6072: adjustment from the current level to the right level after the
                   6073: @code{THEN}.
1.6       pazsan   6074: 
1.26      crook    6075: @cindex locals information on the control-flow stack
                   6076: @cindex control-flow stack items, locals information
                   6077: In a conventional Forth implementation a dest control-flow stack entry
                   6078: is just the target address and an orig entry is just the address to be
                   6079: patched. Our locals implementation adds a word list to every orig or dest
                   6080: item. It is the list of locals visible (or assumed visible) at the point
                   6081: described by the entry. Our implementation also adds a tag to identify
                   6082: the kind of entry, in particular to differentiate between live and dead
                   6083: (reachable and unreachable) orig entries.
1.6       pazsan   6084: 
1.26      crook    6085: A few unusual operations have to be performed on locals word lists:
1.6       pazsan   6086: 
1.26      crook    6087: doc-common-list
                   6088: doc-sub-list?
                   6089: doc-list-size
1.6       pazsan   6090: 
1.26      crook    6091: Several features of our locals word list implementation make these
                   6092: operations easy to implement: The locals word lists are organised as
                   6093: linked lists; the tails of these lists are shared, if the lists
                   6094: contain some of the same locals; and the address of a name is greater
                   6095: than the address of the names behind it in the list.
1.6       pazsan   6096: 
1.26      crook    6097: Another important implementation detail is the variable
                   6098: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
                   6099: determine if they can be reached directly or only through the branch
                   6100: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
                   6101: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
                   6102: definition, by @code{BEGIN} and usually by @code{THEN}.
1.6       pazsan   6103: 
1.26      crook    6104: Counted loops are similar to other loops in most respects, but
                   6105: @code{LEAVE} requires special attention: It performs basically the same
                   6106: service as @code{AHEAD}, but it does not create a control-flow stack
                   6107: entry. Therefore the information has to be stored elsewhere;
                   6108: traditionally, the information was stored in the target fields of the
                   6109: branches created by the @code{LEAVE}s, by organizing these fields into a
                   6110: linked list. Unfortunately, this clever trick does not provide enough
                   6111: space for storing our extended control flow information. Therefore, we
                   6112: introduce another stack, the leave stack. It contains the control-flow
                   6113: stack entries for all unresolved @code{LEAVE}s.
1.6       pazsan   6114: 
1.26      crook    6115: Local names are kept until the end of the colon definition, even if
                   6116: they are no longer visible in any control-flow path. In a few cases
                   6117: this may lead to increased space needs for the locals name area, but
                   6118: usually less than reclaiming this space would cost in code size.
1.6       pazsan   6119: 
                   6120: 
1.26      crook    6121: @node ANS Forth locals,  , Gforth locals, Locals
                   6122: @subsection ANS Forth locals
                   6123: @cindex locals, ANS Forth style
1.6       pazsan   6124: 
1.26      crook    6125: The ANS Forth locals wordset does not define a syntax for locals, but
                   6126: words that make it possible to define various syntaxes. One of the
                   6127: possible syntaxes is a subset of the syntax we used in the Gforth locals
                   6128: wordset, i.e.:
1.6       pazsan   6129: 
                   6130: @example
1.26      crook    6131: @{ local1 local2 ... -- comment @}
1.6       pazsan   6132: @end example
1.23      crook    6133: @noindent
1.26      crook    6134: or
1.6       pazsan   6135: @example
1.26      crook    6136: @{ local1 local2 ... @}
1.6       pazsan   6137: @end example
                   6138: 
1.26      crook    6139: The order of the locals corresponds to the order in a stack comment. The
                   6140: restrictions are:
1.6       pazsan   6141: 
                   6142: @itemize @bullet
                   6143: @item
1.26      crook    6144: Locals can only be cell-sized values (no type specifiers are allowed).
1.6       pazsan   6145: @item
1.26      crook    6146: Locals can be defined only outside control structures.
1.6       pazsan   6147: @item
1.26      crook    6148: Locals can interfere with explicit usage of the return stack. For the
                   6149: exact (and long) rules, see the standard. If you don't use return stack
                   6150: accessing words in a definition using locals, you will be all right. The
                   6151: purpose of this rule is to make locals implementation on the return
                   6152: stack easier.
1.6       pazsan   6153: @item
1.26      crook    6154: The whole definition must be in one line.
                   6155: @end itemize
1.6       pazsan   6156: 
1.26      crook    6157: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
                   6158: Defining Words}). I.e., they are initialized from the stack. Using their
                   6159: name produces their value. Their value can be changed using @code{TO}.
1.6       pazsan   6160: 
1.26      crook    6161: Since this syntax is supported by Gforth directly, you need not do
                   6162: anything to use it. If you want to port a program using this syntax to
                   6163: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
                   6164: syntax on the other system.
1.6       pazsan   6165: 
1.26      crook    6166: Note that a syntax shown in the standard, section A.13 looks
                   6167: similar, but is quite different in having the order of locals
                   6168: reversed. Beware!
1.6       pazsan   6169: 
1.26      crook    6170: The ANS Forth locals wordset itself consists of a word:
1.6       pazsan   6171: 
1.26      crook    6172: doc-(local)
1.6       pazsan   6173: 
1.26      crook    6174: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
                   6175: awful that we strongly recommend not to use it. We have implemented this
                   6176: syntax to make porting to Gforth easy, but do not document it here. The
                   6177: problem with this syntax is that the locals are defined in an order
                   6178: reversed with respect to the standard stack comment notation, making
                   6179: programs harder to read, and easier to misread and miswrite. The only
                   6180: merit of this syntax is that it is easy to implement using the ANS Forth
                   6181: locals wordset.
1.7       pazsan   6182: 
                   6183: 
1.26      crook    6184: @c ----------------------------------------------------------
                   6185: @node Structures, Object-oriented Forth, Locals, Words
                   6186: @section  Structures
                   6187: @cindex structures
                   6188: @cindex records
1.7       pazsan   6189: 
1.26      crook    6190: This section presents the structure package that comes with Gforth. A
                   6191: version of the package implemented in ANS Forth is available in
                   6192: @file{compat/struct.fs}. This package was inspired by a posting on
                   6193: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
                   6194: possibly John Hayes). A version of this section has been published in
                   6195: ???. Marcel Hendrix provided helpful comments.
1.7       pazsan   6196: 
1.26      crook    6197: @menu
                   6198: * Why explicit structure support?::  
                   6199: * Structure Usage::             
                   6200: * Structure Naming Convention::  
                   6201: * Structure Implementation::    
                   6202: * Structure Glossary::          
                   6203: @end menu
1.7       pazsan   6204: 
1.26      crook    6205: @node Why explicit structure support?, Structure Usage, Structures, Structures
                   6206: @subsection Why explicit structure support?
1.7       pazsan   6207: 
1.26      crook    6208: @cindex address arithmetic for structures
                   6209: @cindex structures using address arithmetic
                   6210: If we want to use a structure containing several fields, we could simply
                   6211: reserve memory for it, and access the fields using address arithmetic
1.27      crook    6212: (@pxref{Address Arithmetic}). As an example, consider a structure with
1.26      crook    6213: the following fields
1.7       pazsan   6214: 
1.26      crook    6215: @table @code
                   6216: @item a
                   6217: is a float
                   6218: @item b
                   6219: is a cell
                   6220: @item c
                   6221: is a float
                   6222: @end table
1.7       pazsan   6223: 
1.26      crook    6224: Given the (float-aligned) base address of the structure we get the
                   6225: address of the field
1.13      pazsan   6226: 
1.26      crook    6227: @table @code
                   6228: @item a
                   6229: without doing anything further.
                   6230: @item b
                   6231: with @code{float+}
                   6232: @item c
                   6233: with @code{float+ cell+ faligned}
                   6234: @end table
1.13      pazsan   6235: 
1.26      crook    6236: It is easy to see that this can become quite tiring. 
1.13      pazsan   6237: 
1.26      crook    6238: Moreover, it is not very readable, because seeing a
                   6239: @code{cell+} tells us neither which kind of structure is
                   6240: accessed nor what field is accessed; we have to somehow infer the kind
                   6241: of structure, and then look up in the documentation, which field of
                   6242: that structure corresponds to that offset.
1.13      pazsan   6243: 
1.26      crook    6244: Finally, this kind of address arithmetic also causes maintenance
                   6245: troubles: If you add or delete a field somewhere in the middle of the
                   6246: structure, you have to find and change all computations for the fields
                   6247: afterwards.
1.13      pazsan   6248: 
1.26      crook    6249: So, instead of using @code{cell+} and friends directly, how
                   6250: about storing the offsets in constants:
1.13      pazsan   6251: 
                   6252: @example
1.26      crook    6253: 0 constant a-offset
                   6254: 0 float+ constant b-offset
                   6255: 0 float+ cell+ faligned c-offset
1.13      pazsan   6256: @end example
                   6257: 
1.26      crook    6258: Now we can get the address of field @code{x} with @code{x-offset
                   6259: +}. This is much better in all respects. Of course, you still
                   6260: have to change all later offset definitions if you add a field. You can
                   6261: fix this by declaring the offsets in the following way:
1.13      pazsan   6262: 
                   6263: @example
1.26      crook    6264: 0 constant a-offset
                   6265: a-offset float+ constant b-offset
                   6266: b-offset cell+ faligned constant c-offset
1.13      pazsan   6267: @end example
                   6268: 
1.26      crook    6269: Since we always use the offsets with @code{+}, we could use a defining
                   6270: word @code{cfield} that includes the @code{+} in the action of the
                   6271: defined word:
1.8       pazsan   6272: 
                   6273: @example
1.26      crook    6274: : cfield ( n "name" -- )
                   6275:     create ,
                   6276: does> ( name execution: addr1 -- addr2 )
                   6277:     @@ + ;
1.13      pazsan   6278: 
1.26      crook    6279: 0 cfield a
                   6280: 0 a float+ cfield b
                   6281: 0 b cell+ faligned cfield c
1.13      pazsan   6282: @end example
                   6283: 
1.26      crook    6284: Instead of @code{x-offset +}, we now simply write @code{x}.
                   6285: 
                   6286: The structure field words now can be used quite nicely. However,
                   6287: their definition is still a bit cumbersome: We have to repeat the
                   6288: name, the information about size and alignment is distributed before
                   6289: and after the field definitions etc.  The structure package presented
                   6290: here addresses these problems.
                   6291: 
                   6292: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
                   6293: @subsection Structure Usage
                   6294: @cindex structure usage
1.13      pazsan   6295: 
1.26      crook    6296: @cindex @code{field} usage
                   6297: @cindex @code{struct} usage
                   6298: @cindex @code{end-struct} usage
                   6299: You can define a structure for a (data-less) linked list with:
1.13      pazsan   6300: @example
1.26      crook    6301: struct
                   6302:     cell% field list-next
                   6303: end-struct list%
1.13      pazsan   6304: @end example
                   6305: 
1.26      crook    6306: With the address of the list node on the stack, you can compute the
                   6307: address of the field that contains the address of the next node with
                   6308: @code{list-next}. E.g., you can determine the length of a list
                   6309: with:
1.13      pazsan   6310: 
                   6311: @example
1.26      crook    6312: : list-length ( list -- n )
                   6313: \ "list" is a pointer to the first element of a linked list
                   6314: \ "n" is the length of the list
                   6315:     0 BEGIN ( list1 n1 )
                   6316:         over
                   6317:     WHILE ( list1 n1 )
                   6318:         1+ swap list-next @@ swap
                   6319:     REPEAT
                   6320:     nip ;
1.13      pazsan   6321: @end example
                   6322: 
1.26      crook    6323: You can reserve memory for a list node in the dictionary with
                   6324: @code{list% %allot}, which leaves the address of the list node on the
                   6325: stack. For the equivalent allocation on the heap you can use @code{list%
                   6326: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
                   6327: use @code{list% %allocate}). You can get the the size of a list
                   6328: node with @code{list% %size} and its alignment with @code{list%
                   6329: %alignment}.
1.13      pazsan   6330: 
1.26      crook    6331: Note that in ANS Forth the body of a @code{create}d word is
                   6332: @code{aligned} but not necessarily @code{faligned};
                   6333: therefore, if you do a:
1.13      pazsan   6334: @example
1.26      crook    6335: create @emph{name} foo% %allot
1.8       pazsan   6336: @end example
                   6337: 
1.26      crook    6338: @noindent
                   6339: then the memory alloted for @code{foo%} is
                   6340: guaranteed to start at the body of @code{@emph{name}} only if
                   6341: @code{foo%} contains only character, cell and double fields.
1.20      pazsan   6342: 
1.26      crook    6343: @cindex strcutures containing structures
                   6344: You can include a structure @code{foo%} as a field of
                   6345: another structure, like this:
1.20      pazsan   6346: @example
1.26      crook    6347: struct
                   6348: ...
                   6349:     foo% field ...
                   6350: ...
                   6351: end-struct ...
1.20      pazsan   6352: @end example
                   6353: 
1.26      crook    6354: @cindex structure extension
                   6355: @cindex extended records
                   6356: Instead of starting with an empty structure, you can extend an
                   6357: existing structure. E.g., a plain linked list without data, as defined
                   6358: above, is hardly useful; You can extend it to a linked list of integers,
                   6359: like this:@footnote{This feature is also known as @emph{extended
                   6360: records}. It is the main innovation in the Oberon language; in other
                   6361: words, adding this feature to Modula-2 led Wirth to create a new
                   6362: language, write a new compiler etc.  Adding this feature to Forth just
                   6363: required a few lines of code.}
1.20      pazsan   6364: 
                   6365: @example
1.26      crook    6366: list%
                   6367:     cell% field intlist-int
                   6368: end-struct intlist%
1.20      pazsan   6369: @end example
                   6370: 
1.26      crook    6371: @code{intlist%} is a structure with two fields:
                   6372: @code{list-next} and @code{intlist-int}.
1.20      pazsan   6373: 
1.26      crook    6374: @cindex structures containing arrays
                   6375: You can specify an array type containing @emph{n} elements of
                   6376: type @code{foo%} like this:
1.20      pazsan   6377: 
                   6378: @example
1.26      crook    6379: foo% @emph{n} *
1.20      pazsan   6380: @end example
                   6381: 
1.26      crook    6382: You can use this array type in any place where you can use a normal
                   6383: type, e.g., when defining a @code{field}, or with
                   6384: @code{%allot}.
1.20      pazsan   6385: 
1.26      crook    6386: @cindex first field optimization
                   6387: The first field is at the base address of a structure and the word
                   6388: for this field (e.g., @code{list-next}) actually does not change
                   6389: the address on the stack. You may be tempted to leave it away in the
                   6390: interest of run-time and space efficiency. This is not necessary,
                   6391: because the structure package optimizes this case and compiling such
                   6392: words does not generate any code. So, in the interest of readability
                   6393: and maintainability you should include the word for the field when
                   6394: accessing the field.
1.20      pazsan   6395: 
1.26      crook    6396: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
                   6397: @subsection Structure Naming Convention
                   6398: @cindex structure naming convention
1.20      pazsan   6399: 
1.26      crook    6400: The field names that come to (my) mind are often quite generic, and,
                   6401: if used, would cause frequent name clashes. E.g., many structures
                   6402: probably contain a @code{counter} field. The structure names
                   6403: that come to (my) mind are often also the logical choice for the names
                   6404: of words that create such a structure.
1.20      pazsan   6405: 
1.26      crook    6406: Therefore, I have adopted the following naming conventions: 
1.20      pazsan   6407: 
1.26      crook    6408: @itemize @bullet
                   6409: @cindex field naming convention
                   6410: @item
                   6411: The names of fields are of the form
                   6412: @code{@emph{struct}-@emph{field}}, where
                   6413: @code{@emph{struct}} is the basic name of the structure, and
                   6414: @code{@emph{field}} is the basic name of the field. You can
                   6415: think of field words as converting the (address of the)
                   6416: structure into the (address of the) field.
1.20      pazsan   6417: 
1.26      crook    6418: @cindex structure naming convention
                   6419: @item
                   6420: The names of structures are of the form
                   6421: @code{@emph{struct}%}, where
                   6422: @code{@emph{struct}} is the basic name of the structure.
                   6423: @end itemize
1.20      pazsan   6424: 
1.26      crook    6425: This naming convention does not work that well for fields of extended
                   6426: structures; e.g., the integer list structure has a field
                   6427: @code{intlist-int}, but has @code{list-next}, not
                   6428: @code{intlist-next}.
1.20      pazsan   6429: 
1.26      crook    6430: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
                   6431: @subsection Structure Implementation
                   6432: @cindex structure implementation
                   6433: @cindex implementation of structures
1.20      pazsan   6434: 
1.26      crook    6435: The central idea in the implementation is to pass the data about the
                   6436: structure being built on the stack, not in some global
                   6437: variable. Everything else falls into place naturally once this design
                   6438: decision is made.
1.20      pazsan   6439: 
1.26      crook    6440: The type description on the stack is of the form @emph{align
                   6441: size}. Keeping the size on the top-of-stack makes dealing with arrays
                   6442: very simple.
1.20      pazsan   6443: 
1.26      crook    6444: @code{field} is a defining word that uses @code{Create}
                   6445: and @code{DOES>}. The body of the field contains the offset
                   6446: of the field, and the normal @code{DOES>} action is simply:
1.20      pazsan   6447: 
                   6448: @example
1.26      crook    6449: @ +
1.20      pazsan   6450: @end example
                   6451: 
1.23      crook    6452: @noindent
1.26      crook    6453: i.e., add the offset to the address, giving the stack effect
                   6454: @var{addr1 -- addr2} for a field.
1.20      pazsan   6455: 
1.26      crook    6456: @cindex first field optimization, implementation
                   6457: This simple structure is slightly complicated by the optimization
                   6458: for fields with offset 0, which requires a different
                   6459: @code{DOES>}-part (because we cannot rely on there being
                   6460: something on the stack if such a field is invoked during
                   6461: compilation). Therefore, we put the different @code{DOES>}-parts
                   6462: in separate words, and decide which one to invoke based on the
                   6463: offset. For a zero offset, the field is basically a noop; it is
                   6464: immediate, and therefore no code is generated when it is compiled.
1.20      pazsan   6465: 
1.26      crook    6466: @node Structure Glossary,  , Structure Implementation, Structures
                   6467: @subsection Structure Glossary
                   6468: @cindex structure glossary
1.20      pazsan   6469: 
1.26      crook    6470: doc-%align
                   6471: doc-%alignment
                   6472: doc-%alloc
                   6473: doc-%allocate
                   6474: doc-%allot
                   6475: doc-cell%
                   6476: doc-char%
                   6477: doc-dfloat%
                   6478: doc-double%
                   6479: doc-end-struct
                   6480: doc-field
                   6481: doc-float%
                   6482: doc-naligned
                   6483: doc-sfloat%
                   6484: doc-%size
                   6485: doc-struct
1.23      crook    6486: 
1.26      crook    6487: @c -------------------------------------------------------------
                   6488: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
                   6489: @section Object-oriented Forth
1.20      pazsan   6490: 
1.26      crook    6491: Gforth comes with three packages for object-oriented programming:
                   6492: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
                   6493: is preloaded, so you have to @code{include} them before use. The most
                   6494: important differences between these packages (and others) are discussed
                   6495: in @ref{Comparison with other object models}. All packages are written
                   6496: in ANS Forth and can be used with any other ANS Forth.
1.20      pazsan   6497: 
1.26      crook    6498: @menu
                   6499: * Why object-oriented programming?::
                   6500: * Object-Oriented Terminology::
                   6501: * Objects::
                   6502: * OOF::
                   6503: * Mini-OOF::
                   6504: * Comparison with other object models::  
                   6505: @end menu
1.20      pazsan   6506: 
1.23      crook    6507: 
1.26      crook    6508: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
                   6509: @subsubsection Why object-oriented programming?
                   6510: @cindex object-oriented programming motivation
                   6511: @cindex motivation for object-oriented programming
1.23      crook    6512: 
1.26      crook    6513: Often we have to deal with several data structures (@emph{objects}),
                   6514: that have to be treated similarly in some respects, but differently in
                   6515: others. Graphical objects are the textbook example: circles, triangles,
                   6516: dinosaurs, icons, and others, and we may want to add more during program
                   6517: development. We want to apply some operations to any graphical object,
                   6518: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
                   6519: has to do something different for every kind of object.
                   6520: @comment TODO add some other operations eg perimeter, area
                   6521: @comment and tie in to concrete examples later..
1.23      crook    6522: 
1.26      crook    6523: We could implement @code{draw} as a big @code{CASE}
                   6524: control structure that executes the appropriate code depending on the
                   6525: kind of object to be drawn. This would be not be very elegant, and,
                   6526: moreover, we would have to change @code{draw} every time we add
                   6527: a new kind of graphical object (say, a spaceship).
1.23      crook    6528: 
1.26      crook    6529: What we would rather do is: When defining spaceships, we would tell
                   6530: the system: ``Here's how you @code{draw} a spaceship; you figure
                   6531: out the rest''.
1.23      crook    6532: 
1.26      crook    6533: This is the problem that all systems solve that (rightfully) call
                   6534: themselves object-oriented; the object-oriented packages presented here
                   6535: solve this problem (and not much else).
                   6536: @comment TODO ?list properties of oo systems.. oo vs o-based?
1.23      crook    6537: 
1.26      crook    6538: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
                   6539: @subsubsection Object-Oriented Terminology
                   6540: @cindex object-oriented terminology
                   6541: @cindex terminology for object-oriented programming
1.23      crook    6542: 
1.26      crook    6543: This section is mainly for reference, so you don't have to understand
                   6544: all of it right away.  The terminology is mainly Smalltalk-inspired.  In
                   6545: short:
1.23      crook    6546: 
1.26      crook    6547: @table @emph
                   6548: @cindex class
                   6549: @item class
                   6550: a data structure definition with some extras.
1.23      crook    6551: 
1.26      crook    6552: @cindex object
                   6553: @item object
                   6554: an instance of the data structure described by the class definition.
1.23      crook    6555: 
1.26      crook    6556: @cindex instance variables
                   6557: @item instance variables
                   6558: fields of the data structure.
1.23      crook    6559: 
1.26      crook    6560: @cindex selector
                   6561: @cindex method selector
                   6562: @cindex virtual function
                   6563: @item selector
                   6564: (or @emph{method selector}) a word (e.g.,
                   6565: @code{draw}) that performs an operation on a variety of data
                   6566: structures (classes). A selector describes @emph{what} operation to
                   6567: perform. In C++ terminology: a (pure) virtual function.
1.23      crook    6568: 
1.26      crook    6569: @cindex method
                   6570: @item method
                   6571: the concrete definition that performs the operation
                   6572: described by the selector for a specific class. A method specifies
                   6573: @emph{how} the operation is performed for a specific class.
1.23      crook    6574: 
1.26      crook    6575: @cindex selector invocation
                   6576: @cindex message send
                   6577: @cindex invoking a selector
                   6578: @item selector invocation
                   6579: a call of a selector. One argument of the call (the TOS (top-of-stack))
                   6580: is used for determining which method is used. In Smalltalk terminology:
                   6581: a message (consisting of the selector and the other arguments) is sent
                   6582: to the object.
1.1       anton    6583: 
1.26      crook    6584: @cindex receiving object
                   6585: @item receiving object
                   6586: the object used for determining the method executed by a selector
                   6587: invocation. In the @file{objects.fs} model, it is the object that is on
                   6588: the TOS when the selector is invoked. (@emph{Receiving} comes from
                   6589: the Smalltalk @emph{message} terminology.)
1.1       anton    6590: 
1.26      crook    6591: @cindex child class
                   6592: @cindex parent class
                   6593: @cindex inheritance
                   6594: @item child class
                   6595: a class that has (@emph{inherits}) all properties (instance variables,
                   6596: selectors, methods) from a @emph{parent class}. In Smalltalk
                   6597: terminology: The subclass inherits from the superclass. In C++
                   6598: terminology: The derived class inherits from the base class.
1.1       anton    6599: 
1.26      crook    6600: @end table
1.21      crook    6601: 
1.26      crook    6602: @c If you wonder about the message sending terminology, it comes from
                   6603: @c a time when each object had it's own task and objects communicated via
                   6604: @c message passing; eventually the Smalltalk developers realized that
                   6605: @c they can do most things through simple (indirect) calls. They kept the
                   6606: @c terminology.
1.1       anton    6607: 
                   6608: 
1.26      crook    6609: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
                   6610: @subsection The @file{objects.fs} model
                   6611: @cindex objects
                   6612: @cindex object-oriented programming
1.1       anton    6613: 
1.26      crook    6614: @cindex @file{objects.fs}
                   6615: @cindex @file{oof.fs}
1.1       anton    6616: 
1.26      crook    6617: This section describes the @file{objects.fs} package. This material also has been published in @cite{Yet Another Forth Objects Package} by Anton Ertl and appeared in Forth Dimensions 19(2), pages 37--43 (@url{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
                   6618: @c McKewan's and Zsoter's packages
1.1       anton    6619: 
1.26      crook    6620: This section assumes that you have read @ref{Structures}.
1.1       anton    6621: 
1.26      crook    6622: The techniques on which this model is based have been used to implement
                   6623: the parser generator, Gray, and have also been used in Gforth for
                   6624: implementing the various flavours of word lists (hashed or not,
                   6625: case-sensitive or not, special-purpose word lists for locals etc.).
1.1       anton    6626: 
                   6627: 
1.26      crook    6628: @menu
                   6629: * Properties of the Objects model::  
                   6630: * Basic Objects Usage::         
                   6631: * The Objects base class::            
                   6632: * Creating objects::            
                   6633: * Object-Oriented Programming Style::  
                   6634: * Class Binding::               
                   6635: * Method conveniences::         
                   6636: * Classes and Scoping::         
                   6637: * Object Interfaces::           
                   6638: * Objects Implementation::      
                   6639: * Objects Glossary::            
                   6640: @end menu
1.1       anton    6641: 
1.26      crook    6642: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
                   6643: and Bernd Paysan helped me with the related works section.
1.1       anton    6644: 
1.26      crook    6645: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
                   6646: @subsubsection Properties of the @file{objects.fs} model
                   6647: @cindex @file{objects.fs} properties
1.1       anton    6648: 
1.26      crook    6649: @itemize @bullet
                   6650: @item
                   6651: It is straightforward to pass objects on the stack. Passing
                   6652: selectors on the stack is a little less convenient, but possible.
1.1       anton    6653: 
1.26      crook    6654: @item
                   6655: Objects are just data structures in memory, and are referenced by their
                   6656: address. You can create words for objects with normal defining words
                   6657: like @code{constant}. Likewise, there is no difference between instance
                   6658: variables that contain objects and those that contain other data.
1.1       anton    6659: 
1.26      crook    6660: @item
                   6661: Late binding is efficient and easy to use.
1.21      crook    6662: 
1.26      crook    6663: @item
                   6664: It avoids parsing, and thus avoids problems with state-smartness
                   6665: and reduced extensibility; for convenience there are a few parsing
                   6666: words, but they have non-parsing counterparts. There are also a few
                   6667: defining words that parse. This is hard to avoid, because all standard
                   6668: defining words parse (except @code{:noname}); however, such
                   6669: words are not as bad as many other parsing words, because they are not
                   6670: state-smart.
1.21      crook    6671: 
1.26      crook    6672: @item
                   6673: It does not try to incorporate everything. It does a few things and does
                   6674: them well (IMO). In particular, this model was not designed to support
                   6675: information hiding (although it has features that may help); you can use
                   6676: a separate package for achieving this.
1.21      crook    6677: 
1.26      crook    6678: @item
                   6679: It is layered; you don't have to learn and use all features to use this
                   6680: model. Only a few features are necessary (@xref{Basic Objects Usage},
                   6681: @xref{The Objects base class}, @xref{Creating objects}.), the others
                   6682: are optional and independent of each other.
1.21      crook    6683: 
1.26      crook    6684: @item
                   6685: An implementation in ANS Forth is available.
1.21      crook    6686: 
1.26      crook    6687: @end itemize
1.21      crook    6688: 
                   6689: 
1.26      crook    6690: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
                   6691: @subsubsection Basic @file{objects.fs} Usage
                   6692: @cindex basic objects usage
                   6693: @cindex objects, basic usage
1.21      crook    6694: 
1.26      crook    6695: You can define a class for graphical objects like this:
1.21      crook    6696: 
1.26      crook    6697: @cindex @code{class} usage
                   6698: @cindex @code{end-class} usage
                   6699: @cindex @code{selector} usage
                   6700: @example
                   6701: object class \ "object" is the parent class
                   6702:   selector draw ( x y graphical -- )
                   6703: end-class graphical
                   6704: @end example
1.21      crook    6705: 
1.26      crook    6706: This code defines a class @code{graphical} with an
                   6707: operation @code{draw}.  We can perform the operation
                   6708: @code{draw} on any @code{graphical} object, e.g.:
1.21      crook    6709: 
1.26      crook    6710: @example
                   6711: 100 100 t-rex draw
                   6712: @end example
1.21      crook    6713: 
1.26      crook    6714: @noindent
                   6715: where @code{t-rex} is a word (say, a constant) that produces a
                   6716: graphical object.
1.21      crook    6717: 
1.26      crook    6718: @comment nac TODO add a 2nd operation eg perimeter.. and use for
                   6719: @comment a concrete example
1.21      crook    6720: 
1.26      crook    6721: @cindex abstract class
                   6722: How do we create a graphical object? With the present definitions,
                   6723: we cannot create a useful graphical object. The class
                   6724: @code{graphical} describes graphical objects in general, but not
                   6725: any concrete graphical object type (C++ users would call it an
                   6726: @emph{abstract class}); e.g., there is no method for the selector
                   6727: @code{draw} in the class @code{graphical}.
1.21      crook    6728: 
1.26      crook    6729: For concrete graphical objects, we define child classes of the
                   6730: class @code{graphical}, e.g.:
1.21      crook    6731: 
1.26      crook    6732: @cindex @code{overrides} usage
                   6733: @cindex @code{field} usage in class definition
                   6734: @example
                   6735: graphical class \ "graphical" is the parent class
                   6736:   cell% field circle-radius
1.21      crook    6737: 
1.26      crook    6738: :noname ( x y circle -- )
                   6739:   circle-radius @@ draw-circle ;
                   6740: overrides draw
1.21      crook    6741: 
1.26      crook    6742: :noname ( n-radius circle -- )
                   6743:   circle-radius ! ;
                   6744: overrides construct
1.21      crook    6745: 
1.26      crook    6746: end-class circle
1.21      crook    6747: @end example
                   6748: 
1.26      crook    6749: Here we define a class @code{circle} as a child of @code{graphical},
                   6750: with field @code{circle-radius} (which behaves just like a field
                   6751: (@pxref{Structures}); it defines (using @code{overrides}) new methods
                   6752: for the selectors @code{draw} and @code{construct} (@code{construct} is
                   6753: defined in @code{object}, the parent class of @code{graphical}).
1.21      crook    6754: 
1.26      crook    6755: Now we can create a circle on the heap (i.e.,
                   6756: @code{allocate}d memory) with:
1.21      crook    6757: 
1.26      crook    6758: @cindex @code{heap-new} usage
1.21      crook    6759: @example
1.26      crook    6760: 50 circle heap-new constant my-circle
                   6761: @end example
1.21      crook    6762: 
1.26      crook    6763: @noindent
                   6764: @code{heap-new} invokes @code{construct}, thus
                   6765: initializing the field @code{circle-radius} with 50. We can draw
                   6766: this new circle at (100,100) with:
1.21      crook    6767: 
1.26      crook    6768: @example
                   6769: 100 100 my-circle draw
1.21      crook    6770: @end example
                   6771: 
1.26      crook    6772: @cindex selector invocation, restrictions
                   6773: @cindex class definition, restrictions
                   6774: Note: You can only invoke a selector if the object on the TOS
                   6775: (the receiving object) belongs to the class where the selector was
                   6776: defined or one of its descendents; e.g., you can invoke
                   6777: @code{draw} only for objects belonging to @code{graphical}
                   6778: or its descendents (e.g., @code{circle}).  Immediately before
                   6779: @code{end-class}, the search order has to be the same as
                   6780: immediately after @code{class}.
1.21      crook    6781: 
1.26      crook    6782: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
                   6783: @subsubsection The @file{object.fs} base class
                   6784: @cindex @code{object} class
1.21      crook    6785: 
1.26      crook    6786: When you define a class, you have to specify a parent class.  So how do
                   6787: you start defining classes? There is one class available from the start:
                   6788: @code{object}. It is ancestor for all classes and so is the
                   6789: only class that has no parent. It has two selectors: @code{construct}
                   6790: and @code{print}.
1.21      crook    6791: 
1.26      crook    6792: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
                   6793: @subsubsection Creating objects
                   6794: @cindex creating objects
                   6795: @cindex object creation
                   6796: @cindex object allocation options
1.21      crook    6797: 
1.26      crook    6798: @cindex @code{heap-new} discussion
                   6799: @cindex @code{dict-new} discussion
                   6800: @cindex @code{construct} discussion
                   6801: You can create and initialize an object of a class on the heap with
                   6802: @code{heap-new} ( ... class -- object ) and in the dictionary
                   6803: (allocation with @code{allot}) with @code{dict-new} (
                   6804: ... class -- object ). Both words invoke @code{construct}, which
                   6805: consumes the stack items indicated by "..." above.
1.21      crook    6806: 
1.26      crook    6807: @cindex @code{init-object} discussion
                   6808: @cindex @code{class-inst-size} discussion
                   6809: If you want to allocate memory for an object yourself, you can get its
                   6810: alignment and size with @code{class-inst-size 2@@} ( class --
                   6811: align size ). Once you have memory for an object, you can initialize
                   6812: it with @code{init-object} ( ... class object -- );
                   6813: @code{construct} does only a part of the necessary work.
1.21      crook    6814: 
1.26      crook    6815: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
                   6816: @subsubsection Object-Oriented Programming Style
                   6817: @cindex object-oriented programming style
1.21      crook    6818: 
1.26      crook    6819: This section is not exhaustive.
1.1       anton    6820: 
1.26      crook    6821: @cindex stack effects of selectors
                   6822: @cindex selectors and stack effects
                   6823: In general, it is a good idea to ensure that all methods for the
                   6824: same selector have the same stack effect: when you invoke a selector,
                   6825: you often have no idea which method will be invoked, so, unless all
                   6826: methods have the same stack effect, you will not know the stack effect
                   6827: of the selector invocation.
1.21      crook    6828: 
1.26      crook    6829: One exception to this rule is methods for the selector
                   6830: @code{construct}. We know which method is invoked, because we
                   6831: specify the class to be constructed at the same place. Actually, I
                   6832: defined @code{construct} as a selector only to give the users a
                   6833: convenient way to specify initialization. The way it is used, a
                   6834: mechanism different from selector invocation would be more natural
                   6835: (but probably would take more code and more space to explain).
1.21      crook    6836: 
1.26      crook    6837: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
                   6838: @subsubsection Class Binding
                   6839: @cindex class binding
                   6840: @cindex early binding
1.21      crook    6841: 
1.26      crook    6842: @cindex late binding
                   6843: Normal selector invocations determine the method at run-time depending
                   6844: on the class of the receiving object. This run-time selection is called
                   6845: @var{late binding}.
1.21      crook    6846: 
1.26      crook    6847: Sometimes it's preferable to invoke a different method. For example,
                   6848: you might want to use the simple method for @code{print}ing
                   6849: @code{object}s instead of the possibly long-winded @code{print} method
                   6850: of the receiver class. You can achieve this by replacing the invocation
                   6851: of @code{print} with:
1.21      crook    6852: 
1.26      crook    6853: @cindex @code{[bind]} usage
                   6854: @example
                   6855: [bind] object print
1.21      crook    6856: @end example
                   6857: 
1.26      crook    6858: @noindent
                   6859: in compiled code or:
1.21      crook    6860: 
1.26      crook    6861: @cindex @code{bind} usage
1.21      crook    6862: @example
1.26      crook    6863: bind object print
1.21      crook    6864: @end example
                   6865: 
1.26      crook    6866: @cindex class binding, alternative to
                   6867: @noindent
                   6868: in interpreted code. Alternatively, you can define the method with a
                   6869: name (e.g., @code{print-object}), and then invoke it through the
                   6870: name. Class binding is just a (often more convenient) way to achieve
                   6871: the same effect; it avoids name clutter and allows you to invoke
                   6872: methods directly without naming them first.
                   6873: 
                   6874: @cindex superclass binding
                   6875: @cindex parent class binding
                   6876: A frequent use of class binding is this: When we define a method
                   6877: for a selector, we often want the method to do what the selector does
                   6878: in the parent class, and a little more. There is a special word for
                   6879: this purpose: @code{[parent]}; @code{[parent]
                   6880: @emph{selector}} is equivalent to @code{[bind] @emph{parent
                   6881: selector}}, where @code{@emph{parent}} is the parent
                   6882: class of the current class. E.g., a method definition might look like:
1.21      crook    6883: 
1.26      crook    6884: @cindex @code{[parent]} usage
1.21      crook    6885: @example
1.26      crook    6886: :noname
                   6887:   dup [parent] foo \ do parent's foo on the receiving object
                   6888:   ... \ do some more
                   6889: ; overrides foo
1.21      crook    6890: @end example
                   6891: 
1.26      crook    6892: @cindex class binding as optimization
                   6893: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
                   6894: March 1997), Andrew McKewan presents class binding as an optimization
                   6895: technique. I recommend not using it for this purpose unless you are in
                   6896: an emergency. Late binding is pretty fast with this model anyway, so the
                   6897: benefit of using class binding is small; the cost of using class binding
                   6898: where it is not appropriate is reduced maintainability.
1.21      crook    6899: 
1.26      crook    6900: While we are at programming style questions: You should bind
                   6901: selectors only to ancestor classes of the receiving object. E.g., say,
                   6902: you know that the receiving object is of class @code{foo} or its
                   6903: descendents; then you should bind only to @code{foo} and its
                   6904: ancestors.
1.21      crook    6905: 
1.26      crook    6906: @node Method conveniences, Classes and Scoping, Class Binding, Objects
                   6907: @subsubsection Method conveniences
                   6908: @cindex method conveniences
1.1       anton    6909: 
1.26      crook    6910: In a method you usually access the receiving object pretty often.  If
                   6911: you define the method as a plain colon definition (e.g., with
                   6912: @code{:noname}), you may have to do a lot of stack
                   6913: gymnastics. To avoid this, you can define the method with @code{m:
                   6914: ... ;m}. E.g., you could define the method for
                   6915: @code{draw}ing a @code{circle} with
1.20      pazsan   6916: 
1.26      crook    6917: @cindex @code{this} usage
                   6918: @cindex @code{m:} usage
                   6919: @cindex @code{;m} usage
                   6920: @example
                   6921: m: ( x y circle -- )
                   6922:   ( x y ) this circle-radius @@ draw-circle ;m
                   6923: @end example
1.20      pazsan   6924: 
1.26      crook    6925: @cindex @code{exit} in @code{m: ... ;m}
                   6926: @cindex @code{exitm} discussion
                   6927: @cindex @code{catch} in @code{m: ... ;m}
                   6928: When this method is executed, the receiver object is removed from the
                   6929: stack; you can access it with @code{this} (admittedly, in this
                   6930: example the use of @code{m: ... ;m} offers no advantage). Note
                   6931: that I specify the stack effect for the whole method (i.e. including
                   6932: the receiver object), not just for the code between @code{m:}
                   6933: and @code{;m}. You cannot use @code{exit} in
                   6934: @code{m:...;m}; instead, use
                   6935: @code{exitm}.@footnote{Moreover, for any word that calls
                   6936: @code{catch} and was defined before loading
                   6937: @code{objects.fs}, you have to redefine it like I redefined
                   6938: @code{catch}: @code{: catch this >r catch r> to-this ;}}
1.20      pazsan   6939: 
1.26      crook    6940: @cindex @code{inst-var} usage
                   6941: You will frequently use sequences of the form @code{this
                   6942: @emph{field}} (in the example above: @code{this
                   6943: circle-radius}). If you use the field only in this way, you can
                   6944: define it with @code{inst-var} and eliminate the
                   6945: @code{this} before the field name. E.g., the @code{circle}
                   6946: class above could also be defined with:
1.20      pazsan   6947: 
1.26      crook    6948: @example
                   6949: graphical class
                   6950:   cell% inst-var radius
1.20      pazsan   6951: 
1.26      crook    6952: m: ( x y circle -- )
                   6953:   radius @@ draw-circle ;m
                   6954: overrides draw
1.20      pazsan   6955: 
1.26      crook    6956: m: ( n-radius circle -- )
                   6957:   radius ! ;m
                   6958: overrides construct
1.12      anton    6959: 
1.26      crook    6960: end-class circle
                   6961: @end example
1.12      anton    6962: 
1.26      crook    6963: @code{radius} can only be used in @code{circle} and its
                   6964: descendent classes and inside @code{m:...;m}.
1.12      anton    6965: 
1.26      crook    6966: @cindex @code{inst-value} usage
                   6967: You can also define fields with @code{inst-value}, which is
                   6968: to @code{inst-var} what @code{value} is to
                   6969: @code{variable}.  You can change the value of such a field with
                   6970: @code{[to-inst]}.  E.g., we could also define the class
                   6971: @code{circle} like this:
1.12      anton    6972: 
1.26      crook    6973: @example
                   6974: graphical class
                   6975:   inst-value radius
1.12      anton    6976: 
1.26      crook    6977: m: ( x y circle -- )
                   6978:   radius draw-circle ;m
                   6979: overrides draw
1.12      anton    6980: 
1.26      crook    6981: m: ( n-radius circle -- )
                   6982:   [to-inst] radius ;m
                   6983: overrides construct
1.21      crook    6984: 
1.26      crook    6985: end-class circle
1.12      anton    6986: @end example
                   6987: 
                   6988: 
1.26      crook    6989: @node Classes and Scoping, Object Interfaces, Method conveniences, Objects
                   6990: @subsubsection Classes and Scoping
                   6991: @cindex classes and scoping
                   6992: @cindex scoping and classes
1.12      anton    6993: 
1.26      crook    6994: Inheritance is frequent, unlike structure extension. This exacerbates
                   6995: the problem with the field name convention (@pxref{Structure Naming
                   6996: Convention}): One always has to remember in which class the field was
                   6997: originally defined; changing a part of the class structure would require
                   6998: changes for renaming in otherwise unaffected code.
1.12      anton    6999: 
1.26      crook    7000: @cindex @code{inst-var} visibility
                   7001: @cindex @code{inst-value} visibility
                   7002: To solve this problem, I added a scoping mechanism (which was not in my
                   7003: original charter): A field defined with @code{inst-var} (or
                   7004: @code{inst-value}) is visible only in the class where it is defined and in
                   7005: the descendent classes of this class.  Using such fields only makes
                   7006: sense in @code{m:}-defined methods in these classes anyway.
1.12      anton    7007: 
1.26      crook    7008: This scoping mechanism allows us to use the unadorned field name,
                   7009: because name clashes with unrelated words become much less likely.
1.12      anton    7010: 
1.26      crook    7011: @cindex @code{protected} discussion
                   7012: @cindex @code{private} discussion
                   7013: Once we have this mechanism, we can also use it for controlling the
                   7014: visibility of other words: All words defined after
                   7015: @code{protected} are visible only in the current class and its
                   7016: descendents. @code{public} restores the compilation
                   7017: (i.e. @code{current}) word list that was in effect before. If you
                   7018: have several @code{protected}s without an intervening
                   7019: @code{public} or @code{set-current}, @code{public}
                   7020: will restore the compilation word list in effect before the first of
                   7021: these @code{protected}s.
1.12      anton    7022: 
1.26      crook    7023: @node Object Interfaces, Objects Implementation, Classes and Scoping, Objects
                   7024: @subsubsection Object Interfaces
                   7025: @cindex object interfaces
                   7026: @cindex interfaces for objects
1.12      anton    7027: 
1.26      crook    7028: In this model you can only call selectors defined in the class of the
                   7029: receiving objects or in one of its ancestors. If you call a selector
                   7030: with a receiving object that is not in one of these classes, the
                   7031: result is undefined; if you are lucky, the program crashes
                   7032: immediately.
1.12      anton    7033: 
1.26      crook    7034: @cindex selectors common to hardly-related classes
                   7035: Now consider the case when you want to have a selector (or several)
                   7036: available in two classes: You would have to add the selector to a
                   7037: common ancestor class, in the worst case to @code{object}. You
                   7038: may not want to do this, e.g., because someone else is responsible for
                   7039: this ancestor class.
1.12      anton    7040: 
1.26      crook    7041: The solution for this problem is interfaces. An interface is a
                   7042: collection of selectors. If a class implements an interface, the
                   7043: selectors become available to the class and its descendents. A class
                   7044: can implement an unlimited number of interfaces. For the problem
                   7045: discussed above, we would define an interface for the selector(s), and
                   7046: both classes would implement the interface.
1.12      anton    7047: 
1.26      crook    7048: As an example, consider an interface @code{storage} for
                   7049: writing objects to disk and getting them back, and a class
                   7050: @code{foo} that implements it. The code would look like this:
1.12      anton    7051: 
1.26      crook    7052: @cindex @code{interface} usage
                   7053: @cindex @code{end-interface} usage
                   7054: @cindex @code{implementation} usage
                   7055: @example
                   7056: interface
                   7057:   selector write ( file object -- )
                   7058:   selector read1 ( file object -- )
                   7059: end-interface storage
1.12      anton    7060: 
1.26      crook    7061: bar class
                   7062:   storage implementation
1.12      anton    7063: 
1.26      crook    7064: ... overrides write
                   7065: ... overrides read
                   7066: ...
                   7067: end-class foo
1.12      anton    7068: @end example
                   7069: 
1.26      crook    7070: @noindent
                   7071: (I would add a word @code{read} @var{( file -- object )} that uses
                   7072: @code{read1} internally, but that's beyond the point illustrated
                   7073: here.)
1.12      anton    7074: 
1.26      crook    7075: Note that you cannot use @code{protected} in an interface; and
                   7076: of course you cannot define fields.
1.12      anton    7077: 
1.26      crook    7078: In the Neon model, all selectors are available for all classes;
                   7079: therefore it does not need interfaces. The price you pay in this model
                   7080: is slower late binding, and therefore, added complexity to avoid late
                   7081: binding.
1.12      anton    7082: 
1.26      crook    7083: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
                   7084: @subsubsection @file{objects.fs} Implementation
                   7085: @cindex @file{objects.fs} implementation
1.12      anton    7086: 
1.26      crook    7087: @cindex @code{object-map} discussion
                   7088: An object is a piece of memory, like one of the data structures
                   7089: described with @code{struct...end-struct}. It has a field
                   7090: @code{object-map} that points to the method map for the object's
                   7091: class.
1.12      anton    7092: 
1.26      crook    7093: @cindex method map
                   7094: @cindex virtual function table
                   7095: The @emph{method map}@footnote{This is Self terminology; in C++
                   7096: terminology: virtual function table.} is an array that contains the
                   7097: execution tokens (@var{xt}s) of the methods for the object's class. Each
                   7098: selector contains an offset into a method map.
1.12      anton    7099: 
1.26      crook    7100: @cindex @code{selector} implementation, class
                   7101: @code{selector} is a defining word that uses
                   7102: @code{CREATE} and @code{DOES>}. The body of the
                   7103: selector contains the offset; the @code{does>} action for a
                   7104: class selector is, basically:
1.21      crook    7105: 
1.26      crook    7106: @example
                   7107: ( object addr ) @@ over object-map @@ + @@ execute
                   7108: @end example
1.12      anton    7109: 
1.26      crook    7110: Since @code{object-map} is the first field of the object, it
                   7111: does not generate any code. As you can see, calling a selector has a
                   7112: small, constant cost.
1.12      anton    7113: 
1.26      crook    7114: @cindex @code{current-interface} discussion
                   7115: @cindex class implementation and representation
                   7116: A class is basically a @code{struct} combined with a method
                   7117: map. During the class definition the alignment and size of the class
                   7118: are passed on the stack, just as with @code{struct}s, so
                   7119: @code{field} can also be used for defining class
                   7120: fields. However, passing more items on the stack would be
                   7121: inconvenient, so @code{class} builds a data structure in memory,
                   7122: which is accessed through the variable
                   7123: @code{current-interface}. After its definition is complete, the
                   7124: class is represented on the stack by a pointer (e.g., as parameter for
                   7125: a child class definition).
1.1       anton    7126: 
1.26      crook    7127: A new class starts off with the alignment and size of its parent,
                   7128: and a copy of the parent's method map. Defining new fields extends the
                   7129: size and alignment; likewise, defining new selectors extends the
                   7130: method map. @code{overrides} just stores a new @var{xt} in the method
                   7131: map at the offset given by the selector.
1.20      pazsan   7132: 
1.26      crook    7133: @cindex class binding, implementation
                   7134: Class binding just gets the @var{xt} at the offset given by the selector
                   7135: from the class's method map and @code{compile,}s (in the case of
                   7136: @code{[bind]}) it.
1.21      crook    7137: 
1.26      crook    7138: @cindex @code{this} implementation
                   7139: @cindex @code{catch} and @code{this}
                   7140: @cindex @code{this} and @code{catch}
                   7141: I implemented @code{this} as a @code{value}. At the
                   7142: start of an @code{m:...;m} method the old @code{this} is
                   7143: stored to the return stack and restored at the end; and the object on
                   7144: the TOS is stored @code{TO this}. This technique has one
                   7145: disadvantage: If the user does not leave the method via
                   7146: @code{;m}, but via @code{throw} or @code{exit},
                   7147: @code{this} is not restored (and @code{exit} may
                   7148: crash). To deal with the @code{throw} problem, I have redefined
                   7149: @code{catch} to save and restore @code{this}; the same
                   7150: should be done with any word that can catch an exception. As for
                   7151: @code{exit}, I simply forbid it (as a replacement, there is
                   7152: @code{exitm}).
1.21      crook    7153: 
1.26      crook    7154: @cindex @code{inst-var} implementation
                   7155: @code{inst-var} is just the same as @code{field}, with
                   7156: a different @code{DOES>} action:
                   7157: @example
                   7158: @@ this +
                   7159: @end example
                   7160: Similar for @code{inst-value}.
1.21      crook    7161: 
1.26      crook    7162: @cindex class scoping implementation
                   7163: Each class also has a word list that contains the words defined with
                   7164: @code{inst-var} and @code{inst-value}, and its protected
                   7165: words. It also has a pointer to its parent. @code{class} pushes
                   7166: the word lists of the class and all its ancestors onto the search order stack,
                   7167: and @code{end-class} drops them.
1.21      crook    7168: 
1.26      crook    7169: @cindex interface implementation
                   7170: An interface is like a class without fields, parent and protected
                   7171: words; i.e., it just has a method map. If a class implements an
                   7172: interface, its method map contains a pointer to the method map of the
                   7173: interface. The positive offsets in the map are reserved for class
                   7174: methods, therefore interface map pointers have negative
                   7175: offsets. Interfaces have offsets that are unique throughout the
                   7176: system, unlike class selectors, whose offsets are only unique for the
                   7177: classes where the selector is available (invokable).
1.21      crook    7178: 
1.26      crook    7179: This structure means that interface selectors have to perform one
                   7180: indirection more than class selectors to find their method. Their body
                   7181: contains the interface map pointer offset in the class method map, and
                   7182: the method offset in the interface method map. The
                   7183: @code{does>} action for an interface selector is, basically:
1.21      crook    7184: 
                   7185: @example
1.26      crook    7186: ( object selector-body )
                   7187: 2dup selector-interface @@ ( object selector-body object interface-offset )
                   7188: swap object-map @@ + @@ ( object selector-body map )
                   7189: swap selector-offset @@ + @@ execute
1.21      crook    7190: @end example
                   7191: 
1.26      crook    7192: where @code{object-map} and @code{selector-offset} are
                   7193: first fields and generate no code.
                   7194: 
                   7195: As a concrete example, consider the following code:
1.21      crook    7196: 
1.26      crook    7197: @example
                   7198: interface
                   7199:   selector if1sel1
                   7200:   selector if1sel2
                   7201: end-interface if1
1.21      crook    7202: 
1.26      crook    7203: object class
                   7204:   if1 implementation
                   7205:   selector cl1sel1
                   7206:   cell% inst-var cl1iv1
1.21      crook    7207: 
1.26      crook    7208: ' m1 overrides construct
                   7209: ' m2 overrides if1sel1
                   7210: ' m3 overrides if1sel2
                   7211: ' m4 overrides cl1sel2
                   7212: end-class cl1
1.21      crook    7213: 
1.26      crook    7214: create obj1 object dict-new drop
                   7215: create obj2 cl1    dict-new drop
                   7216: @end example
1.21      crook    7217: 
1.26      crook    7218: The data structure created by this code (including the data structure
                   7219: for @code{object}) is shown in the <a
                   7220: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
                   7221: @comment nac TODO add this diagram..
1.21      crook    7222: 
1.26      crook    7223: @node Objects Glossary,  , Objects Implementation, Objects
                   7224: @subsubsection @file{objects.fs} Glossary
                   7225: @cindex @file{objects.fs} Glossary
1.21      crook    7226: 
1.26      crook    7227: doc---objects-bind
                   7228: doc---objects-<bind>
                   7229: doc---objects-bind'
                   7230: doc---objects-[bind]
                   7231: doc---objects-class
                   7232: doc---objects-class->map
                   7233: doc---objects-class-inst-size
                   7234: doc---objects-class-override!
                   7235: doc---objects-construct
                   7236: doc---objects-current'
                   7237: doc---objects-[current]
                   7238: doc---objects-current-interface
                   7239: doc---objects-dict-new
                   7240: doc---objects-drop-order
                   7241: doc---objects-end-class
                   7242: doc---objects-end-class-noname
                   7243: doc---objects-end-interface
                   7244: doc---objects-end-interface-noname
                   7245: doc---objects-exitm
                   7246: doc---objects-heap-new
                   7247: doc---objects-implementation
                   7248: doc---objects-init-object
                   7249: doc---objects-inst-value
                   7250: doc---objects-inst-var
                   7251: doc---objects-interface
                   7252: doc---objects-;m
                   7253: doc---objects-m:
                   7254: doc---objects-method
                   7255: doc---objects-object
                   7256: doc---objects-overrides
                   7257: doc---objects-[parent]
                   7258: doc---objects-print
                   7259: doc---objects-protected
                   7260: doc---objects-public
                   7261: doc---objects-push-order
                   7262: doc---objects-selector
                   7263: doc---objects-this
                   7264: doc---objects-<to-inst>
                   7265: doc---objects-[to-inst]
                   7266: doc---objects-to-this
                   7267: doc---objects-xt-new
1.21      crook    7268: 
1.26      crook    7269: @c -------------------------------------------------------------
                   7270: @node OOF, Mini-OOF, Objects, Object-oriented Forth
                   7271: @subsection The @file{oof.fs} model
                   7272: @cindex oof
                   7273: @cindex object-oriented programming
1.21      crook    7274: 
1.26      crook    7275: @cindex @file{objects.fs}
                   7276: @cindex @file{oof.fs}
1.21      crook    7277: 
1.26      crook    7278: This section describes the @file{oof.fs} package.
1.21      crook    7279: 
1.26      crook    7280: The package described in this section has been used in bigFORTH since 1991, and
                   7281: used for two large applications: a chromatographic system used to
                   7282: create new medicaments, and a graphic user interface library (MINOS).
1.21      crook    7283: 
1.26      crook    7284: You can find a description (in German) of @file{oof.fs} in @cite{Object
                   7285: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
                   7286: 10(2), 1994.
1.21      crook    7287: 
1.26      crook    7288: @menu
                   7289: * Properties of the OOF model::
                   7290: * Basic OOF Usage::
                   7291: * The OOF base class::
                   7292: * Class Declaration::
                   7293: * Class Implementation::
                   7294: @end menu
1.21      crook    7295: 
1.26      crook    7296: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
                   7297: @subsubsection Properties of the @file{oof.fs} model
                   7298: @cindex @file{oof.fs} properties
1.21      crook    7299: 
1.26      crook    7300: @itemize @bullet
                   7301: @item
                   7302: This model combines object oriented programming with information
                   7303: hiding. It helps you writing large application, where scoping is
                   7304: necessary, because it provides class-oriented scoping.
1.21      crook    7305: 
1.26      crook    7306: @item
                   7307: Named objects, object pointers, and object arrays can be created,
                   7308: selector invocation uses the ``object selector'' syntax. Selector invocation
                   7309: to objects and/or selectors on the stack is a bit less convenient, but
                   7310: possible.
1.21      crook    7311: 
1.26      crook    7312: @item
                   7313: Selector invocation and instance variable usage of the active object is
                   7314: straightforward, since both make use of the active object.
1.21      crook    7315: 
1.26      crook    7316: @item
                   7317: Late binding is efficient and easy to use.
1.21      crook    7318: 
1.26      crook    7319: @item
                   7320: State-smart objects parse selectors. However, extensibility is provided
                   7321: using a (parsing) selector @code{postpone} and a selector @code{'}.
1.21      crook    7322: 
                   7323: @item
1.26      crook    7324: An implementation in ANS Forth is available.
                   7325: 
1.21      crook    7326: @end itemize
                   7327: 
                   7328: 
1.26      crook    7329: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
                   7330: @subsubsection Basic @file{oof.fs} Usage
                   7331: @cindex @file{oof.fs} usage
                   7332: 
                   7333: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
1.21      crook    7334: 
1.26      crook    7335: You can define a class for graphical objects like this:
1.21      crook    7336: 
1.26      crook    7337: @cindex @code{class} usage
                   7338: @cindex @code{class;} usage
                   7339: @cindex @code{method} usage
                   7340: @example
                   7341: object class graphical \ "object" is the parent class
                   7342:   method draw ( x y graphical -- )
                   7343: class;
                   7344: @end example
1.21      crook    7345: 
1.26      crook    7346: This code defines a class @code{graphical} with an
                   7347: operation @code{draw}.  We can perform the operation
                   7348: @code{draw} on any @code{graphical} object, e.g.:
1.21      crook    7349: 
1.26      crook    7350: @example
                   7351: 100 100 t-rex draw
                   7352: @end example
1.21      crook    7353: 
1.26      crook    7354: @noindent
                   7355: where @code{t-rex} is an object or object pointer, created with e.g.
                   7356: @code{graphical : t-rex}.
1.21      crook    7357: 
1.26      crook    7358: @cindex abstract class
                   7359: How do we create a graphical object? With the present definitions,
                   7360: we cannot create a useful graphical object. The class
                   7361: @code{graphical} describes graphical objects in general, but not
                   7362: any concrete graphical object type (C++ users would call it an
                   7363: @emph{abstract class}); e.g., there is no method for the selector
                   7364: @code{draw} in the class @code{graphical}.
1.21      crook    7365: 
1.26      crook    7366: For concrete graphical objects, we define child classes of the
                   7367: class @code{graphical}, e.g.:
1.21      crook    7368: 
                   7369: @example
1.26      crook    7370: graphical class circle \ "graphical" is the parent class
                   7371:   cell var circle-radius
                   7372: how:
                   7373:   : draw ( x y -- )
                   7374:     circle-radius @@ draw-circle ;
                   7375: 
                   7376:   : init ( n-radius -- (
                   7377:     circle-radius ! ;
                   7378: class;
                   7379: @end example
                   7380: 
                   7381: Here we define a class @code{circle} as a child of @code{graphical},
                   7382: with a field @code{circle-radius}; it defines new methods for the
                   7383: selectors @code{draw} and @code{init} (@code{init} is defined in
                   7384: @code{object}, the parent class of @code{graphical}).
1.21      crook    7385: 
1.26      crook    7386: Now we can create a circle in the dictionary with:
1.21      crook    7387: 
1.26      crook    7388: @example
                   7389: 50 circle : my-circle
1.21      crook    7390: @end example
                   7391: 
1.26      crook    7392: @noindent
                   7393: @code{:} invokes @code{init}, thus initializing the field
                   7394: @code{circle-radius} with 50. We can draw this new circle at (100,100)
                   7395: with:
1.21      crook    7396: 
                   7397: @example
1.26      crook    7398: 100 100 my-circle draw
1.21      crook    7399: @end example
                   7400: 
1.26      crook    7401: @cindex selector invocation, restrictions
                   7402: @cindex class definition, restrictions
                   7403: Note: You can only invoke a selector if the receiving object belongs to
                   7404: the class where the selector was defined or one of its descendents;
                   7405: e.g., you can invoke @code{draw} only for objects belonging to
                   7406: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
                   7407: mechanism will check if you try to invoke a selector that is not
                   7408: defined in this class hierarchy, so you'll get an error at compilation
                   7409: time.
                   7410: 
                   7411: 
                   7412: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
                   7413: @subsubsection The @file{oof.fs} base class
                   7414: @cindex @file{oof.fs} base class
                   7415: 
                   7416: When you define a class, you have to specify a parent class.  So how do
                   7417: you start defining classes? There is one class available from the start:
                   7418: @code{object}. You have to use it as ancestor for all classes. It is the
                   7419: only class that has no parent. Classes are also objects, except that
                   7420: they don't have instance variables; class manipulation such as
                   7421: inheritance or changing definitions of a class is handled through
                   7422: selectors of the class @code{object}.
                   7423: 
                   7424: @code{object} provides a number of selectors:
                   7425: 
1.21      crook    7426: @itemize @bullet
                   7427: @item
1.26      crook    7428: @code{class} for subclassing, @code{definitions} to add definitions
                   7429: later on, and @code{class?} to get type informations (is the class a
                   7430: subclass of the class passed on the stack?).
                   7431: doc---object-class
                   7432: doc---object-definitions
                   7433: doc---object-class?
                   7434: 
1.21      crook    7435: @item
1.26      crook    7436: @code{init} and @code{dispose} as constructor and destructor of the
                   7437: object. @code{init} is invocated after the object's memory is allocated,
                   7438: while @code{dispose} also handles deallocation. Thus if you redefine
                   7439: @code{dispose}, you have to call the parent's dispose with @code{super
                   7440: dispose}, too.
                   7441: doc---object-init
                   7442: doc---object-dispose
                   7443: 
1.21      crook    7444: @item
1.26      crook    7445: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
                   7446: @code{[]} to create named and unnamed objects and object arrays or
                   7447: object pointers.
                   7448: doc---object-new
                   7449: doc---object-new[]
                   7450: doc---object-:
                   7451: doc---object-ptr
                   7452: doc---object-asptr
                   7453: doc---object-[]
1.21      crook    7454: 
1.26      crook    7455: @item
                   7456: @code{::} and @code{super} for explicit scoping. You should use explicit
                   7457: scoping only for super classes or classes with the same set of instance
                   7458: variables. Explicitly-scoped selectors use early binding.
                   7459: doc---object-::
                   7460: doc---object-super
1.21      crook    7461: 
1.26      crook    7462: @item
                   7463: @code{self} to get the address of the object
                   7464: doc---object-self
1.21      crook    7465: 
                   7466: @item
1.26      crook    7467: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
                   7468: pointers and instance defers.
                   7469: doc---object-bind
                   7470: doc---object-bound
                   7471: doc---object-link
                   7472: doc---object-is
                   7473: 
1.21      crook    7474: @item
1.26      crook    7475: @code{'} to obtain selector tokens, @code{send} to invocate selectors
                   7476: form the stack, and @code{postpone} to generate selector invocation code.
                   7477: doc---object-'
                   7478: doc---object-postpone
                   7479: 
1.21      crook    7480: @item
1.26      crook    7481: @code{with} and @code{endwith} to select the active object from the
                   7482: stack, and enable its scope. Using @code{with} and @code{endwith}
                   7483: also allows you to create code using selector @code{postpone} without being
                   7484: trapped by the state-smart objects.
                   7485: doc---object-with
                   7486: doc---object-endwith
                   7487: 
1.21      crook    7488: @end itemize
                   7489: 
1.26      crook    7490: @node Class Declaration, Class Implementation, The OOF base class, OOF
                   7491: @subsubsection Class Declaration
                   7492: @cindex class declaration
                   7493: 
                   7494: @itemize @bullet
                   7495: @item
                   7496: Instance variables
                   7497: doc---oof-var
1.21      crook    7498: 
1.26      crook    7499: @item
                   7500: Object pointers
                   7501: doc---oof-ptr
                   7502: doc---oof-asptr
1.21      crook    7503: 
1.26      crook    7504: @item
                   7505: Instance defers
                   7506: doc---oof-defer
1.21      crook    7507: 
1.26      crook    7508: @item
                   7509: Method selectors
                   7510: doc---oof-early
                   7511: doc---oof-method
1.21      crook    7512: 
1.26      crook    7513: @item
                   7514: Class-wide variables
                   7515: doc---oof-static
1.21      crook    7516: 
1.26      crook    7517: @item
                   7518: End declaration
                   7519: doc---oof-how:
                   7520: doc---oof-class;
1.21      crook    7521: 
1.26      crook    7522: @end itemize
1.21      crook    7523: 
1.26      crook    7524: @c -------------------------------------------------------------
                   7525: @node Class Implementation,  , Class Declaration, OOF
                   7526: @subsubsection Class Implementation
                   7527: @cindex class implementation
1.21      crook    7528: 
1.26      crook    7529: @c -------------------------------------------------------------
                   7530: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
                   7531: @subsection The @file{mini-oof.fs} model
                   7532: @cindex mini-oof
1.1       anton    7533: 
1.26      crook    7534: Gforth's third object oriented Forth package is a 12-liner. It uses a
                   7535: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
                   7536: and reduces to the bare minimum of features. This is based on a posting
                   7537: of Bernd Paysan in comp.arch.
1.1       anton    7538: 
                   7539: @menu
1.26      crook    7540: * Basic Mini-OOF Usage::
                   7541: * Mini-OOF Example::
                   7542: * Mini-OOF Implementation::
1.1       anton    7543: @end menu
                   7544: 
1.26      crook    7545: @c -------------------------------------------------------------
                   7546: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
                   7547: @subsubsection Basic @file{mini-oof.fs} Usage
                   7548: @cindex mini-oof usage
1.1       anton    7549: 
1.28    ! crook    7550: There is a base class (@code{class}, which allocates one cell for the
        !          7551: object pointer) plus seven other words: to define a method, a variable,
        !          7552: a class; to end a class, to resolve binding, to allocate an object and
        !          7553: to compile a class method.
1.26      crook    7554: @comment TODO better description of the last one
1.1       anton    7555: 
1.26      crook    7556: doc-object
                   7557: doc-method
                   7558: doc-var
                   7559: doc-class
                   7560: doc-end-class
                   7561: doc-defines
                   7562: doc-new
                   7563: doc-::
1.1       anton    7564: 
1.21      crook    7565: 
1.26      crook    7566: @c -------------------------------------------------------------
                   7567: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
                   7568: @subsubsection Mini-OOF Example
                   7569: @cindex mini-oof example
1.21      crook    7570: 
1.26      crook    7571: A short example shows how to use this package. This example, in slightly
                   7572: extended form, is supplied as @file{moof-exm.fs}
                   7573: @comment nac TODO could flesh this out with some comments from the Forthwrite article
1.21      crook    7574: 
1.26      crook    7575: @example
                   7576: object class
                   7577:   method init
                   7578:   method draw
                   7579: end-class graphical
                   7580: @end example
1.21      crook    7581: 
1.26      crook    7582: This code defines a class @code{graphical} with an
                   7583: operation @code{draw}.  We can perform the operation
                   7584: @code{draw} on any @code{graphical} object, e.g.:
1.1       anton    7585: 
1.26      crook    7586: @example
                   7587: 100 100 t-rex draw
                   7588: @end example
1.1       anton    7589: 
1.26      crook    7590: where @code{t-rex} is an object or object pointer, created with e.g.
                   7591: @code{graphical new Constant t-rex}.
1.1       anton    7592: 
1.26      crook    7593: For concrete graphical objects, we define child classes of the
                   7594: class @code{graphical}, e.g.:
1.21      crook    7595: 
                   7596: @example
1.26      crook    7597: graphical class
                   7598:   cell var circle-radius
                   7599: end-class circle \ "graphical" is the parent class
1.21      crook    7600: 
1.26      crook    7601: :noname ( x y -- )
                   7602:   circle-radius @@ draw-circle ; circle defines draw
                   7603: :noname ( r -- )
                   7604:   circle-radius ! ; circle defines init
1.21      crook    7605: @end example
                   7606: 
1.26      crook    7607: There is no implicit init method, so we have to define one. The creation
                   7608: code of the object now has to call init explicitely.
1.21      crook    7609: 
1.26      crook    7610: @example
                   7611: circle new Constant my-circle
                   7612: 50 my-circle init
                   7613: @end example
1.21      crook    7614: 
1.26      crook    7615: It is also possible to add a function to create named objects with
                   7616: automatic call of @code{init}, given that all objects have @code{init}
                   7617: on the same place:
1.1       anton    7618: 
                   7619: @example
1.26      crook    7620: : new: ( .. o "name" -- )
                   7621:     new dup Constant init ;
                   7622: 80 circle new: large-circle
1.1       anton    7623: @end example
                   7624: 
1.26      crook    7625: We can draw this new circle at (100,100) with:
1.1       anton    7626: 
                   7627: @example
1.26      crook    7628: 100 100 my-circle draw
1.1       anton    7629: @end example
                   7630: 
1.26      crook    7631: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
                   7632: @subsubsection @file{mini-oof.fs} Implementation
1.1       anton    7633: 
1.26      crook    7634: Object-oriented systems with late binding typically use a
                   7635: ``vtable''-approach: the first variable in each object is a pointer to a
                   7636: table, which contains the methods as function pointers. The vtable
                   7637: may also contain other information.
1.1       anton    7638: 
1.26      crook    7639: So first, let's declare methods:
1.1       anton    7640: 
1.26      crook    7641: @example
                   7642: : method ( m v -- m' v ) Create  over , swap cell+ swap
                   7643:   DOES> ( ... o -- ... ) @ over @ + @ execute ;
                   7644: @end example
1.1       anton    7645: 
1.26      crook    7646: During method declaration, the number of methods and instance
                   7647: variables is on the stack (in address units). @code{method} creates
                   7648: one method and increments the method number. To execute a method, it
                   7649: takes the object, fetches the vtable pointer, adds the offset, and
                   7650: executes the @var{xt} stored there. Each method takes the object it is
                   7651: invoked from as top of stack parameter. The method itself should
                   7652: consume that object.
1.1       anton    7653: 
1.26      crook    7654: Now, we also have to declare instance variables
1.21      crook    7655: 
1.26      crook    7656: @example
                   7657: : var ( m v size -- m v' ) Create  over , +
                   7658:   DOES> ( o -- addr ) @ + ;
                   7659: @end example
1.21      crook    7660: 
1.26      crook    7661: As before, a word is created with the current offset. Instance
                   7662: variables can have different sizes (cells, floats, doubles, chars), so
                   7663: all we do is take the size and add it to the offset. If your machine
                   7664: has alignment restrictions, put the proper @code{aligned} or
                   7665: @code{faligned} before the variable, to adjust the variable
                   7666: offset. That's why it is on the top of stack.
1.2       jwilke   7667: 
1.26      crook    7668: We need a starting point (the base object) and some syntactic sugar:
1.21      crook    7669: 
1.26      crook    7670: @example
                   7671: Create object  1 cells , 2 cells ,
                   7672: : class ( class -- class methods vars ) dup 2@ ;
                   7673: @end example
1.21      crook    7674: 
1.26      crook    7675: For inheritance, the vtable of the parent object has to be
                   7676: copied when a new, derived class is declared. This gives all the
                   7677: methods of the parent class, which can be overridden, though.
1.21      crook    7678: 
1.2       jwilke   7679: @example
1.26      crook    7680: : end-class  ( class methods vars -- )
                   7681:   Create  here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
                   7682:   cell+ dup cell+ r> rot @ 2 cells /string move ;
                   7683: @end example
                   7684: 
                   7685: The first line creates the vtable, initialized with
                   7686: @code{noop}s. The second line is the inheritance mechanism, it
                   7687: copies the xts from the parent vtable.
1.2       jwilke   7688: 
1.26      crook    7689: We still have no way to define new methods, let's do that now:
1.2       jwilke   7690: 
1.26      crook    7691: @example
                   7692: : defines ( xt class -- ) ' >body @ + ! ;
1.2       jwilke   7693: @end example
                   7694: 
1.26      crook    7695: To allocate a new object, we need a word, too:
1.2       jwilke   7696: 
1.26      crook    7697: @example
                   7698: : new ( class -- o )  here over @ allot swap over ! ;
                   7699: @end example
1.2       jwilke   7700: 
1.26      crook    7701: Sometimes derived classes want to access the method of the
                   7702: parent object. There are two ways to achieve this with Mini-OOF:
                   7703: first, you could use named words, and second, you could look up the
                   7704: vtable of the parent object.
1.2       jwilke   7705: 
1.26      crook    7706: @example
                   7707: : :: ( class "name" -- ) ' >body @ + @ compile, ;
                   7708: @end example
1.2       jwilke   7709: 
                   7710: 
1.26      crook    7711: Nothing can be more confusing than a good example, so here is
                   7712: one. First let's declare a text object (called
                   7713: @code{button}), that stores text and position:
1.2       jwilke   7714: 
1.26      crook    7715: @example
                   7716: object class
                   7717:   cell var text
                   7718:   cell var len
                   7719:   cell var x
                   7720:   cell var y
                   7721:   method init
                   7722:   method draw
                   7723: end-class button
                   7724: @end example
1.2       jwilke   7725: 
1.26      crook    7726: @noindent
                   7727: Now, implement the two methods, @code{draw} and @code{init}:
1.2       jwilke   7728: 
1.26      crook    7729: @example
                   7730: :noname ( o -- )
                   7731:  >r r@ x @ r@ y @ at-xy  r@ text @ r> len @ type ;
                   7732:  button defines draw
                   7733: :noname ( addr u o -- )
                   7734:  >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
                   7735:  button defines init
                   7736: @end example
1.2       jwilke   7737: 
1.26      crook    7738: @noindent
                   7739: To demonstrate inheritance, we define a class @code{bold-button}, with no
                   7740: new data and no new methods:
1.2       jwilke   7741: 
1.26      crook    7742: @example
                   7743: button class
                   7744: end-class bold-button
1.1       anton    7745: 
1.26      crook    7746: : bold   27 emit ." [1m" ;
                   7747: : normal 27 emit ." [0m" ;
                   7748: @end example
1.1       anton    7749: 
1.26      crook    7750: @noindent
                   7751: The class @code{bold-button} has a different draw method to
                   7752: @code{button}, but the new method is defined in terms of the draw method
                   7753: for @code{button}:
1.1       anton    7754: 
1.26      crook    7755: @example
                   7756: :noname bold [ button :: draw ] normal ; bold-button defines draw
                   7757: @end example
1.1       anton    7758: 
1.26      crook    7759: @noindent
                   7760: Finally, create two objects and apply methods:
1.1       anton    7761: 
1.26      crook    7762: @example
                   7763: button new Constant foo
                   7764: s" thin foo" foo init
                   7765: page
                   7766: foo draw
                   7767: bold-button new Constant bar
                   7768: s" fat bar" bar init
                   7769: 1 bar y !
                   7770: bar draw
                   7771: @end example
1.1       anton    7772: 
                   7773: 
1.26      crook    7774: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
                   7775: @subsubsection Comparison with other object models
                   7776: @cindex comparison of object models
                   7777: @cindex object models, comparison
1.1       anton    7778: 
1.26      crook    7779: Many object-oriented Forth extensions have been proposed (@cite{A survey
                   7780: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
                   7781: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
                   7782: relation of the object models described here to two well-known and two
                   7783: closely-related (by the use of method maps) models.
1.1       anton    7784: 
1.26      crook    7785: @cindex Neon model
                   7786: The most popular model currently seems to be the Neon model (see
                   7787: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
                   7788: 1997) by Andrew McKewan) but this model has a number of limitations
                   7789: @footnote{A longer version of this critique can be
                   7790: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
                   7791: Dimensions, May 1997) by Anton Ertl.}:
1.1       anton    7792: 
1.26      crook    7793: @itemize @bullet
                   7794: @item
                   7795: It uses a @code{@emph{selector
                   7796: object}} syntax, which makes it unnatural to pass objects on the
                   7797: stack.
1.1       anton    7798: 
1.26      crook    7799: @item
                   7800: It requires that the selector parses the input stream (at
                   7801: compile time); this leads to reduced extensibility and to bugs that are+
                   7802: hard to find.
1.1       anton    7803: 
1.26      crook    7804: @item
                   7805: It allows using every selector to every object;
                   7806: this eliminates the need for classes, but makes it harder to create
                   7807: efficient implementations. 
                   7808: @end itemize
1.1       anton    7809: 
1.26      crook    7810: @cindex Pountain's object-oriented model
                   7811: Another well-known publication is @cite{Object-Oriented Forth} (Academic
                   7812: Press, London, 1987) by Dick Pountain. However, it is not really about
                   7813: object-oriented programming, because it hardly deals with late
                   7814: binding. Instead, it focuses on features like information hiding and
                   7815: overloading that are characteristic of modular languages like Ada (83).
1.1       anton    7816: 
1.26      crook    7817: @cindex Zsoter's object-oriented model
                   7818: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
                   7819: Andras Zsoter describes a model that makes heavy use of an active object
                   7820: (like @code{this} in @file{objects.fs}): The active object is not only
                   7821: used for accessing all fields, but also specifies the receiving object
                   7822: of every selector invocation; you have to change the active object
                   7823: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
                   7824: changes more or less implicitly at @code{m: ... ;m}. Such a change at
                   7825: the method entry point is unnecessary with the Zsoter's model, because
                   7826: the receiving object is the active object already. On the other hand, the explicit
                   7827: change is absolutely necessary in that model, because otherwise no one
                   7828: could ever change the active object. An ANS Forth implementation of this
                   7829: model is available at @url{http://www.forth.org/fig/oopf.html}.
1.1       anton    7830: 
1.26      crook    7831: @cindex @file{oof.fs}, differences to other models
                   7832: The @file{oof.fs} model combines information hiding and overloading
                   7833: resolution (by keeping names in various word lists) with object-oriented
                   7834: programming. It sets the active object implicitly on method entry, but
                   7835: also allows explicit changing (with @code{>o...o>} or with
                   7836: @code{with...endwith}). It uses parsing and state-smart objects and
                   7837: classes for resolving overloading and for early binding: the object or
                   7838: class parses the selector and determines the method from this. If the
                   7839: selector is not parsed by an object or class, it performs a call to the
                   7840: selector for the active object (late binding), like Zsoter's model.
                   7841: Fields are always accessed through the active object. The big
                   7842: disadvantage of this model is the parsing and the state-smartness, which
                   7843: reduces extensibility and increases the opportunities for subtle bugs;
                   7844: essentially, you are only safe if you never tick or @code{postpone} an
                   7845: object or class (Bernd disagrees, but I (Anton) am not convinced).
1.1       anton    7846: 
1.26      crook    7847: @cindex @file{mini-oof.fs}, differences to other models
                   7848: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
                   7849: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
                   7850: @file{oof.fs} models.
1.1       anton    7851: 
1.26      crook    7852: @c -------------------------------------------------------------
                   7853: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
1.21      crook    7854: @section Passing Commands to the Operating System
                   7855: @cindex operating system - passing commands
                   7856: @cindex shell commands
                   7857: 
                   7858: Gforth allows you to pass an arbitrary string to the host operating
                   7859: system shell (if such a thing exists) for execution.
                   7860: 
                   7861: doc-sh
                   7862: doc-system
                   7863: doc-$?
1.23      crook    7864: doc-getenv
1.21      crook    7865: 
1.26      crook    7866: @c -------------------------------------------------------------
1.21      crook    7867: @node Miscellaneous Words,  , Passing Commands to the OS, Words
                   7868: @section Miscellaneous Words
                   7869: @cindex miscellaneous words
                   7870: 
1.26      crook    7871: These section lists the ANS Forth words that are not documented
1.21      crook    7872: elsewhere in this manual. Ultimately, they all need proper homes.
                   7873: 
                   7874: doc-ms
                   7875: doc-time&date
1.27      crook    7876: 
1.21      crook    7877: doc-[compile]
                   7878: 
                   7879: 
1.26      crook    7880: The following ANS Forth words are not currently supported by Gforth 
1.27      crook    7881: (@pxref{ANS conformance}):
1.21      crook    7882: 
                   7883: @code{EDITOR} 
                   7884: @code{EKEY} 
                   7885: @code{EKEY>CHAR} 
                   7886: @code{EKEY?} 
                   7887: @code{EMIT?} 
                   7888: @code{FORGET} 
                   7889: 
1.24      anton    7890: @c ******************************************************************
                   7891: @node Error messages, Tools, Words, Top
                   7892: @chapter Error messages
                   7893: @cindex error messages
                   7894: @cindex backtrace
                   7895: 
                   7896: A typical Gforth error message looks like this:
                   7897: 
                   7898: @example
                   7899: in file included from :-1
                   7900: in file included from ./yyy.fs:1
                   7901: ./xxx.fs:4: Invalid memory address
                   7902: bar
                   7903: ^^^
1.25      anton    7904: $400E664C @@
                   7905: $400E6664 foo
1.24      anton    7906: @end example
                   7907: 
                   7908: The message identifying the error is @code{Invalid memory address}.  The
                   7909: error happened when text-interpreting line 4 of the file
                   7910: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
                   7911: word on the line where the error happened, is pointed out (with
                   7912: @code{^^^}).
                   7913: 
                   7914: The file containing the error was included in line 1 of @file{./yyy.fs},
                   7915: and @file{yyy.fs} was included from a non-file (in this case, by giving
                   7916: @file{yyy.fs} as command-line parameter to Gforth).
                   7917: 
                   7918: At the end of the error message you find a return stack dump that can be
                   7919: interpreted as a backtrace (possibly empty). On top you find the top of
                   7920: the return stack when the @code{throw} happened, and at the bottom you
                   7921: find the return stack entry just above the return stack of the topmost
                   7922: text interpreter.
                   7923: 
                   7924: To the right of most return stack entries you see a guess for the word
                   7925: that pushed that return stack entry as its return address. This gives a
                   7926: backtrace. In our case we see that @code{bar} called @code{foo}, and
                   7927: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
                   7928: address} exception).
                   7929: 
                   7930: Note that the backtrace is not perfect: We don't know which return stack
                   7931: entries are return addresses (so we may get false positives); and in
                   7932: some cases (e.g., for @code{abort"}) we cannot determine from the return
                   7933: address the word that pushed the return address, so for some return
                   7934: addresses you see no names in the return stack dump.
1.25      anton    7935: 
                   7936: @cindex @code{catch} and backtraces
                   7937: The return stack dump represents the return stack at the time when a
                   7938: specific @code{throw} was executed.  In programs that make use of
                   7939: @code{catch}, it is not necessarily clear which @code{throw} should be
                   7940: used for the return stack dump (e.g., consider one @code{throw} that
                   7941: indicates an error, which is caught, and during recovery another error
                   7942: happens; which @code{throw} should be used for the stack dump).  Gforth
                   7943: presents the return stack dump for the first @code{throw} after the last
                   7944: executed (not returned-to) @code{catch}; this works well in the usual
                   7945: case.
                   7946: 
                   7947: @cindex @code{gforth-fast} and backtraces
                   7948: @cindex @code{gforth-fast}, difference from @code{gforth}
                   7949: @cindex backtraces with @code{gforth-fast}
                   7950: @cindex return stack dump with @code{gforth-fast}
                   7951: @code{gforth} is able to do a return stack dump for throws generated
                   7952: from primitives (e.g., invalid memory address, stack empty etc.);
                   7953: @code{gforth-fast} is only able to do a return stack dump from a
                   7954: directly called @code{throw} (including @code{abort} etc.).  This is the
                   7955: only difference (apart from a speed difference of about 30%) between
                   7956: @code{gforth} and @code{gforth-fast}.  Given an exception caused by a
                   7957: primitive in @code{gforth-fast}, you will typically see no return stack
                   7958: dump at all; however, if the exception is caught by @code{catch} (e.g.,
                   7959: for restoring some state), and then @code{throw}n again, the return
                   7960: stack dump will be for the first such @code{throw}.
1.2       jwilke   7961: 
1.5       anton    7962: @c ******************************************************************
1.24      anton    7963: @node Tools, ANS conformance, Error messages, Top
1.1       anton    7964: @chapter Tools
                   7965: 
                   7966: @menu
                   7967: * ANS Report::                  Report the words used, sorted by wordset.
                   7968: @end menu
                   7969: 
                   7970: See also @ref{Emacs and Gforth}.
                   7971: 
                   7972: @node ANS Report,  , Tools, Tools
                   7973: @section @file{ans-report.fs}: Report the words used, sorted by wordset
                   7974: @cindex @file{ans-report.fs}
                   7975: @cindex report the words used in your program
                   7976: @cindex words used in your program
                   7977: 
                   7978: If you want to label a Forth program as ANS Forth Program, you must
                   7979: document which wordsets the program uses; for extension wordsets, it is
                   7980: helpful to list the words the program requires from these wordsets
                   7981: (because Forth systems are allowed to provide only some words of them).
                   7982: 
                   7983: The @file{ans-report.fs} tool makes it easy for you to determine which
                   7984: words from which wordset and which non-ANS words your application
                   7985: uses. You simply have to include @file{ans-report.fs} before loading the
                   7986: program you want to check. After loading your program, you can get the
                   7987: report with @code{print-ans-report}. A typical use is to run this as
                   7988: batch job like this:
                   7989: @example
                   7990: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
                   7991: @end example
                   7992: 
                   7993: The output looks like this (for @file{compat/control.fs}):
                   7994: @example
                   7995: The program uses the following words
                   7996: from CORE :
                   7997: : POSTPONE THEN ; immediate ?dup IF 0= 
                   7998: from BLOCK-EXT :
                   7999: \ 
                   8000: from FILE :
                   8001: ( 
                   8002: @end example
                   8003: 
                   8004: @subsection Caveats
                   8005: 
                   8006: Note that @file{ans-report.fs} just checks which words are used, not whether
                   8007: they are used in an ANS Forth conforming way!
                   8008: 
                   8009: Some words are defined in several wordsets in the
                   8010: standard. @file{ans-report.fs} reports them for only one of the
                   8011: wordsets, and not necessarily the one you expect. It depends on usage
                   8012: which wordset is the right one to specify. E.g., if you only use the
                   8013: compilation semantics of @code{S"}, it is a Core word; if you also use
                   8014: its interpretation semantics, it is a File word.
                   8015: 
                   8016: @c ******************************************************************
                   8017: @node ANS conformance, Model, Tools, Top
                   8018: @chapter ANS conformance
                   8019: @cindex ANS conformance of Gforth
                   8020: 
                   8021: To the best of our knowledge, Gforth is an
                   8022: 
                   8023: ANS Forth System
                   8024: @itemize @bullet
                   8025: @item providing the Core Extensions word set
                   8026: @item providing the Block word set
                   8027: @item providing the Block Extensions word set
                   8028: @item providing the Double-Number word set
                   8029: @item providing the Double-Number Extensions word set
                   8030: @item providing the Exception word set
                   8031: @item providing the Exception Extensions word set
                   8032: @item providing the Facility word set
                   8033: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
                   8034: @item providing the File Access word set
                   8035: @item providing the File Access Extensions word set
                   8036: @item providing the Floating-Point word set
                   8037: @item providing the Floating-Point Extensions word set
                   8038: @item providing the Locals word set
                   8039: @item providing the Locals Extensions word set
                   8040: @item providing the Memory-Allocation word set
                   8041: @item providing the Memory-Allocation Extensions word set (that one's easy)
                   8042: @item providing the Programming-Tools word set
                   8043: @item providing @code{;CODE}, @code{AHEAD}, @code{ASSEMBLER}, @code{BYE}, @code{CODE}, @code{CS-PICK}, @code{CS-ROLL}, @code{STATE}, @code{[ELSE]}, @code{[IF]}, @code{[THEN]} from the Programming-Tools Extensions word set
                   8044: @item providing the Search-Order word set
                   8045: @item providing the Search-Order Extensions word set
                   8046: @item providing the String word set
                   8047: @item providing the String Extensions word set (another easy one)
                   8048: @end itemize
                   8049: 
                   8050: @cindex system documentation
                   8051: In addition, ANS Forth systems are required to document certain
                   8052: implementation choices. This chapter tries to meet these
                   8053: requirements. In many cases it gives a way to ask the system for the
                   8054: information instead of providing the information directly, in
                   8055: particular, if the information depends on the processor, the operating
                   8056: system or the installation options chosen, or if they are likely to
                   8057: change during the maintenance of Gforth.
                   8058: 
                   8059: @comment The framework for the rest has been taken from pfe.
                   8060: 
                   8061: @menu
                   8062: * The Core Words::              
                   8063: * The optional Block word set::  
                   8064: * The optional Double Number word set::  
                   8065: * The optional Exception word set::  
                   8066: * The optional Facility word set::  
                   8067: * The optional File-Access word set::  
                   8068: * The optional Floating-Point word set::  
                   8069: * The optional Locals word set::  
                   8070: * The optional Memory-Allocation word set::  
                   8071: * The optional Programming-Tools word set::  
                   8072: * The optional Search-Order word set::  
                   8073: @end menu
                   8074: 
                   8075: 
                   8076: @c =====================================================================
                   8077: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
                   8078: @comment  node-name,  next,  previous,  up
                   8079: @section The Core Words
                   8080: @c =====================================================================
                   8081: @cindex core words, system documentation
                   8082: @cindex system documentation, core words
                   8083: 
                   8084: @menu
                   8085: * core-idef::                   Implementation Defined Options                   
                   8086: * core-ambcond::                Ambiguous Conditions                
                   8087: * core-other::                  Other System Documentation                  
                   8088: @end menu
                   8089: 
                   8090: @c ---------------------------------------------------------------------
                   8091: @node core-idef, core-ambcond, The Core Words, The Core Words
                   8092: @subsection Implementation Defined Options
                   8093: @c ---------------------------------------------------------------------
                   8094: @cindex core words, implementation-defined options
                   8095: @cindex implementation-defined options, core words
                   8096: 
                   8097: 
                   8098: @table @i
                   8099: @item (Cell) aligned addresses:
                   8100: @cindex cell-aligned addresses
                   8101: @cindex aligned addresses
                   8102: processor-dependent. Gforth's alignment words perform natural alignment
                   8103: (e.g., an address aligned for a datum of size 8 is divisible by
                   8104: 8). Unaligned accesses usually result in a @code{-23 THROW}.
                   8105: 
                   8106: @item @code{EMIT} and non-graphic characters:
                   8107: @cindex @code{EMIT} and non-graphic characters
                   8108: @cindex non-graphic characters and @code{EMIT}
                   8109: The character is output using the C library function (actually, macro)
                   8110: @code{putc}.
                   8111: 
                   8112: @item character editing of @code{ACCEPT} and @code{EXPECT}:
                   8113: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
                   8114: @cindex editing in @code{ACCEPT} and @code{EXPECT}
                   8115: @cindex @code{ACCEPT}, editing
                   8116: @cindex @code{EXPECT}, editing
                   8117: This is modeled on the GNU readline library (@pxref{Readline
                   8118: Interaction, , Command Line Editing, readline, The GNU Readline
                   8119: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
                   8120: producing a full word completion every time you type it (instead of
1.28    ! crook    8121: producing the common prefix of all completions). @xref{Command-line editing}.
1.1       anton    8122: 
                   8123: @item character set:
                   8124: @cindex character set
                   8125: The character set of your computer and display device. Gforth is
                   8126: 8-bit-clean (but some other component in your system may make trouble).
                   8127: 
                   8128: @item Character-aligned address requirements:
                   8129: @cindex character-aligned address requirements
                   8130: installation-dependent. Currently a character is represented by a C
                   8131: @code{unsigned char}; in the future we might switch to @code{wchar_t}
                   8132: (Comments on that requested).
                   8133: 
                   8134: @item character-set extensions and matching of names:
                   8135: @cindex character-set extensions and matching of names
1.26      crook    8136: @cindex case-sensitivity for name lookup
                   8137: @cindex name lookup, case-sensitivity
                   8138: @cindex locale and case-sensitivity
1.21      crook    8139: Any character except the ASCII NUL character can be used in a
1.1       anton    8140: name. Matching is case-insensitive (except in @code{TABLE}s). The
                   8141: matching is performed using the C function @code{strncasecmp}, whose
                   8142: function is probably influenced by the locale. E.g., the @code{C} locale
                   8143: does not know about accents and umlauts, so they are matched
                   8144: case-sensitively in that locale. For portability reasons it is best to
                   8145: write programs such that they work in the @code{C} locale. Then one can
                   8146: use libraries written by a Polish programmer (who might use words
                   8147: containing ISO Latin-2 encoded characters) and by a French programmer
                   8148: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
                   8149: funny results for some of the words (which ones, depends on the font you
                   8150: are using)). Also, the locale you prefer may not be available in other
                   8151: operating systems. Hopefully, Unicode will solve these problems one day.
                   8152: 
                   8153: @item conditions under which control characters match a space delimiter:
                   8154: @cindex space delimiters
                   8155: @cindex control characters as delimiters
                   8156: If @code{WORD} is called with the space character as a delimiter, all
                   8157: white-space characters (as identified by the C macro @code{isspace()})
                   8158: are delimiters. @code{PARSE}, on the other hand, treats space like other
                   8159: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
                   8160: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
                   8161: interpreter (aka text interpreter) by default, treats all white-space
                   8162: characters as delimiters.
                   8163: 
1.26      crook    8164: @item format of the control-flow stack:
                   8165: @cindex control-flow stack, format
                   8166: The data stack is used as control-flow stack. The size of a control-flow
1.1       anton    8167: stack item in cells is given by the constant @code{cs-item-size}. At the
                   8168: time of this writing, an item consists of a (pointer to a) locals list
                   8169: (third), an address in the code (second), and a tag for identifying the
                   8170: item (TOS). The following tags are used: @code{defstart},
                   8171: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
                   8172: @code{scopestart}.
                   8173: 
                   8174: @item conversion of digits > 35
                   8175: @cindex digits > 35
                   8176: The characters @code{[\]^_'} are the digits with the decimal value
                   8177: 36@minus{}41. There is no way to input many of the larger digits.
                   8178: 
                   8179: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
                   8180: @cindex @code{EXPECT}, display after end of input
                   8181: @cindex @code{ACCEPT}, display after end of input
                   8182: The cursor is moved to the end of the entered string. If the input is
                   8183: terminated using the @kbd{Return} key, a space is typed.
                   8184: 
                   8185: @item exception abort sequence of @code{ABORT"}:
                   8186: @cindex exception abort sequence of @code{ABORT"}
                   8187: @cindex @code{ABORT"}, exception abort sequence
                   8188: The error string is stored into the variable @code{"error} and a
                   8189: @code{-2 throw} is performed.
                   8190: 
                   8191: @item input line terminator:
                   8192: @cindex input line terminator
                   8193: @cindex line terminator on input
1.26      crook    8194: @cindex newline character on input
1.1       anton    8195: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
                   8196: lines. One of these characters is typically produced when you type the
                   8197: @kbd{Enter} or @kbd{Return} key.
                   8198: 
                   8199: @item maximum size of a counted string:
                   8200: @cindex maximum size of a counted string
                   8201: @cindex counted string, maximum size
                   8202: @code{s" /counted-string" environment? drop .}. Currently 255 characters
                   8203: on all ports, but this may change.
                   8204: 
                   8205: @item maximum size of a parsed string:
                   8206: @cindex maximum size of a parsed string
                   8207: @cindex parsed string, maximum size
                   8208: Given by the constant @code{/line}. Currently 255 characters.
                   8209: 
                   8210: @item maximum size of a definition name, in characters:
                   8211: @cindex maximum size of a definition name, in characters
                   8212: @cindex name, maximum length
                   8213: 31
                   8214: 
                   8215: @item maximum string length for @code{ENVIRONMENT?}, in characters:
                   8216: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
                   8217: @cindex @code{ENVIRONMENT?} string length, maximum
                   8218: 31
                   8219: 
                   8220: @item method of selecting the user input device:
                   8221: @cindex user input device, method of selecting
                   8222: The user input device is the standard input. There is currently no way to
                   8223: change it from within Gforth. However, the input can typically be
                   8224: redirected in the command line that starts Gforth.
                   8225: 
                   8226: @item method of selecting the user output device:
                   8227: @cindex user output device, method of selecting
                   8228: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
1.10      anton    8229: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
                   8230: output when the user output device is a terminal, otherwise the output
                   8231: is buffered.
1.1       anton    8232: 
                   8233: @item methods of dictionary compilation:
                   8234: What are we expected to document here?
                   8235: 
                   8236: @item number of bits in one address unit:
                   8237: @cindex number of bits in one address unit
                   8238: @cindex address unit, size in bits
                   8239: @code{s" address-units-bits" environment? drop .}. 8 in all current
                   8240: ports.
                   8241: 
                   8242: @item number representation and arithmetic:
                   8243: @cindex number representation and arithmetic
                   8244: Processor-dependent. Binary two's complement on all current ports.
                   8245: 
                   8246: @item ranges for integer types:
                   8247: @cindex ranges for integer types
                   8248: @cindex integer types, ranges
                   8249: Installation-dependent. Make environmental queries for @code{MAX-N},
                   8250: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
                   8251: unsigned (and positive) types is 0. The lower bound for signed types on
                   8252: two's complement and one's complement machines machines can be computed
                   8253: by adding 1 to the upper bound.
                   8254: 
                   8255: @item read-only data space regions:
                   8256: @cindex read-only data space regions
                   8257: @cindex data-space, read-only regions
                   8258: The whole Forth data space is writable.
                   8259: 
                   8260: @item size of buffer at @code{WORD}:
                   8261: @cindex size of buffer at @code{WORD}
                   8262: @cindex @code{WORD} buffer size
                   8263: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   8264: shared with the pictured numeric output string. If overwriting
                   8265: @code{PAD} is acceptable, it is as large as the remaining dictionary
                   8266: space, although only as much can be sensibly used as fits in a counted
                   8267: string.
                   8268: 
                   8269: @item size of one cell in address units:
                   8270: @cindex cell size
                   8271: @code{1 cells .}.
                   8272: 
                   8273: @item size of one character in address units:
                   8274: @cindex char size
                   8275: @code{1 chars .}. 1 on all current ports.
                   8276: 
                   8277: @item size of the keyboard terminal buffer:
                   8278: @cindex size of the keyboard terminal buffer
                   8279: @cindex terminal buffer, size
                   8280: Varies. You can determine the size at a specific time using @code{lp@@
                   8281: tib - .}. It is shared with the locals stack and TIBs of files that
                   8282: include the current file. You can change the amount of space for TIBs
                   8283: and locals stack at Gforth startup with the command line option
                   8284: @code{-l}.
                   8285: 
                   8286: @item size of the pictured numeric output buffer:
                   8287: @cindex size of the pictured numeric output buffer
                   8288: @cindex pictured numeric output buffer, size
                   8289: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   8290: shared with @code{WORD}.
                   8291: 
                   8292: @item size of the scratch area returned by @code{PAD}:
                   8293: @cindex size of the scratch area returned by @code{PAD}
                   8294: @cindex @code{PAD} size
                   8295: The remainder of dictionary space. @code{unused pad here - - .}.
                   8296: 
                   8297: @item system case-sensitivity characteristics:
                   8298: @cindex case-sensitivity characteristics
1.26      crook    8299: Dictionary searches are case-insensitive (except in
1.1       anton    8300: @code{TABLE}s). However, as explained above under @i{character-set
                   8301: extensions}, the matching for non-ASCII characters is determined by the
                   8302: locale you are using. In the default @code{C} locale all non-ASCII
                   8303: characters are matched case-sensitively.
                   8304: 
                   8305: @item system prompt:
                   8306: @cindex system prompt
                   8307: @cindex prompt
                   8308: @code{ ok} in interpret state, @code{ compiled} in compile state.
                   8309: 
                   8310: @item division rounding:
                   8311: @cindex division rounding
                   8312: installation dependent. @code{s" floored" environment? drop .}. We leave
                   8313: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
                   8314: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
                   8315: 
                   8316: @item values of @code{STATE} when true:
                   8317: @cindex @code{STATE} values
                   8318: -1.
                   8319: 
                   8320: @item values returned after arithmetic overflow:
                   8321: On two's complement machines, arithmetic is performed modulo
                   8322: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   8323: arithmetic (with appropriate mapping for signed types). Division by zero
                   8324: typically results in a @code{-55 throw} (Floating-point unidentified
                   8325: fault), although a @code{-10 throw} (divide by zero) would be more
                   8326: appropriate.
                   8327: 
                   8328: @item whether the current definition can be found after @t{DOES>}:
                   8329: @cindex @t{DOES>}, visibility of current definition
                   8330: No.
                   8331: 
                   8332: @end table
                   8333: 
                   8334: @c ---------------------------------------------------------------------
                   8335: @node core-ambcond, core-other, core-idef, The Core Words
                   8336: @subsection Ambiguous conditions
                   8337: @c ---------------------------------------------------------------------
                   8338: @cindex core words, ambiguous conditions
                   8339: @cindex ambiguous conditions, core words
                   8340: 
                   8341: @table @i
                   8342: 
                   8343: @item a name is neither a word nor a number:
                   8344: @cindex name not found
1.26      crook    8345: @cindex undefined word
1.1       anton    8346: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
                   8347: preserves the data and FP stack, so you don't lose more work than
                   8348: necessary.
                   8349: 
                   8350: @item a definition name exceeds the maximum length allowed:
1.26      crook    8351: @cindex word name too long
1.1       anton    8352: @code{-19 throw} (Word name too long)
                   8353: 
                   8354: @item addressing a region not inside the various data spaces of the forth system:
                   8355: @cindex Invalid memory address
                   8356: The stacks, code space and name space are accessible. Machine code space is
                   8357: typically readable. Accessing other addresses gives results dependent on
                   8358: the operating system. On decent systems: @code{-9 throw} (Invalid memory
                   8359: address).
                   8360: 
                   8361: @item argument type incompatible with parameter:
1.26      crook    8362: @cindex argument type mismatch
1.1       anton    8363: This is usually not caught. Some words perform checks, e.g., the control
                   8364: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
                   8365: mismatch).
                   8366: 
                   8367: @item attempting to obtain the execution token of a word with undefined execution semantics:
                   8368: @cindex Interpreting a compile-only word, for @code{'} etc.
                   8369: @cindex execution token of words with undefined execution semantics
                   8370: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
                   8371: get an execution token for @code{compile-only-error} (which performs a
                   8372: @code{-14 throw} when executed).
                   8373: 
                   8374: @item dividing by zero:
                   8375: @cindex dividing by zero
                   8376: @cindex floating point unidentified fault, integer division
1.24      anton    8377: On better platforms, this produces a @code{-10 throw} (Division by
                   8378: zero); on other systems, this typically results in a @code{-55 throw}
                   8379: (Floating-point unidentified fault).
1.1       anton    8380: 
                   8381: @item insufficient data stack or return stack space:
                   8382: @cindex insufficient data stack or return stack space
                   8383: @cindex stack overflow
1.26      crook    8384: @cindex address alignment exception, stack overflow
1.1       anton    8385: @cindex Invalid memory address, stack overflow
                   8386: Depending on the operating system, the installation, and the invocation
                   8387: of Gforth, this is either checked by the memory management hardware, or
1.24      anton    8388: it is not checked. If it is checked, you typically get a @code{-3 throw}
                   8389: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
                   8390: throw} (Invalid memory address) (depending on the platform and how you
                   8391: achieved the overflow) as soon as the overflow happens. If it is not
                   8392: checked, overflows typically result in mysterious illegal memory
                   8393: accesses, producing @code{-9 throw} (Invalid memory address) or
                   8394: @code{-23 throw} (Address alignment exception); they might also destroy
                   8395: the internal data structure of @code{ALLOCATE} and friends, resulting in
                   8396: various errors in these words.
1.1       anton    8397: 
                   8398: @item insufficient space for loop control parameters:
                   8399: @cindex insufficient space for loop control parameters
                   8400: like other return stack overflows.
                   8401: 
                   8402: @item insufficient space in the dictionary:
                   8403: @cindex insufficient space in the dictionary
                   8404: @cindex dictionary overflow
1.12      anton    8405: If you try to allot (either directly with @code{allot}, or indirectly
                   8406: with @code{,}, @code{create} etc.) more memory than available in the
                   8407: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
                   8408: to access memory beyond the end of the dictionary, the results are
                   8409: similar to stack overflows.
1.1       anton    8410: 
                   8411: @item interpreting a word with undefined interpretation semantics:
                   8412: @cindex interpreting a word with undefined interpretation semantics
                   8413: @cindex Interpreting a compile-only word
                   8414: For some words, we have defined interpretation semantics. For the
                   8415: others: @code{-14 throw} (Interpreting a compile-only word).
                   8416: 
                   8417: @item modifying the contents of the input buffer or a string literal:
                   8418: @cindex modifying the contents of the input buffer or a string literal
                   8419: These are located in writable memory and can be modified.
                   8420: 
                   8421: @item overflow of the pictured numeric output string:
                   8422: @cindex overflow of the pictured numeric output string
                   8423: @cindex pictured numeric output string, overflow
1.24      anton    8424: @code{-17 throw} (Pictured numeric ouput string overflow).
1.1       anton    8425: 
                   8426: @item parsed string overflow:
                   8427: @cindex parsed string overflow
                   8428: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
                   8429: 
                   8430: @item producing a result out of range:
                   8431: @cindex result out of range
                   8432: On two's complement machines, arithmetic is performed modulo
                   8433: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   8434: arithmetic (with appropriate mapping for signed types). Division by zero
1.24      anton    8435: typically results in a @code{-10 throw} (divide by zero) or @code{-55
                   8436: throw} (floating point unidentified fault). @code{convert} and
                   8437: @code{>number} currently overflow silently.
1.1       anton    8438: 
                   8439: @item reading from an empty data or return stack:
                   8440: @cindex stack empty
                   8441: @cindex stack underflow
1.24      anton    8442: @cindex return stack underflow
1.1       anton    8443: The data stack is checked by the outer (aka text) interpreter after
                   8444: every word executed. If it has underflowed, a @code{-4 throw} (Stack
                   8445: underflow) is performed. Apart from that, stacks may be checked or not,
1.24      anton    8446: depending on operating system, installation, and invocation. If they are
                   8447: caught by a check, they typically result in @code{-4 throw} (Stack
                   8448: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
                   8449: (Invalid memory address), depending on the platform and which stack
                   8450: underflows and by how much. Note that even if the system uses checking
                   8451: (through the MMU), your program may have to underflow by a significant
                   8452: number of stack items to trigger the reaction (the reason for this is
                   8453: that the MMU, and therefore the checking, works with a page-size
                   8454: granularity).  If there is no checking, the symptoms resulting from an
                   8455: underflow are similar to those from an overflow.  Unbalanced return
                   8456: stack errors result in a variaty of symptoms, including @code{-9 throw}
                   8457: (Invalid memory address) and Illegal Instruction (typically @code{-260
                   8458: throw}).
1.1       anton    8459: 
                   8460: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
                   8461: @cindex unexpected end of the input buffer
                   8462: @cindex zero-length string as a name
                   8463: @cindex Attempt to use zero-length string as a name
                   8464: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
                   8465: use zero-length string as a name). Words like @code{'} probably will not
                   8466: find what they search. Note that it is possible to create zero-length
                   8467: names with @code{nextname} (should it not?).
                   8468: 
                   8469: @item @code{>IN} greater than input buffer:
                   8470: @cindex @code{>IN} greater than input buffer
                   8471: The next invocation of a parsing word returns a string with length 0.
                   8472: 
                   8473: @item @code{RECURSE} appears after @code{DOES>}:
                   8474: @cindex @code{RECURSE} appears after @code{DOES>}
                   8475: Compiles a recursive call to the defining word, not to the defined word.
                   8476: 
                   8477: @item argument input source different than current input source for @code{RESTORE-INPUT}:
                   8478: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
1.26      crook    8479: @cindex argument type mismatch, @code{RESTORE-INPUT}
1.1       anton    8480: @cindex @code{RESTORE-INPUT}, Argument type mismatch
                   8481: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
                   8482: the end of the file was reached), its source-id may be
                   8483: reused. Therefore, restoring an input source specification referencing a
                   8484: closed file may lead to unpredictable results instead of a @code{-12
                   8485: THROW}.
                   8486: 
                   8487: In the future, Gforth may be able to restore input source specifications
                   8488: from other than the current input source.
                   8489: 
                   8490: @item data space containing definitions gets de-allocated:
                   8491: @cindex data space containing definitions gets de-allocated
                   8492: Deallocation with @code{allot} is not checked. This typically results in
                   8493: memory access faults or execution of illegal instructions.
                   8494: 
                   8495: @item data space read/write with incorrect alignment:
                   8496: @cindex data space read/write with incorrect alignment
                   8497: @cindex alignment faults
1.26      crook    8498: @cindex address alignment exception
1.1       anton    8499: Processor-dependent. Typically results in a @code{-23 throw} (Address
1.12      anton    8500: alignment exception). Under Linux-Intel on a 486 or later processor with
1.1       anton    8501: alignment turned on, incorrect alignment results in a @code{-9 throw}
                   8502: (Invalid memory address). There are reportedly some processors with
1.12      anton    8503: alignment restrictions that do not report violations.
1.1       anton    8504: 
                   8505: @item data space pointer not properly aligned, @code{,}, @code{C,}:
                   8506: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
                   8507: Like other alignment errors.
                   8508: 
                   8509: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
                   8510: Like other stack underflows.
                   8511: 
                   8512: @item loop control parameters not available:
                   8513: @cindex loop control parameters not available
                   8514: Not checked. The counted loop words simply assume that the top of return
                   8515: stack items are loop control parameters and behave accordingly.
                   8516: 
                   8517: @item most recent definition does not have a name (@code{IMMEDIATE}):
                   8518: @cindex most recent definition does not have a name (@code{IMMEDIATE})
                   8519: @cindex last word was headerless
                   8520: @code{abort" last word was headerless"}.
                   8521: 
                   8522: @item name not defined by @code{VALUE} used by @code{TO}:
                   8523: @cindex name not defined by @code{VALUE} used by @code{TO}
                   8524: @cindex @code{TO} on non-@code{VALUE}s
                   8525: @cindex Invalid name argument, @code{TO}
                   8526: @code{-32 throw} (Invalid name argument) (unless name is a local or was
                   8527: defined by @code{CONSTANT}; in the latter case it just changes the constant).
                   8528: 
                   8529: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
                   8530: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
1.26      crook    8531: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
1.1       anton    8532: @code{-13 throw} (Undefined word)
                   8533: 
                   8534: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
                   8535: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
                   8536: Gforth behaves as if they were of the same type. I.e., you can predict
                   8537: the behaviour by interpreting all parameters as, e.g., signed.
                   8538: 
                   8539: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
                   8540: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
                   8541: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
                   8542: compilation semantics of @code{TO}.
                   8543: 
                   8544: @item String longer than a counted string returned by @code{WORD}:
1.26      crook    8545: @cindex string longer than a counted string returned by @code{WORD}
1.1       anton    8546: @cindex @code{WORD}, string overflow
                   8547: Not checked. The string will be ok, but the count will, of course,
                   8548: contain only the least significant bits of the length.
                   8549: 
                   8550: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
                   8551: @cindex @code{LSHIFT}, large shift counts
                   8552: @cindex @code{RSHIFT}, large shift counts
                   8553: Processor-dependent. Typical behaviours are returning 0 and using only
                   8554: the low bits of the shift count.
                   8555: 
                   8556: @item word not defined via @code{CREATE}:
                   8557: @cindex @code{>BODY} of non-@code{CREATE}d words
                   8558: @code{>BODY} produces the PFA of the word no matter how it was defined.
                   8559: 
                   8560: @cindex @code{DOES>} of non-@code{CREATE}d words
                   8561: @code{DOES>} changes the execution semantics of the last defined word no
                   8562: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
                   8563: @code{CREATE , DOES>}.
                   8564: 
                   8565: @item words improperly used outside @code{<#} and @code{#>}:
                   8566: Not checked. As usual, you can expect memory faults.
                   8567: 
                   8568: @end table
                   8569: 
                   8570: 
                   8571: @c ---------------------------------------------------------------------
                   8572: @node core-other,  , core-ambcond, The Core Words
                   8573: @subsection Other system documentation
                   8574: @c ---------------------------------------------------------------------
                   8575: @cindex other system documentation, core words
                   8576: @cindex core words, other system documentation
                   8577: 
                   8578: @table @i
                   8579: @item nonstandard words using @code{PAD}:
                   8580: @cindex @code{PAD} use by nonstandard words
                   8581: None.
                   8582: 
                   8583: @item operator's terminal facilities available:
                   8584: @cindex operator's terminal facilities available
                   8585: After processing the command line, Gforth goes into interactive mode,
                   8586: and you can give commands to Gforth interactively. The actual facilities
                   8587: available depend on how you invoke Gforth.
                   8588: 
                   8589: @item program data space available:
                   8590: @cindex program data space available
                   8591: @cindex data space available
                   8592: @code{UNUSED .} gives the remaining dictionary space. The total
                   8593: dictionary space can be specified with the @code{-m} switch
                   8594: (@pxref{Invoking Gforth}) when Gforth starts up.
                   8595: 
                   8596: @item return stack space available:
                   8597: @cindex return stack space available
                   8598: You can compute the total return stack space in cells with
                   8599: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
                   8600: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
                   8601: 
                   8602: @item stack space available:
                   8603: @cindex stack space available
                   8604: You can compute the total data stack space in cells with
                   8605: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
                   8606: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
                   8607: 
                   8608: @item system dictionary space required, in address units:
                   8609: @cindex system dictionary space required, in address units
                   8610: Type @code{here forthstart - .} after startup. At the time of this
                   8611: writing, this gives 80080 (bytes) on a 32-bit system.
                   8612: @end table
                   8613: 
                   8614: 
                   8615: @c =====================================================================
                   8616: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
                   8617: @section The optional Block word set
                   8618: @c =====================================================================
                   8619: @cindex system documentation, block words
                   8620: @cindex block words, system documentation
                   8621: 
                   8622: @menu
                   8623: * block-idef::                  Implementation Defined Options
                   8624: * block-ambcond::               Ambiguous Conditions               
                   8625: * block-other::                 Other System Documentation                 
                   8626: @end menu
                   8627: 
                   8628: 
                   8629: @c ---------------------------------------------------------------------
                   8630: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
                   8631: @subsection Implementation Defined Options
                   8632: @c ---------------------------------------------------------------------
                   8633: @cindex implementation-defined options, block words
                   8634: @cindex block words, implementation-defined options
                   8635: 
                   8636: @table @i
                   8637: @item the format for display by @code{LIST}:
                   8638: @cindex @code{LIST} display format
                   8639: First the screen number is displayed, then 16 lines of 64 characters,
                   8640: each line preceded by the line number.
                   8641: 
                   8642: @item the length of a line affected by @code{\}:
                   8643: @cindex length of a line affected by @code{\}
                   8644: @cindex @code{\}, line length in blocks
                   8645: 64 characters.
                   8646: @end table
                   8647: 
                   8648: 
                   8649: @c ---------------------------------------------------------------------
                   8650: @node block-ambcond, block-other, block-idef, The optional Block word set
                   8651: @subsection Ambiguous conditions
                   8652: @c ---------------------------------------------------------------------
                   8653: @cindex block words, ambiguous conditions
                   8654: @cindex ambiguous conditions, block words
                   8655: 
                   8656: @table @i
                   8657: @item correct block read was not possible:
                   8658: @cindex block read not possible
                   8659: Typically results in a @code{throw} of some OS-derived value (between
                   8660: -512 and -2048). If the blocks file was just not long enough, blanks are
                   8661: supplied for the missing portion.
                   8662: 
                   8663: @item I/O exception in block transfer:
                   8664: @cindex I/O exception in block transfer
                   8665: @cindex block transfer, I/O exception
                   8666: Typically results in a @code{throw} of some OS-derived value (between
                   8667: -512 and -2048).
                   8668: 
                   8669: @item invalid block number:
                   8670: @cindex invalid block number
                   8671: @cindex block number invalid
                   8672: @code{-35 throw} (Invalid block number)
                   8673: 
                   8674: @item a program directly alters the contents of @code{BLK}:
                   8675: @cindex @code{BLK}, altering @code{BLK}
                   8676: The input stream is switched to that other block, at the same
                   8677: position. If the storing to @code{BLK} happens when interpreting
                   8678: non-block input, the system will get quite confused when the block ends.
                   8679: 
                   8680: @item no current block buffer for @code{UPDATE}:
                   8681: @cindex @code{UPDATE}, no current block buffer
                   8682: @code{UPDATE} has no effect.
                   8683: 
                   8684: @end table
                   8685: 
                   8686: @c ---------------------------------------------------------------------
                   8687: @node block-other,  , block-ambcond, The optional Block word set
                   8688: @subsection Other system documentation
                   8689: @c ---------------------------------------------------------------------
                   8690: @cindex other system documentation, block words
                   8691: @cindex block words, other system documentation
                   8692: 
                   8693: @table @i
                   8694: @item any restrictions a multiprogramming system places on the use of buffer addresses:
                   8695: No restrictions (yet).
                   8696: 
                   8697: @item the number of blocks available for source and data:
                   8698: depends on your disk space.
                   8699: 
                   8700: @end table
                   8701: 
                   8702: 
                   8703: @c =====================================================================
                   8704: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
                   8705: @section The optional Double Number word set
                   8706: @c =====================================================================
                   8707: @cindex system documentation, double words
                   8708: @cindex double words, system documentation
                   8709: 
                   8710: @menu
                   8711: * double-ambcond::              Ambiguous Conditions              
                   8712: @end menu
                   8713: 
                   8714: 
                   8715: @c ---------------------------------------------------------------------
                   8716: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
                   8717: @subsection Ambiguous conditions
                   8718: @c ---------------------------------------------------------------------
                   8719: @cindex double words, ambiguous conditions
                   8720: @cindex ambiguous conditions, double words
                   8721: 
                   8722: @table @i
                   8723: @item @var{d} outside of range of @var{n} in @code{D>S}:
                   8724: @cindex @code{D>S}, @var{d} out of range of @var{n} 
                   8725: The least significant cell of @var{d} is produced.
                   8726: 
                   8727: @end table
                   8728: 
                   8729: 
                   8730: @c =====================================================================
                   8731: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
                   8732: @section The optional Exception word set
                   8733: @c =====================================================================
                   8734: @cindex system documentation, exception words
                   8735: @cindex exception words, system documentation
                   8736: 
                   8737: @menu
                   8738: * exception-idef::              Implementation Defined Options              
                   8739: @end menu
                   8740: 
                   8741: 
                   8742: @c ---------------------------------------------------------------------
                   8743: @node exception-idef,  , The optional Exception word set, The optional Exception word set
                   8744: @subsection Implementation Defined Options
                   8745: @c ---------------------------------------------------------------------
                   8746: @cindex implementation-defined options, exception words
                   8747: @cindex exception words, implementation-defined options
                   8748: 
                   8749: @table @i
                   8750: @item @code{THROW}-codes used in the system:
                   8751: @cindex @code{THROW}-codes used in the system
                   8752: The codes -256@minus{}-511 are used for reporting signals. The mapping
                   8753: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
                   8754: codes -512@minus{}-2047 are used for OS errors (for file and memory
                   8755: allocation operations). The mapping from OS error numbers to throw codes
                   8756: is -512@minus{}@code{errno}. One side effect of this mapping is that
                   8757: undefined OS errors produce a message with a strange number; e.g.,
                   8758: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
                   8759: @end table
                   8760: 
                   8761: @c =====================================================================
                   8762: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
                   8763: @section The optional Facility word set
                   8764: @c =====================================================================
                   8765: @cindex system documentation, facility words
                   8766: @cindex facility words, system documentation
                   8767: 
                   8768: @menu
                   8769: * facility-idef::               Implementation Defined Options               
                   8770: * facility-ambcond::            Ambiguous Conditions            
                   8771: @end menu
                   8772: 
                   8773: 
                   8774: @c ---------------------------------------------------------------------
                   8775: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
                   8776: @subsection Implementation Defined Options
                   8777: @c ---------------------------------------------------------------------
                   8778: @cindex implementation-defined options, facility words
                   8779: @cindex facility words, implementation-defined options
                   8780: 
                   8781: @table @i
                   8782: @item encoding of keyboard events (@code{EKEY}):
                   8783: @cindex keyboard events, encoding in @code{EKEY}
                   8784: @cindex @code{EKEY}, encoding of keyboard events
                   8785: Not yet implemented.
                   8786: 
                   8787: @item duration of a system clock tick:
                   8788: @cindex duration of a system clock tick
                   8789: @cindex clock tick duration
                   8790: System dependent. With respect to @code{MS}, the time is specified in
                   8791: microseconds. How well the OS and the hardware implement this, is
                   8792: another question.
                   8793: 
                   8794: @item repeatability to be expected from the execution of @code{MS}:
                   8795: @cindex repeatability to be expected from the execution of @code{MS}
                   8796: @cindex @code{MS}, repeatability to be expected
                   8797: System dependent. On Unix, a lot depends on load. If the system is
                   8798: lightly loaded, and the delay is short enough that Gforth does not get
                   8799: swapped out, the performance should be acceptable. Under MS-DOS and
                   8800: other single-tasking systems, it should be good.
                   8801: 
                   8802: @end table
                   8803: 
                   8804: 
                   8805: @c ---------------------------------------------------------------------
                   8806: @node facility-ambcond,  , facility-idef, The optional Facility word set
                   8807: @subsection Ambiguous conditions
                   8808: @c ---------------------------------------------------------------------
                   8809: @cindex facility words, ambiguous conditions
                   8810: @cindex ambiguous conditions, facility words
                   8811: 
                   8812: @table @i
                   8813: @item @code{AT-XY} can't be performed on user output device:
                   8814: @cindex @code{AT-XY} can't be performed on user output device
                   8815: Largely terminal dependent. No range checks are done on the arguments.
                   8816: No errors are reported. You may see some garbage appearing, you may see
                   8817: simply nothing happen.
                   8818: 
                   8819: @end table
                   8820: 
                   8821: 
                   8822: @c =====================================================================
                   8823: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
                   8824: @section The optional File-Access word set
                   8825: @c =====================================================================
                   8826: @cindex system documentation, file words
                   8827: @cindex file words, system documentation
                   8828: 
                   8829: @menu
                   8830: * file-idef::                   Implementation Defined Options
                   8831: * file-ambcond::                Ambiguous Conditions                
                   8832: @end menu
                   8833: 
                   8834: @c ---------------------------------------------------------------------
                   8835: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
                   8836: @subsection Implementation Defined Options
                   8837: @c ---------------------------------------------------------------------
                   8838: @cindex implementation-defined options, file words
                   8839: @cindex file words, implementation-defined options
                   8840: 
                   8841: @table @i
                   8842: @item file access methods used:
                   8843: @cindex file access methods used
                   8844: @code{R/O}, @code{R/W} and @code{BIN} work as you would
                   8845: expect. @code{W/O} translates into the C file opening mode @code{w} (or
                   8846: @code{wb}): The file is cleared, if it exists, and created, if it does
                   8847: not (with both @code{open-file} and @code{create-file}).  Under Unix
                   8848: @code{create-file} creates a file with 666 permissions modified by your
                   8849: umask.
                   8850: 
                   8851: @item file exceptions:
                   8852: @cindex file exceptions
                   8853: The file words do not raise exceptions (except, perhaps, memory access
                   8854: faults when you pass illegal addresses or file-ids).
                   8855: 
                   8856: @item file line terminator:
                   8857: @cindex file line terminator
                   8858: System-dependent. Gforth uses C's newline character as line
                   8859: terminator. What the actual character code(s) of this are is
                   8860: system-dependent.
                   8861: 
                   8862: @item file name format:
                   8863: @cindex file name format
                   8864: System dependent. Gforth just uses the file name format of your OS.
                   8865: 
                   8866: @item information returned by @code{FILE-STATUS}:
                   8867: @cindex @code{FILE-STATUS}, returned information
                   8868: @code{FILE-STATUS} returns the most powerful file access mode allowed
                   8869: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
                   8870: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
                   8871: along with the returned mode.
                   8872: 
                   8873: @item input file state after an exception when including source:
                   8874: @cindex exception when including source
                   8875: All files that are left via the exception are closed.
                   8876: 
                   8877: @item @var{ior} values and meaning:
                   8878: @cindex @var{ior} values and meaning
                   8879: The @var{ior}s returned by the file and memory allocation words are
                   8880: intended as throw codes. They typically are in the range
                   8881: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   8882: @var{ior}s is -512@minus{}@var{errno}.
                   8883: 
                   8884: @item maximum depth of file input nesting:
                   8885: @cindex maximum depth of file input nesting
                   8886: @cindex file input nesting, maximum depth
                   8887: limited by the amount of return stack, locals/TIB stack, and the number
                   8888: of open files available. This should not give you troubles.
                   8889: 
                   8890: @item maximum size of input line:
                   8891: @cindex maximum size of input line
                   8892: @cindex input line size, maximum
                   8893: @code{/line}. Currently 255.
                   8894: 
                   8895: @item methods of mapping block ranges to files:
                   8896: @cindex mapping block ranges to files
                   8897: @cindex files containing blocks
                   8898: @cindex blocks in files
                   8899: By default, blocks are accessed in the file @file{blocks.fb} in the
                   8900: current working directory. The file can be switched with @code{USE}.
                   8901: 
                   8902: @item number of string buffers provided by @code{S"}:
                   8903: @cindex @code{S"}, number of string buffers
                   8904: 1
                   8905: 
                   8906: @item size of string buffer used by @code{S"}:
                   8907: @cindex @code{S"}, size of string buffer
                   8908: @code{/line}. currently 255.
                   8909: 
                   8910: @end table
                   8911: 
                   8912: @c ---------------------------------------------------------------------
                   8913: @node file-ambcond,  , file-idef, The optional File-Access word set
                   8914: @subsection Ambiguous conditions
                   8915: @c ---------------------------------------------------------------------
                   8916: @cindex file words, ambiguous conditions
                   8917: @cindex ambiguous conditions, file words
                   8918: 
                   8919: @table @i
                   8920: @item attempting to position a file outside its boundaries:
                   8921: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
                   8922: @code{REPOSITION-FILE} is performed as usual: Afterwards,
                   8923: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
                   8924: 
                   8925: @item attempting to read from file positions not yet written:
                   8926: @cindex reading from file positions not yet written
                   8927: End-of-file, i.e., zero characters are read and no error is reported.
                   8928: 
                   8929: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
                   8930: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid 
                   8931: An appropriate exception may be thrown, but a memory fault or other
                   8932: problem is more probable.
                   8933: 
                   8934: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
                   8935: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
                   8936: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
                   8937: The @var{ior} produced by the operation, that discovered the problem, is
                   8938: thrown.
                   8939: 
                   8940: @item named file cannot be opened (@code{INCLUDED}):
                   8941: @cindex @code{INCLUDED}, named file cannot be opened
                   8942: The @var{ior} produced by @code{open-file} is thrown.
                   8943: 
                   8944: @item requesting an unmapped block number:
                   8945: @cindex unmapped block numbers
                   8946: There are no unmapped legal block numbers. On some operating systems,
                   8947: writing a block with a large number may overflow the file system and
                   8948: have an error message as consequence.
                   8949: 
                   8950: @item using @code{source-id} when @code{blk} is non-zero:
                   8951: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
                   8952: @code{source-id} performs its function. Typically it will give the id of
                   8953: the source which loaded the block. (Better ideas?)
                   8954: 
                   8955: @end table
                   8956: 
                   8957: 
                   8958: @c =====================================================================
                   8959: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
                   8960: @section The optional Floating-Point word set
                   8961: @c =====================================================================
                   8962: @cindex system documentation, floating-point words
                   8963: @cindex floating-point words, system documentation
                   8964: 
                   8965: @menu
                   8966: * floating-idef::               Implementation Defined Options
                   8967: * floating-ambcond::            Ambiguous Conditions            
                   8968: @end menu
                   8969: 
                   8970: 
                   8971: @c ---------------------------------------------------------------------
                   8972: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
                   8973: @subsection Implementation Defined Options
                   8974: @c ---------------------------------------------------------------------
                   8975: @cindex implementation-defined options, floating-point words
                   8976: @cindex floating-point words, implementation-defined options
                   8977: 
                   8978: @table @i
                   8979: @item format and range of floating point numbers:
                   8980: @cindex format and range of floating point numbers
                   8981: @cindex floating point numbers, format and range
                   8982: System-dependent; the @code{double} type of C.
                   8983: 
                   8984: @item results of @code{REPRESENT} when @var{float} is out of range:
                   8985: @cindex  @code{REPRESENT}, results when @var{float} is out of range
                   8986: System dependent; @code{REPRESENT} is implemented using the C library
                   8987: function @code{ecvt()} and inherits its behaviour in this respect.
                   8988: 
                   8989: @item rounding or truncation of floating-point numbers:
                   8990: @cindex rounding of floating-point numbers
                   8991: @cindex truncation of floating-point numbers
                   8992: @cindex floating-point numbers, rounding or truncation
                   8993: System dependent; the rounding behaviour is inherited from the hosting C
                   8994: compiler. IEEE-FP-based (i.e., most) systems by default round to
                   8995: nearest, and break ties by rounding to even (i.e., such that the last
                   8996: bit of the mantissa is 0).
                   8997: 
                   8998: @item size of floating-point stack:
                   8999: @cindex floating-point stack size
                   9000: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
                   9001: the floating-point stack (in floats). You can specify this on startup
                   9002: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
                   9003: 
                   9004: @item width of floating-point stack:
                   9005: @cindex floating-point stack width 
                   9006: @code{1 floats}.
                   9007: 
                   9008: @end table
                   9009: 
                   9010: 
                   9011: @c ---------------------------------------------------------------------
                   9012: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
                   9013: @subsection Ambiguous conditions
                   9014: @c ---------------------------------------------------------------------
                   9015: @cindex floating-point words, ambiguous conditions
                   9016: @cindex ambiguous conditions, floating-point words
                   9017: 
                   9018: @table @i
                   9019: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
                   9020: @cindex @code{df@@} or @code{df!} used with an address that is not double-float  aligned
                   9021: System-dependent. Typically results in a @code{-23 THROW} like other
                   9022: alignment violations.
                   9023: 
                   9024: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
                   9025: @cindex @code{f@@} used with an address that is not float aligned
                   9026: @cindex @code{f!} used with an address that is not float aligned
                   9027: System-dependent. Typically results in a @code{-23 THROW} like other
                   9028: alignment violations.
                   9029: 
                   9030: @item floating-point result out of range:
                   9031: @cindex floating-point result out of range
                   9032: System-dependent. Can result in a @code{-55 THROW} (Floating-point
                   9033: unidentified fault), or can produce a special value representing, e.g.,
                   9034: Infinity.
                   9035: 
                   9036: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
                   9037: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned
                   9038: System-dependent. Typically results in an alignment fault like other
                   9039: alignment violations.
                   9040: 
                   9041: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
                   9042: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
                   9043: The floating-point number is converted into decimal nonetheless.
                   9044: 
                   9045: @item Both arguments are equal to zero (@code{FATAN2}):
                   9046: @cindex @code{FATAN2}, both arguments are equal to zero
                   9047: System-dependent. @code{FATAN2} is implemented using the C library
                   9048: function @code{atan2()}.
                   9049: 
                   9050: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
                   9051: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
                   9052: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
                   9053: because of small errors and the tan will be a very large (or very small)
                   9054: but finite number.
                   9055: 
                   9056: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
                   9057: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
                   9058: The result is rounded to the nearest float.
                   9059: 
                   9060: @item dividing by zero:
                   9061: @cindex dividing by zero, floating-point
                   9062: @cindex floating-point dividing by zero
                   9063: @cindex floating-point unidentified fault, FP divide-by-zero
                   9064: @code{-55 throw} (Floating-point unidentified fault)
                   9065: 
                   9066: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
                   9067: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
                   9068: System dependent. On IEEE-FP based systems the number is converted into
                   9069: an infinity.
                   9070: 
                   9071: @item @var{float}<1 (@code{FACOSH}):
                   9072: @cindex @code{FACOSH}, @var{float}<1
                   9073: @cindex floating-point unidentified fault, @code{FACOSH}
                   9074: @code{-55 throw} (Floating-point unidentified fault)
                   9075: 
                   9076: @item @var{float}=<-1 (@code{FLNP1}):
                   9077: @cindex @code{FLNP1}, @var{float}=<-1
                   9078: @cindex floating-point unidentified fault, @code{FLNP1}
                   9079: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   9080: negative infinity is typically produced for @var{float}=-1.
                   9081: 
                   9082: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
                   9083: @cindex @code{FLN}, @var{float}=<0
                   9084: @cindex @code{FLOG}, @var{float}=<0
                   9085: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
                   9086: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   9087: negative infinity is typically produced for @var{float}=0.
                   9088: 
                   9089: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
                   9090: @cindex @code{FASINH}, @var{float}<0
                   9091: @cindex @code{FSQRT}, @var{float}<0
                   9092: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
                   9093: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
                   9094: produces values for these inputs on my Linux box (Bug in the C library?)
                   9095: 
                   9096: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
                   9097: @cindex @code{FACOS}, |@var{float}|>1
                   9098: @cindex @code{FASIN}, |@var{float}|>1
                   9099: @cindex @code{FATANH}, |@var{float}|>1
                   9100: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
                   9101: @code{-55 throw} (Floating-point unidentified fault).
                   9102: 
                   9103: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
                   9104: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
                   9105: @cindex floating-point unidentified fault, @code{F>D}
                   9106: @code{-55 throw} (Floating-point unidentified fault).
                   9107: 
                   9108: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
                   9109: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
                   9110: This does not happen.
                   9111: @end table
                   9112: 
                   9113: @c =====================================================================
                   9114: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
                   9115: @section The optional Locals word set
                   9116: @c =====================================================================
                   9117: @cindex system documentation, locals words
                   9118: @cindex locals words, system documentation
                   9119: 
                   9120: @menu
                   9121: * locals-idef::                 Implementation Defined Options                 
                   9122: * locals-ambcond::              Ambiguous Conditions              
                   9123: @end menu
                   9124: 
                   9125: 
                   9126: @c ---------------------------------------------------------------------
                   9127: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
                   9128: @subsection Implementation Defined Options
                   9129: @c ---------------------------------------------------------------------
                   9130: @cindex implementation-defined options, locals words
                   9131: @cindex locals words, implementation-defined options
                   9132: 
                   9133: @table @i
                   9134: @item maximum number of locals in a definition:
                   9135: @cindex maximum number of locals in a definition
                   9136: @cindex locals, maximum number in a definition
                   9137: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
                   9138: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
                   9139: characters. The number of locals in a definition is bounded by the size
                   9140: of locals-buffer, which contains the names of the locals.
                   9141: 
                   9142: @end table
                   9143: 
                   9144: 
                   9145: @c ---------------------------------------------------------------------
                   9146: @node locals-ambcond,  , locals-idef, The optional Locals word set
                   9147: @subsection Ambiguous conditions
                   9148: @c ---------------------------------------------------------------------
                   9149: @cindex locals words, ambiguous conditions
                   9150: @cindex ambiguous conditions, locals words
                   9151: 
                   9152: @table @i
                   9153: @item executing a named local in interpretation state:
                   9154: @cindex local in interpretation state
                   9155: @cindex Interpreting a compile-only word, for a local
                   9156: Locals have no interpretation semantics. If you try to perform the
                   9157: interpretation semantics, you will get a @code{-14 throw} somewhere
                   9158: (Interpreting a compile-only word). If you perform the compilation
                   9159: semantics, the locals access will be compiled (irrespective of state).
                   9160: 
                   9161: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
                   9162: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
                   9163: @cindex @code{TO} on non-@code{VALUE}s and non-locals
                   9164: @cindex Invalid name argument, @code{TO}
                   9165: @code{-32 throw} (Invalid name argument)
                   9166: 
                   9167: @end table
                   9168: 
                   9169: 
                   9170: @c =====================================================================
                   9171: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
                   9172: @section The optional Memory-Allocation word set
                   9173: @c =====================================================================
                   9174: @cindex system documentation, memory-allocation words
                   9175: @cindex memory-allocation words, system documentation
                   9176: 
                   9177: @menu
                   9178: * memory-idef::                 Implementation Defined Options                 
                   9179: @end menu
                   9180: 
                   9181: 
                   9182: @c ---------------------------------------------------------------------
                   9183: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
                   9184: @subsection Implementation Defined Options
                   9185: @c ---------------------------------------------------------------------
                   9186: @cindex implementation-defined options, memory-allocation words
                   9187: @cindex memory-allocation words, implementation-defined options
                   9188: 
                   9189: @table @i
                   9190: @item values and meaning of @var{ior}:
                   9191: @cindex  @var{ior} values and meaning
                   9192: The @var{ior}s returned by the file and memory allocation words are
                   9193: intended as throw codes. They typically are in the range
                   9194: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   9195: @var{ior}s is -512@minus{}@var{errno}.
                   9196: 
                   9197: @end table
                   9198: 
                   9199: @c =====================================================================
                   9200: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
                   9201: @section The optional Programming-Tools word set
                   9202: @c =====================================================================
                   9203: @cindex system documentation, programming-tools words
                   9204: @cindex programming-tools words, system documentation
                   9205: 
                   9206: @menu
                   9207: * programming-idef::            Implementation Defined Options            
                   9208: * programming-ambcond::         Ambiguous Conditions         
                   9209: @end menu
                   9210: 
                   9211: 
                   9212: @c ---------------------------------------------------------------------
                   9213: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
                   9214: @subsection Implementation Defined Options
                   9215: @c ---------------------------------------------------------------------
                   9216: @cindex implementation-defined options, programming-tools words
                   9217: @cindex programming-tools words, implementation-defined options
                   9218: 
                   9219: @table @i
                   9220: @item ending sequence for input following @code{;CODE} and @code{CODE}:
                   9221: @cindex @code{;CODE} ending sequence
                   9222: @cindex @code{CODE} ending sequence
                   9223: @code{END-CODE}
                   9224: 
                   9225: @item manner of processing input following @code{;CODE} and @code{CODE}:
                   9226: @cindex @code{;CODE}, processing input
                   9227: @cindex @code{CODE}, processing input
                   9228: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
                   9229: the input is processed by the text interpreter, (starting) in interpret
                   9230: state.
                   9231: 
                   9232: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
                   9233: @cindex @code{ASSEMBLER}, search order capability
                   9234: The ANS Forth search order word set.
                   9235: 
                   9236: @item source and format of display by @code{SEE}:
                   9237: @cindex @code{SEE}, source and format of output
                   9238: The source for @code{see} is the intermediate code used by the inner
                   9239: interpreter.  The current @code{see} tries to output Forth source code
                   9240: as well as possible.
                   9241: 
                   9242: @end table
                   9243: 
                   9244: @c ---------------------------------------------------------------------
                   9245: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
                   9246: @subsection Ambiguous conditions
                   9247: @c ---------------------------------------------------------------------
                   9248: @cindex programming-tools words, ambiguous conditions
                   9249: @cindex ambiguous conditions, programming-tools words
                   9250: 
                   9251: @table @i
                   9252: 
1.21      crook    9253: @item deleting the compilation word list (@code{FORGET}):
                   9254: @cindex @code{FORGET}, deleting the compilation word list
1.1       anton    9255: Not implemented (yet).
                   9256: 
1.26      crook    9257: @item fewer than @var{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
                   9258: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow-stack
                   9259: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow-stack
1.1       anton    9260: @cindex control-flow stack underflow
                   9261: This typically results in an @code{abort"} with a descriptive error
                   9262: message (may change into a @code{-22 throw} (Control structure mismatch)
                   9263: in the future). You may also get a memory access error. If you are
                   9264: unlucky, this ambiguous condition is not caught.
                   9265: 
                   9266: @item @var{name} can't be found (@code{FORGET}):
                   9267: @cindex @code{FORGET}, @var{name} can't be found
                   9268: Not implemented (yet).
                   9269: 
                   9270: @item @var{name} not defined via @code{CREATE}:
                   9271: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
                   9272: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
                   9273: the execution semantics of the last defined word no matter how it was
                   9274: defined.
                   9275: 
                   9276: @item @code{POSTPONE} applied to @code{[IF]}:
                   9277: @cindex @code{POSTPONE} applied to @code{[IF]}
                   9278: @cindex @code{[IF]} and @code{POSTPONE}
                   9279: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
                   9280: equivalent to @code{[IF]}.
                   9281: 
                   9282: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
                   9283: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
                   9284: Continue in the same state of conditional compilation in the next outer
                   9285: input source. Currently there is no warning to the user about this.
                   9286: 
                   9287: @item removing a needed definition (@code{FORGET}):
                   9288: @cindex @code{FORGET}, removing a needed definition
                   9289: Not implemented (yet).
                   9290: 
                   9291: @end table
                   9292: 
                   9293: 
                   9294: @c =====================================================================
                   9295: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
                   9296: @section The optional Search-Order word set
                   9297: @c =====================================================================
                   9298: @cindex system documentation, search-order words
                   9299: @cindex search-order words, system documentation
                   9300: 
                   9301: @menu
                   9302: * search-idef::                 Implementation Defined Options                 
                   9303: * search-ambcond::              Ambiguous Conditions              
                   9304: @end menu
                   9305: 
                   9306: 
                   9307: @c ---------------------------------------------------------------------
                   9308: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
                   9309: @subsection Implementation Defined Options
                   9310: @c ---------------------------------------------------------------------
                   9311: @cindex implementation-defined options, search-order words
                   9312: @cindex search-order words, implementation-defined options
                   9313: 
                   9314: @table @i
                   9315: @item maximum number of word lists in search order:
                   9316: @cindex maximum number of word lists in search order
                   9317: @cindex search order, maximum depth
                   9318: @code{s" wordlists" environment? drop .}. Currently 16.
                   9319: 
                   9320: @item minimum search order:
                   9321: @cindex minimum search order
                   9322: @cindex search order, minimum
                   9323: @code{root root}.
                   9324: 
                   9325: @end table
                   9326: 
                   9327: @c ---------------------------------------------------------------------
                   9328: @node search-ambcond,  , search-idef, The optional Search-Order word set
                   9329: @subsection Ambiguous conditions
                   9330: @c ---------------------------------------------------------------------
                   9331: @cindex search-order words, ambiguous conditions
                   9332: @cindex ambiguous conditions, search-order words
                   9333: 
                   9334: @table @i
1.21      crook    9335: @item changing the compilation word list (during compilation):
                   9336: @cindex changing the compilation word list (during compilation)
                   9337: @cindex compilation word list, change before definition ends
                   9338: The word is entered into the word list that was the compilation word list
1.1       anton    9339: at the start of the definition. Any changes to the name field (e.g.,
                   9340: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
                   9341: are applied to the latest defined word (as reported by @code{last} or
1.21      crook    9342: @code{lastxt}), if possible, irrespective of the compilation word list.
1.1       anton    9343: 
                   9344: @item search order empty (@code{previous}):
                   9345: @cindex @code{previous}, search order empty
1.26      crook    9346: @cindex vocstack empty, @code{previous}
1.1       anton    9347: @code{abort" Vocstack empty"}.
                   9348: 
                   9349: @item too many word lists in search order (@code{also}):
                   9350: @cindex @code{also}, too many word lists in search order
1.26      crook    9351: @cindex vocstack full, @code{also}
1.1       anton    9352: @code{abort" Vocstack full"}.
                   9353: 
                   9354: @end table
                   9355: 
                   9356: @c ***************************************************************
                   9357: @node Model, Integrating Gforth, ANS conformance, Top
                   9358: @chapter Model
                   9359: 
                   9360: This chapter has yet to be written. It will contain information, on
                   9361: which internal structures you can rely.
                   9362: 
                   9363: @c ***************************************************************
                   9364: @node Integrating Gforth, Emacs and Gforth, Model, Top
                   9365: @chapter Integrating Gforth into C programs
                   9366: 
                   9367: This is not yet implemented.
                   9368: 
                   9369: Several people like to use Forth as scripting language for applications
                   9370: that are otherwise written in C, C++, or some other language.
                   9371: 
                   9372: The Forth system ATLAST provides facilities for embedding it into
                   9373: applications; unfortunately it has several disadvantages: most
                   9374: importantly, it is not based on ANS Forth, and it is apparently dead
                   9375: (i.e., not developed further and not supported). The facilities
1.21      crook    9376: provided by Gforth in this area are inspired by ATLAST's facilities, so
1.1       anton    9377: making the switch should not be hard.
                   9378: 
                   9379: We also tried to design the interface such that it can easily be
                   9380: implemented by other Forth systems, so that we may one day arrive at a
                   9381: standardized interface. Such a standard interface would allow you to
                   9382: replace the Forth system without having to rewrite C code.
                   9383: 
                   9384: You embed the Gforth interpreter by linking with the library
                   9385: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
                   9386: global symbols in this library that belong to the interface, have the
                   9387: prefix @code{forth_}. (Global symbols that are used internally have the
                   9388: prefix @code{gforth_}).
                   9389: 
                   9390: You can include the declarations of Forth types and the functions and
                   9391: variables of the interface with @code{#include <forth.h>}.
                   9392: 
                   9393: Types.
                   9394: 
                   9395: Variables.
                   9396: 
                   9397: Data and FP Stack pointer. Area sizes.
                   9398: 
                   9399: functions.
                   9400: 
                   9401: forth_init(imagefile)
                   9402: forth_evaluate(string) exceptions?
                   9403: forth_goto(address) (or forth_execute(xt)?)
                   9404: forth_continue() (a corountining mechanism)
                   9405: 
                   9406: Adding primitives.
                   9407: 
                   9408: No checking.
                   9409: 
                   9410: Signals?
                   9411: 
                   9412: Accessing the Stacks
                   9413: 
1.26      crook    9414: @c ******************************************************************
1.1       anton    9415: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
                   9416: @chapter Emacs and Gforth
                   9417: @cindex Emacs and Gforth
                   9418: 
                   9419: @cindex @file{gforth.el}
                   9420: @cindex @file{forth.el}
                   9421: @cindex Rydqvist, Goran
                   9422: @cindex comment editing commands
                   9423: @cindex @code{\}, editing with Emacs
                   9424: @cindex debug tracer editing commands
                   9425: @cindex @code{~~}, removal with Emacs
                   9426: @cindex Forth mode in Emacs
                   9427: Gforth comes with @file{gforth.el}, an improved version of
                   9428: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.26      crook    9429: improvements are:
                   9430: 
                   9431: @itemize @bullet
                   9432: @item
                   9433: A better (but still not perfect) handling of indentation.
                   9434: @item
                   9435: Comment paragraph filling (@kbd{M-q})
                   9436: @item
                   9437: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
                   9438: @item
                   9439: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
                   9440: @end itemize
                   9441: 
                   9442: I left the stuff I do not use alone, even though some of it only makes
                   9443: sense for TILE. To get a description of these features, enter Forth mode
                   9444: and type @kbd{C-h m}.
1.1       anton    9445: 
                   9446: @cindex source location of error or debugging output in Emacs
                   9447: @cindex error output, finding the source location in Emacs
                   9448: @cindex debugging output, finding the source location in Emacs
                   9449: In addition, Gforth supports Emacs quite well: The source code locations
                   9450: given in error messages, debugging output (from @code{~~}) and failed
                   9451: assertion messages are in the right format for Emacs' compilation mode
                   9452: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
                   9453: Manual}) so the source location corresponding to an error or other
                   9454: message is only a few keystrokes away (@kbd{C-x `} for the next error,
                   9455: @kbd{C-c C-c} for the error under the cursor).
                   9456: 
                   9457: @cindex @file{TAGS} file
                   9458: @cindex @file{etags.fs}
                   9459: @cindex viewing the source of a word in Emacs
1.26      crook    9460: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file will
                   9461: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
1.1       anton    9462: contains the definitions of all words defined afterwards. You can then
                   9463: find the source for a word using @kbd{M-.}. Note that emacs can use
                   9464: several tags files at the same time (e.g., one for the Gforth sources
                   9465: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
                   9466: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
                   9467: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
                   9468: @file{/usr/local/share/gforth/0.2.0/TAGS}).
                   9469: 
                   9470: @cindex @file{.emacs}
                   9471: To get all these benefits, add the following lines to your @file{.emacs}
                   9472: file:
                   9473: 
                   9474: @example
                   9475: (autoload 'forth-mode "gforth.el")
                   9476: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
                   9477: @end example
                   9478: 
1.26      crook    9479: @c ******************************************************************
1.1       anton    9480: @node Image Files, Engine, Emacs and Gforth, Top
                   9481: @chapter Image Files
1.26      crook    9482: @cindex image file
                   9483: @cindex @file{.fi} files
1.1       anton    9484: @cindex precompiled Forth code
                   9485: @cindex dictionary in persistent form
                   9486: @cindex persistent form of dictionary
                   9487: 
                   9488: An image file is a file containing an image of the Forth dictionary,
                   9489: i.e., compiled Forth code and data residing in the dictionary.  By
                   9490: convention, we use the extension @code{.fi} for image files.
                   9491: 
                   9492: @menu
1.18      anton    9493: * Image Licensing Issues::      Distribution terms for images.
                   9494: * Image File Background::       Why have image files?
                   9495: * Non-Relocatable Image Files::  don't always work.
                   9496: * Data-Relocatable Image Files::  are better.
1.1       anton    9497: * Fully Relocatable Image Files::  better yet.
1.18      anton    9498: * Stack and Dictionary Sizes::  Setting the default sizes for an image.
                   9499: * Running Image Files::         @code{gforth -i @var{file}} or @var{file}.
                   9500: * Modifying the Startup Sequence::  and turnkey applications.
1.1       anton    9501: @end menu
                   9502: 
1.18      anton    9503: @node Image Licensing Issues, Image File Background, Image Files, Image Files
                   9504: @section Image Licensing Issues
                   9505: @cindex license for images
                   9506: @cindex image license
                   9507: 
                   9508: An image created with @code{gforthmi} (@pxref{gforthmi}) or
                   9509: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
                   9510: original image; i.e., according to copyright law it is a derived work of
                   9511: the original image.
                   9512: 
                   9513: Since Gforth is distributed under the GNU GPL, the newly created image
                   9514: falls under the GNU GPL, too. In particular, this means that if you
                   9515: distribute the image, you have to make all of the sources for the image
                   9516: available, including those you wrote.  For details see @ref{License, ,
                   9517: GNU General Public License (Section 3)}.
                   9518: 
                   9519: If you create an image with @code{cross} (@pxref{cross.fs}), the image
                   9520: contains only code compiled from the sources you gave it; if none of
                   9521: these sources is under the GPL, the terms discussed above do not apply
                   9522: to the image. However, if your image needs an engine (a gforth binary)
                   9523: that is under the GPL, you should make sure that you distribute both in
                   9524: a way that is at most a @emph{mere aggregation}, if you don't want the
                   9525: terms of the GPL to apply to the image.
                   9526: 
                   9527: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
1.1       anton    9528: @section Image File Background
                   9529: @cindex image file background
                   9530: 
                   9531: Our Forth system consists not only of primitives, but also of
                   9532: definitions written in Forth. Since the Forth compiler itself belongs to
                   9533: those definitions, it is not possible to start the system with the
                   9534: primitives and the Forth source alone. Therefore we provide the Forth
1.26      crook    9535: code as an image file in nearly executable form. When Gforth starts up,
                   9536: a C routine loads the image file into memory, optionally relocates the
                   9537: addresses, then sets up the memory (stacks etc.) according to
                   9538: information in the image file, and (finally) starts executing Forth
                   9539: code.
1.1       anton    9540: 
                   9541: The image file variants represent different compromises between the
                   9542: goals of making it easy to generate image files and making them
                   9543: portable.
                   9544: 
                   9545: @cindex relocation at run-time
1.26      crook    9546: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
1.1       anton    9547: run-time. This avoids many of the complications discussed below (image
                   9548: files are data relocatable without further ado), but costs performance
                   9549: (one addition per memory access).
                   9550: 
                   9551: @cindex relocation at load-time
1.26      crook    9552: By contrast, the Gforth loader performs relocation at image load time. The
                   9553: loader also has to replace tokens that represent primitive calls with the
1.1       anton    9554: appropriate code-field addresses (or code addresses in the case of
                   9555: direct threading).
                   9556: 
                   9557: There are three kinds of image files, with different degrees of
                   9558: relocatability: non-relocatable, data-relocatable, and fully relocatable
                   9559: image files.
                   9560: 
                   9561: @cindex image file loader
                   9562: @cindex relocating loader
                   9563: @cindex loader for image files
                   9564: These image file variants have several restrictions in common; they are
                   9565: caused by the design of the image file loader:
                   9566: 
                   9567: @itemize @bullet
                   9568: @item
                   9569: There is only one segment; in particular, this means, that an image file
                   9570: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
1.26      crook    9571: them). The contents of the stacks are not represented, either.
1.1       anton    9572: 
                   9573: @item
                   9574: The only kinds of relocation supported are: adding the same offset to
                   9575: all cells that represent data addresses; and replacing special tokens
                   9576: with code addresses or with pieces of machine code.
                   9577: 
                   9578: If any complex computations involving addresses are performed, the
                   9579: results cannot be represented in the image file. Several applications that
                   9580: use such computations come to mind:
                   9581: @itemize @minus
                   9582: @item
                   9583: Hashing addresses (or data structures which contain addresses) for table
                   9584: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
                   9585: purpose, you will have no problem, because the hash tables are
                   9586: recomputed automatically when the system is started. If you use your own
                   9587: hash tables, you will have to do something similar.
                   9588: 
                   9589: @item
                   9590: There's a cute implementation of doubly-linked lists that uses
                   9591: @code{XOR}ed addresses. You could represent such lists as singly-linked
                   9592: in the image file, and restore the doubly-linked representation on
                   9593: startup.@footnote{In my opinion, though, you should think thrice before
                   9594: using a doubly-linked list (whatever implementation).}
                   9595: 
                   9596: @item
                   9597: The code addresses of run-time routines like @code{docol:} cannot be
                   9598: represented in the image file (because their tokens would be replaced by
                   9599: machine code in direct threaded implementations). As a workaround,
                   9600: compute these addresses at run-time with @code{>code-address} from the
                   9601: executions tokens of appropriate words (see the definitions of
                   9602: @code{docol:} and friends in @file{kernel.fs}).
                   9603: 
                   9604: @item
                   9605: On many architectures addresses are represented in machine code in some
                   9606: shifted or mangled form. You cannot put @code{CODE} words that contain
                   9607: absolute addresses in this form in a relocatable image file. Workarounds
                   9608: are representing the address in some relative form (e.g., relative to
                   9609: the CFA, which is present in some register), or loading the address from
                   9610: a place where it is stored in a non-mangled form.
                   9611: @end itemize
                   9612: @end itemize
                   9613: 
                   9614: @node  Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
                   9615: @section Non-Relocatable Image Files
                   9616: @cindex non-relocatable image files
1.26      crook    9617: @cindex image file, non-relocatable
1.1       anton    9618: 
                   9619: These files are simple memory dumps of the dictionary. They are specific
                   9620: to the executable (i.e., @file{gforth} file) they were created
                   9621: with. What's worse, they are specific to the place on which the
                   9622: dictionary resided when the image was created. Now, there is no
                   9623: guarantee that the dictionary will reside at the same place the next
                   9624: time you start Gforth, so there's no guarantee that a non-relocatable
                   9625: image will work the next time (Gforth will complain instead of crashing,
                   9626: though).
                   9627: 
                   9628: You can create a non-relocatable image file with
                   9629: 
                   9630: doc-savesystem
                   9631: 
                   9632: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
                   9633: @section Data-Relocatable Image Files
                   9634: @cindex data-relocatable image files
1.26      crook    9635: @cindex image file, data-relocatable
1.1       anton    9636: 
                   9637: These files contain relocatable data addresses, but fixed code addresses
                   9638: (instead of tokens). They are specific to the executable (i.e.,
                   9639: @file{gforth} file) they were created with. For direct threading on some
                   9640: architectures (e.g., the i386), data-relocatable images do not work. You
                   9641: get a data-relocatable image, if you use @file{gforthmi} with a
                   9642: Gforth binary that is not doubly indirect threaded (@pxref{Fully
                   9643: Relocatable Image Files}).
                   9644: 
                   9645: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
                   9646: @section Fully Relocatable Image Files
                   9647: @cindex fully relocatable image files
1.26      crook    9648: @cindex image file, fully relocatable
1.1       anton    9649: 
                   9650: @cindex @file{kern*.fi}, relocatability
                   9651: @cindex @file{gforth.fi}, relocatability
                   9652: These image files have relocatable data addresses, and tokens for code
                   9653: addresses. They can be used with different binaries (e.g., with and
                   9654: without debugging) on the same machine, and even across machines with
                   9655: the same data formats (byte order, cell size, floating point
                   9656: format). However, they are usually specific to the version of Gforth
                   9657: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
                   9658: are fully relocatable.
                   9659: 
                   9660: There are two ways to create a fully relocatable image file:
                   9661: 
                   9662: @menu
                   9663: * gforthmi::            The normal way
                   9664: * cross.fs::                    The hard way
                   9665: @end menu
                   9666: 
                   9667: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
                   9668: @subsection @file{gforthmi}
                   9669: @cindex @file{comp-i.fs}
                   9670: @cindex @file{gforthmi}
                   9671: 
                   9672: You will usually use @file{gforthmi}. If you want to create an
                   9673: image @var{file} that contains everything you would load by invoking
1.27      crook    9674: Gforth with @code{gforth @var{options}}, you simply say:
1.1       anton    9675: @example
                   9676: gforthmi @var{file} @var{options}
                   9677: @end example
                   9678: 
                   9679: E.g., if you want to create an image @file{asm.fi} that has the file
                   9680: @file{asm.fs} loaded in addition to the usual stuff, you could do it
                   9681: like this:
                   9682: 
                   9683: @example
                   9684: gforthmi asm.fi asm.fs
                   9685: @end example
                   9686: 
1.27      crook    9687: @file{gforthmi} is implemented as a sh script and works like this: It
                   9688: produces two non-relocatable images for different addresses and then
                   9689: compares them. Its output reflects this: first you see the output (if
                   9690: any) of the two Gforth invocations that produce the nonrelocatable image
                   9691: files, then you see the output of the comparing program: It displays the
                   9692: offset used for data addresses and the offset used for code addresses;
1.1       anton    9693: moreover, for each cell that cannot be represented correctly in the
                   9694: image files, it displays a line like the following one:
                   9695: 
                   9696: @example
                   9697:      78DC         BFFFFA50         BFFFFA40
                   9698: @end example
                   9699: 
                   9700: This means that at offset $78dc from @code{forthstart}, one input image
                   9701: contains $bffffa50, and the other contains $bffffa40. Since these cells
                   9702: cannot be represented correctly in the output image, you should examine
                   9703: these places in the dictionary and verify that these cells are dead
                   9704: (i.e., not read before they are written).
                   9705: 
1.27      crook    9706: If you type @file{gforthmi} with no arguments, it prints some usage
                   9707: instructions.
                   9708: 
1.1       anton    9709: @cindex @code{savesystem} during @file{gforthmi}
                   9710: @cindex @code{bye} during @file{gforthmi}
                   9711: @cindex doubly indirect threaded code
                   9712: @cindex environment variable @code{GFORTHD}
                   9713: @cindex @code{GFORTHD} environment variable
                   9714: @cindex @code{gforth-ditc}
                   9715: There are a few wrinkles: After processing the passed @var{options}, the
                   9716: words @code{savesystem} and @code{bye} must be visible. A special doubly
                   9717: indirect threaded version of the @file{gforth} executable is used for
                   9718: creating the nonrelocatable images; you can pass the exact filename of
                   9719: this executable through the environment variable @code{GFORTHD}
                   9720: (default: @file{gforth-ditc}); if you pass a version that is not doubly
                   9721: indirect threaded, you will not get a fully relocatable image, but a
1.27      crook    9722: data-relocatable image (because there is no code address offset). The
                   9723: normal @file{gforth} executable is used for creating the relocatable
                   9724: image; you can pass the exact filename of this executable through the
                   9725: environment variable @code{GFORTH}.
1.1       anton    9726: 
                   9727: @node cross.fs,  , gforthmi, Fully Relocatable Image Files
                   9728: @subsection @file{cross.fs}
                   9729: @cindex @file{cross.fs}
                   9730: @cindex cross-compiler
                   9731: @cindex metacompiler
                   9732: 
                   9733: You can also use @code{cross}, a batch compiler that accepts a Forth-like
                   9734: programming language. This @code{cross} language has to be documented
                   9735: yet.
                   9736: 
                   9737: @cindex target compiler
                   9738: @code{cross} also allows you to create image files for machines with
                   9739: different data sizes and data formats than the one used for generating
                   9740: the image file. You can also use it to create an application image that
                   9741: does not contain a Forth compiler. These features are bought with
                   9742: restrictions and inconveniences in programming. E.g., addresses have to
                   9743: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
                   9744: order to make the code relocatable.
                   9745: 
                   9746: 
                   9747: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
                   9748: @section Stack and Dictionary Sizes
                   9749: @cindex image file, stack and dictionary sizes
                   9750: @cindex dictionary size default
                   9751: @cindex stack size default
                   9752: 
                   9753: If you invoke Gforth with a command line flag for the size
                   9754: (@pxref{Invoking Gforth}), the size you specify is stored in the
                   9755: dictionary. If you save the dictionary with @code{savesystem} or create
                   9756: an image with @file{gforthmi}, this size will become the default
                   9757: for the resulting image file. E.g., the following will create a
1.21      crook    9758: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
1.1       anton    9759: 
                   9760: @example
                   9761: gforthmi gforth.fi -m 1M
                   9762: @end example
                   9763: 
                   9764: In other words, if you want to set the default size for the dictionary
                   9765: and the stacks of an image, just invoke @file{gforthmi} with the
                   9766: appropriate options when creating the image.
                   9767: 
                   9768: @cindex stack size, cache-friendly
                   9769: Note: For cache-friendly behaviour (i.e., good performance), you should
                   9770: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
                   9771: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
                   9772: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
                   9773: 
                   9774: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
                   9775: @section Running Image Files
                   9776: @cindex running image files
                   9777: @cindex invoking image files
                   9778: @cindex image file invocation
                   9779: 
                   9780: @cindex -i, invoke image file
                   9781: @cindex --image file, invoke image file
                   9782: You can invoke Gforth with an image file @var{image} instead of the
                   9783: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
                   9784: @example
                   9785: gforth -i @var{image}
                   9786: @end example
                   9787: 
                   9788: @cindex executable image file
1.26      crook    9789: @cindex image file, executable
1.1       anton    9790: If your operating system supports starting scripts with a line of the
                   9791: form @code{#! ...}, you just have to type the image file name to start
                   9792: Gforth with this image file (note that the file extension @code{.fi} is
                   9793: just a convention). I.e., to run Gforth with the image file @var{image},
                   9794: you can just type @var{image} instead of @code{gforth -i @var{image}}.
1.27      crook    9795: This works because every @code{.fi} file starts with a line of this
                   9796: format:
                   9797: 
                   9798: @example
                   9799: #! /usr/local/bin/gforth-0.4.0 -i
                   9800: @end example
                   9801: 
                   9802: The file and pathname for the Gforth engine specified on this line is
                   9803: the specific Gforth executable that it was built against; i.e. the value
                   9804: of the environment variable @code{GFORTH} at the time that
                   9805: @file{gforthmi} was executed.
1.1       anton    9806: 
1.27      crook    9807: You can make use of the same shell capability to make a Forth source
                   9808: file into an executable. For example, if you place this text in a file:
1.26      crook    9809: 
                   9810: @example
                   9811: #! /usr/local/bin/gforth
                   9812: 
                   9813: ." Hello, world" CR
                   9814: bye
                   9815: @end example
                   9816: 
                   9817: @noindent
1.27      crook    9818: and then make the file executable (chmod +x in Unix), you can run it
1.26      crook    9819: directly from the command line. The sequence @code{#!} is used in two
                   9820: ways; firstly, it is recognised as a ``magic sequence'' by the operating
                   9821: system, secondly it is treated as a comment character by Gforth. Because
                   9822: of the second usage, a space is required between @code{#!} and the path
                   9823: to the executable.
1.27      crook    9824: 
                   9825: The disadvantage of this latter technique, compared with using
                   9826: @file{gforthmi}, is that it is slower; the Forth source code is compiled
                   9827: on-the-fly, each time the program is invoked.
                   9828: 
1.26      crook    9829: @comment TODO describe the #! magic with reference to the Power Tools book.
                   9830: 
1.1       anton    9831: doc-#!
                   9832: 
                   9833: @node Modifying the Startup Sequence,  , Running Image Files, Image Files
                   9834: @section Modifying the Startup Sequence
                   9835: @cindex startup sequence for image file
                   9836: @cindex image file initialization sequence
                   9837: @cindex initialization sequence of image file
                   9838: 
                   9839: You can add your own initialization to the startup sequence through the
1.26      crook    9840: deferred word @code{'cold}. @code{'cold} is invoked just before the
                   9841: image-specific command line processing (by default, loading files and
                   9842: evaluating (@code{-e}) strings) starts.
1.1       anton    9843: 
                   9844: A sequence for adding your initialization usually looks like this:
                   9845: 
                   9846: @example
                   9847: :noname
                   9848:     Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
                   9849:     ... \ your stuff
                   9850: ; IS 'cold
                   9851: @end example
                   9852: 
                   9853: @cindex turnkey image files
1.26      crook    9854: @cindex image file, turnkey applications
1.1       anton    9855: You can make a turnkey image by letting @code{'cold} execute a word
                   9856: (your turnkey application) that never returns; instead, it exits Gforth
                   9857: via @code{bye} or @code{throw}.
                   9858: 
                   9859: @cindex command-line arguments, access
                   9860: @cindex arguments on the command line, access
                   9861: You can access the (image-specific) command-line arguments through the
1.26      crook    9862: variables @code{argc} and @code{argv}. @code{arg} provides convenient
1.1       anton    9863: access to @code{argv}.
                   9864: 
1.26      crook    9865: If @code{'cold} exits normally, Gforth processes the command-line
                   9866: arguments as files to be loaded and strings to be evaluated.  Therefore,
                   9867: @code{'cold} should remove the arguments it has used in this case.
                   9868: 
                   9869: doc-'cold
1.1       anton    9870: doc-argc
                   9871: doc-argv
                   9872: doc-arg
                   9873: 
                   9874: 
                   9875: @c ******************************************************************
1.13      pazsan   9876: @node Engine, Binding to System Library, Image Files, Top
1.1       anton    9877: @chapter Engine
                   9878: @cindex engine
                   9879: @cindex virtual machine
                   9880: 
1.26      crook    9881: Reading this chapter is not necessary for programming with Gforth. It
1.1       anton    9882: may be helpful for finding your way in the Gforth sources.
                   9883: 
                   9884: The ideas in this section have also been published in the papers
                   9885: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
                   9886: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
                   9887: Ertl, presented at EuroForth '93; the latter is available at
                   9888: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
                   9889: 
                   9890: @menu
                   9891: * Portability::                 
                   9892: * Threading::                   
                   9893: * Primitives::                  
                   9894: * Performance::                 
                   9895: @end menu
                   9896: 
                   9897: @node Portability, Threading, Engine, Engine
                   9898: @section Portability
                   9899: @cindex engine portability
                   9900: 
1.26      crook    9901: An important goal of the Gforth Project is availability across a wide
                   9902: range of personal machines. fig-Forth, and, to a lesser extent, F83,
                   9903: achieved this goal by manually coding the engine in assembly language
                   9904: for several then-popular processors. This approach is very
                   9905: labor-intensive and the results are short-lived due to progress in
                   9906: computer architecture.
1.1       anton    9907: 
                   9908: @cindex C, using C for the engine
                   9909: Others have avoided this problem by coding in C, e.g., Mitch Bradley
                   9910: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
                   9911: particularly popular for UNIX-based Forths due to the large variety of
                   9912: architectures of UNIX machines. Unfortunately an implementation in C
                   9913: does not mix well with the goals of efficiency and with using
                   9914: traditional techniques: Indirect or direct threading cannot be expressed
                   9915: in C, and switch threading, the fastest technique available in C, is
                   9916: significantly slower. Another problem with C is that it is very
                   9917: cumbersome to express double integer arithmetic.
                   9918: 
                   9919: @cindex GNU C for the engine
                   9920: @cindex long long
                   9921: Fortunately, there is a portable language that does not have these
                   9922: limitations: GNU C, the version of C processed by the GNU C compiler
                   9923: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
                   9924: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
                   9925: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
                   9926: threading possible, its @code{long long} type (@pxref{Long Long, ,
                   9927: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
                   9928: double numbers@footnote{Unfortunately, long longs are not implemented
                   9929: properly on all machines (e.g., on alpha-osf1, long longs are only 64
                   9930: bits, the same size as longs (and pointers), but they should be twice as
1.4       anton    9931: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
1.1       anton    9932: C Manual}). So, we had to implement doubles in C after all. Still, on
                   9933: most machines we can use long longs and achieve better performance than
                   9934: with the emulation package.}. GNU C is available for free on all
                   9935: important (and many unimportant) UNIX machines, VMS, 80386s running
                   9936: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
                   9937: on all these machines.
                   9938: 
                   9939: Writing in a portable language has the reputation of producing code that
                   9940: is slower than assembly. For our Forth engine we repeatedly looked at
                   9941: the code produced by the compiler and eliminated most compiler-induced
                   9942: inefficiencies by appropriate changes in the source code.
                   9943: 
                   9944: @cindex explicit register declarations
                   9945: @cindex --enable-force-reg, configuration flag
                   9946: @cindex -DFORCE_REG
                   9947: However, register allocation cannot be portably influenced by the
                   9948: programmer, leading to some inefficiencies on register-starved
                   9949: machines. We use explicit register declarations (@pxref{Explicit Reg
                   9950: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
                   9951: improve the speed on some machines. They are turned on by using the
                   9952: configuration flag @code{--enable-force-reg} (@code{gcc} switch
                   9953: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
                   9954: machine, but also on the compiler version: On some machines some
                   9955: compiler versions produce incorrect code when certain explicit register
                   9956: declarations are used. So by default @code{-DFORCE_REG} is not used.
                   9957: 
                   9958: @node Threading, Primitives, Portability, Engine
                   9959: @section Threading
                   9960: @cindex inner interpreter implementation
                   9961: @cindex threaded code implementation
                   9962: 
                   9963: @cindex labels as values
                   9964: GNU C's labels as values extension (available since @code{gcc-2.0},
                   9965: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
                   9966: makes it possible to take the address of @var{label} by writing
                   9967: @code{&&@var{label}}.  This address can then be used in a statement like
                   9968: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
                   9969: @code{goto x}.
                   9970: 
1.26      crook    9971: @cindex @code{NEXT}, indirect threaded
1.1       anton    9972: @cindex indirect threaded inner interpreter
                   9973: @cindex inner interpreter, indirect threaded
1.26      crook    9974: With this feature an indirect threaded @code{NEXT} looks like:
1.1       anton    9975: @example
                   9976: cfa = *ip++;
                   9977: ca = *cfa;
                   9978: goto *ca;
                   9979: @end example
                   9980: @cindex instruction pointer
                   9981: For those unfamiliar with the names: @code{ip} is the Forth instruction
                   9982: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
                   9983: execution token and points to the code field of the next word to be
                   9984: executed; The @code{ca} (code address) fetched from there points to some
                   9985: executable code, e.g., a primitive or the colon definition handler
                   9986: @code{docol}.
                   9987: 
1.26      crook    9988: @cindex @code{NEXT}, direct threaded
1.1       anton    9989: @cindex direct threaded inner interpreter
                   9990: @cindex inner interpreter, direct threaded
                   9991: Direct threading is even simpler:
                   9992: @example
                   9993: ca = *ip++;
                   9994: goto *ca;
                   9995: @end example
                   9996: 
                   9997: Of course we have packaged the whole thing neatly in macros called
1.26      crook    9998: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
1.1       anton    9999: 
                   10000: @menu
                   10001: * Scheduling::                  
                   10002: * Direct or Indirect Threaded?::  
                   10003: * DOES>::                       
                   10004: @end menu
                   10005: 
                   10006: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
                   10007: @subsection Scheduling
                   10008: @cindex inner interpreter optimization
                   10009: 
                   10010: There is a little complication: Pipelined and superscalar processors,
                   10011: i.e., RISC and some modern CISC machines can process independent
                   10012: instructions while waiting for the results of an instruction. The
                   10013: compiler usually reorders (schedules) the instructions in a way that
                   10014: achieves good usage of these delay slots. However, on our first tries
                   10015: the compiler did not do well on scheduling primitives. E.g., for
                   10016: @code{+} implemented as
                   10017: @example
                   10018: n=sp[0]+sp[1];
                   10019: sp++;
                   10020: sp[0]=n;
                   10021: NEXT;
                   10022: @end example
1.26      crook    10023: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
1.1       anton    10024: scheduling. After a little thought the problem becomes clear: The
1.21      crook    10025: compiler cannot know that @code{sp} and @code{ip} point to different
                   10026: addresses (and the version of @code{gcc} we used would not know it even
                   10027: if it was possible), so it could not move the load of the cfa above the
                   10028: store to the TOS. Indeed the pointers could be the same, if code on or
                   10029: very near the top of stack were executed. In the interest of speed we
                   10030: chose to forbid this probably unused ``feature'' and helped the compiler
1.26      crook    10031: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
1.21      crook    10032: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.1       anton    10033: @example
                   10034: n=sp[0]+sp[1];
                   10035: sp++;
                   10036: NEXT_P1;
                   10037: sp[0]=n;
                   10038: NEXT_P2;
                   10039: @end example
                   10040: This can be scheduled optimally by the compiler.
                   10041: 
                   10042: This division can be turned off with the switch @code{-DCISC_NEXT}. This
                   10043: switch is on by default on machines that do not profit from scheduling
                   10044: (e.g., the 80386), in order to preserve registers.
                   10045: 
                   10046: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
                   10047: @subsection Direct or Indirect Threaded?
                   10048: @cindex threading, direct or indirect?
                   10049: 
                   10050: @cindex -DDIRECT_THREADED
                   10051: Both! After packaging the nasty details in macro definitions we
                   10052: realized that we could switch between direct and indirect threading by
                   10053: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
                   10054: defining a few machine-specific macros for the direct-threading case.
                   10055: On the Forth level we also offer access words that hide the
                   10056: differences between the threading methods (@pxref{Threading Words}).
                   10057: 
                   10058: Indirect threading is implemented completely machine-independently.
                   10059: Direct threading needs routines for creating jumps to the executable
1.21      crook    10060: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
                   10061: machine-dependent, but they do not amount to many source lines. Therefore,
                   10062: even porting direct threading to a new machine requires little effort.
1.1       anton    10063: 
                   10064: @cindex --enable-indirect-threaded, configuration flag
                   10065: @cindex --enable-direct-threaded, configuration flag
                   10066: The default threading method is machine-dependent. You can enforce a
                   10067: specific threading method when building Gforth with the configuration
                   10068: flag @code{--enable-direct-threaded} or
                   10069: @code{--enable-indirect-threaded}. Note that direct threading is not
                   10070: supported on all machines.
                   10071: 
                   10072: @node DOES>,  , Direct or Indirect Threaded?, Threading
                   10073: @subsection DOES>
                   10074: @cindex @code{DOES>} implementation
                   10075: 
1.26      crook    10076: @cindex @code{dodoes} routine
                   10077: @cindex @code{DOES>}-code
1.1       anton    10078: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
                   10079: the chunk of code executed by every word defined by a
                   10080: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
                   10081: the Forth code to be executed, i.e. the code after the
1.26      crook    10082: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
1.1       anton    10083: 
1.21      crook    10084: In fig-Forth the code field points directly to the @code{dodoes} and the
1.26      crook    10085: @code{DOES>}code address is stored in the cell after the code address (i.e. at
                   10086: @code{@var{CFA} cell+}). It may seem that this solution is illegal in
1.1       anton    10087: the Forth-79 and all later standards, because in fig-Forth this address
                   10088: lies in the body (which is illegal in these standards). However, by
                   10089: making the code field larger for all words this solution becomes legal
                   10090: again. We use this approach for the indirect threaded version and for
                   10091: direct threading on some machines. Leaving a cell unused in most words
                   10092: is a bit wasteful, but on the machines we are targeting this is hardly a
                   10093: problem. The other reason for having a code field size of two cells is
                   10094: to avoid having different image files for direct and indirect threaded
                   10095: systems (direct threaded systems require two-cell code fields on many
                   10096: machines).
                   10097: 
1.26      crook    10098: @cindex @code{DOES>}-handler
1.1       anton    10099: The other approach is that the code field points or jumps to the cell
1.26      crook    10100: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
                   10101: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
                   10102: @code{DOES>}-code address by computing the code address, i.e., the address of
1.1       anton    10103: the jump to dodoes, and add the length of that jump field. A variant of
                   10104: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
                   10105: return address (which can be found in the return register on RISCs) is
1.26      crook    10106: the @code{DOES>}-code address. Since the two cells available in the code field
1.1       anton    10107: are used up by the jump to the code address in direct threading on many
                   10108: architectures, we use this approach for direct threading on these
                   10109: architectures. We did not want to add another cell to the code field.
                   10110: 
                   10111: @node Primitives, Performance, Threading, Engine
                   10112: @section Primitives
                   10113: @cindex primitives, implementation
                   10114: @cindex virtual machine instructions, implementation
                   10115: 
                   10116: @menu
                   10117: * Automatic Generation::        
                   10118: * TOS Optimization::            
                   10119: * Produced code::               
                   10120: @end menu
                   10121: 
                   10122: @node Automatic Generation, TOS Optimization, Primitives, Primitives
                   10123: @subsection Automatic Generation
                   10124: @cindex primitives, automatic generation
                   10125: 
                   10126: @cindex @file{prims2x.fs}
                   10127: Since the primitives are implemented in a portable language, there is no
                   10128: longer any need to minimize the number of primitives. On the contrary,
                   10129: having many primitives has an advantage: speed. In order to reduce the
                   10130: number of errors in primitives and to make programming them easier, we
                   10131: provide a tool, the primitive generator (@file{prims2x.fs}), that
                   10132: automatically generates most (and sometimes all) of the C code for a
                   10133: primitive from the stack effect notation.  The source for a primitive
                   10134: has the following form:
                   10135: 
                   10136: @cindex primitive source format
                   10137: @format
                   10138: @var{Forth-name}       @var{stack-effect}      @var{category}  [@var{pronounc.}]
                   10139: [@code{""}@var{glossary entry}@code{""}]
                   10140: @var{C code}
                   10141: [@code{:}
                   10142: @var{Forth code}]
                   10143: @end format
                   10144: 
                   10145: The items in brackets are optional. The category and glossary fields
                   10146: are there for generating the documentation, the Forth code is there
                   10147: for manual implementations on machines without GNU C. E.g., the source
                   10148: for the primitive @code{+} is:
                   10149: @example
                   10150: +    n1 n2 -- n    core    plus
                   10151: n = n1+n2;
                   10152: @end example
                   10153: 
                   10154: This looks like a specification, but in fact @code{n = n1+n2} is C
                   10155: code. Our primitive generation tool extracts a lot of information from
                   10156: the stack effect notations@footnote{We use a one-stack notation, even
                   10157: though we have separate data and floating-point stacks; The separate
                   10158: notation can be generated easily from the unified notation.}: The number
                   10159: of items popped from and pushed on the stack, their type, and by what
                   10160: name they are referred to in the C code. It then generates a C code
                   10161: prelude and postlude for each primitive. The final C code for @code{+}
                   10162: looks like this:
                   10163: 
                   10164: @example
                   10165: I_plus:        /* + ( n1 n2 -- n ) */  /* label, stack effect */
                   10166: /*  */                          /* documentation */
                   10167: @{
                   10168: DEF_CA                          /* definition of variable ca (indirect threading) */
                   10169: Cell n1;                        /* definitions of variables */
                   10170: Cell n2;
                   10171: Cell n;
                   10172: n1 = (Cell) sp[1];              /* input */
                   10173: n2 = (Cell) TOS;
                   10174: sp += 1;                        /* stack adjustment */
                   10175: NAME("+")                       /* debugging output (with -DDEBUG) */
                   10176: @{
                   10177: n = n1+n2;                      /* C code taken from the source */
                   10178: @}
                   10179: NEXT_P1;                        /* NEXT part 1 */
                   10180: TOS = (Cell)n;                  /* output */
                   10181: NEXT_P2;                        /* NEXT part 2 */
                   10182: @}
                   10183: @end example
                   10184: 
                   10185: This looks long and inefficient, but the GNU C compiler optimizes quite
                   10186: well and produces optimal code for @code{+} on, e.g., the R3000 and the
                   10187: HP RISC machines: Defining the @code{n}s does not produce any code, and
                   10188: using them as intermediate storage also adds no cost.
                   10189: 
1.26      crook    10190: There are also other optimizations that are not illustrated by this
                   10191: example: assignments between simple variables are usually for free (copy
1.1       anton    10192: propagation). If one of the stack items is not used by the primitive
                   10193: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
                   10194: (dead code elimination). On the other hand, there are some things that
                   10195: the compiler does not do, therefore they are performed by
                   10196: @file{prims2x.fs}: The compiler does not optimize code away that stores
                   10197: a stack item to the place where it just came from (e.g., @code{over}).
                   10198: 
                   10199: While programming a primitive is usually easy, there are a few cases
                   10200: where the programmer has to take the actions of the generator into
                   10201: account, most notably @code{?dup}, but also words that do not (always)
1.26      crook    10202: fall through to @code{NEXT}.
1.1       anton    10203: 
                   10204: @node TOS Optimization, Produced code, Automatic Generation, Primitives
                   10205: @subsection TOS Optimization
                   10206: @cindex TOS optimization for primitives
                   10207: @cindex primitives, keeping the TOS in a register
                   10208: 
                   10209: An important optimization for stack machine emulators, e.g., Forth
                   10210: engines, is keeping  one or more of the top stack items in
                   10211: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
                   10212: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
                   10213: @itemize @bullet
                   10214: @item
                   10215: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
                   10216: due to fewer loads from and stores to the stack.
                   10217: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
                   10218: @var{y<n}, due to additional moves between registers.
                   10219: @end itemize
                   10220: 
                   10221: @cindex -DUSE_TOS
                   10222: @cindex -DUSE_NO_TOS
                   10223: In particular, keeping one item in a register is never a disadvantage,
                   10224: if there are enough registers. Keeping two items in registers is a
                   10225: disadvantage for frequent words like @code{?branch}, constants,
                   10226: variables, literals and @code{i}. Therefore our generator only produces
                   10227: code that keeps zero or one items in registers. The generated C code
                   10228: covers both cases; the selection between these alternatives is made at
                   10229: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
                   10230: code for @code{+} is just a simple variable name in the one-item case,
                   10231: otherwise it is a macro that expands into @code{sp[0]}. Note that the
                   10232: GNU C compiler tries to keep simple variables like @code{TOS} in
                   10233: registers, and it usually succeeds, if there are enough registers.
                   10234: 
                   10235: @cindex -DUSE_FTOS
                   10236: @cindex -DUSE_NO_FTOS
                   10237: The primitive generator performs the TOS optimization for the
                   10238: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
                   10239: operations the benefit of this optimization is even larger:
                   10240: floating-point operations take quite long on most processors, but can be
                   10241: performed in parallel with other operations as long as their results are
                   10242: not used. If the FP-TOS is kept in a register, this works. If
                   10243: it is kept on the stack, i.e., in memory, the store into memory has to
                   10244: wait for the result of the floating-point operation, lengthening the
                   10245: execution time of the primitive considerably.
                   10246: 
                   10247: The TOS optimization makes the automatic generation of primitives a
                   10248: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
                   10249: @code{TOS} is not sufficient. There are some special cases to
                   10250: consider:
                   10251: @itemize @bullet
                   10252: @item In the case of @code{dup ( w -- w w )} the generator must not
                   10253: eliminate the store to the original location of the item on the stack,
                   10254: if the TOS optimization is turned on.
                   10255: @item Primitives with stack effects of the form @code{--}
                   10256: @var{out1}...@var{outy} must store the TOS to the stack at the start.
                   10257: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
                   10258: must load the TOS from the stack at the end. But for the null stack
                   10259: effect @code{--} no stores or loads should be generated.
                   10260: @end itemize
                   10261: 
                   10262: @node Produced code,  , TOS Optimization, Primitives
                   10263: @subsection Produced code
                   10264: @cindex primitives, assembly code listing
                   10265: 
                   10266: @cindex @file{engine.s}
                   10267: To see what assembly code is produced for the primitives on your machine
                   10268: with your compiler and your flag settings, type @code{make engine.s} and
                   10269: look at the resulting file @file{engine.s}.
                   10270: 
                   10271: @node  Performance,  , Primitives, Engine
                   10272: @section Performance
                   10273: @cindex performance of some Forth interpreters
                   10274: @cindex engine performance
                   10275: @cindex benchmarking Forth systems
                   10276: @cindex Gforth performance
                   10277: 
                   10278: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
                   10279: impossible to write a significantly faster engine.
                   10280: 
                   10281: On register-starved machines like the 386 architecture processors
                   10282: improvements are possible, because @code{gcc} does not utilize the
                   10283: registers as well as a human, even with explicit register declarations;
                   10284: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
                   10285: and hand-tuned it for the 486; this system is 1.19 times faster on the
                   10286: Sieve benchmark on a 486DX2/66 than Gforth compiled with
                   10287: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
                   10288: 
                   10289: @cindex Win32Forth performance
                   10290: @cindex NT Forth performance
                   10291: @cindex eforth performance
                   10292: @cindex ThisForth performance
                   10293: @cindex PFE performance
                   10294: @cindex TILE performance
                   10295: However, this potential advantage of assembly language implementations
                   10296: is not necessarily realized in complete Forth systems: We compared
                   10297: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
                   10298: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
                   10299: 1994) and Eforth (with and without peephole (aka pinhole) optimization
                   10300: of the threaded code); all these systems were written in assembly
                   10301: language. We also compared Gforth with three systems written in C:
                   10302: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
                   10303: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
1.21      crook    10304: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
                   10305: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
1.1       anton    10306: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
                   10307: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
                   10308: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
                   10309: 486DX2/66 with similar memory performance under Windows NT. Marcel
                   10310: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
                   10311: added the peephole optimizer, ran the benchmarks and reported the
                   10312: results.
                   10313:  
                   10314: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
                   10315: matrix multiplication come from the Stanford integer benchmarks and have
                   10316: been translated into Forth by Martin Fraeman; we used the versions
                   10317: included in the TILE Forth package, but with bigger data set sizes; and
                   10318: a recursive Fibonacci number computation for benchmarking calling
                   10319: performance. The following table shows the time taken for the benchmarks
                   10320: scaled by the time taken by Gforth (in other words, it shows the speedup
                   10321: factor that Gforth achieved over the other systems).
                   10322: 
                   10323: @example
                   10324: relative      Win32-    NT       eforth       This-
                   10325:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
                   10326: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
                   10327: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
                   10328: matmul    1.00  1.47  1.35   1.46  0.74  1.58        4.09
                   10329: fib       1.00  1.52  1.34   1.22  0.86  1.74  2.99  4.30
                   10330: @end example
                   10331: 
1.26      crook    10332: You may be quite surprised by the good performance of Gforth when
                   10333: compared with systems written in assembly language. One important reason
                   10334: for the disappointing performance of these other systems is probably
                   10335: that they are not written optimally for the 486 (e.g., they use the
                   10336: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
                   10337: but costly method for relocating the Forth image: like @code{cforth}, it
                   10338: computes the actual addresses at run time, resulting in two address
                   10339: computations per @code{NEXT} (@pxref{Image File Background}).
                   10340: 
                   10341: Only Eforth with the peephole optimizer has a performance that is
                   10342: comparable to Gforth. The speedups achieved with peephole optimization
                   10343: of threaded code are quite remarkable. Adding a peephole optimizer to
                   10344: Gforth should cause similar speedups.
1.1       anton    10345: 
                   10346: The speedup of Gforth over PFE, ThisForth and TILE can be easily
                   10347: explained with the self-imposed restriction of the latter systems to
                   10348: standard C, which makes efficient threading impossible (however, the
1.4       anton    10349: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
1.1       anton    10350: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
                   10351: Moreover, current C compilers have a hard time optimizing other aspects
                   10352: of the ThisForth and the TILE source.
                   10353: 
1.26      crook    10354: The performance of Gforth on 386 architecture processors varies widely
                   10355: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
                   10356: allocate any of the virtual machine registers into real machine
                   10357: registers by itself and would not work correctly with explicit register
                   10358: declarations, giving a 1.3 times slower engine (on a 486DX2/66 running
                   10359: the Sieve) than the one measured above.
1.1       anton    10360: 
1.26      crook    10361: Note that there have been several releases of Win32Forth since the
                   10362: release presented here, so the results presented above may have little
1.1       anton    10363: predictive value for the performance of Win32Forth today.
                   10364: 
                   10365: @cindex @file{Benchres}
                   10366: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
                   10367: Maierhofer (presented at EuroForth '95), an indirect threaded version of
                   10368: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
                   10369: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
                   10370: threaded version used here. The paper available at
                   10371: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
                   10372: it also contains numbers for some native code systems. You can find a
                   10373: newer version of these measurements at
                   10374: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
                   10375: find numbers for Gforth on various machines in @file{Benchres}.
                   10376: 
1.26      crook    10377: @c ******************************************************************
1.13      pazsan   10378: @node Binding to System Library, Cross Compiler, Engine, Top
1.14      pazsan   10379: @chapter Binding to System Library
1.13      pazsan   10380: 
                   10381: @node Cross Compiler, Bugs, Binding to System Library, Top
1.14      pazsan   10382: @chapter Cross Compiler
1.13      pazsan   10383: 
                   10384: Cross Compiler
                   10385: 
                   10386: @menu
                   10387: * Using the Cross Compiler::
                   10388: * How the Cross Compiler Works::
                   10389: @end menu
                   10390: 
1.21      crook    10391: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
1.14      pazsan   10392: @section Using the Cross Compiler
1.13      pazsan   10393: 
1.21      crook    10394: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
1.14      pazsan   10395: @section How the Cross Compiler Works
1.13      pazsan   10396: 
                   10397: @node Bugs, Origin, Cross Compiler, Top
1.21      crook    10398: @appendix Bugs
1.1       anton    10399: @cindex bug reporting
                   10400: 
1.21      crook    10401: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
1.1       anton    10402: 
                   10403: If you find a bug, please send a bug report to
1.21      crook    10404: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should include this
                   10405: information:
                   10406: 
                   10407: @itemize @bullet
                   10408: @item
                   10409: The Gforth version used (it is announced at the start of an
                   10410: interactive Gforth session).
                   10411: @item
                   10412: The machine and operating system (on Unix
                   10413: systems @code{uname -a} will report this information).
                   10414: @item
                   10415: The installation options (send the file @file{config.status}).
                   10416: @item
                   10417: A complete list of changes (if any) you (or your installer) have made to the
                   10418: Gforth sources.
                   10419: @item
                   10420: A program (or a sequence of keyboard commands) that reproduces the bug.
                   10421: @item
                   10422: A description of what you think constitutes the buggy behaviour.
                   10423: @end itemize
1.1       anton    10424: 
                   10425: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
                   10426: to Report Bugs, gcc.info, GNU C Manual}.
                   10427: 
                   10428: 
1.21      crook    10429: @node Origin, Forth-related information, Bugs, Top
                   10430: @appendix Authors and Ancestors of Gforth
1.1       anton    10431: 
                   10432: @section Authors and Contributors
                   10433: @cindex authors of Gforth
                   10434: @cindex contributors to Gforth
                   10435: 
                   10436: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
                   10437: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
                   10438: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
                   10439: with their continuous feedback. Lennart Benshop contributed
                   10440: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
                   10441: support for calling C libraries. Helpful comments also came from Paul
                   10442: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.12      anton    10443: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
                   10444: release of Gforth-0.2.1 there were also helpful comments from many
                   10445: others; thank you all, sorry for not listing you here (but digging
1.23      crook    10446: through my mailbox to extract your names is on my to-do list). Since the
                   10447: release of Gforth-0.4.0 Neal Crook worked on the manual.
1.1       anton    10448: 
                   10449: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
                   10450: and autoconf, among others), and to the creators of the Internet: Gforth
1.21      crook    10451: was developed across the Internet, and its authors did not meet
1.20      pazsan   10452: physically for the first 4 years of development.
1.1       anton    10453: 
                   10454: @section Pedigree
1.26      crook    10455: @cindex pedigree of Gforth
1.1       anton    10456: 
1.20      pazsan   10457: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
1.1       anton    10458: Dirk Zoller) will cross-fertilize each other. Of course, a significant
                   10459: part of the design of Gforth was prescribed by ANS Forth.
                   10460: 
1.20      pazsan   10461: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
1.1       anton    10462: 32 bit native code version of VolksForth for the Atari ST, written
                   10463: mostly by Dietrich Weineck.
                   10464: 
                   10465: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
                   10466: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
                   10467: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
                   10468: 
                   10469: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
                   10470: Forth-83 standard. !! Pedigree? When?
                   10471: 
                   10472: A team led by Bill Ragsdale implemented fig-Forth on many processors in
                   10473: 1979. Robert Selzer and Bill Ragsdale developed the original
                   10474: implementation of fig-Forth for the 6502 based on microForth.
                   10475: 
                   10476: The principal architect of microForth was Dean Sanderson. microForth was
                   10477: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
                   10478: the 1802, and subsequently implemented on the 8080, the 6800 and the
                   10479: Z80.
                   10480: 
                   10481: All earlier Forth systems were custom-made, usually by Charles Moore,
                   10482: who discovered (as he puts it) Forth during the late 60s. The first full
                   10483: Forth existed in 1971.
                   10484: 
                   10485: A part of the information in this section comes from @cite{The Evolution
                   10486: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
                   10487: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
                   10488: Notices 28(3), 1993.  You can find more historical and genealogical
                   10489: information about Forth there.
                   10490: 
1.21      crook    10491: @node Forth-related information, Word Index, Origin, Top
                   10492: @appendix Other Forth-related information
                   10493: @cindex Forth-related information
                   10494: 
                   10495: @menu
                   10496: * Internet resources::
                   10497: * Books::
                   10498: * The Forth Interest Group::
                   10499: * Conferences::
                   10500: @end menu
                   10501: 
                   10502: 
                   10503: @node Internet resources, Books, Forth-related information, Forth-related information
                   10504: @section Internet resources
1.26      crook    10505: @cindex internet resources
1.21      crook    10506: 
                   10507: @cindex comp.lang.forth
                   10508: @cindex frequently asked questions
                   10509: There is an active newsgroup (comp.lang.forth) discussing Forth and
                   10510: Forth-related issues. A frequently-asked-questions (FAQ) list
                   10511: is posted to the newsgroup regulary, and archived at these sites:
                   10512: 
                   10513: @itemize @bullet
                   10514: @item
                   10515: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
                   10516: @item
                   10517: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
                   10518: @end itemize
                   10519: 
                   10520: The FAQ list should be considered mandatory reading before posting to
                   10521: the newsgroup.
                   10522: 
                   10523: Here are some other web sites holding Forth-related material:
                   10524: 
                   10525: @itemize @bullet
                   10526: @item
                   10527: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
                   10528: @item
                   10529: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
                   10530: @item
                   10531: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
                   10532: @item
                   10533: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
                   10534: Research page, including links to the Journal of Forth Application and
                   10535: Research (JFAR) and a searchable Forth bibliography.
                   10536: @end itemize
                   10537: 
                   10538: 
                   10539: @node Books, The Forth Interest Group, Internet resources, Forth-related information
                   10540: @section Books
1.26      crook    10541: @cindex books on Forth
1.21      crook    10542: 
                   10543: As the Standard is relatively new, there are not many books out yet. It
                   10544: is not recommended to learn Forth by using Gforth and a book that is not
                   10545: written for ANS Forth, as you will not know your mistakes from the
                   10546: deviations of the book. However, books based on the Forth-83 standard
                   10547: should be ok, because ANS Forth is primarily an extension of Forth-83.
                   10548: 
                   10549: @cindex standard document for ANS Forth
                   10550: @cindex ANS Forth document
                   10551: The definite reference if you want to write ANS Forth programs is, of
1.26      crook    10552: course, the ANS Forth document. It is available in printed form from the
1.21      crook    10553: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
                   10554: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
                   10555: $200. You can also get it from Global Engineering Documents (Tel.: USA
                   10556: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
                   10557: 
                   10558: @cite{dpANS6}, the last draft of the standard, which was then submitted
                   10559: to ANSI for publication is available electronically and for free in some
                   10560: MS Word format, and it has been converted to HTML
                   10561: (@url{http://www.taygeta.com/forth/dpans.html}; this is my favourite
                   10562: format); this HTML version also includes the answers to Requests for
                   10563: Interpretation (RFIs). Some pointers to these versions can be found
                   10564: through @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
                   10565: 
1.26      crook    10566: @cindex introductory book on Forth
                   10567: @cindex book on Forth, introductory
1.21      crook    10568: @cindex Woehr, Jack: @cite{Forth: The New Model}
                   10569: @cindex @cite{Forth: The new model} (book)
                   10570: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
                   10571: introductory book based on a draft version of the standard. It does not
                   10572: cover the whole standard. It also contains interesting background
                   10573: information (Jack Woehr was in the ANS Forth Technical Committee). It is
                   10574: not appropriate for complete newbies, but programmers experienced in
                   10575: other languages should find it ok.
                   10576: 
                   10577: @cindex Conklin, Edward K., and Elizabeth Rather: @cite{Forth Programmer's Handbook}
                   10578: @cindex Rather, Elizabeth and Edward K. Conklin: @cite{Forth Programmer's Handbook}
                   10579: @cindex @cite{Forth Programmer's Handbook} (book)
                   10580: @cite{Forth Programmer's Handbook} by Edward K. Conklin, Elizabeth
                   10581: D. Rather and the technical staff of Forth, Inc. (Forth, Inc., 1997;
                   10582: ISBN 0-9662156-0-5) contains little introductory material. The majority
                   10583: of the book is similar to @ref{Words}, but the book covers most of the
                   10584: standard words and some non-standard words (whereas this manual is
                   10585: quite incomplete). In addition, the book contains a chapter on
                   10586: programming style. The major drawback of this book is that it usually
                   10587: does not identify what is standard and what is specific to the Forth
                   10588: system described in the book (probably one of Forth, Inc.'s systems).
                   10589: Fortunately, many of the non-standard programming practices described in
                   10590: the book work in Gforth, too.  Still, this drawback makes the book
                   10591: hardly more useful than a pre-ANS book.
                   10592: 
                   10593: @node The Forth Interest Group, Conferences, Books, Forth-related information
                   10594: @section The Forth Interest Group
                   10595: @cindex Forth interest group (FIG)
                   10596: 
                   10597: The Forth Interest Group (FIG) is a world-wide, non-profit,
1.26      crook    10598: member-supported organisation. It publishes a regular magazine,
                   10599: @var{FORTH Dimensions}, and offers other benefits of membership. You can
                   10600: contact the FIG through their office email address:
                   10601: @email{office@@forth.org} or by visiting their web site at
                   10602: @url{http://www.forth.org/}. This web site also includes links to FIG
                   10603: chapters in other countries and American cities
1.21      crook    10604: (@url{http://www.forth.org/chapters.html}).
                   10605: 
                   10606: @node Conferences, , The Forth Interest Group, Forth-related information
                   10607: @section Conferences
                   10608: @cindex Conferences
                   10609: 
                   10610: There are several regular conferences related to Forth. They are all
1.26      crook    10611: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
                   10612: news group:
1.21      crook    10613: 
                   10614: @itemize @bullet
                   10615: @item
                   10616: FORML -- the Forth modification laboratory convenes every year near
                   10617: Monterey, California.
                   10618: @item
                   10619: The Rochester Forth Conference -- an annual conference traditionally
                   10620: held in Rochester, New York.
                   10621: @item
                   10622: EuroForth -- this European conference takes place annually.
                   10623: @end itemize
                   10624: 
                   10625: 
                   10626: @node Word Index, Concept Index, Forth-related information, Top
1.1       anton    10627: @unnumbered Word Index
                   10628: 
1.26      crook    10629: This index is a list of Forth words that have ``glossary'' entries
                   10630: within this manual. Each word is listed with its stack effect and
                   10631: wordset.
1.1       anton    10632: 
                   10633: @printindex fn
                   10634: 
                   10635: @node Concept Index,  , Word Index, Top
                   10636: @unnumbered Concept and Word Index
                   10637: 
1.26      crook    10638: Not all entries listed in this index are present verbatim in the
                   10639: text. This index also duplicates, in abbreviated form, all of the words
                   10640: listed in the Word Index (only the names are listed for the words here).
1.1       anton    10641: 
                   10642: @printindex cp
                   10643: 
                   10644: @contents
                   10645: @bye
                   10646: 

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>