File:  [gforth] / gforth / doc / gforth.ds
Revision 1.88: download - view: text, annotated - select for diffs
Sat Sep 23 15:47:05 2000 UTC (21 years, 2 months ago) by anton
Branches: MAIN
CVS tags: HEAD
changed FSF address in copyright messages

    1: \input texinfo   @c -*-texinfo-*-
    2: @comment The source is gforth.ds, from which gforth.texi is generated
    3: 
    4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
    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: @c          Not an improvement IMO - anton
   12: @c          and anyway, this should be taken up
   13: @c          with Karl Berry (the texinfo guy) - anton
   14: @comment .. would be useful to have a word that identified all deferred words
   15: @comment should semantics stuff in intro be moved to another section
   16: 
   17: @c POSTPONE, COMPILE, [COMPILE], LITERAL should have their own section
   18: 
   19: @comment %**start of header (This is for running Texinfo on a region.)
   20: @setfilename gforth.info
   21: @settitle Gforth Manual
   22: @dircategory GNU programming tools
   23: @direntry
   24: * Gforth: (gforth).             A fast interpreter for the Forth language.
   25: @end direntry
   26: @c The Texinfo manual also recommends doing this, but for Gforth it may
   27: @c  not make much sense
   28: @c @dircategory Individual utilities
   29: @c @direntry
   30: @c * Gforth: (gforth)Invoking Gforth.      gforth, gforth-fast, gforthmi
   31: @c @end direntry
   32: 
   33: @comment @setchapternewpage odd
   34: @comment TODO this gets left in by HTML converter
   35: @macro progstyle {}
   36: Programming style note:
   37: @end macro
   38: 
   39: @macro assignment {}
   40: @table @i
   41: @item Assignment:
   42: @end macro
   43: @macro endassignment {}
   44: @end table
   45: @end macro
   46: 
   47: @comment %**end of header (This is for running Texinfo on a region.)
   48: 
   49: 
   50: @comment ----------------------------------------------------------
   51: @comment macros for beautifying glossary entries
   52: @comment if these are used, need to strip them out for HTML converter
   53: @comment else they get repeated verbatim in HTML output.
   54: @comment .. not working yet.
   55: 
   56: @macro GLOSS-START {}
   57: @iftex
   58: @ninerm
   59: @end iftex
   60: @end macro
   61: 
   62: @macro GLOSS-END {}
   63: @iftex
   64: @rm
   65: @end iftex
   66: @end macro
   67: 
   68: @comment ----------------------------------------------------------
   69: 
   70: 
   71: @include version.texi
   72: 
   73: @ifnottex
   74: This file documents Gforth @value{VERSION}
   75: 
   76: Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
   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: @ignore
   83:      Permission is granted to process this file through TeX and print the
   84:      results, provided the printed document carries a copying permission
   85:      notice identical to this one except for the removal of this paragraph
   86:      (this paragraph not being relevant to the printed manual).
   87:      
   88: @end ignore
   89:      Permission is granted to copy and distribute modified versions of this
   90:      manual under the conditions for verbatim copying, provided also that the
   91:      sections entitled "Distribution" and "General Public License" are
   92:      included exactly as in the original, and provided that the entire
   93:      resulting derived work is distributed under the terms of a permission
   94:      notice identical to this one.
   95:      
   96:      Permission is granted to copy and distribute translations of this manual
   97:      into another language, under the above conditions for modified versions,
   98:      except that the sections entitled "Distribution" and "General Public
   99:      License" may be included in a translation approved by the author instead
  100:      of in the original English.
  101: @end ifnottex
  102: 
  103: @finalout
  104: @titlepage
  105: @sp 10
  106: @center @titlefont{Gforth Manual}
  107: @sp 2
  108: @center for version @value{VERSION}
  109: @sp 2
  110: @center Neal Crook
  111: @center Anton Ertl
  112: @center Bernd Paysan
  113: @center Jens Wilke
  114: @sp 3
  115: @center This manual is permanently under construction and was last updated on 15-Mar-2000
  116: 
  117: @comment  The following two commands start the copyright page.
  118: @page
  119: @vskip 0pt plus 1filll
  120: Copyright @copyright{} 1995--2000 Free Software Foundation, Inc.
  121: 
  122: @comment !! Published by ... or You can get a copy of this manual ...
  123: 
  124:      Permission is granted to make and distribute verbatim copies of
  125:      this manual provided the copyright notice and this permission notice
  126:      are preserved on all copies.
  127:      
  128:      Permission is granted to copy and distribute modified versions of this
  129:      manual under the conditions for verbatim copying, provided also that the
  130:      sections entitled "Distribution" and "General Public License" are
  131:      included exactly as in the original, and provided that the entire
  132:      resulting derived work is distributed under the terms of a permission
  133:      notice identical to this one.
  134:      
  135:      Permission is granted to copy and distribute translations of this manual
  136:      into another language, under the above conditions for modified versions,
  137:      except that the sections entitled "Distribution" and "General Public
  138:      License" may be included in a translation approved by the author instead
  139:      of in the original English.
  140: @end titlepage
  141: 
  142: @node Top, License, (dir), (dir)
  143: @ifnottex
  144: Gforth is a free implementation of ANS Forth available on many
  145: personal machines. This manual corresponds to version @value{VERSION}.
  146: @end ifnottex
  147: 
  148: @menu
  149: * License::                     The GPL
  150: * Goals::                       About the Gforth Project
  151: * Gforth Environment::          Starting (and exiting) Gforth
  152: * Tutorial::                    Hands-on Forth Tutorial
  153: * Introduction::                An introduction to ANS Forth
  154: * Words::                       Forth words available in Gforth
  155: * Error messages::              How to interpret them
  156: * Tools::                       Programming tools
  157: * ANS conformance::             Implementation-defined options etc.
  158: * Standard vs Extensions::      Should I use extensions?
  159: * Model::                       The abstract machine of Gforth
  160: * Integrating Gforth::          Forth as scripting language for applications
  161: * Emacs and Gforth::            The Gforth Mode
  162: * Image Files::                 @code{.fi} files contain compiled code
  163: * Engine::                      The inner interpreter and the primitives
  164: * Binding to System Library::   
  165: * Cross Compiler::              The Cross Compiler
  166: * Bugs::                        How to report them
  167: * Origin::                      Authors and ancestors of Gforth
  168: * Forth-related information::   Books and places to look on the WWW
  169: * Word Index::                  An item for each Forth word
  170: * Concept Index::               A menu covering many topics
  171: 
  172: @detailmenu --- The Detailed Node Listing ---
  173: 
  174: Gforth Environment
  175: 
  176: * Invoking Gforth::             Getting in
  177: * Leaving Gforth::              Getting out
  178: * Command-line editing::        
  179: * Environment variables::       that affect how Gforth starts up
  180: * Gforth Files::                What gets installed and where
  181: * Startup speed::               When 35ms is not fast enough ...
  182: 
  183: Forth Tutorial
  184: 
  185: * Starting Gforth Tutorial::    
  186: * Syntax Tutorial::             
  187: * Crash Course Tutorial::       
  188: * Stack Tutorial::              
  189: * Arithmetics Tutorial::        
  190: * Stack Manipulation Tutorial::  
  191: * Using files for Forth code Tutorial::  
  192: * Comments Tutorial::           
  193: * Colon Definitions Tutorial::  
  194: * Decompilation Tutorial::      
  195: * Stack-Effect Comments Tutorial::  
  196: * Types Tutorial::              
  197: * Factoring Tutorial::          
  198: * Designing the stack effect Tutorial::  
  199: * Local Variables Tutorial::    
  200: * Conditional execution Tutorial::  
  201: * Flags and Comparisons Tutorial::  
  202: * General Loops Tutorial::      
  203: * Counted loops Tutorial::      
  204: * Recursion Tutorial::          
  205: * Leaving definitions or loops Tutorial::  
  206: * Return Stack Tutorial::       
  207: * Memory Tutorial::             
  208: * Characters and Strings Tutorial::  
  209: * Alignment Tutorial::          
  210: * Files Tutorial::              
  211: * Interpretation and Compilation Semantics and Immediacy Tutorial::  
  212: * Execution Tokens Tutorial::   
  213: * Exceptions Tutorial::         
  214: * Defining Words Tutorial::     
  215: * Arrays and Records Tutorial::  
  216: * POSTPONE Tutorial::           
  217: * Literal Tutorial::            
  218: * Advanced macros Tutorial::    
  219: * Compilation Tokens Tutorial::  
  220: * Wordlists and Search Order Tutorial::  
  221: 
  222: An Introduction to ANS Forth
  223: 
  224: * Introducing the Text Interpreter::  
  225: * Stacks and Postfix notation::  
  226: * Your first definition::       
  227: * How does that work?::         
  228: * Forth is written in Forth::   
  229: * Review - elements of a Forth system::  
  230: * Where to go next::            
  231: * Exercises::                   
  232: 
  233: Forth Words
  234: 
  235: * Notation::                    
  236: * Case insensitivity::          
  237: * Comments::                    
  238: * Boolean Flags::               
  239: * Arithmetic::                  
  240: * Stack Manipulation::          
  241: * Memory::                      
  242: * Control Structures::          
  243: * Defining Words::              
  244: * Interpretation and Compilation Semantics::  
  245: * Tokens for Words::            
  246: * Compiling words::             
  247: * The Text Interpreter::        
  248: * Word Lists::                  
  249: * Environmental Queries::       
  250: * Files::                       
  251: * Blocks::                      
  252: * Other I/O::                   
  253: * Locals::                      
  254: * Structures::                  
  255: * Object-oriented Forth::       
  256: * Programming Tools::           
  257: * Assembler and Code Words::    
  258: * Threading Words::             
  259: * Passing Commands to the OS::  
  260: * Keeping track of Time::       
  261: * Miscellaneous Words::         
  262: 
  263: Arithmetic
  264: 
  265: * Single precision::            
  266: * Double precision::            Double-cell integer arithmetic
  267: * Bitwise operations::          
  268: * Numeric comparison::          
  269: * Mixed precision::             Operations with single and double-cell integers
  270: * Floating Point::              
  271: 
  272: Stack Manipulation
  273: 
  274: * Data stack::                  
  275: * Floating point stack::        
  276: * Return stack::                
  277: * Locals stack::                
  278: * Stack pointer manipulation::  
  279: 
  280: Memory
  281: 
  282: * Memory model::                
  283: * Dictionary allocation::       
  284: * Heap Allocation::             
  285: * Memory Access::               
  286: * Address arithmetic::          
  287: * Memory Blocks::               
  288: 
  289: Control Structures
  290: 
  291: * Selection::                   IF ... ELSE ... ENDIF
  292: * Simple Loops::                BEGIN ...
  293: * Counted Loops::               DO
  294: * Arbitrary control structures::  
  295: * Calls and returns::           
  296: * Exception Handling::          
  297: 
  298: Defining Words
  299: 
  300: * CREATE::                      
  301: * Variables::                   Variables and user variables
  302: * Constants::                   
  303: * Values::                      Initialised variables
  304: * Colon Definitions::           
  305: * Anonymous Definitions::       Definitions without names
  306: * Supplying names::             Passing definition names as strings
  307: * User-defined Defining Words::  
  308: * Deferred words::              Allow forward references
  309: * Aliases::                     
  310: 
  311: User-defined Defining Words
  312: 
  313: * CREATE..DOES> applications::  
  314: * CREATE..DOES> details::       
  315: * Advanced does> usage example::  
  316: 
  317: Interpretation and Compilation Semantics
  318: 
  319: * Combined words::              
  320: 
  321: Tokens for Words
  322: 
  323: * Execution token::             represents execution/interpretation semantics
  324: * Compilation token::           represents compilation semantics
  325: * Name token::                  represents named words
  326: 
  327: Compiling words
  328: 
  329: * Literals::                    Compiling data values
  330: * Macros::                      Compiling words
  331: 
  332: The Text Interpreter
  333: 
  334: * Input Sources::               
  335: * Number Conversion::           
  336: * Interpret/Compile states::    
  337: * Interpreter Directives::      
  338: 
  339: Word Lists
  340: 
  341: * Vocabularies::                
  342: * Why use word lists?::         
  343: * Word list example::           
  344: 
  345: Files
  346: 
  347: * Forth source files::          
  348: * General files::               
  349: * Search Paths::                
  350: 
  351: Search Paths
  352: 
  353: * Source Search Paths::         
  354: * General Search Paths::        
  355: 
  356: Other I/O
  357: 
  358: * Simple numeric output::       Predefined formats
  359: * Formatted numeric output::    Formatted (pictured) output
  360: * String Formats::              How Forth stores strings in memory
  361: * Displaying characters and strings::  Other stuff
  362: * Input::                       Input
  363: 
  364: Locals
  365: 
  366: * Gforth locals::               
  367: * ANS Forth locals::            
  368: 
  369: Gforth locals
  370: 
  371: * Where are locals visible by name?::  
  372: * How long do locals live?::    
  373: * Locals programming style::    
  374: * Locals implementation::       
  375: 
  376: Structures
  377: 
  378: * Why explicit structure support?::  
  379: * Structure Usage::             
  380: * Structure Naming Convention::  
  381: * Structure Implementation::    
  382: * Structure Glossary::          
  383: 
  384: Object-oriented Forth
  385: 
  386: * Why object-oriented programming?::  
  387: * Object-Oriented Terminology::  
  388: * Objects::                     
  389: * OOF::                         
  390: * Mini-OOF::                    
  391: * Comparison with other object models::  
  392: 
  393: The @file{objects.fs} model
  394: 
  395: * Properties of the Objects model::  
  396: * Basic Objects Usage::         
  397: * The Objects base class::      
  398: * Creating objects::            
  399: * Object-Oriented Programming Style::  
  400: * Class Binding::               
  401: * Method conveniences::         
  402: * Classes and Scoping::         
  403: * Dividing classes::            
  404: * Object Interfaces::           
  405: * Objects Implementation::      
  406: * Objects Glossary::            
  407: 
  408: The @file{oof.fs} model
  409: 
  410: * Properties of the OOF model::  
  411: * Basic OOF Usage::             
  412: * The OOF base class::          
  413: * Class Declaration::           
  414: * Class Implementation::        
  415: 
  416: The @file{mini-oof.fs} model
  417: 
  418: * Basic Mini-OOF Usage::        
  419: * Mini-OOF Example::            
  420: * Mini-OOF Implementation::     
  421: 
  422: Programming Tools
  423: 
  424: * Examining::                   
  425: * Forgetting words::            
  426: * Debugging::                   Simple and quick.
  427: * Assertions::                  Making your programs self-checking.
  428: * Singlestep Debugger::         Executing your program word by word.
  429: 
  430: Assembler and Code Words
  431: 
  432: * Code and ;code::              
  433: * Common Assembler::            Assembler Syntax
  434: * Common Disassembler::         
  435: * 386 Assembler::               Deviations and special cases
  436: * Alpha Assembler::             Deviations and special cases
  437: * MIPS assembler::              Deviations and special cases
  438: * Other assemblers::            How to write them
  439: 
  440: Tools
  441: 
  442: * ANS Report::                  Report the words used, sorted by wordset.
  443: 
  444: ANS conformance
  445: 
  446: * The Core Words::              
  447: * The optional Block word set::  
  448: * The optional Double Number word set::  
  449: * The optional Exception word set::  
  450: * The optional Facility word set::  
  451: * The optional File-Access word set::  
  452: * The optional Floating-Point word set::  
  453: * The optional Locals word set::  
  454: * The optional Memory-Allocation word set::  
  455: * The optional Programming-Tools word set::  
  456: * The optional Search-Order word set::  
  457: 
  458: The Core Words
  459: 
  460: * core-idef::                   Implementation Defined Options                   
  461: * core-ambcond::                Ambiguous Conditions                
  462: * core-other::                  Other System Documentation                  
  463: 
  464: The optional Block word set
  465: 
  466: * block-idef::                  Implementation Defined Options
  467: * block-ambcond::               Ambiguous Conditions               
  468: * block-other::                 Other System Documentation                 
  469: 
  470: The optional Double Number word set
  471: 
  472: * double-ambcond::              Ambiguous Conditions              
  473: 
  474: The optional Exception word set
  475: 
  476: * exception-idef::              Implementation Defined Options              
  477: 
  478: The optional Facility word set
  479: 
  480: * facility-idef::               Implementation Defined Options               
  481: * facility-ambcond::            Ambiguous Conditions            
  482: 
  483: The optional File-Access word set
  484: 
  485: * file-idef::                   Implementation Defined Options
  486: * file-ambcond::                Ambiguous Conditions                
  487: 
  488: The optional Floating-Point word set
  489: 
  490: * floating-idef::               Implementation Defined Options
  491: * floating-ambcond::            Ambiguous Conditions            
  492: 
  493: The optional Locals word set
  494: 
  495: * locals-idef::                 Implementation Defined Options                 
  496: * locals-ambcond::              Ambiguous Conditions              
  497: 
  498: The optional Memory-Allocation word set
  499: 
  500: * memory-idef::                 Implementation Defined Options                 
  501: 
  502: The optional Programming-Tools word set
  503: 
  504: * programming-idef::            Implementation Defined Options            
  505: * programming-ambcond::         Ambiguous Conditions         
  506: 
  507: The optional Search-Order word set
  508: 
  509: * search-idef::                 Implementation Defined Options                 
  510: * search-ambcond::              Ambiguous Conditions              
  511: 
  512: Image Files
  513: 
  514: * Image Licensing Issues::      Distribution terms for images.
  515: * Image File Background::       Why have image files?
  516: * Non-Relocatable Image Files::  don't always work.
  517: * Data-Relocatable Image Files::  are better.
  518: * Fully Relocatable Image Files::  better yet.
  519: * Stack and Dictionary Sizes::  Setting the default sizes for an image.
  520: * Running Image Files::         @code{gforth -i @i{file}} or @i{file}.
  521: * Modifying the Startup Sequence::  and turnkey applications.
  522: 
  523: Fully Relocatable Image Files
  524: 
  525: * gforthmi::                    The normal way
  526: * cross.fs::                    The hard way
  527: 
  528: Engine
  529: 
  530: * Portability::                 
  531: * Threading::                   
  532: * Primitives::                  
  533: * Performance::                 
  534: 
  535: Threading
  536: 
  537: * Scheduling::                  
  538: * Direct or Indirect Threaded?::  
  539: * DOES>::                       
  540: 
  541: Primitives
  542: 
  543: * Automatic Generation::        
  544: * TOS Optimization::            
  545: * Produced code::               
  546: 
  547: Cross Compiler
  548: 
  549: * Using the Cross Compiler::    
  550: * How the Cross Compiler Works::  
  551: 
  552: @end detailmenu
  553: @end menu
  554: 
  555: @node License, Goals, Top, Top
  556: @unnumbered GNU GENERAL PUBLIC LICENSE
  557: @center Version 2, June 1991
  558: 
  559: @display
  560: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
  561: 59 Temple Place, Suite 330, Boston, MA 02111, USA
  562: 
  563: Everyone is permitted to copy and distribute verbatim copies
  564: of this license document, but changing it is not allowed.
  565: @end display
  566: 
  567: @unnumberedsec Preamble
  568: 
  569:   The licenses for most software are designed to take away your
  570: freedom to share and change it.  By contrast, the GNU General Public
  571: License is intended to guarantee your freedom to share and change free
  572: software---to make sure the software is free for all its users.  This
  573: General Public License applies to most of the Free Software
  574: Foundation's software and to any other program whose authors commit to
  575: using it.  (Some other Free Software Foundation software is covered by
  576: the GNU Library General Public License instead.)  You can apply it to
  577: your programs, too.
  578: 
  579:   When we speak of free software, we are referring to freedom, not
  580: price.  Our General Public Licenses are designed to make sure that you
  581: have the freedom to distribute copies of free software (and charge for
  582: this service if you wish), that you receive source code or can get it
  583: if you want it, that you can change the software or use pieces of it
  584: in new free programs; and that you know you can do these things.
  585: 
  586:   To protect your rights, we need to make restrictions that forbid
  587: anyone to deny you these rights or to ask you to surrender the rights.
  588: These restrictions translate to certain responsibilities for you if you
  589: distribute copies of the software, or if you modify it.
  590: 
  591:   For example, if you distribute copies of such a program, whether
  592: gratis or for a fee, you must give the recipients all the rights that
  593: you have.  You must make sure that they, too, receive or can get the
  594: source code.  And you must show them these terms so they know their
  595: rights.
  596: 
  597:   We protect your rights with two steps: (1) copyright the software, and
  598: (2) offer you this license which gives you legal permission to copy,
  599: distribute and/or modify the software.
  600: 
  601:   Also, for each author's protection and ours, we want to make certain
  602: that everyone understands that there is no warranty for this free
  603: software.  If the software is modified by someone else and passed on, we
  604: want its recipients to know that what they have is not the original, so
  605: that any problems introduced by others will not reflect on the original
  606: authors' reputations.
  607: 
  608:   Finally, any free program is threatened constantly by software
  609: patents.  We wish to avoid the danger that redistributors of a free
  610: program will individually obtain patent licenses, in effect making the
  611: program proprietary.  To prevent this, we have made it clear that any
  612: patent must be licensed for everyone's free use or not licensed at all.
  613: 
  614:   The precise terms and conditions for copying, distribution and
  615: modification follow.
  616: 
  617: @iftex
  618: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  619: @end iftex
  620: @ifnottex
  621: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  622: @end ifnottex
  623: 
  624: @enumerate 0
  625: @item
  626: This License applies to any program or other work which contains
  627: a notice placed by the copyright holder saying it may be distributed
  628: under the terms of this General Public License.  The ``Program'', below,
  629: refers to any such program or work, and a ``work based on the Program''
  630: means either the Program or any derivative work under copyright law:
  631: that is to say, a work containing the Program or a portion of it,
  632: either verbatim or with modifications and/or translated into another
  633: language.  (Hereinafter, translation is included without limitation in
  634: the term ``modification''.)  Each licensee is addressed as ``you''.
  635: 
  636: Activities other than copying, distribution and modification are not
  637: covered by this License; they are outside its scope.  The act of
  638: running the Program is not restricted, and the output from the Program
  639: is covered only if its contents constitute a work based on the
  640: Program (independent of having been made by running the Program).
  641: Whether that is true depends on what the Program does.
  642: 
  643: @item
  644: You may copy and distribute verbatim copies of the Program's
  645: source code as you receive it, in any medium, provided that you
  646: conspicuously and appropriately publish on each copy an appropriate
  647: copyright notice and disclaimer of warranty; keep intact all the
  648: notices that refer to this License and to the absence of any warranty;
  649: and give any other recipients of the Program a copy of this License
  650: along with the Program.
  651: 
  652: You may charge a fee for the physical act of transferring a copy, and
  653: you may at your option offer warranty protection in exchange for a fee.
  654: 
  655: @item
  656: You may modify your copy or copies of the Program or any portion
  657: of it, thus forming a work based on the Program, and copy and
  658: distribute such modifications or work under the terms of Section 1
  659: above, provided that you also meet all of these conditions:
  660: 
  661: @enumerate a
  662: @item
  663: You must cause the modified files to carry prominent notices
  664: stating that you changed the files and the date of any change.
  665: 
  666: @item
  667: You must cause any work that you distribute or publish, that in
  668: whole or in part contains or is derived from the Program or any
  669: part thereof, to be licensed as a whole at no charge to all third
  670: parties under the terms of this License.
  671: 
  672: @item
  673: If the modified program normally reads commands interactively
  674: when run, you must cause it, when started running for such
  675: interactive use in the most ordinary way, to print or display an
  676: announcement including an appropriate copyright notice and a
  677: notice that there is no warranty (or else, saying that you provide
  678: a warranty) and that users may redistribute the program under
  679: these conditions, and telling the user how to view a copy of this
  680: License.  (Exception: if the Program itself is interactive but
  681: does not normally print such an announcement, your work based on
  682: the Program is not required to print an announcement.)
  683: @end enumerate
  684: 
  685: These requirements apply to the modified work as a whole.  If
  686: identifiable sections of that work are not derived from the Program,
  687: and can be reasonably considered independent and separate works in
  688: themselves, then this License, and its terms, do not apply to those
  689: sections when you distribute them as separate works.  But when you
  690: distribute the same sections as part of a whole which is a work based
  691: on the Program, the distribution of the whole must be on the terms of
  692: this License, whose permissions for other licensees extend to the
  693: entire whole, and thus to each and every part regardless of who wrote it.
  694: 
  695: Thus, it is not the intent of this section to claim rights or contest
  696: your rights to work written entirely by you; rather, the intent is to
  697: exercise the right to control the distribution of derivative or
  698: collective works based on the Program.
  699: 
  700: In addition, mere aggregation of another work not based on the Program
  701: with the Program (or with a work based on the Program) on a volume of
  702: a storage or distribution medium does not bring the other work under
  703: the scope of this License.
  704: 
  705: @item
  706: You may copy and distribute the Program (or a work based on it,
  707: under Section 2) in object code or executable form under the terms of
  708: Sections 1 and 2 above provided that you also do one of the following:
  709: 
  710: @enumerate a
  711: @item
  712: Accompany it with the complete corresponding machine-readable
  713: source code, which must be distributed under the terms of Sections
  714: 1 and 2 above on a medium customarily used for software interchange; or,
  715: 
  716: @item
  717: Accompany it with a written offer, valid for at least three
  718: years, to give any third party, for a charge no more than your
  719: cost of physically performing source distribution, a complete
  720: machine-readable copy of the corresponding source code, to be
  721: distributed under the terms of Sections 1 and 2 above on a medium
  722: customarily used for software interchange; or,
  723: 
  724: @item
  725: Accompany it with the information you received as to the offer
  726: to distribute corresponding source code.  (This alternative is
  727: allowed only for noncommercial distribution and only if you
  728: received the program in object code or executable form with such
  729: an offer, in accord with Subsection b above.)
  730: @end enumerate
  731: 
  732: The source code for a work means the preferred form of the work for
  733: making modifications to it.  For an executable work, complete source
  734: code means all the source code for all modules it contains, plus any
  735: associated interface definition files, plus the scripts used to
  736: control compilation and installation of the executable.  However, as a
  737: special exception, the source code distributed need not include
  738: anything that is normally distributed (in either source or binary
  739: form) with the major components (compiler, kernel, and so on) of the
  740: operating system on which the executable runs, unless that component
  741: itself accompanies the executable.
  742: 
  743: If distribution of executable or object code is made by offering
  744: access to copy from a designated place, then offering equivalent
  745: access to copy the source code from the same place counts as
  746: distribution of the source code, even though third parties are not
  747: compelled to copy the source along with the object code.
  748: 
  749: @item
  750: You may not copy, modify, sublicense, or distribute the Program
  751: except as expressly provided under this License.  Any attempt
  752: otherwise to copy, modify, sublicense or distribute the Program is
  753: void, and will automatically terminate your rights under this License.
  754: However, parties who have received copies, or rights, from you under
  755: this License will not have their licenses terminated so long as such
  756: parties remain in full compliance.
  757: 
  758: @item
  759: You are not required to accept this License, since you have not
  760: signed it.  However, nothing else grants you permission to modify or
  761: distribute the Program or its derivative works.  These actions are
  762: prohibited by law if you do not accept this License.  Therefore, by
  763: modifying or distributing the Program (or any work based on the
  764: Program), you indicate your acceptance of this License to do so, and
  765: all its terms and conditions for copying, distributing or modifying
  766: the Program or works based on it.
  767: 
  768: @item
  769: Each time you redistribute the Program (or any work based on the
  770: Program), the recipient automatically receives a license from the
  771: original licensor to copy, distribute or modify the Program subject to
  772: these terms and conditions.  You may not impose any further
  773: restrictions on the recipients' exercise of the rights granted herein.
  774: You are not responsible for enforcing compliance by third parties to
  775: this License.
  776: 
  777: @item
  778: If, as a consequence of a court judgment or allegation of patent
  779: infringement or for any other reason (not limited to patent issues),
  780: conditions are imposed on you (whether by court order, agreement or
  781: otherwise) that contradict the conditions of this License, they do not
  782: excuse you from the conditions of this License.  If you cannot
  783: distribute so as to satisfy simultaneously your obligations under this
  784: License and any other pertinent obligations, then as a consequence you
  785: may not distribute the Program at all.  For example, if a patent
  786: license would not permit royalty-free redistribution of the Program by
  787: all those who receive copies directly or indirectly through you, then
  788: the only way you could satisfy both it and this License would be to
  789: refrain entirely from distribution of the Program.
  790: 
  791: If any portion of this section is held invalid or unenforceable under
  792: any particular circumstance, the balance of the section is intended to
  793: apply and the section as a whole is intended to apply in other
  794: circumstances.
  795: 
  796: It is not the purpose of this section to induce you to infringe any
  797: patents or other property right claims or to contest validity of any
  798: such claims; this section has the sole purpose of protecting the
  799: integrity of the free software distribution system, which is
  800: implemented by public license practices.  Many people have made
  801: generous contributions to the wide range of software distributed
  802: through that system in reliance on consistent application of that
  803: system; it is up to the author/donor to decide if he or she is willing
  804: to distribute software through any other system and a licensee cannot
  805: impose that choice.
  806: 
  807: This section is intended to make thoroughly clear what is believed to
  808: be a consequence of the rest of this License.
  809: 
  810: @item
  811: If the distribution and/or use of the Program is restricted in
  812: certain countries either by patents or by copyrighted interfaces, the
  813: original copyright holder who places the Program under this License
  814: may add an explicit geographical distribution limitation excluding
  815: those countries, so that distribution is permitted only in or among
  816: countries not thus excluded.  In such case, this License incorporates
  817: the limitation as if written in the body of this License.
  818: 
  819: @item
  820: The Free Software Foundation may publish revised and/or new versions
  821: of the General Public License from time to time.  Such new versions will
  822: be similar in spirit to the present version, but may differ in detail to
  823: address new problems or concerns.
  824: 
  825: Each version is given a distinguishing version number.  If the Program
  826: specifies a version number of this License which applies to it and ``any
  827: later version'', you have the option of following the terms and conditions
  828: either of that version or of any later version published by the Free
  829: Software Foundation.  If the Program does not specify a version number of
  830: this License, you may choose any version ever published by the Free Software
  831: Foundation.
  832: 
  833: @item
  834: If you wish to incorporate parts of the Program into other free
  835: programs whose distribution conditions are different, write to the author
  836: to ask for permission.  For software which is copyrighted by the Free
  837: Software Foundation, write to the Free Software Foundation; we sometimes
  838: make exceptions for this.  Our decision will be guided by the two goals
  839: of preserving the free status of all derivatives of our free software and
  840: of promoting the sharing and reuse of software generally.
  841: 
  842: @iftex
  843: @heading NO WARRANTY
  844: @end iftex
  845: @ifnottex
  846: @center NO WARRANTY
  847: @end ifnottex
  848: 
  849: @item
  850: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
  851: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
  852: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
  853: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
  854: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  855: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
  856: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
  857: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
  858: REPAIR OR CORRECTION.
  859: 
  860: @item
  861: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
  862: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
  863: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
  864: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
  865: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
  866: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
  867: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
  868: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
  869: POSSIBILITY OF SUCH DAMAGES.
  870: @end enumerate
  871: 
  872: @iftex
  873: @heading END OF TERMS AND CONDITIONS
  874: @end iftex
  875: @ifnottex
  876: @center END OF TERMS AND CONDITIONS
  877: @end ifnottex
  878: 
  879: @page
  880: @unnumberedsec How to Apply These Terms to Your New Programs
  881: 
  882:   If you develop a new program, and you want it to be of the greatest
  883: possible use to the public, the best way to achieve this is to make it
  884: free software which everyone can redistribute and change under these terms.
  885: 
  886:   To do so, attach the following notices to the program.  It is safest
  887: to attach them to the start of each source file to most effectively
  888: convey the exclusion of warranty; and each file should have at least
  889: the ``copyright'' line and a pointer to where the full notice is found.
  890: 
  891: @smallexample
  892: @var{one line to give the program's name and a brief idea of what it does.}
  893: Copyright (C) 19@var{yy}  @var{name of author}
  894: 
  895: This program is free software; you can redistribute it and/or modify 
  896: it under the terms of the GNU General Public License as published by 
  897: the Free Software Foundation; either version 2 of the License, or 
  898: (at your option) any later version.
  899: 
  900: This program is distributed in the hope that it will be useful,
  901: but WITHOUT ANY WARRANTY; without even the implied warranty of
  902: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  903: GNU General Public License for more details.
  904: 
  905: You should have received a copy of the GNU General Public License
  906: along with this program; if not, write to the Free Software
  907: Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
  908: @end smallexample
  909: 
  910: Also add information on how to contact you by electronic and paper mail.
  911: 
  912: If the program is interactive, make it output a short notice like this
  913: when it starts in an interactive mode:
  914: 
  915: @smallexample
  916: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
  917: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
  918: type `show w'.  
  919: This is free software, and you are welcome to redistribute it 
  920: under certain conditions; type `show c' for details.
  921: @end smallexample
  922: 
  923: The hypothetical commands @samp{show w} and @samp{show c} should show
  924: the appropriate parts of the General Public License.  Of course, the
  925: commands you use may be called something other than @samp{show w} and
  926: @samp{show c}; they could even be mouse-clicks or menu items---whatever
  927: suits your program.
  928: 
  929: You should also get your employer (if you work as a programmer) or your
  930: school, if any, to sign a ``copyright disclaimer'' for the program, if
  931: necessary.  Here is a sample; alter the names:
  932: 
  933: @smallexample
  934: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
  935: `Gnomovision' (which makes passes at compilers) written by James Hacker.
  936: 
  937: @var{signature of Ty Coon}, 1 April 1989
  938: Ty Coon, President of Vice
  939: @end smallexample
  940: 
  941: This General Public License does not permit incorporating your program into
  942: proprietary programs.  If your program is a subroutine library, you may
  943: consider it more useful to permit linking proprietary applications with the
  944: library.  If this is what you want to do, use the GNU Library General
  945: Public License instead of this License.
  946: 
  947: @iftex
  948: @unnumbered Preface
  949: @cindex Preface
  950: This manual documents Gforth. Some introductory material is provided for
  951: readers who are unfamiliar with Forth or who are migrating to Gforth
  952: from other Forth compilers. However, this manual is primarily a
  953: reference manual.
  954: @end iftex
  955: 
  956: @comment TODO much more blurb here.
  957: 
  958: @c ******************************************************************
  959: @node Goals, Gforth Environment, License, Top
  960: @comment node-name,     next,           previous, up
  961: @chapter Goals of Gforth
  962: @cindex goals of the Gforth project
  963: The goal of the Gforth Project is to develop a standard model for
  964: ANS Forth. This can be split into several subgoals:
  965: 
  966: @itemize @bullet
  967: @item
  968: Gforth should conform to the ANS Forth Standard.
  969: @item
  970: It should be a model, i.e. it should define all the
  971: implementation-dependent things.
  972: @item
  973: It should become standard, i.e. widely accepted and used. This goal
  974: is the most difficult one.
  975: @end itemize
  976: 
  977: To achieve these goals Gforth should be
  978: @itemize @bullet
  979: @item
  980: Similar to previous models (fig-Forth, F83)
  981: @item
  982: Powerful. It should provide for all the things that are considered
  983: necessary today and even some that are not yet considered necessary.
  984: @item
  985: Efficient. It should not get the reputation of being exceptionally
  986: slow.
  987: @item
  988: Free.
  989: @item
  990: Available on many machines/easy to port.
  991: @end itemize
  992: 
  993: Have we achieved these goals? Gforth conforms to the ANS Forth
  994: standard. It may be considered a model, but we have not yet documented
  995: which parts of the model are stable and which parts we are likely to
  996: change. It certainly has not yet become a de facto standard, but it
  997: appears to be quite popular. It has some similarities to and some
  998: differences from previous models. It has some powerful features, but not
  999: yet everything that we envisioned. We certainly have achieved our
 1000: execution speed goals (@pxref{Performance})@footnote{However, in 1998
 1001: the bar was raised when the major commercial Forth vendors switched to
 1002: native code compilers.}.  It is free and available on many machines.
 1003: 
 1004: @c ******************************************************************
 1005: @node Gforth Environment, Tutorial, Goals, Top
 1006: @chapter Gforth Environment
 1007: @cindex Gforth environment
 1008: 
 1009: Note: ultimately, the Gforth man page will be auto-generated from the
 1010: material in this chapter.
 1011: 
 1012: @menu
 1013: * Invoking Gforth::             Getting in
 1014: * Leaving Gforth::              Getting out
 1015: * Command-line editing::        
 1016: * Environment variables::       that affect how Gforth starts up
 1017: * Gforth Files::                What gets installed and where
 1018: * Startup speed::               When 35ms is not fast enough ...
 1019: @end menu
 1020: 
 1021: For related information about the creation of images see @ref{Image Files}.
 1022: 
 1023: @comment ----------------------------------------------
 1024: @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
 1025: @section Invoking Gforth
 1026: @cindex invoking Gforth
 1027: @cindex running Gforth
 1028: @cindex command-line options
 1029: @cindex options on the command line
 1030: @cindex flags on the command line
 1031: 
 1032: Gforth is made up of two parts; an executable ``engine'' (named
 1033: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
 1034: will usually just say @code{gforth} -- this automatically loads the
 1035: default image file @file{gforth.fi}. In many other cases the default
 1036: Gforth image will be invoked like this:
 1037: @example
 1038: gforth [file | -e forth-code] ...
 1039: @end example
 1040: @noindent
 1041: This interprets the contents of the files and the Forth code in the order they
 1042: are given.
 1043: 
 1044: In addition to the @file{gforth} engine, there is also an engine called
 1045: @file{gforth-fast}, which is faster, but gives less informative error
 1046: messages (@pxref{Error messages}).
 1047: 
 1048: In general, the command line looks like this:
 1049: 
 1050: @example
 1051: gforth[-fast] [engine options] [image options]
 1052: @end example
 1053: 
 1054: The engine options must come before the rest of the command
 1055: line. They are:
 1056: 
 1057: @table @code
 1058: @cindex -i, command-line option
 1059: @cindex --image-file, command-line option
 1060: @item --image-file @i{file}
 1061: @itemx -i @i{file}
 1062: Loads the Forth image @i{file} instead of the default
 1063: @file{gforth.fi} (@pxref{Image Files}).
 1064: 
 1065: @cindex --appl-image, command-line option
 1066: @item --appl-image @i{file}
 1067: Loads the image @i{file} and leaves all further command-line arguments
 1068: to the image (instead of processing them as engine options).  This is
 1069: useful for building executable application images on Unix, built with
 1070: @code{gforthmi --application ...}.
 1071: 
 1072: @cindex --path, command-line option
 1073: @cindex -p, command-line option
 1074: @item --path @i{path}
 1075: @itemx -p @i{path}
 1076: Uses @i{path} for searching the image file and Forth source code files
 1077: instead of the default in the environment variable @code{GFORTHPATH} or
 1078: the path specified at installation time (e.g.,
 1079: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
 1080: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
 1081: 
 1082: @cindex --dictionary-size, command-line option
 1083: @cindex -m, command-line option
 1084: @cindex @i{size} parameters for command-line options
 1085: @cindex size of the dictionary and the stacks
 1086: @item --dictionary-size @i{size}
 1087: @itemx -m @i{size}
 1088: Allocate @i{size} space for the Forth dictionary space instead of
 1089: using the default specified in the image (typically 256K). The
 1090: @i{size} specification for this and subsequent options consists of
 1091: an integer and a unit (e.g.,
 1092: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
 1093: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
 1094: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
 1095: @code{e} is used.
 1096: 
 1097: @cindex --data-stack-size, command-line option
 1098: @cindex -d, command-line option
 1099: @item --data-stack-size @i{size}
 1100: @itemx -d @i{size}
 1101: Allocate @i{size} space for the data stack instead of using the
 1102: default specified in the image (typically 16K).
 1103: 
 1104: @cindex --return-stack-size, command-line option
 1105: @cindex -r, command-line option
 1106: @item --return-stack-size @i{size}
 1107: @itemx -r @i{size}
 1108: Allocate @i{size} space for the return stack instead of using the
 1109: default specified in the image (typically 15K).
 1110: 
 1111: @cindex --fp-stack-size, command-line option
 1112: @cindex -f, command-line option
 1113: @item --fp-stack-size @i{size}
 1114: @itemx -f @i{size}
 1115: Allocate @i{size} space for the floating point stack instead of
 1116: using the default specified in the image (typically 15.5K). In this case
 1117: the unit specifier @code{e} refers to floating point numbers.
 1118: 
 1119: @cindex --locals-stack-size, command-line option
 1120: @cindex -l, command-line option
 1121: @item --locals-stack-size @i{size}
 1122: @itemx -l @i{size}
 1123: Allocate @i{size} space for the locals stack instead of using the
 1124: default specified in the image (typically 14.5K).
 1125: 
 1126: @cindex -h, command-line option
 1127: @cindex --help, command-line option
 1128: @item --help
 1129: @itemx -h
 1130: Print a message about the command-line options
 1131: 
 1132: @cindex -v, command-line option
 1133: @cindex --version, command-line option
 1134: @item --version
 1135: @itemx -v
 1136: Print version and exit
 1137: 
 1138: @cindex --debug, command-line option
 1139: @item --debug
 1140: Print some information useful for debugging on startup.
 1141: 
 1142: @cindex --offset-image, command-line option
 1143: @item --offset-image
 1144: Start the dictionary at a slightly different position than would be used
 1145: otherwise (useful for creating data-relocatable images,
 1146: @pxref{Data-Relocatable Image Files}).
 1147: 
 1148: @cindex --no-offset-im, command-line option
 1149: @item --no-offset-im
 1150: Start the dictionary at the normal position.
 1151: 
 1152: @cindex --clear-dictionary, command-line option
 1153: @item --clear-dictionary
 1154: Initialize all bytes in the dictionary to 0 before loading the image
 1155: (@pxref{Data-Relocatable Image Files}).
 1156: 
 1157: @cindex --die-on-signal, command-line-option
 1158: @item --die-on-signal
 1159: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
 1160: or the segmentation violation SIGSEGV) by translating it into a Forth
 1161: @code{THROW}. With this option, Gforth exits if it receives such a
 1162: signal. This option is useful when the engine and/or the image might be
 1163: severely broken (such that it causes another signal before recovering
 1164: from the first); this option avoids endless loops in such cases.
 1165: @end table
 1166: 
 1167: @cindex loading files at startup
 1168: @cindex executing code on startup
 1169: @cindex batch processing with Gforth
 1170: As explained above, the image-specific command-line arguments for the
 1171: default image @file{gforth.fi} consist of a sequence of filenames and
 1172: @code{-e @var{forth-code}} options that are interpreted in the sequence
 1173: in which they are given. The @code{-e @var{forth-code}} or
 1174: @code{--evaluate @var{forth-code}} option evaluates the Forth
 1175: code. This option takes only one argument; if you want to evaluate more
 1176: Forth words, you have to quote them or use @code{-e} several times. To exit
 1177: after processing the command line (instead of entering interactive mode)
 1178: append @code{-e bye} to the command line.
 1179: 
 1180: @cindex versions, invoking other versions of Gforth
 1181: If you have several versions of Gforth installed, @code{gforth} will
 1182: invoke the version that was installed last. @code{gforth-@i{version}}
 1183: invokes a specific version. If your environment contains the variable
 1184: @code{GFORTHPATH}, you may want to override it by using the
 1185: @code{--path} option.
 1186: 
 1187: Not yet implemented:
 1188: On startup the system first executes the system initialization file
 1189: (unless the option @code{--no-init-file} is given; note that the system
 1190: resulting from using this option may not be ANS Forth conformant). Then
 1191: the user initialization file @file{.gforth.fs} is executed, unless the
 1192: option @code{--no-rc} is given; this file is searched for in @file{.},
 1193: then in @file{~}, then in the normal path (see above).
 1194: 
 1195: 
 1196: 
 1197: @comment ----------------------------------------------
 1198: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
 1199: @section Leaving Gforth
 1200: @cindex Gforth - leaving
 1201: @cindex leaving Gforth
 1202: 
 1203: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
 1204: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
 1205: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
 1206: data are discarded.  For ways of saving the state of the system before
 1207: leaving Gforth see @ref{Image Files}.
 1208: 
 1209: doc-bye
 1210: 
 1211: 
 1212: @comment ----------------------------------------------
 1213: @node Command-line editing, Environment variables, Leaving Gforth, Gforth Environment
 1214: @section Command-line editing
 1215: @cindex command-line editing
 1216: 
 1217: Gforth maintains a history file that records every line that you type to
 1218: the text interpreter. This file is preserved between sessions, and is
 1219: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
 1220: repeatedly you can recall successively older commands from this (or
 1221: previous) session(s). The full list of command-line editing facilities is:
 1222: 
 1223: @itemize @bullet
 1224: @item
 1225: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
 1226: commands from the history buffer.
 1227: @item
 1228: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
 1229: from the history buffer.
 1230: @item
 1231: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
 1232: @item
 1233: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
 1234: @item
 1235: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
 1236: closing up the line.
 1237: @item
 1238: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
 1239: @item
 1240: @kbd{Ctrl-a} to move the cursor to the start of the line.
 1241: @item
 1242: @kbd{Ctrl-e} to move the cursor to the end of the line.
 1243: @item
 1244: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
 1245: line.
 1246: @item
 1247: @key{TAB} to step through all possible full-word completions of the word
 1248: currently being typed.
 1249: @item
 1250: @kbd{Ctrl-d} on an empty line line to terminate Gforth (gracefully,
 1251: using @code{bye}). 
 1252: @item
 1253: @kbd{Ctrl-x} (or @code{Ctrl-d} on a non-empty line) to delete the
 1254: character under the cursor.
 1255: @end itemize
 1256: 
 1257: When editing, displayable characters are inserted to the left of the
 1258: cursor position; the line is always in ``insert'' (as opposed to
 1259: ``overstrike'') mode.
 1260: 
 1261: @cindex history file
 1262: @cindex @file{.gforth-history}
 1263: On Unix systems, the history file is @file{~/.gforth-history} by
 1264: default@footnote{i.e. it is stored in the user's home directory.}. You
 1265: can find out the name and location of your history file using:
 1266: 
 1267: @example 
 1268: history-file type \ Unix-class systems
 1269: 
 1270: history-file type \ Other systems
 1271: history-dir  type
 1272: @end example
 1273: 
 1274: If you enter long definitions by hand, you can use a text editor to
 1275: paste them out of the history file into a Forth source file for reuse at
 1276: a later time.
 1277: 
 1278: Gforth never trims the size of the history file, so you should do this
 1279: periodically, if necessary.
 1280: 
 1281: @comment this is all defined in history.fs
 1282: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
 1283: @comment chosen?
 1284: 
 1285: 
 1286: @comment ----------------------------------------------
 1287: @node Environment variables, Gforth Files, Command-line editing, Gforth Environment
 1288: @section Environment variables
 1289: @cindex environment variables
 1290: 
 1291: Gforth uses these environment variables:
 1292: 
 1293: @itemize @bullet
 1294: @item
 1295: @cindex @code{GFORTHHIST} -- environment variable
 1296: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
 1297: open/create the history file, @file{.gforth-history}. Default:
 1298: @code{$HOME}.
 1299: 
 1300: @item
 1301: @cindex @code{GFORTHPATH} -- environment variable
 1302: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
 1303: for Forth source-code files.
 1304: 
 1305: @item
 1306: @cindex @code{GFORTH} -- environment variable
 1307: @code{GFORTH} -- used by @file{gforthmi}, @xref{gforthmi}.
 1308: 
 1309: @item
 1310: @cindex @code{GFORTHD} -- environment variable
 1311: @code{GFORTHD} -- used by @file{gforthmi}, @xref{gforthmi}.
 1312: 
 1313: @item
 1314: @cindex @code{TMP}, @code{TEMP} - environment variable
 1315: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
 1316: location for the history file.
 1317: @end itemize
 1318: 
 1319: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
 1320: @comment mentioning these.
 1321: 
 1322: All the Gforth environment variables default to sensible values if they
 1323: are not set.
 1324: 
 1325: 
 1326: @comment ----------------------------------------------
 1327: @node Gforth Files, Startup speed, Environment variables, Gforth Environment
 1328: @section Gforth files
 1329: @cindex Gforth files
 1330: 
 1331: When you install Gforth on a Unix system, it installs files in these
 1332: locations by default:
 1333: 
 1334: @itemize @bullet
 1335: @item
 1336: @file{/usr/local/bin/gforth}
 1337: @item
 1338: @file{/usr/local/bin/gforthmi}
 1339: @item
 1340: @file{/usr/local/man/man1/gforth.1} - man page.
 1341: @item
 1342: @file{/usr/local/info} - the Info version of this manual.
 1343: @item
 1344: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
 1345: @item
 1346: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
 1347: @item
 1348: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
 1349: @item
 1350: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
 1351: @end itemize
 1352: 
 1353: You can select different places for installation by using
 1354: @code{configure} options (listed with @code{configure --help}).
 1355: 
 1356: @comment ----------------------------------------------
 1357: @node Startup speed,  , Gforth Files, Gforth Environment
 1358: @section Startup speed
 1359: @cindex Startup speed
 1360: @cindex speed, startup
 1361: 
 1362: If Gforth is used for CGI scripts or in shell scripts, its startup
 1363: speed may become a problem.  On a 300MHz 21064a under Linux-2.2.13 with
 1364: glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
 1365: system time.
 1366: 
 1367: If startup speed is a problem, you may consider the following ways to
 1368: improve it; or you may consider ways to reduce the number of startups
 1369: (for example, by using Fast-CGI).
 1370: 
 1371: The first step to improve startup speed is to statically link Gforth, by
 1372: building it with @code{XLDFLAGS=-static}.  This requires more memory for
 1373: the code and will therefore slow down the first invocation, but
 1374: subsequent invocations avoid the dynamic linking overhead.  Another
 1375: disadvantage is that Gforth won't profit from library upgrades.  As a
 1376: result, @code{gforth-static -e bye} takes about 17.1ms user and
 1377: 8.2ms system time.
 1378: 
 1379: The next step to improve startup speed is to use a non-relocatable image
 1380: (@pxref{Non-Relocatable Image Files}).  You can create this image with
 1381: @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
 1382: @code{gforth -i gforthnr.fi ...}.  This avoids the relocation overhead
 1383: and a part of the copy-on-write overhead.  The disadvantage is that the
 1384: non-relocatable image does not work if the OS gives Gforth a different
 1385: address for the dictionary, for whatever reason; so you better provide a
 1386: fallback on a relocatable image.  @code{gforth-static -i gforthnr.fi -e
 1387: bye} takes about 15.3ms user and 7.5ms system time.
 1388: 
 1389: The final step is to disable dictionary hashing in Gforth.  Gforth
 1390: builds the hash table on startup, which takes much of the startup
 1391: overhead. You can do this by commenting out the @code{include hash.fs}
 1392: in @file{startup.fs} and everything that requires @file{hash.fs} (at the
 1393: moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
 1394: The disadvantages are that functionality like @code{table} and
 1395: @code{ekey} is missing and that text interpretation (e.g., compiling)
 1396: now takes much longer. So, you should only use this method if there is
 1397: no significant text interpretation to perform (the script should be
 1398: compiled into the image, amongst other things).  @code{gforth-static -i
 1399: gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
 1400: 
 1401: @c ******************************************************************
 1402: @node Tutorial, Introduction, Gforth Environment, Top
 1403: @chapter Forth Tutorial
 1404: @cindex Tutorial
 1405: @cindex Forth Tutorial
 1406: 
 1407: @c Topics from nac's Introduction that could be mentioned:
 1408: @c press <ret> after each line
 1409: @c Prompt
 1410: @c numbers vs. words in dictionary on text interpretation
 1411: @c what happens on redefinition
 1412: @c parsing words (in particular, defining words)
 1413: 
 1414: The difference of this chapter from the Introduction
 1415: (@pxref{Introduction}) is that this tutorial is more fast-paced, should
 1416: be used while sitting in front of a computer, and covers much more
 1417: material, but does not explain how the Forth system works.
 1418: 
 1419: This tutorial can be used with any ANS-compliant Forth; any
 1420: Gforth-specific features are marked as such and you can skip them if you
 1421: work with another Forth.  This tutorial does not explain all features of
 1422: Forth, just enough to get you started and give you some ideas about the
 1423: facilities available in Forth.  Read the rest of the manual and the
 1424: standard when you are through this.
 1425: 
 1426: The intended way to use this tutorial is that you work through it while
 1427: sitting in front of the console, take a look at the examples and predict
 1428: what they will do, then try them out; if the outcome is not as expected,
 1429: find out why (e.g., by trying out variations of the example), so you
 1430: understand what's going on.  There are also some assignments that you
 1431: should solve.
 1432: 
 1433: This tutorial assumes that you have programmed before and know what,
 1434: e.g., a loop is.
 1435: 
 1436: @c !! explain compat library
 1437: 
 1438: @menu
 1439: * Starting Gforth Tutorial::    
 1440: * Syntax Tutorial::             
 1441: * Crash Course Tutorial::       
 1442: * Stack Tutorial::              
 1443: * Arithmetics Tutorial::        
 1444: * Stack Manipulation Tutorial::  
 1445: * Using files for Forth code Tutorial::  
 1446: * Comments Tutorial::           
 1447: * Colon Definitions Tutorial::  
 1448: * Decompilation Tutorial::      
 1449: * Stack-Effect Comments Tutorial::  
 1450: * Types Tutorial::              
 1451: * Factoring Tutorial::          
 1452: * Designing the stack effect Tutorial::  
 1453: * Local Variables Tutorial::    
 1454: * Conditional execution Tutorial::  
 1455: * Flags and Comparisons Tutorial::  
 1456: * General Loops Tutorial::      
 1457: * Counted loops Tutorial::      
 1458: * Recursion Tutorial::          
 1459: * Leaving definitions or loops Tutorial::  
 1460: * Return Stack Tutorial::       
 1461: * Memory Tutorial::             
 1462: * Characters and Strings Tutorial::  
 1463: * Alignment Tutorial::          
 1464: * Files Tutorial::              
 1465: * Interpretation and Compilation Semantics and Immediacy Tutorial::  
 1466: * Execution Tokens Tutorial::   
 1467: * Exceptions Tutorial::         
 1468: * Defining Words Tutorial::     
 1469: * Arrays and Records Tutorial::  
 1470: * POSTPONE Tutorial::           
 1471: * Literal Tutorial::            
 1472: * Advanced macros Tutorial::    
 1473: * Compilation Tokens Tutorial::  
 1474: * Wordlists and Search Order Tutorial::  
 1475: @end menu
 1476: 
 1477: @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
 1478: @section Starting Gforth
 1479: @cindex starting Gforth tutorial
 1480: You can start Gforth by typing its name:
 1481: 
 1482: @example
 1483: gforth
 1484: @end example
 1485: 
 1486: That puts you into interactive mode; you can leave Gforth by typing
 1487: @code{bye}.  While in Gforth, you can edit the command line and access
 1488: the command line history with cursor keys, similar to bash.
 1489: 
 1490: 
 1491: @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
 1492: @section Syntax
 1493: @cindex syntax tutorial
 1494: 
 1495: A @dfn{word} is a sequence of arbitrary characters (expcept white
 1496: space).  Words are separated by white space.  E.g., each of the
 1497: following lines contains exactly one word:
 1498: 
 1499: @example
 1500: word
 1501: !@@#$%^&*()
 1502: 1234567890
 1503: 5!a
 1504: @end example
 1505: 
 1506: A frequent beginner's error is to leave away necessary white space,
 1507: resulting in an error like @samp{Undefined word}; so if you see such an
 1508: error, check if you have put spaces wherever necessary.
 1509: 
 1510: @example
 1511: ." hello, world" \ correct
 1512: ."hello, world"  \ gives an "Undefined word" error
 1513: @end example
 1514: 
 1515: Gforth and most other Forth systems ignore differences in case (they are
 1516: case-insensitive), i.e., @samp{word} is the same as @samp{Word}.  If
 1517: your system is case-sensitive, you may have to type all the examples
 1518: given here in upper case.
 1519: 
 1520: 
 1521: @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
 1522: @section Crash Course
 1523: 
 1524: Type
 1525: 
 1526: @example
 1527: 0 0 !
 1528: here execute
 1529: ' catch >body 20 erase abort
 1530: ' (quit) >body 20 erase
 1531: @end example
 1532: 
 1533: The last two examples are guaranteed to destroy parts of Gforth (and
 1534: most other systems), so you better leave Gforth afterwards (if it has
 1535: not finished by itself).  On some systems you may have to kill gforth
 1536: from outside (e.g., in Unix with @code{kill}).
 1537: 
 1538: Now that you know how to produce crashes (and that there's not much to
 1539: them), let's learn how to produce meaningful programs.
 1540: 
 1541: 
 1542: @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
 1543: @section Stack
 1544: @cindex stack tutorial
 1545: 
 1546: The most obvious feature of Forth is the stack.  When you type in a
 1547: number, it is pushed on the stack.  You can display the content of the
 1548: stack with @code{.s}.
 1549: 
 1550: @example
 1551: 1 2 .s
 1552: 3 .s
 1553: @end example
 1554: 
 1555: @code{.s} displays the top-of-stack to the right, i.e., the numbers
 1556: appear in @code{.s} output as they appeared in the input.
 1557: 
 1558: You can print the top of stack element with @code{.}.
 1559: 
 1560: @example
 1561: 1 2 3 . . .
 1562: @end example
 1563: 
 1564: In general, words consume their stack arguments (@code{.s} is an
 1565: exception).
 1566: 
 1567: @assignment
 1568: What does the stack contain after @code{5 6 7 .}?
 1569: @endassignment
 1570: 
 1571: 
 1572: @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
 1573: @section Arithmetics
 1574: @cindex arithmetics tutorial
 1575: 
 1576: The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
 1577: operate on the top two stack items:
 1578: 
 1579: @example
 1580: 2 2 .s
 1581: + .s
 1582: .
 1583: 2 1 - .
 1584: 7 3 mod .
 1585: @end example
 1586: 
 1587: The operands of @code{-}, @code{/}, and @code{mod} are in the same order
 1588: as in the corresponding infix expression (this is generally the case in
 1589: Forth).
 1590: 
 1591: Parentheses are superfluous (and not available), because the order of
 1592: the words unambiguously determines the order of evaluation and the
 1593: operands:
 1594: 
 1595: @example
 1596: 3 4 + 5 * .
 1597: 3 4 5 * + .
 1598: @end example
 1599: 
 1600: @assignment
 1601: What are the infix expressions corresponding to the Forth code above?
 1602: Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
 1603: known as Postfix or RPN (Reverse Polish Notation).}.
 1604: @endassignment
 1605: 
 1606: To change the sign, use @code{negate}:
 1607: 
 1608: @example
 1609: 2 negate .
 1610: @end example
 1611: 
 1612: @assignment
 1613: Convert -(-3)*4-5 to Forth.
 1614: @endassignment
 1615: 
 1616: @code{/mod} performs both @code{/} and @code{mod}.
 1617: 
 1618: @example
 1619: 7 3 /mod . .
 1620: @end example
 1621: 
 1622: Reference: @ref{Arithmetic}.
 1623: 
 1624: 
 1625: @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
 1626: @section Stack Manipulation
 1627: @cindex stack manipulation tutorial
 1628: 
 1629: Stack manipulation words rearrange the data on the stack.
 1630: 
 1631: @example
 1632: 1 .s drop .s
 1633: 1 .s dup .s drop drop .s
 1634: 1 2 .s over .s drop drop drop
 1635: 1 2 .s swap .s drop drop
 1636: 1 2 3 .s rot .s drop drop drop
 1637: @end example
 1638: 
 1639: These are the most important stack manipulation words.  There are also
 1640: variants that manipulate twice as many stack items:
 1641: 
 1642: @example
 1643: 1 2 3 4 .s 2swap .s 2drop 2drop
 1644: @end example
 1645: 
 1646: Two more stack manipulation words are:
 1647: 
 1648: @example
 1649: 1 2 .s nip .s drop
 1650: 1 2 .s tuck .s 2drop drop
 1651: @end example
 1652: 
 1653: @assignment
 1654: Replace @code{nip} and @code{tuck} with combinations of other stack
 1655: manipulation words.
 1656: 
 1657: @example
 1658: Given:          How do you get:
 1659: 1 2 3           3 2 1           
 1660: 1 2 3           1 2 3 2                 
 1661: 1 2 3           1 2 3 3                 
 1662: 1 2 3           1 3 3           
 1663: 1 2 3           2 1 3           
 1664: 1 2 3 4         4 3 2 1         
 1665: 1 2 3           1 2 3 1 2 3             
 1666: 1 2 3 4         1 2 3 4 1 2             
 1667: 1 2 3
 1668: 1 2 3           1 2 3 4                 
 1669: 1 2 3           1 3             
 1670: @end example
 1671: @endassignment
 1672: 
 1673: @example
 1674: 5 dup * .
 1675: @end example
 1676: 
 1677: @assignment
 1678: Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
 1679: Write a piece of Forth code that expects two numbers on the stack
 1680: (@var{a} and @var{b}, with @var{b} on top) and computes
 1681: @code{(a-b)(a+1)}.
 1682: @endassignment
 1683: 
 1684: Reference: @ref{Stack Manipulation}.
 1685: 
 1686: 
 1687: @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
 1688: @section Using files for Forth code
 1689: @cindex loading Forth code, tutorial
 1690: @cindex files containing Forth code, tutorial
 1691: 
 1692: While working at the Forth command line is convenient for one-line
 1693: examples and short one-off code, you probably want to store your source
 1694: code in files for convenient editing and persistence.  You can use your
 1695: favourite editor (Gforth includes Emacs support, @pxref{Emacs and
 1696: Gforth}) to create @var{file} and use
 1697: 
 1698: @example
 1699: s" @var{file}" included
 1700: @end example
 1701: 
 1702: to load it into your Forth system.  The file name extension I use for
 1703: Forth files is @samp{.fs}.
 1704: 
 1705: You can easily start Gforth with some files loaded like this:
 1706: 
 1707: @example
 1708: gforth @var{file1} @var{file2}
 1709: @end example
 1710: 
 1711: If an error occurs during loading these files, Gforth terminates,
 1712: whereas an error during @code{INCLUDED} within Gforth usually gives you
 1713: a Gforth command line.  Starting the Forth system every time gives you a
 1714: clean start every time, without interference from the results of earlier
 1715: tries.
 1716: 
 1717: I often put all the tests in a file, then load the code and run the
 1718: tests with
 1719: 
 1720: @example
 1721: gforth @var{code} @var{tests} -e bye
 1722: @end example
 1723: 
 1724: (often by performing this command with @kbd{C-x C-e} in Emacs).  The
 1725: @code{-e bye} ensures that Gforth terminates afterwards so that I can
 1726: restart this command without ado.
 1727: 
 1728: The advantage of this approach is that the tests can be repeated easily
 1729: every time the program ist changed, making it easy to catch bugs
 1730: introduced by the change.
 1731: 
 1732: Reference: @ref{Forth source files}.
 1733: 
 1734: 
 1735: @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
 1736: @section Comments
 1737: @cindex comments tutorial
 1738: 
 1739: @example
 1740: \ That's a comment; it ends at the end of the line
 1741: ( Another comment; it ends here: )  .s
 1742: @end example
 1743: 
 1744: @code{\} and @code{(} are ordinary Forth words and therefore have to be
 1745: separated with white space from the following text.
 1746: 
 1747: @example
 1748: \This gives an "Undefined word" error
 1749: @end example
 1750: 
 1751: The first @code{)} ends a comment started with @code{(}, so you cannot
 1752: nest @code{(}-comments; and you cannot comment out text containing a
 1753: @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
 1754: avoid @code{)} in word names.}.
 1755: 
 1756: I use @code{\}-comments for descriptive text and for commenting out code
 1757: of one or more line; I use @code{(}-comments for describing the stack
 1758: effect, the stack contents, or for commenting out sub-line pieces of
 1759: code.
 1760: 
 1761: The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
 1762: these uses by commenting out a region with @kbd{C-x \}, uncommenting a
 1763: region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
 1764: with @kbd{M-q}.
 1765: 
 1766: Reference: @ref{Comments}.
 1767: 
 1768: 
 1769: @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
 1770: @section Colon Definitions
 1771: @cindex colon definitions, tutorial
 1772: @cindex definitions, tutorial
 1773: @cindex procedures, tutorial
 1774: @cindex functions, tutorial
 1775: 
 1776: are similar to procedures and functions in other programming languages.
 1777: 
 1778: @example
 1779: : squared ( n -- n^2 )
 1780:    dup * ;
 1781: 5 squared .
 1782: 7 squared .
 1783: @end example
 1784: 
 1785: @code{:} starts the colon definition; its name is @code{squared}.  The
 1786: following comment describes its stack effect.  The words @code{dup *}
 1787: are not executed, but compiled into the definition.  @code{;} ends the
 1788: colon definition.
 1789: 
 1790: The newly-defined word can be used like any other word, including using
 1791: it in other definitions:
 1792: 
 1793: @example
 1794: : cubed ( n -- n^3 )
 1795:    dup squared * ;
 1796: -5 cubed .
 1797: : fourth-power ( n -- n^4 )
 1798:    squared squared ;
 1799: 3 fourth-power .
 1800: @end example
 1801: 
 1802: @assignment
 1803: Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
 1804: @code{/mod} in terms of other Forth words, and check if they work (hint:
 1805: test your tests on the originals first).  Don't let the
 1806: @samp{redefined}-Messages spook you, they are just warnings.
 1807: @endassignment
 1808: 
 1809: Reference: @ref{Colon Definitions}.
 1810: 
 1811: 
 1812: @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
 1813: @section Decompilation
 1814: @cindex decompilation tutorial
 1815: @cindex see tutorial
 1816: 
 1817: You can decompile colon definitions with @code{see}:
 1818: 
 1819: @example
 1820: see squared
 1821: see cubed
 1822: @end example
 1823: 
 1824: In Gforth @code{see} shows you a reconstruction of the source code from
 1825: the executable code.  Informations that were present in the source, but
 1826: not in the executable code, are lost (e.g., comments).
 1827: 
 1828: You can also decompile the predefined words:
 1829: 
 1830: @example
 1831: see .
 1832: see +
 1833: @end example
 1834: 
 1835: 
 1836: @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
 1837: @section Stack-Effect Comments
 1838: @cindex stack-effect comments, tutorial
 1839: @cindex --, tutorial
 1840: By convention the comment after the name of a definition describes the
 1841: stack effect: The part in from of the @samp{--} describes the state of
 1842: the stack before the execution of the definition, i.e., the parameters
 1843: that are passed into the colon definition; the part behind the @samp{--}
 1844: is the state of the stack after the execution of the definition, i.e.,
 1845: the results of the definition.  The stack comment only shows the top
 1846: stack items that the definition accesses and/or changes.
 1847: 
 1848: You should put a correct stack effect on every definition, even if it is
 1849: just @code{( -- )}.  You should also add some descriptive comment to
 1850: more complicated words (I usually do this in the lines following
 1851: @code{:}).  If you don't do this, your code becomes unreadable (because
 1852: you have to work through every definition before you can undertsand
 1853: any).
 1854: 
 1855: @assignment
 1856: The stack effect of @code{swap} can be written like this: @code{x1 x2 --
 1857: x2 x1}.  Describe the stack effect of @code{-}, @code{drop}, @code{dup},
 1858: @code{over}, @code{rot}, @code{nip}, and @code{tuck}.  Hint: When you
 1859: are done, you can compare your stack effects to those in this manual
 1860: (@pxref{Word Index}).
 1861: @endassignment
 1862: 
 1863: Sometimes programmers put comments at various places in colon
 1864: definitions that describe the contents of the stack at that place (stack
 1865: comments); i.e., they are like the first part of a stack-effect
 1866: comment. E.g.,
 1867: 
 1868: @example
 1869: : cubed ( n -- n^3 )
 1870:    dup squared  ( n n^2 ) * ;
 1871: @end example
 1872: 
 1873: In this case the stack comment is pretty superfluous, because the word
 1874: is simple enough.  If you think it would be a good idea to add such a
 1875: comment to increase readability, you should also consider factoring the
 1876: word into several simpler words (@pxref{Factoring Tutorial,,
 1877: Factoring}), which typically eliminates the need for the stack comment;
 1878: however, if you decide not to refactor it, then having such a comment is
 1879: better than not having it.
 1880: 
 1881: The names of the stack items in stack-effect and stack comments in the
 1882: standard, in this manual, and in many programs specify the type through
 1883: a type prefix, similar to Fortran and Hungarian notation.  The most
 1884: frequent prefixes are:
 1885: 
 1886: @table @code
 1887: @item n
 1888: signed integer
 1889: @item u
 1890: unsigned integer
 1891: @item c
 1892: character
 1893: @item f
 1894: Boolean flags, i.e. @code{false} or @code{true}.
 1895: @item a-addr,a-
 1896: Cell-aligned address
 1897: @item c-addr,c-
 1898: Char-aligned address (note that a Char may have two bytes in Windows NT)
 1899: @item xt
 1900: Execution token, same size as Cell
 1901: @item w,x
 1902: Cell, can contain an integer or an address.  It usually takes 32, 64 or
 1903: 16 bits (depending on your platform and Forth system). A cell is more
 1904: commonly known as machine word, but the term @emph{word} already means
 1905: something different in Forth.
 1906: @item d
 1907: signed double-cell integer
 1908: @item ud
 1909: unsigned double-cell integer
 1910: @item r
 1911: Float (on the FP stack)
 1912: @end table
 1913: 
 1914: You can find a more complete list in @ref{Notation}.
 1915: 
 1916: @assignment
 1917: Write stack-effect comments for all definitions you have written up to
 1918: now.
 1919: @endassignment
 1920: 
 1921: 
 1922: @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
 1923: @section Types
 1924: @cindex types tutorial
 1925: 
 1926: In Forth the names of the operations are not overloaded; so similar
 1927: operations on different types need different names; e.g., @code{+} adds
 1928: integers, and you have to use @code{f+} to add floating-point numbers.
 1929: The following prefixes are often used for related operations on
 1930: different types:
 1931: 
 1932: @table @code
 1933: @item (none)
 1934: signed integer
 1935: @item u
 1936: unsigned integer
 1937: @item c
 1938: character
 1939: @item d
 1940: signed double-cell integer
 1941: @item ud, du
 1942: unsigned double-cell integer
 1943: @item 2
 1944: two cells (not-necessarily double-cell numbers)
 1945: @item m, um
 1946: mixed single-cell and double-cell operations
 1947: @item f
 1948: floating-point (note that in stack comments @samp{f} represents flags,
 1949: and @samp{r} represents FP numbers).
 1950: @end table
 1951: 
 1952: If there are no differences between the signed and the unsigned variant
 1953: (e.g., for @code{+}), there is only the prefix-less variant.
 1954: 
 1955: Forth does not perform type checking, neither at compile time, nor at
 1956: run time.  If you use the wrong oeration, the data are interpreted
 1957: incorrectly:
 1958: 
 1959: @example
 1960: -1 u.
 1961: @end example
 1962: 
 1963: If you have only experience with type-checked languages until now, and
 1964: have heard how important type-checking is, don't panic!  In my
 1965: experience (and that of other Forthers), type errors in Forth code are
 1966: usually easy to find (once you get used to it), the increased vigilance
 1967: of the programmer tends to catch some harder errors in addition to most
 1968: type errors, and you never have to work around the type system, so in
 1969: most situations the lack of type-checking seems to be a win (projects to
 1970: add type checking to Forth have not caught on).
 1971: 
 1972: 
 1973: @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
 1974: @section Factoring
 1975: @cindex factoring tutorial
 1976: 
 1977: If you try to write longer definitions, you will soon find it hard to
 1978: keep track of the stack contents.  Therefore, good Forth programmers
 1979: tend to write only short definitions (e.g., three lines).  The art of
 1980: finding meaningful short definitions is known as factoring (as in
 1981: factoring polynomials).
 1982: 
 1983: Well-factored programs offer additional advantages: smaller, more
 1984: general words, are easier to test and debug and can be reused more and
 1985: better than larger, specialized words.
 1986: 
 1987: So, if you run into difficulties with stack management, when writing
 1988: code, try to define meaningful factors for the word, and define the word
 1989: in terms of those.  Even if a factor contains only two words, it is
 1990: often helpful.
 1991: 
 1992: Good factoring is not easy, and it takes some practice to get the knack
 1993: for it; but even experienced Forth programmers often don't find the
 1994: right solution right away, but only when rewriting the program.  So, if
 1995: you don't come up with a good solution immediately, keep trying, don't
 1996: despair.
 1997: 
 1998: @c example !!
 1999: 
 2000: 
 2001: @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
 2002: @section Designing the stack effect
 2003: @cindex Stack effect design, tutorial
 2004: @cindex design of stack effects, tutorial
 2005: 
 2006: In other languages you can use an arbitrary order of parameters for a
 2007: function; and since there is only one result, you don't have to deal with
 2008: the order of results, either.
 2009: 
 2010: In Forth (and other stack-based languages, e.g., Postscript) the
 2011: parameter and result order of a definition is important and should be
 2012: designed well.  The general guideline is to design the stack effect such
 2013: that the word is simple to use in most cases, even if that complicates
 2014: the implementation of the word.  Some concrete rules are:
 2015: 
 2016: @itemize @bullet
 2017: 
 2018: @item
 2019: Words consume all of their parameters (e.g., @code{.}).
 2020: 
 2021: @item
 2022: If there is a convention on the order of parameters (e.g., from
 2023: mathematics or another programming language), stick with it (e.g.,
 2024: @code{-}).
 2025: 
 2026: @item
 2027: If one parameter usually requires only a short computation (e.g., it is
 2028: a constant), pass it on the top of the stack.  Conversely, parameters
 2029: that usually require a long sequence of code to compute should be passed
 2030: as the bottom (i.e., first) parameter.  This makes the code easier to
 2031: read, because reader does not need to keep track of the bottom item
 2032: through a long sequence of code (or, alternatively, through stack
 2033: manipulations). E.g., @code{!} (store, @pxref{Memory}) expects the
 2034: address on top of the stack because it is usually simpler to compute
 2035: than the stored value (often the address is just a variable).
 2036: 
 2037: @item
 2038: Similarly, results that are usually consumed quickly should be returned
 2039: on the top of stack, whereas a result that is often used in long
 2040: computations should be passed as bottom result.  E.g., the file words
 2041: like @code{open-file} return the error code on the top of stack, because
 2042: it is usually consumed quickly by @code{throw}; moreover, the error code
 2043: has to be checked before doing anything with the other results.
 2044: 
 2045: @end itemize
 2046: 
 2047: These rules are just general guidelines, don't lose sight of the overall
 2048: goal to make the words easy to use.  E.g., if the convention rule
 2049: conflicts with the computation-length rule, you might decide in favour
 2050: of the convention if the word will be used rarely, and in favour of the
 2051: computation-length rule if the word will be used frequently (because
 2052: with frequent use the cost of breaking the computation-length rule would
 2053: be quite high, and frequent use makes it easier to remember an
 2054: unconventional order).
 2055: 
 2056: @c example !! structure package
 2057: 
 2058: 
 2059: @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
 2060: @section Local Variables
 2061: @cindex local variables, tutorial
 2062: 
 2063: You can define local variables (@emph{locals}) in a colon definition:
 2064: 
 2065: @example
 2066: : swap @{ a b -- b a @}
 2067:   b a ;
 2068: 1 2 swap .s 2drop
 2069: @end example
 2070: 
 2071: (If your Forth system does not support this syntax, include
 2072: @file{compat/anslocals.fs} first).
 2073: 
 2074: In this example @code{@{ a b -- b a @}} is the locals definition; it
 2075: takes two cells from the stack, puts the top of stack in @code{b} and
 2076: the next stack element in @code{a}.  @code{--} starts a comment ending
 2077: with @code{@}}.  After the locals definition, using the name of the
 2078: local will push its value on the stack.  You can leave the comment
 2079: part (@code{-- b a}) away:
 2080: 
 2081: @example
 2082: : swap ( x1 x2 -- x2 x1 )
 2083:   @{ a b @} b a ;
 2084: @end example
 2085: 
 2086: In Gforth you can have several locals definitions, anywhere in a colon
 2087: definition; in contrast, in a standard program you can have only one
 2088: locals definition per colon definition, and that locals definition must
 2089: be outside any controll structure.
 2090: 
 2091: With locals you can write slightly longer definitions without running
 2092: into stack trouble.  However, I recommend trying to write colon
 2093: definitions without locals for exercise purposes to help you gain the
 2094: essential factoring skills.
 2095: 
 2096: @assignment
 2097: Rewrite your definitions until now with locals
 2098: @endassignment
 2099: 
 2100: Reference: @ref{Locals}.
 2101: 
 2102: 
 2103: @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
 2104: @section Conditional execution
 2105: @cindex conditionals, tutorial
 2106: @cindex if, tutorial
 2107: 
 2108: In Forth you can use control structures only inside colon definitions.
 2109: An @code{if}-structure looks like this:
 2110: 
 2111: @example
 2112: : abs ( n1 -- +n2 )
 2113:     dup 0 < if
 2114:         negate
 2115:     endif ;
 2116: 5 abs .
 2117: -5 abs .
 2118: @end example
 2119: 
 2120: @code{if} takes a flag from the stack.  If the flag is non-zero (true),
 2121: the following code is performed, otherwise execution continues after the
 2122: @code{endif} (or @code{else}).  @code{<} compares the top two stack
 2123: elements and prioduces a flag:
 2124: 
 2125: @example
 2126: 1 2 < .
 2127: 2 1 < .
 2128: 1 1 < .
 2129: @end example
 2130: 
 2131: Actually the standard name for @code{endif} is @code{then}.  This
 2132: tutorial presents the examples using @code{endif}, because this is often
 2133: less confusing for people familiar with other programming languages
 2134: where @code{then} has a different meaning.  If your system does not have
 2135: @code{endif}, define it with
 2136: 
 2137: @example
 2138: : endif postpone then ; immediate
 2139: @end example
 2140: 
 2141: You can optionally use an @code{else}-part:
 2142: 
 2143: @example
 2144: : min ( n1 n2 -- n )
 2145:   2dup < if
 2146:     drop
 2147:   else
 2148:     nip
 2149:   endif ;
 2150: 2 3 min .
 2151: 3 2 min .
 2152: @end example
 2153: 
 2154: @assignment
 2155: Write @code{min} without @code{else}-part (hint: what's the definition
 2156: of @code{nip}?).
 2157: @endassignment
 2158: 
 2159: Reference: @ref{Selection}.
 2160: 
 2161: 
 2162: @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
 2163: @section Flags and Comparisons
 2164: @cindex flags tutorial
 2165: @cindex comparison tutorial
 2166: 
 2167: In a false-flag all bits are clear (0 when interpreted as integer).  In
 2168: a canonical true-flag all bits are set (-1 as a twos-complement signed
 2169: integer); in many contexts (e.g., @code{if}) any non-zero value is
 2170: treated as true flag.
 2171: 
 2172: @example
 2173: false .
 2174: true .
 2175: true hex u. decimal
 2176: @end example
 2177: 
 2178: Comparison words produce canonical flags:
 2179: 
 2180: @example
 2181: 1 1 = .
 2182: 1 0= .
 2183: 0 1 < .
 2184: 0 0 < .
 2185: -1 1 u< . \ type error, u< interprets -1 as large unsigned number
 2186: -1 1 < .
 2187: @end example
 2188: 
 2189: Gforth supports all combinations of the prefixes @code{0 u d d0 du f f0}
 2190: (or none) and the comparisons @code{= <> < > <= >=}.  Only a part of
 2191: these combinations are standard (for details see the standard,
 2192: @ref{Numeric comparison}, @ref{Floating Point} or @ref{Word Index}).
 2193: 
 2194: You can use @code{and or xor invert} can be used as operations on
 2195: canonical flags.  Actually they are bitwise operations:
 2196: 
 2197: @example
 2198: 1 2 and .
 2199: 1 2 or .
 2200: 1 3 xor .
 2201: 1 invert .
 2202: @end example
 2203: 
 2204: You can convert a zero/non-zero flag into a canonical flag with
 2205: @code{0<>} (and complement it on the way with @code{0=}).
 2206: 
 2207: @example
 2208: 1 0= .
 2209: 1 0<> .
 2210: @end example
 2211: 
 2212: You can use the all-bits-set feature of canonical flags and the bitwise
 2213: operation of the Boolean operations to avoid @code{if}s:
 2214: 
 2215: @example
 2216: : foo ( n1 -- n2 )
 2217:   0= if
 2218:     14
 2219:   else
 2220:     0
 2221:   endif ;
 2222: 0 foo .
 2223: 1 foo .
 2224: 
 2225: : foo ( n1 -- n2 )
 2226:   0= 14 and ;
 2227: 0 foo .
 2228: 1 foo .
 2229: @end example
 2230: 
 2231: @assignment
 2232: Write @code{min} without @code{if}.
 2233: @endassignment
 2234: 
 2235: For reference, see @ref{Boolean Flags}, @ref{Numeric comparison}, and
 2236: @ref{Bitwise operations}.
 2237: 
 2238: 
 2239: @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
 2240: @section General Loops
 2241: @cindex loops, indefinite, tutorial
 2242: 
 2243: The endless loop is the most simple one:
 2244: 
 2245: @example
 2246: : endless ( -- )
 2247:   0 begin
 2248:     dup . 1+
 2249:   again ;
 2250: endless
 2251: @end example
 2252: 
 2253: Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth).  @code{begin}
 2254: does nothing at run-time, @code{again} jumps back to @code{begin}.
 2255: 
 2256: A loop with one exit at any place looks like this:
 2257: 
 2258: @example
 2259: : log2 ( +n1 -- n2 )
 2260: \ logarithmus dualis of n1>0, rounded down to the next integer
 2261:   assert( dup 0> )
 2262:   2/ 0 begin
 2263:     over 0> while
 2264:       1+ swap 2/ swap
 2265:   repeat
 2266:   nip ;
 2267: 7 log2 .
 2268: 8 log2 .
 2269: @end example
 2270: 
 2271: At run-time @code{while} consumes a flag; if it is 0, execution
 2272: continues behind the @code{repeat}; if the flag is non-zero, execution
 2273: continues behind the @code{while}.  @code{Repeat} jumps back to
 2274: @code{begin}, just like @code{again}.
 2275: 
 2276: In Forth there are many combinations/abbreviations, like @code{1+}.
 2277: However, @code{2/} is not one of them; it shifts it's argument right by
 2278: one bit (arithmetic shift right):
 2279: 
 2280: @example
 2281: -5 2 / .
 2282: -5 2/ .
 2283: @end example
 2284: 
 2285: @code{assert(} is no standard word, but you can get it on systems other
 2286: then Gforth by including @file{compat/assert.fs}.  You can see what it
 2287: does by trying
 2288: 
 2289: @example
 2290: 0 log2 .
 2291: @end example
 2292: 
 2293: Here's a loop with an exit at the end:
 2294: 
 2295: @example
 2296: : log2 ( +n1 -- n2 )
 2297: \ logarithmus dualis of n1>0, rounded down to the next integer
 2298:   assert( dup 0 > )
 2299:   -1 begin
 2300:     1+ swap 2/ swap
 2301:     over 0 <=
 2302:   until
 2303:   nip ;
 2304: @end example
 2305: 
 2306: @code{Until} consumes a flag; if it is non-zero, execution continues at
 2307: the @code{begin}, otherwise after the @code{until}.
 2308: 
 2309: @assignment
 2310: Write a definition for computing the greatest common divisor.
 2311: @endassignment
 2312: 
 2313: Reference: @ref{Simple Loops}.
 2314: 
 2315: 
 2316: @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
 2317: @section Counted loops
 2318: @cindex loops, counted, tutorial
 2319: 
 2320: @example
 2321: : ^ ( n1 u -- n )
 2322: \ n = the uth power of u1
 2323:   1 swap 0 u+do
 2324:     over *
 2325:   loop
 2326:   nip ;
 2327: 3 2 ^ .
 2328: 4 3 ^ .
 2329: @end example
 2330: 
 2331: @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
 2332: have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
 2333: performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
 2334: times (or not at all, if @code{u3-u4<0}).
 2335: 
 2336: You can see the stack effect design rules at work in the stack effect of
 2337: the loop start words: Since the start value of the loop is more
 2338: frequently constant than the end value, the start value is passed on
 2339: the top-of-stack.
 2340: 
 2341: You can access the counter of a counted loop with @code{i}:
 2342: 
 2343: @example
 2344: : fac ( u -- u! )
 2345:   1 swap 1+ 1 u+do
 2346:     i *
 2347:   loop ;
 2348: 5 fac .
 2349: 7 fac .
 2350: @end example
 2351: 
 2352: There is also @code{+do}, which expects signed numbers (important for
 2353: deciding whether to enter the loop).
 2354: 
 2355: @assignment
 2356: Write a definition for computing the nth Fibonacci number.
 2357: @endassignment
 2358: 
 2359: You can also use increments other than 1:
 2360: 
 2361: @example
 2362: : up2 ( n1 n2 -- )
 2363:   +do
 2364:     i .
 2365:   2 +loop ;
 2366: 10 0 up2
 2367: 
 2368: : down2 ( n1 n2 -- )
 2369:   -do
 2370:     i .
 2371:   2 -loop ;
 2372: 0 10 down2
 2373: @end example
 2374: 
 2375: Reference: @ref{Counted Loops}.
 2376: 
 2377: 
 2378: @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
 2379: @section Recursion
 2380: @cindex recursion tutorial
 2381: 
 2382: Usually the name of a definition is not visible in the definition; but
 2383: earlier definitions are usually visible:
 2384: 
 2385: @example
 2386: 1 0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
 2387: : / ( n1 n2 -- n )
 2388:   dup 0= if
 2389:     -10 throw \ report division by zero
 2390:   endif
 2391:   /           \ old version
 2392: ;
 2393: 1 0 /
 2394: @end example
 2395: 
 2396: For recursive definitions you can use @code{recursive} (non-standard) or
 2397: @code{recurse}:
 2398: 
 2399: @example
 2400: : fac1 ( n -- n! ) recursive
 2401:  dup 0> if
 2402:    dup 1- fac1 *
 2403:  else
 2404:    drop 1
 2405:  endif ;
 2406: 7 fac1 .
 2407: 
 2408: : fac2 ( n -- n! )
 2409:  dup 0> if
 2410:    dup 1- recurse *
 2411:  else
 2412:    drop 1
 2413:  endif ;
 2414: 8 fac2 .
 2415: @end example
 2416: 
 2417: @assignment
 2418: Write a recursive definition for computing the nth Fibonacci number.
 2419: @endassignment
 2420: 
 2421: Reference (including indirect recursion): @xref{Calls and returns}.
 2422: 
 2423: 
 2424: @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
 2425: @section Leaving definitions or loops
 2426: @cindex leaving definitions, tutorial
 2427: @cindex leaving loops, tutorial
 2428: 
 2429: @code{EXIT} exits the current definition right away.  For every counted
 2430: loop that is left in this way, an @code{UNLOOP} has to be performed
 2431: before the @code{EXIT}:
 2432: 
 2433: @c !! real examples
 2434: @example
 2435: : ...
 2436:  ... u+do
 2437:    ... if
 2438:      ... unloop exit
 2439:    endif
 2440:    ...
 2441:  loop
 2442:  ... ;
 2443: @end example
 2444: 
 2445: @code{LEAVE} leaves the innermost counted loop right away:
 2446: 
 2447: @example
 2448: : ...
 2449:  ... u+do
 2450:    ... if
 2451:      ... leave
 2452:    endif
 2453:    ...
 2454:  loop
 2455:  ... ;
 2456: @end example
 2457: 
 2458: @c !! example
 2459: 
 2460: Reference: @ref{Calls and returns}, @ref{Counted Loops}.
 2461: 
 2462: 
 2463: @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
 2464: @section Return Stack
 2465: @cindex return stack tutorial
 2466: 
 2467: In addition to the data stack Forth also has a second stack, the return
 2468: stack; most Forth systems store the return addresses of procedure calls
 2469: there (thus its name).  Programmers can also use this stack:
 2470: 
 2471: @example
 2472: : foo ( n1 n2 -- )
 2473:  .s
 2474:  >r .s
 2475:  r@@ .
 2476:  >r .s
 2477:  r@@ .
 2478:  r> .
 2479:  r@@ .
 2480:  r> . ;
 2481: 1 2 foo
 2482: @end example
 2483: 
 2484: @code{>r} takes an element from the data stack and pushes it onto the
 2485: return stack; conversely, @code{r>} moves an elementm from the return to
 2486: the data stack; @code{r@@} pushes a copy of the top of the return stack
 2487: on the return stack.
 2488: 
 2489: Forth programmers usually use the return stack for storing data
 2490: temporarily, if using the data stack alone would be too complex, and
 2491: factoring and locals are not an option:
 2492: 
 2493: @example
 2494: : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
 2495:  rot >r rot r> ;
 2496: @end example
 2497: 
 2498: The return address of the definition and the loop control parameters of
 2499: counted loops usually reside on the return stack, so you have to take
 2500: all items, that you have pushed on the return stack in a colon
 2501: definition or counted loop, from the return stack before the definition
 2502: or loop ends.  You cannot access items that you pushed on the return
 2503: stack outside some definition or loop within the definition of loop.
 2504: 
 2505: If you miscount the return stack items, this usually ends in a crash:
 2506: 
 2507: @example
 2508: : crash ( n -- )
 2509:   >r ;
 2510: 5 crash
 2511: @end example
 2512: 
 2513: You cannot mix using locals and using the return stack (according to the
 2514: standard; Gforth has no problem).  However, they solve the same
 2515: problems, so this shouldn't be an issue.
 2516: 
 2517: @assignment
 2518: Can you rewrite any of the definitions you wrote until now in a better
 2519: way using the return stack?
 2520: @endassignment
 2521: 
 2522: Reference: @ref{Return stack}.
 2523: 
 2524: 
 2525: @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
 2526: @section Memory
 2527: @cindex memory access/allocation tutorial
 2528: 
 2529: You can create a global variable @code{v} with
 2530: 
 2531: @example
 2532: variable v ( -- addr )
 2533: @end example
 2534: 
 2535: @code{v} pushes the address of a cell in memory on the stack.  This cell
 2536: was reserved by @code{variable}.  You can use @code{!} (store) to store
 2537: values into this cell and @code{@@} (fetch) to load the value from the
 2538: stack into memory:
 2539: 
 2540: @example
 2541: v .
 2542: 5 v ! .s
 2543: v @@ .
 2544: @end example
 2545: 
 2546: You can see a raw dump of memory with @code{dump}:
 2547: 
 2548: @example
 2549: v 1 cells .s dump
 2550: @end example
 2551: 
 2552: @code{Cells ( n1 -- n2 )} gives you the number of bytes (or, more
 2553: generally, address units (aus)) that @code{n1 cells} occupy.  You can
 2554: also reserve more memory:
 2555: 
 2556: @example
 2557: create v2 20 cells allot
 2558: v2 20 cells dump
 2559: @end example
 2560: 
 2561: creates a word @code{v2} and reserves 20 uninitialized cells; the
 2562: address pushed by @code{v2} points to the start of these 20 cells.  You
 2563: can use address arithmetic to access these cells:
 2564: 
 2565: @example
 2566: 3 v2 5 cells + !
 2567: v2 20 cells dump
 2568: @end example
 2569: 
 2570: You can reserve and initialize memory with @code{,}:
 2571: 
 2572: @example
 2573: create v3
 2574:   5 , 4 , 3 , 2 , 1 ,
 2575: v3 @@ .
 2576: v3 cell+ @@ .
 2577: v3 2 cells + @@ .
 2578: v3 5 cells dump
 2579: @end example
 2580: 
 2581: @assignment
 2582: Write a definition @code{vsum ( addr u -- n )} that computes the sum of
 2583: @code{u} cells, with the first of these cells at @code{addr}, the next
 2584: one at @code{addr cell+} etc.
 2585: @endassignment
 2586: 
 2587: You can also reserve memory without creating a new word:
 2588: 
 2589: @example
 2590: here 10 cells allot .
 2591: here .
 2592: @end example
 2593: 
 2594: @code{Here} pushes the start address of the memory area.  You should
 2595: store it somewhere, or you will have a hard time finding the memory area
 2596: again.
 2597: 
 2598: @code{Allot} manages dictionary memory.  The dictionary memory contains
 2599: the system's data structures for words etc. on Gforth and most other
 2600: Forth systems.  It is managed like a stack: You can free the memory that
 2601: you have just @code{allot}ed with
 2602: 
 2603: @example
 2604: -10 cells allot
 2605: here .
 2606: @end example
 2607: 
 2608: Note that you cannot do this if you have created a new word in the
 2609: meantime (because then your @code{allot}ed memory is no longer on the
 2610: top of the dictionary ``stack'').
 2611: 
 2612: Alternatively, you can use @code{allocate} and @code{free} which allow
 2613: freeing memory in any order:
 2614: 
 2615: @example
 2616: 10 cells allocate throw .s
 2617: 20 cells allocate throw .s
 2618: swap
 2619: free throw
 2620: free throw
 2621: @end example
 2622: 
 2623: The @code{throw}s deal with errors (e.g., out of memory).
 2624: 
 2625: And there is also a
 2626: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
 2627: garbage collector}, which eliminates the need to @code{free} memory
 2628: explicitly.
 2629: 
 2630: Reference: @ref{Memory}.
 2631: 
 2632: 
 2633: @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
 2634: @section Characters and Strings
 2635: @cindex strings tutorial
 2636: @cindex characters tutorial
 2637: 
 2638: On the stack characters take up a cell, like numbers.  In memory they
 2639: have their own size (one 8-bit byte on most systems), and therefore
 2640: require their own words for memory access:
 2641: 
 2642: @example
 2643: create v4 
 2644:   104 c, 97 c, 108 c, 108 c, 111 c,
 2645: v4 4 chars + c@@ .
 2646: v4 5 chars dump
 2647: @end example
 2648: 
 2649: The preferred representation of strings on the stack is @code{addr
 2650: u-count}, where @code{addr} is the address of the first character and
 2651: @code{u-count} is the number of characters in the string.
 2652: 
 2653: @example
 2654: v4 5 type
 2655: @end example
 2656: 
 2657: You get a string constant with
 2658: 
 2659: @example
 2660: s" hello, world" .s
 2661: type
 2662: @end example
 2663: 
 2664: Make sure you have a space between @code{s"} and the string; @code{s"}
 2665: is a normal Forth word and must be delimited with white space (try what
 2666: happens when you remove the space).
 2667: 
 2668: However, this interpretive use of @code{s"} is quite restricted: the
 2669: string exists only until the next call of @code{s"} (some Forth systems
 2670: keep more than one of these strings, but usually they still have a
 2671: limited lifetime).
 2672: 
 2673: @example
 2674: s" hello," s" world" .s
 2675: type
 2676: type
 2677: @end example
 2678: 
 2679: You can also use @code{s"} in a definition, and the resulting
 2680: strings then live forever (well, for as long as the definition):
 2681: 
 2682: @example
 2683: : foo s" hello," s" world" ;
 2684: foo .s
 2685: type
 2686: type
 2687: @end example
 2688: 
 2689: @assignment
 2690: @code{Emit ( c -- )} types @code{c} as character (not a number).
 2691: Implement @code{type ( addr u -- )}.
 2692: @endassignment
 2693: 
 2694: Reference: @ref{Memory Blocks}.
 2695: 
 2696: 
 2697: @node Alignment Tutorial, Files Tutorial, Characters and Strings Tutorial, Tutorial
 2698: @section Alignment
 2699: @cindex alignment tutorial
 2700: @cindex memory alignment tutorial
 2701: 
 2702: On many processors cells have to be aligned in memory, if you want to
 2703: access them with @code{@@} and @code{!} (and even if the processor does
 2704: not require alignment, access to aligned cells is faster).
 2705: 
 2706: @code{Create} aligns @code{here} (i.e., the place where the next
 2707: allocation will occur, and that the @code{create}d word points to).
 2708: Likewise, the memory produced by @code{allocate} starts at an aligned
 2709: address.  Adding a number of @code{cells} to an aligned address produces
 2710: another aligned address.
 2711: 
 2712: However, address arithmetic involving @code{char+} and @code{chars} can
 2713: create an address that is not cell-aligned.  @code{Aligned ( addr --
 2714: a-addr )} produces the next aligned address:
 2715: 
 2716: @example
 2717: v3 char+ aligned .s @@ .
 2718: v3 char+ .s @@ .
 2719: @end example
 2720: 
 2721: Similarly, @code{align} advances @code{here} to the next aligned
 2722: address:
 2723: 
 2724: @example
 2725: create v5 97 c,
 2726: here .
 2727: align here .
 2728: 1000 ,
 2729: @end example
 2730: 
 2731: Note that you should use aligned addresses even if your processor does
 2732: not require them, if you want your program to be portable.
 2733: 
 2734: Reference: @ref{Address arithmetic}.
 2735: 
 2736: 
 2737: @node Files Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Alignment Tutorial, Tutorial
 2738: @section Files
 2739: @cindex files tutorial
 2740: 
 2741: This section gives a short introduction into how to use files inside
 2742: Forth. It's broken up into five easy steps:
 2743: 
 2744: @enumerate 1
 2745: @item Opened an ASCII text file for input
 2746: @item Opened a file for output
 2747: @item Read input file until string matched (or some other condition matched)
 2748: @item Wrote some lines from input ( modified or not) to output
 2749: @item Closed the files.
 2750: @end enumerate
 2751: 
 2752: @subsection Open file for input
 2753: 
 2754: @example
 2755: s" foo.in"  r/o open-file throw Value fd-in
 2756: @end example
 2757: 
 2758: @subsection Create file for output
 2759: 
 2760: @example
 2761: s" foo.out" w/o create-file throw Value fd-out
 2762: @end example
 2763: 
 2764: The available file modes are r/o for read-only access, r/w for
 2765: read-write access, and w/o for write-only access. You could open both
 2766: files with r/w, too, if you like. All file words return error codes; for
 2767: most applications, it's best to pass there error codes with @code{throw}
 2768: to the outer error handler.
 2769: 
 2770: If you want words for opening and assigning, define them as follows:
 2771: 
 2772: @example
 2773: 0 Value fd-in
 2774: 0 Value fd-out
 2775: : open-input ( addr u -- )  r/o open-file throw to fd-in ;
 2776: : open-output ( addr u -- )  w/o create-file throw to fd-out ;
 2777: @end example
 2778: 
 2779: Usage example:
 2780: 
 2781: @example
 2782: s" foo.in" open-input
 2783: s" foo.out" open-output
 2784: @end example
 2785: 
 2786: @subsection Scan file for a particular line
 2787: 
 2788: @example
 2789: 256 Constant max-line
 2790: Create line-buffer  max-line 2 + allot
 2791: 
 2792: : scan-file ( addr u -- )
 2793:   begin
 2794:       line-buffer max-line fd-in read-line throw
 2795:   while
 2796:          >r 2dup line-buffer r> compare 0=
 2797:      until
 2798:   else
 2799:      drop
 2800:   then
 2801:   2drop ;
 2802: @end example
 2803: 
 2804: @code{read-line ( addr u1 fd -- u2 flag ior )} reads up to u1 bytes into
 2805: the buffer at addr, and returns the number of bytes read, a flag that's
 2806: true when the end of file is reached, and an error code.
 2807: 
 2808: @code{compare ( addr1 u1 addr2 u2 -- n )} compares two strings and
 2809: returns zero if both strings are equal. It returns a positive number if
 2810: the first string is lexically greater, a negative if the second string
 2811: is lexically greater.
 2812: 
 2813: We haven't seen this loop here; it has two exits. Since the @code{while}
 2814: exits with the number of bytes read on the stack, we have to clean up
 2815: that separately; that's after the @code{else}.
 2816: 
 2817: Usage example:
 2818: 
 2819: @example
 2820: s" The text I search is here" scan-file
 2821: @end example
 2822: 
 2823: @subsection Copy input to output
 2824: 
 2825: @example
 2826: : copy-file ( -- )
 2827:   begin
 2828:       line-buffer max-line fd-in read-line throw
 2829:   while
 2830:       line-buffer swap fd-out write-file throw
 2831:   repeat ;
 2832: @end example
 2833: 
 2834: @subsection Close files
 2835: 
 2836: @example
 2837: fd-in close-file throw
 2838: fd-out close-file throw
 2839: @end example
 2840: 
 2841: Likewise, you can put that into definitions, too:
 2842: 
 2843: @example
 2844: : close-input ( -- )  fd-in close-file throw ;
 2845: : close-output ( -- )  fd-out close-file throw ;
 2846: @end example
 2847: 
 2848: @assignment
 2849: How could you modify @code{copy-file} so that it copies until a second line is
 2850: matched? Can you write a program that extracts a section of a text file,
 2851: given the line that starts and the line that terminates that section?
 2852: @endassignment
 2853: 
 2854: @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Files Tutorial, Tutorial
 2855: @section Interpretation and Compilation Semantics and Immediacy
 2856: @cindex semantics tutorial
 2857: @cindex interpretation semantics tutorial
 2858: @cindex compilation semantics tutorial
 2859: @cindex immediate, tutorial
 2860: 
 2861: When a word is compiled, it behaves differently from being interpreted.
 2862: E.g., consider @code{+}:
 2863: 
 2864: @example
 2865: 1 2 + .
 2866: : foo + ;
 2867: @end example
 2868: 
 2869: These two behaviours are known as compilation and interpretation
 2870: semantics.  For normal words (e.g., @code{+}), the compilation semantics
 2871: is to append the interpretation semantics to the currently defined word
 2872: (@code{foo} in the example above).  I.e., when @code{foo} is executed
 2873: later, the interpretation semantics of @code{+} (i.e., adding two
 2874: numbers) will be performed.
 2875: 
 2876: However, there are words with non-default compilation semantics, e.g.,
 2877: the control-flow words like @code{if}.  You can use @code{immediate} to
 2878: change the compilation semantics of the last defined word to be equal to
 2879: the interpretation semantics:
 2880: 
 2881: @example
 2882: : [FOO] ( -- )
 2883:  5 . ; immediate
 2884: 
 2885: [FOO]
 2886: : bar ( -- )
 2887:   [FOO] ;
 2888: bar
 2889: see bar
 2890: @end example
 2891: 
 2892: Two conventions to mark words with non-default compilation semnatics are
 2893: names with brackets (more frequently used) and to write them all in
 2894: upper case (less frequently used).
 2895: 
 2896: In Gforth (and many other systems) you can also remove the
 2897: interpretation semantics with @code{compile-only} (the compilation
 2898: semantics is derived from the original interpretation semantics):
 2899: 
 2900: @example
 2901: : flip ( -- )
 2902:  6 . ; compile-only \ but not immediate
 2903: flip
 2904: 
 2905: : flop ( -- )
 2906:  flip ;
 2907: flop
 2908: @end example
 2909: 
 2910: In this example the interpretation semantics of @code{flop} is equal to
 2911: the original interpretation semantics of @code{flip}.
 2912: 
 2913: The text interpreter has two states: in interpret state, it performs the
 2914: interpretation semantics of words it encounters; in compile state, it
 2915: performs the compilation semantics of these words.
 2916: 
 2917: Among other things, @code{:} switches into compile state, and @code{;}
 2918: switches back to interpret state.  They contain the factors @code{]}
 2919: (switch to compile state) and @code{[} (switch to interpret state), that
 2920: do nothing but switch the state.
 2921: 
 2922: @example
 2923: : xxx ( -- )
 2924:   [ 5 . ]
 2925: ;
 2926: 
 2927: xxx
 2928: see xxx
 2929: @end example
 2930: 
 2931: These brackets are also the source of the naming convention mentioned
 2932: above.
 2933: 
 2934: Reference: @ref{Interpretation and Compilation Semantics}.
 2935: 
 2936: 
 2937: @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
 2938: @section Execution Tokens
 2939: @cindex execution tokens tutorial
 2940: @cindex XT tutorial
 2941: 
 2942: @code{' word} gives you the execution token (XT) of a word.  The XT is a
 2943: cell representing the interpretation semantics of a word.  You can
 2944: execute this semantics with @code{execute}:
 2945: 
 2946: @example
 2947: ' + .s
 2948: 1 2 rot execute .
 2949: @end example
 2950: 
 2951: The XT is similar to a function pointer in C.  However, parameter
 2952: passing through the stack makes it a little more flexible:
 2953: 
 2954: @example
 2955: : map-array ( ... addr u xt -- ... )
 2956: \ executes xt ( ... x -- ... ) for every element of the array starting
 2957: \ at addr and containing u elements
 2958:   @{ xt @}
 2959:   cells over + swap ?do
 2960:     i @@ xt execute
 2961:   1 cells +loop ;
 2962: 
 2963: create a 3 , 4 , 2 , -1 , 4 ,
 2964: a 5 ' . map-array .s
 2965: 0 a 5 ' + map-array .
 2966: s" max-n" environment? drop .s
 2967: a 5 ' min map-array .
 2968: @end example
 2969: 
 2970: You can use map-array with the XTs of words that consume one element
 2971: more than they produce.  In theory you can also use it with other XTs,
 2972: but the stack effect then depends on the size of the array, which is
 2973: hard to understand.
 2974: 
 2975: Since XTs are cell-sized, you can store them in memory and manipulate
 2976: them on the stack like other cells.  You can also compile the XT into a
 2977: word with @code{compile,}:
 2978: 
 2979: @example
 2980: : foo1 ( n1 n2 -- n )
 2981:    [ ' + compile, ] ;
 2982: see foo
 2983: @end example
 2984: 
 2985: This is non-standard, because @code{compile,} has no compilation
 2986: semantics in the standard, but it works in good Forth systems.  For the
 2987: broken ones, use
 2988: 
 2989: @example
 2990: : [compile,] compile, ; immediate
 2991: 
 2992: : foo1 ( n1 n2 -- n )
 2993:    [ ' + ] [compile,] ;
 2994: see foo
 2995: @end example
 2996: 
 2997: @code{'} is a word with default compilation semantics; it parses the
 2998: next word when its interpretation semantics are executed, not during
 2999: compilation:
 3000: 
 3001: @example
 3002: : foo ( -- xt )
 3003:   ' ;
 3004: see foo
 3005: : bar ( ... "word" -- ... )
 3006:   ' execute ;
 3007: see bar
 3008: 1 2 bar + .
 3009: @end example
 3010: 
 3011: You often want to parse a word during compilation and compile its XT so
 3012: it will be pushed on the stack at run-time.  @code{[']} does this:
 3013: 
 3014: @example
 3015: : xt-+ ( -- xt )
 3016:   ['] + ;
 3017: see xt-+
 3018: 1 2 xt-+ execute .
 3019: @end example
 3020: 
 3021: Many programmers tend to see @code{'} and the word it parses as one
 3022: unit, and expect it to behave like @code{[']} when compiled, and are
 3023: confused by the actual behaviour.  If you are, just remember that the
 3024: Forth system just takes @code{'} as one unit and has no idea that it is
 3025: a parsing word (attempts to convenience programmers in this issue have
 3026: usually resulted in even worse pitfalls, see
 3027: @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,
 3028: @code{State}-smartness---Why it is evil and How to Exorcise it}).
 3029: 
 3030: Note that the state of the interpreter does not come into play when
 3031: creating and executing XTs.  I.e., even when you execute @code{'} in
 3032: compile state, it still gives you the interpretation semantics.  And
 3033: whatever that state is, @code{execute} performs the semantics
 3034: represented by the XT (i.e., for XTs produced with @code{'} the
 3035: interpretation semantics).
 3036: 
 3037: Reference: @ref{Tokens for Words}.
 3038: 
 3039: 
 3040: @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
 3041: @section Exceptions
 3042: @cindex exceptions tutorial
 3043: 
 3044: @code{throw ( n -- )} causes an exception unless n is zero.
 3045: 
 3046: @example
 3047: 100 throw .s
 3048: 0 throw .s
 3049: @end example
 3050: 
 3051: @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
 3052: it catches exceptions and pushes the number of the exception on the
 3053: stack (or 0, if the xt executed without exception).  If there was an
 3054: exception, the stacks have the same depth as when entering @code{catch}:
 3055: 
 3056: @example
 3057: .s
 3058: 3 0 ' / catch .s
 3059: 3 2 ' / catch .s
 3060: @end example
 3061: 
 3062: @assignment
 3063: Try the same with @code{execute} instead of @code{catch}.
 3064: @endassignment
 3065: 
 3066: @code{Throw} always jumps to the dynamically next enclosing
 3067: @code{catch}, even if it has to leave several call levels to achieve
 3068: this:
 3069: 
 3070: @example
 3071: : foo 100 throw ;
 3072: : foo1 foo ." after foo" ;
 3073: : bar ['] foo1 catch ;
 3074: bar .
 3075: @end example
 3076: 
 3077: It is often important to restore a value upon leaving a definition, even
 3078: if the definition is left through an exception.  You can ensure this
 3079: like this:
 3080: 
 3081: @example
 3082: : ...
 3083:    save-x
 3084:    ['] word-changing-x catch ( ... n )
 3085:    restore-x
 3086:    ( ... n ) throw ;
 3087: @end example
 3088: 
 3089: Gforth provides an alternative syntax in addition to @code{catch}:
 3090: @code{try ... recover ... endtry}.  If the code between @code{try} and
 3091: @code{recover} has an exception, the stack depths are restored, the
 3092: exception number is pushed on the stack, and the code between
 3093: @code{recover} and @code{endtry} is performed.  E.g., the definition for
 3094: @code{catch} is
 3095: 
 3096: @example
 3097: : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
 3098:   try
 3099:     execute 0
 3100:   recover
 3101:     nip
 3102:   endtry ;
 3103: @end example
 3104: 
 3105: The equivalent to the restoration code above is
 3106: 
 3107: @example
 3108: : ...
 3109:   save-x
 3110:   try
 3111:     word-changing-x
 3112:   end-try
 3113:   restore-x
 3114:   throw ;
 3115: @end example
 3116: 
 3117: As you can see, the @code{recover} part is optional.
 3118: 
 3119: Reference: @ref{Exception Handling}.
 3120: 
 3121: 
 3122: @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
 3123: @section Defining Words
 3124: @cindex defining words tutorial
 3125: @cindex does> tutorial
 3126: @cindex create...does> tutorial
 3127: 
 3128: @c before semantics?
 3129: 
 3130: @code{:}, @code{create}, and @code{variable} are definition words: They
 3131: define other words.  @code{Constant} is another definition word:
 3132: 
 3133: @example
 3134: 5 constant foo
 3135: foo .
 3136: @end example
 3137: 
 3138: You can also use the prefixes @code{2} (double-cell) and @code{f}
 3139: (floating point) with @code{variable} and @code{constant}.
 3140: 
 3141: You can also define your own defining words.  E.g.:
 3142: 
 3143: @example
 3144: : variable ( "name" -- )
 3145:   create 0 , ;
 3146: @end example
 3147: 
 3148: You can also define defining words that create words that do something
 3149: other than just producing their address:
 3150: 
 3151: @example
 3152: : constant ( n "name" -- )
 3153:   create ,
 3154: does> ( -- n )
 3155:   ( addr ) @@ ;
 3156: 
 3157: 5 constant foo
 3158: foo .
 3159: @end example
 3160: 
 3161: The definition of @code{constant} above ends at the @code{does>}; i.e.,
 3162: @code{does>} replaces @code{;}, but it also does something else: It
 3163: changes the last defined word such that it pushes the address of the
 3164: body of the word and then performs the code after the @code{does>}
 3165: whenever it is called.
 3166: 
 3167: In the example above, @code{constant} uses @code{,} to store 5 into the
 3168: body of @code{foo}.  When @code{foo} executes, it pushes the address of
 3169: the body onto the stack, then (in the code after the @code{does>})
 3170: fetches the 5 from there.
 3171: 
 3172: The stack comment near the @code{does>} reflects the stack effect of the
 3173: defined word, not the stack effect of the code after the @code{does>}
 3174: (the difference is that the code expects the address of the body that
 3175: the stack comment does not show).
 3176: 
 3177: You can use these definition words to do factoring in cases that involve
 3178: (other) definition words.  E.g., a field offset is always added to an
 3179: address.  Instead of defining
 3180: 
 3181: @example
 3182: 2 cells constant offset-field1
 3183: @end example
 3184: 
 3185: and using this like
 3186: 
 3187: @example
 3188: ( addr ) offset-field1 +
 3189: @end example
 3190: 
 3191: you can define a definition word
 3192: 
 3193: @example
 3194: : simple-field ( n "name" -- )
 3195:   create ,
 3196: does> ( n1 -- n1+n )
 3197:   ( addr ) @@ + ;
 3198: @end example
 3199: 
 3200: Definition and use of field offsets now look like this:
 3201: 
 3202: @example
 3203: 2 cells simple-field field1
 3204: create mystruct 4 cells allot
 3205: mystruct .s field1 .s drop
 3206: @end example
 3207: 
 3208: If you want to do something with the word without performing the code
 3209: after the @code{does>}, you can access the body of a @code{create}d word
 3210: with @code{>body ( xt -- addr )}:
 3211: 
 3212: @example
 3213: : value ( n "name" -- )
 3214:   create ,
 3215: does> ( -- n1 )
 3216:   @@ ;
 3217: : to ( n "name" -- )
 3218:   ' >body ! ;
 3219: 
 3220: 5 value foo
 3221: foo .
 3222: 7 to foo
 3223: foo .
 3224: @end example
 3225: 
 3226: @assignment
 3227: Define @code{defer ( "name" -- )}, which creates a word that stores an
 3228: XT (at the start the XT of @code{abort}), and upon execution
 3229: @code{execute}s the XT.  Define @code{is ( xt "name" -- )} that stores
 3230: @code{xt} into @code{name}, a word defined with @code{defer}.  Indirect
 3231: recursion is one application of @code{defer}.
 3232: @endassignment
 3233: 
 3234: Reference: @ref{User-defined Defining Words}.
 3235: 
 3236: 
 3237: @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
 3238: @section Arrays and Records
 3239: @cindex arrays tutorial
 3240: @cindex records tutorial
 3241: @cindex structs tutorial
 3242: 
 3243: Forth has no standard words for defining data structures such as arrays
 3244: and records (structs in C terminology), but you can build them yourself
 3245: based on address arithmetic.  You can also define words for defining
 3246: arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
 3247: 
 3248: One of the first projects a Forth newcomer sets out upon when learning
 3249: about defining words is an array defining word (possibly for
 3250: n-dimensional arrays).  Go ahead and do it, I did it, too; you will
 3251: learn something from it.  However, don't be disappointed when you later
 3252: learn that you have little use for these words (inappropriate use would
 3253: be even worse).  I have not yet found a set of useful array words yet;
 3254: the needs are just too diverse, and named, global arrays (the result of
 3255: naive use of defining words) are often not flexible enough (e.g.,
 3256: consider how to pass them as parameters).  Another such project is a set
 3257: of words to help dealing with strings.
 3258: 
 3259: On the other hand, there is a useful set of record words, and it has
 3260: been defined in @file{compat/struct.fs}; these words are predefined in
 3261: Gforth.  They are explained in depth elsewhere in this manual (see
 3262: @pxref{Structures}).  The @code{simple-field} example above is
 3263: simplified variant of fields in this package.
 3264: 
 3265: 
 3266: @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
 3267: @section @code{POSTPONE}
 3268: @cindex postpone tutorial
 3269: 
 3270: You can compile the compilation semantics (instead of compiling the
 3271: interpretation semantics) of a word with @code{POSTPONE}:
 3272: 
 3273: @example
 3274: : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
 3275:  POSTPONE + ; immediate
 3276: : foo ( n1 n2 -- n )
 3277:  MY-+ ;
 3278: 1 2 foo .
 3279: see foo
 3280: @end example
 3281: 
 3282: During the definition of @code{foo} the text interpreter performs the
 3283: compilation semantics of @code{MY-+}, which performs the compilation
 3284: semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
 3285: 
 3286: This example also displays separate stack comments for the compilation
 3287: semantics and for the stack effect of the compiled code.  For words with
 3288: default compilation semantics these stack effects are usually not
 3289: displayed; the stack effect of the compilation semantics is always
 3290: @code{( -- )} for these words, the stack effect for the compiled code is
 3291: the stack effect of the interpretation semantics.
 3292: 
 3293: Note that the state of the interpreter does not come into play when
 3294: performing the compilation semantics in this way.  You can also perform
 3295: it interpretively, e.g.:
 3296: 
 3297: @example
 3298: : foo2 ( n1 n2 -- n )
 3299:  [ MY-+ ] ;
 3300: 1 2 foo .
 3301: see foo
 3302: @end example
 3303: 
 3304: However, there are some broken Forth systems where this does not always
 3305: work, and therefore this practice was been declared non-standard in
 3306: 1999.
 3307: @c !! repair.fs
 3308: 
 3309: Here is another example for using @code{POSTPONE}:
 3310: 
 3311: @example
 3312: : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
 3313:  POSTPONE negate POSTPONE + ; immediate compile-only
 3314: : bar ( n1 n2 -- n )
 3315:   MY-- ;
 3316: 2 1 bar .
 3317: see bar
 3318: @end example
 3319: 
 3320: You can define @code{ENDIF} in this way:
 3321: 
 3322: @example
 3323: : ENDIF ( Compilation: orig -- )
 3324:   POSTPONE then ; immediate
 3325: @end example
 3326: 
 3327: @assignment
 3328: Write @code{MY-2DUP} that has compilation semantics equivalent to
 3329: @code{2dup}, but compiles @code{over over}.
 3330: @endassignment
 3331: 
 3332: @c !! @xref{Macros} for reference
 3333: 
 3334: 
 3335: @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
 3336: @section @code{Literal}
 3337: @cindex literal tutorial
 3338: 
 3339: You cannot @code{POSTPONE} numbers:
 3340: 
 3341: @example
 3342: : [FOO] POSTPONE 500 ; immediate
 3343: @end example
 3344: 
 3345: Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
 3346: 
 3347: @example
 3348: : [FOO] ( compilation: --; run-time: -- n )
 3349:   500 POSTPONE literal ; immediate
 3350: 
 3351: : flip [FOO] ;
 3352: flip .
 3353: see flip
 3354: @end example
 3355: 
 3356: @code{LITERAL} consumes a number at compile-time (when it's compilation
 3357: semantics are executed) and pushes it at run-time (when the code it
 3358: compiled is executed).  A frequent use of @code{LITERAL} is to compile a
 3359: number computed at compile time into the current word:
 3360: 
 3361: @example
 3362: : bar ( -- n )
 3363:   [ 2 2 + ] literal ;
 3364: see bar
 3365: @end example
 3366: 
 3367: @assignment
 3368: Write @code{]L} which allows writing the example above as @code{: bar (
 3369: -- n ) [ 2 2 + ]L ;}
 3370: @endassignment
 3371: 
 3372: @c !! @xref{Macros} for reference
 3373: 
 3374: 
 3375: @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
 3376: @section Advanced macros
 3377: @cindex macros, advanced tutorial
 3378: @cindex run-time code generation, tutorial
 3379: 
 3380: Reconsider @code{map-array} from @ref{Execution Tokens Tutorial,,
 3381: Execution Tokens}.  It frequently performs @code{execute}, a relatively
 3382: expensive operation in some Forth implementations.  You can use
 3383: @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
 3384: and produce a word that contains the word to be performed directly:
 3385: 
 3386: @c use ]] ... [[
 3387: @example
 3388: : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
 3389: \ at run-time, execute xt ( ... x -- ... ) for each element of the
 3390: \ array beginning at addr and containing u elements
 3391:   @{ xt @}
 3392:   POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
 3393:     POSTPONE i POSTPONE @@ xt compile,
 3394:   1 cells POSTPONE literal POSTPONE +loop ;
 3395: 
 3396: : sum-array ( addr u -- n )
 3397:  0 rot rot [ ' + compile-map-array ] ;
 3398: see sum-array
 3399: a 5 sum-array .
 3400: @end example
 3401: 
 3402: You can use the full power of Forth for generating the code; here's an
 3403: example where the code is generated in a loop:
 3404: 
 3405: @example
 3406: : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
 3407: \ n2=n1+(addr1)*n, addr2=addr1+cell
 3408:   POSTPONE tuck POSTPONE @@
 3409:   POSTPONE literal POSTPONE * POSTPONE +
 3410:   POSTPONE swap POSTPONE cell+ ;
 3411: 
 3412: : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
 3413: \ n=v1*v2 (inner product), where the v_i are represented as addr_i u
 3414:   0 postpone literal postpone swap
 3415:   [ ' compile-vmul-step compile-map-array ]
 3416:   postpone drop ;
 3417: see compile-vmul
 3418: 
 3419: : a-vmul ( addr -- n )
 3420: \ n=a*v, where v is a vector that's as long as a and starts at addr
 3421:  [ a 5 compile-vmul ] ;
 3422: see a-vmul
 3423: a a-vmul .
 3424: @end example
 3425: 
 3426: This example uses @code{compile-map-array} to show off, but you could
 3427: also use @code{map-array} instead (try it now!).
 3428: 
 3429: You can use this technique for efficient multiplication of large
 3430: matrices.  In matrix multiplication, you multiply every line of one
 3431: matrix with every column of the other matrix.  You can generate the code
 3432: for one line once, and use it for every column.  The only downside of
 3433: this technique is that it is cumbersome to recover the memory consumed
 3434: by the generated code when you are done (and in more complicated cases
 3435: it is not possible portably).
 3436: 
 3437: @c !! @xref{Macros} for reference
 3438: 
 3439: 
 3440: @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
 3441: @section Compilation Tokens
 3442: @cindex compilation tokens, tutorial
 3443: @cindex CT, tutorial
 3444: 
 3445: This section is Gforth-specific.  You can skip it.
 3446: 
 3447: @code{' word compile,} compiles the interpretation semantics.  For words
 3448: with default compilation semantics this is the same as performing the
 3449: compilation semantics.  To represent the compilation semantics of other
 3450: words (e.g., words like @code{if} that have no interpretation
 3451: semantics), Gforth has the concept of a compilation token (CT,
 3452: consisting of two cells), and words @code{comp'} and @code{[comp']}.
 3453: You can perform the compilation semantics represented by a CT with
 3454: @code{execute}:
 3455: 
 3456: @example
 3457: : foo2 ( n1 n2 -- n )
 3458:    [ comp' + execute ] ;
 3459: see foo
 3460: @end example
 3461: 
 3462: You can compile the compilation semantics represented by a CT with
 3463: @code{postpone,}:
 3464: 
 3465: @example
 3466: : foo3 ( -- )
 3467:   [ comp' + postpone, ] ;
 3468: see foo3
 3469: @end example
 3470: 
 3471: @code{[ comp' word postpone, ]} is equivalent to @code{POSTPONE word}.
 3472: @code{comp'} is particularly useful for words that have no
 3473: interpretation semantics:
 3474: 
 3475: @example
 3476: ' if
 3477: comp' if .s 2drop
 3478: @end example
 3479: 
 3480: Reference: @ref{Tokens for Words}.
 3481: 
 3482: 
 3483: @node Wordlists and Search Order Tutorial,  , Compilation Tokens Tutorial, Tutorial
 3484: @section Wordlists and Search Order
 3485: @cindex wordlists tutorial
 3486: @cindex search order, tutorial
 3487: 
 3488: The dictionary is not just a memory area that allows you to allocate
 3489: memory with @code{allot}, it also contains the Forth words, arranged in
 3490: several wordlists.  When searching for a word in a wordlist,
 3491: conceptually you start searching at the youngest and proceed towards
 3492: older words (in reality most systems nowadays use hash-tables); i.e., if
 3493: you define a word with the same name as an older word, the new word
 3494: shadows the older word.
 3495: 
 3496: Which wordlists are searched in which order is determined by the search
 3497: order.  You can display the search order with @code{order}.  It displays
 3498: first the search order, starting with the wordlist searched first, then
 3499: it displays the wordlist that will contain newly defined words.
 3500: 
 3501: You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
 3502: 
 3503: @example
 3504: wordlist constant mywords
 3505: @end example
 3506: 
 3507: @code{Set-current ( wid -- )} sets the wordlist that will contain newly
 3508: defined words (the @emph{current} wordlist):
 3509: 
 3510: @example
 3511: mywords set-current
 3512: order
 3513: @end example
 3514: 
 3515: Gforth does not display a name for the wordlist in @code{mywords}
 3516: because this wordlist was created anonymously with @code{wordlist}.
 3517: 
 3518: You can get the current wordlist with @code{get-current ( -- wid)}.  If
 3519: you want to put something into a specific wordlist without overall
 3520: effect on the current wordlist, this typically looks like this:
 3521: 
 3522: @example
 3523: get-current mywords set-current ( wid )
 3524: create someword
 3525: ( wid ) set-current
 3526: @end example
 3527: 
 3528: You can write the search order with @code{set-order ( wid1 .. widn n --
 3529: )} and read it with @code{get-order ( -- wid1 .. widn n )}.  The first
 3530: searched wordlist is topmost.
 3531: 
 3532: @example
 3533: get-order mywords swap 1+ set-order
 3534: order
 3535: @end example
 3536: 
 3537: Yes, the order of wordlists in the output of @code{order} is reversed
 3538: from stack comments and the output of @code{.s} and thus unintuitive.
 3539: 
 3540: @assignment
 3541: Define @code{>order ( wid -- )} with adds @code{wid} as first searched
 3542: wordlist to the search order.  Define @code{previous ( -- )}, which
 3543: removes the first searched wordlist from the search order.  Experiment
 3544: with boundary conditions (you will see some crashes or situations that
 3545: are hard or impossible to leave).
 3546: @endassignment
 3547: 
 3548: The search order is a powerful foundation for providing features similar
 3549: to Modula-2 modules and C++ namespaces.  However, trying to modularize
 3550: programs in this way has disadvantages for debugging and reuse/factoring
 3551: that overcome the advantages in my experience (I don't do huge projects,
 3552: though).  These disadvantages are not so clear in other
 3553: languages/programming environments, because these languages are not so
 3554: strong in debugging and reuse.
 3555: 
 3556: @c !! example
 3557: 
 3558: Reference: @ref{Word Lists}.
 3559: 
 3560: @c ******************************************************************
 3561: @node Introduction, Words, Tutorial, Top
 3562: @comment node-name,     next,           previous, up
 3563: @chapter An Introduction to ANS Forth
 3564: @cindex Forth - an introduction
 3565: 
 3566: The difference of this chapter from the Tutorial (@pxref{Tutorial}) is
 3567: that it is slower-paced in its examples, but uses them to dive deep into
 3568: explaining Forth internals (not covered by the Tutorial).  Apart from
 3569: that, this chapter covers far less material.  It is suitable for reading
 3570: without using a computer.
 3571: 
 3572: The primary purpose of this manual is to document Gforth. However, since
 3573: Forth is not a widely-known language and there is a lack of up-to-date
 3574: teaching material, it seems worthwhile to provide some introductory
 3575: material.  For other sources of Forth-related
 3576: information, see @ref{Forth-related information}.
 3577: 
 3578: The examples in this section should work on any ANS Forth; the
 3579: output shown was produced using Gforth. Each example attempts to
 3580: reproduce the exact output that Gforth produces. If you try out the
 3581: examples (and you should), what you should type is shown @kbd{like this}
 3582: and Gforth's response is shown @code{like this}. The single exception is
 3583: that, where the example shows @key{RET} it means that you should
 3584: press the ``carriage return'' key. Unfortunately, some output formats for
 3585: this manual cannot show the difference between @kbd{this} and
 3586: @code{this} which will make trying out the examples harder (but not
 3587: impossible).
 3588: 
 3589: Forth is an unusual language. It provides an interactive development
 3590: environment which includes both an interpreter and compiler. Forth
 3591: programming style encourages you to break a problem down into many
 3592: @cindex factoring
 3593: small fragments (@dfn{factoring}), and then to develop and test each
 3594: fragment interactively. Forth advocates assert that breaking the
 3595: edit-compile-test cycle used by conventional programming languages can
 3596: lead to great productivity improvements.
 3597: 
 3598: @menu
 3599: * Introducing the Text Interpreter::  
 3600: * Stacks and Postfix notation::  
 3601: * Your first definition::       
 3602: * How does that work?::         
 3603: * Forth is written in Forth::   
 3604: * Review - elements of a Forth system::  
 3605: * Where to go next::            
 3606: * Exercises::                   
 3607: @end menu
 3608: 
 3609: @comment ----------------------------------------------
 3610: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
 3611: @section Introducing the Text Interpreter
 3612: @cindex text interpreter
 3613: @cindex outer interpreter
 3614: 
 3615: @c IMO this is too detailed and the pace is too slow for
 3616: @c an introduction.  If you know German, take a look at
 3617: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html 
 3618: @c to see how I do it - anton 
 3619: 
 3620: @c nac-> Where I have accepted your comments 100% and modified the text
 3621: @c accordingly, I have deleted your comments. Elsewhere I have added a
 3622: @c response like this to attempt to rationalise what I have done. Of
 3623: @c course, this is a very clumsy mechanism for something that would be
 3624: @c done far more efficiently over a beer. Please delete any dialogue
 3625: @c you consider closed.
 3626: 
 3627: When you invoke the Forth image, you will see a startup banner printed
 3628: and nothing else (if you have Gforth installed on your system, try
 3629: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
 3630: its command line interpreter, which is called the @dfn{Text Interpreter}
 3631: (also known as the @dfn{Outer Interpreter}).  (You will learn a lot
 3632: about the text interpreter as you read through this chapter, for more
 3633: detail @pxref{The Text Interpreter}).
 3634: 
 3635: Although it's not obvious, Forth is actually waiting for your
 3636: input. Type a number and press the @key{RET} key:
 3637: 
 3638: @example
 3639: @kbd{45@key{RET}}  ok
 3640: @end example
 3641: 
 3642: Rather than give you a prompt to invite you to input something, the text
 3643: interpreter prints a status message @i{after} it has processed a line
 3644: of input. The status message in this case (``@code{ ok}'' followed by
 3645: carriage-return) indicates that the text interpreter was able to process
 3646: all of your input successfully. Now type something illegal:
 3647: 
 3648: @example
 3649: @kbd{qwer341@key{RET}}
 3650: :1: Undefined word
 3651: qwer341
 3652: ^^^^^^^
 3653: $400D2BA8 Bounce
 3654: $400DBDA8 no.extensions
 3655: @end example
 3656: 
 3657: The exact text, other than the ``Undefined word'' may differ slightly on
 3658: your system, but the effect is the same; when the text interpreter
 3659: detects an error, it discards any remaining text on a line, resets
 3660: certain internal state and prints an error message. For a detailed description of error messages see @ref{Error
 3661: messages}.
 3662: 
 3663: The text interpreter waits for you to press carriage-return, and then
 3664: processes your input line. Starting at the beginning of the line, it
 3665: breaks the line into groups of characters separated by spaces. For each
 3666: group of characters in turn, it makes two attempts to do something:
 3667: 
 3668: @itemize @bullet
 3669: @item
 3670: @cindex name dictionary
 3671: It tries to treat it as a command. It does this by searching a @dfn{name
 3672: dictionary}. If the group of characters matches an entry in the name
 3673: dictionary, the name dictionary provides the text interpreter with
 3674: information that allows the text interpreter perform some actions. In
 3675: Forth jargon, we say that the group
 3676: @cindex word
 3677: @cindex definition
 3678: @cindex execution token
 3679: @cindex xt
 3680: of characters names a @dfn{word}, that the dictionary search returns an
 3681: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
 3682: word, and that the text interpreter executes the xt. Often, the terms
 3683: @dfn{word} and @dfn{definition} are used interchangeably.
 3684: @item
 3685: If the text interpreter fails to find a match in the name dictionary, it
 3686: tries to treat the group of characters as a number in the current number
 3687: base (when you start up Forth, the current number base is base 10). If
 3688: the group of characters legitimately represents a number, the text
 3689: interpreter pushes the number onto a stack (we'll learn more about that
 3690: in the next section).
 3691: @end itemize
 3692: 
 3693: If the text interpreter is unable to do either of these things with any
 3694: group of characters, it discards the group of characters and the rest of
 3695: the line, then prints an error message. If the text interpreter reaches
 3696: the end of the line without error, it prints the status message ``@code{ ok}''
 3697: followed by carriage-return.
 3698: 
 3699: This is the simplest command we can give to the text interpreter:
 3700: 
 3701: @example
 3702: @key{RET}  ok
 3703: @end example
 3704: 
 3705: The text interpreter did everything we asked it to do (nothing) without
 3706: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
 3707: command:
 3708: 
 3709: @example
 3710: @kbd{12 dup fred dup@key{RET}}
 3711: :1: Undefined word
 3712: 12 dup fred dup
 3713:        ^^^^
 3714: $400D2BA8 Bounce
 3715: $400DBDA8 no.extensions
 3716: @end example
 3717: 
 3718: When you press the carriage-return key, the text interpreter starts to
 3719: work its way along the line:
 3720: 
 3721: @itemize @bullet
 3722: @item
 3723: When it gets to the space after the @code{2}, it takes the group of
 3724: characters @code{12} and looks them up in the name
 3725: dictionary@footnote{We can't tell if it found them or not, but assume
 3726: for now that it did not}. There is no match for this group of characters
 3727: in the name dictionary, so it tries to treat them as a number. It is
 3728: able to do this successfully, so it puts the number, 12, ``on the stack''
 3729: (whatever that means).
 3730: @item
 3731: The text interpreter resumes scanning the line and gets the next group
 3732: of characters, @code{dup}. It looks it up in the name dictionary and
 3733: (you'll have to take my word for this) finds it, and executes the word
 3734: @code{dup} (whatever that means).
 3735: @item
 3736: Once again, the text interpreter resumes scanning the line and gets the
 3737: group of characters @code{fred}. It looks them up in the name
 3738: dictionary, but can't find them. It tries to treat them as a number, but
 3739: they don't represent any legal number.
 3740: @end itemize
 3741: 
 3742: At this point, the text interpreter gives up and prints an error
 3743: message. The error message shows exactly how far the text interpreter
 3744: got in processing the line. In particular, it shows that the text
 3745: interpreter made no attempt to do anything with the final character
 3746: group, @code{dup}, even though we have good reason to believe that the
 3747: text interpreter would have no problem looking that word up and
 3748: executing it a second time.
 3749: 
 3750: 
 3751: @comment ----------------------------------------------
 3752: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
 3753: @section Stacks, postfix notation and parameter passing
 3754: @cindex text interpreter
 3755: @cindex outer interpreter
 3756: 
 3757: In procedural programming languages (like C and Pascal), the
 3758: building-block of programs is the @dfn{function} or @dfn{procedure}. These
 3759: functions or procedures are called with @dfn{explicit parameters}. For
 3760: example, in C we might write:
 3761: 
 3762: @example
 3763: total = total + new_volume(length,height,depth);
 3764: @end example
 3765: 
 3766: @noindent
 3767: where new_volume is a function-call to another piece of code, and total,
 3768: length, height and depth are all variables. length, height and depth are
 3769: parameters to the function-call.
 3770: 
 3771: In Forth, the equivalent of the function or procedure is the
 3772: @dfn{definition} and parameters are implicitly passed between
 3773: definitions using a shared stack that is visible to the
 3774: programmer. Although Forth does support variables, the existence of the
 3775: stack means that they are used far less often than in most other
 3776: programming languages. When the text interpreter encounters a number, it
 3777: will place (@dfn{push}) it on the stack. There are several stacks (the
 3778: actual number is implementation-dependent ...) and the particular stack
 3779: used for any operation is implied unambiguously by the operation being
 3780: performed. The stack used for all integer operations is called the @dfn{data
 3781: stack} and, since this is the stack used most commonly, references to
 3782: ``the data stack'' are often abbreviated to ``the stack''.
 3783: 
 3784: The stacks have a last-in, first-out (LIFO) organisation. If you type:
 3785: 
 3786: @example
 3787: @kbd{1 2 3@key{RET}}  ok
 3788: @end example
 3789: 
 3790: Then this instructs the text interpreter to placed three numbers on the
 3791: (data) stack. An analogy for the behaviour of the stack is to take a
 3792: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
 3793: the table. The 3 was the last card onto the pile (``last-in'') and if
 3794: you take a card off the pile then, unless you're prepared to fiddle a
 3795: bit, the card that you take off will be the 3 (``first-out''). The
 3796: number that will be first-out of the stack is called the @dfn{top of
 3797: stack}, which
 3798: @cindex TOS definition
 3799: is often abbreviated to @dfn{TOS}.
 3800: 
 3801: To understand how parameters are passed in Forth, consider the
 3802: behaviour of the definition @code{+} (pronounced ``plus''). You will not
 3803: be surprised to learn that this definition performs addition. More
 3804: precisely, it adds two number together and produces a result. Where does
 3805: it get the two numbers from? It takes the top two numbers off the
 3806: stack. Where does it place the result? On the stack. You can act-out the
 3807: behaviour of @code{+} with your playing cards like this:
 3808: 
 3809: @itemize @bullet
 3810: @item
 3811: Pick up two cards from the stack on the table
 3812: @item
 3813: Stare at them intently and ask yourself ``what @i{is} the sum of these two
 3814: numbers''
 3815: @item
 3816: Decide that the answer is 5
 3817: @item
 3818: Shuffle the two cards back into the pack and find a 5
 3819: @item
 3820: Put a 5 on the remaining ace that's on the table.
 3821: @end itemize
 3822: 
 3823: If you don't have a pack of cards handy but you do have Forth running,
 3824: you can use the definition @code{.s} to show the current state of the stack,
 3825: without affecting the stack. Type:
 3826: 
 3827: @example
 3828: @kbd{clearstack 1 2 3@key{RET}} ok
 3829: @kbd{.s@key{RET}} <3> 1 2 3  ok
 3830: @end example
 3831: 
 3832: The text interpreter looks up the word @code{clearstack} and executes
 3833: it; it tidies up the stack and removes any entries that may have been
 3834: left on it by earlier examples. The text interpreter pushes each of the
 3835: three numbers in turn onto the stack. Finally, the text interpreter
 3836: looks up the word @code{.s} and executes it. The effect of executing
 3837: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
 3838: followed by a list of all the items on the stack; the item on the far
 3839: right-hand side is the TOS.
 3840: 
 3841: You can now type:
 3842: 
 3843: @example
 3844: @kbd{+ .s@key{RET}} <2> 1 5  ok
 3845: @end example
 3846: 
 3847: @noindent
 3848: which is correct; there are now 2 items on the stack and the result of
 3849: the addition is 5.
 3850: 
 3851: If you're playing with cards, try doing a second addition: pick up the
 3852: two cards, work out that their sum is 6, shuffle them into the pack,
 3853: look for a 6 and place that on the table. You now have just one item on
 3854: the stack. What happens if you try to do a third addition? Pick up the
 3855: first card, pick up the second card -- ah! There is no second card. This
 3856: is called a @dfn{stack underflow} and consitutes an error. If you try to
 3857: do the same thing with Forth it will report an error (probably a Stack
 3858: Underflow or an Invalid Memory Address error).
 3859: 
 3860: The opposite situation to a stack underflow is a @dfn{stack overflow},
 3861: which simply accepts that there is a finite amount of storage space
 3862: reserved for the stack. To stretch the playing card analogy, if you had
 3863: enough packs of cards and you piled the cards up on the table, you would
 3864: eventually be unable to add another card; you'd hit the ceiling. Gforth
 3865: allows you to set the maximum size of the stacks. In general, the only
 3866: time that you will get a stack overflow is because a definition has a
 3867: bug in it and is generating data on the stack uncontrollably.
 3868: 
 3869: There's one final use for the playing card analogy. If you model your
 3870: stack using a pack of playing cards, the maximum number of items on
 3871: your stack will be 52 (I assume you didn't use the Joker). The maximum
 3872: @i{value} of any item on the stack is 13 (the King). In fact, the only
 3873: possible numbers are positive integer numbers 1 through 13; you can't
 3874: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
 3875: think about some of the cards, you can accommodate different
 3876: numbers. For example, you could think of the Jack as representing 0,
 3877: the Queen as representing -1 and the King as representing -2. Your
 3878: @i{range} remains unchanged (you can still only represent a total of 13
 3879: numbers) but the numbers that you can represent are -2 through 10.
 3880: 
 3881: In that analogy, the limit was the amount of information that a single
 3882: stack entry could hold, and Forth has a similar limit. In Forth, the
 3883: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
 3884: implementation dependent and affects the maximum value that a stack
 3885: entry can hold. A Standard Forth provides a cell size of at least
 3886: 16-bits, and most desktop systems use a cell size of 32-bits.
 3887: 
 3888: Forth does not do any type checking for you, so you are free to
 3889: manipulate and combine stack items in any way you wish. A convenient way
 3890: of treating stack items is as 2's complement signed integers, and that
 3891: is what Standard words like @code{+} do. Therefore you can type:
 3892: 
 3893: @example
 3894: @kbd{-5 12 + .s@key{RET}} <1> 7  ok
 3895: @end example
 3896: 
 3897: If you use numbers and definitions like @code{+} in order to turn Forth
 3898: into a great big pocket calculator, you will realise that it's rather
 3899: different from a normal calculator. Rather than typing 2 + 3 = you had
 3900: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
 3901: result). The terminology used to describe this difference is to say that
 3902: your calculator uses @dfn{Infix Notation} (parameters and operators are
 3903: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
 3904: operators are separate), also called @dfn{Reverse Polish Notation}.
 3905: 
 3906: Whilst postfix notation might look confusing to begin with, it has
 3907: several important advantages:
 3908: 
 3909: @itemize @bullet
 3910: @item
 3911: it is unambiguous
 3912: @item
 3913: it is more concise
 3914: @item
 3915: it fits naturally with a stack-based system
 3916: @end itemize
 3917: 
 3918: To examine these claims in more detail, consider these sums:
 3919: 
 3920: @example
 3921: 6 + 5 * 4 =
 3922: 4 * 5 + 6 =
 3923: @end example
 3924: 
 3925: If you're just learning maths or your maths is very rusty, you will
 3926: probably come up with the answer 44 for the first and 26 for the
 3927: second. If you are a bit of a whizz at maths you will remember the
 3928: @i{convention} that multiplication takes precendence over addition, and
 3929: you'd come up with the answer 26 both times. To explain the answer 26
 3930: to someone who got the answer 44, you'd probably rewrite the first sum
 3931: like this:
 3932: 
 3933: @example
 3934: 6 + (5 * 4) =
 3935: @end example
 3936: 
 3937: If what you really wanted was to perform the addition before the
 3938: multiplication, you would have to use parentheses to force it.
 3939: 
 3940: If you did the first two sums on a pocket calculator you would probably
 3941: get the right answers, unless you were very cautious and entered them using
 3942: these keystroke sequences:
 3943: 
 3944: 6 + 5 = * 4 =
 3945: 4 * 5 = + 6 =
 3946: 
 3947: Postfix notation is unambiguous because the order that the operators
 3948: are applied is always explicit; that also means that parentheses are
 3949: never required. The operators are @i{active} (the act of quoting the
 3950: operator makes the operation occur) which removes the need for ``=''.
 3951: 
 3952: The sum 6 + 5 * 4 can be written (in postfix notation) in two
 3953: equivalent ways:
 3954: 
 3955: @example
 3956: 6 5 4 * +      or:
 3957: 5 4 * 6 +
 3958: @end example
 3959: 
 3960: An important thing that you should notice about this notation is that
 3961: the @i{order} of the numbers does not change; if you want to subtract
 3962: 2 from 10 you type @code{10 2 -}.
 3963: 
 3964: The reason that Forth uses postfix notation is very simple to explain: it
 3965: makes the implementation extremely simple, and it follows naturally from
 3966: using the stack as a mechanism for passing parameters. Another way of
 3967: thinking about this is to realise that all Forth definitions are
 3968: @i{active}; they execute as they are encountered by the text
 3969: interpreter. The result of this is that the syntax of Forth is trivially
 3970: simple.
 3971: 
 3972: 
 3973: 
 3974: @comment ----------------------------------------------
 3975: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
 3976: @section Your first Forth definition
 3977: @cindex first definition
 3978: 
 3979: Until now, the examples we've seen have been trivial; we've just been
 3980: using Forth as a bigger-than-pocket calculator. Also, each calculation
 3981: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
 3982: again@footnote{That's not quite true. If you press the up-arrow key on
 3983: your keyboard you should be able to scroll back to any earlier command,
 3984: edit it and re-enter it.} In this section we'll see how to add new
 3985: words to Forth's vocabulary.
 3986: 
 3987: The easiest way to create a new word is to use a @dfn{colon
 3988: definition}. We'll define a few and try them out before worrying too
 3989: much about how they work. Try typing in these examples; be careful to
 3990: copy the spaces accurately:
 3991: 
 3992: @example
 3993: : add-two 2 + . ;
 3994: : greet ." Hello and welcome" ;
 3995: : demo 5 add-two ;
 3996: @end example
 3997: 
 3998: @noindent
 3999: Now try them out:
 4000: 
 4001: @example
 4002: @kbd{greet@key{RET}} Hello and welcome  ok
 4003: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome  ok
 4004: @kbd{4 add-two@key{RET}} 6  ok
 4005: @kbd{demo@key{RET}} 7  ok
 4006: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11  ok
 4007: @end example
 4008: 
 4009: The first new thing that we've introduced here is the pair of words
 4010: @code{:} and @code{;}. These are used to start and terminate a new
 4011: definition, respectively. The first word after the @code{:} is the name
 4012: for the new definition.
 4013: 
 4014: As you can see from the examples, a definition is built up of words that
 4015: have already been defined; Forth makes no distinction between
 4016: definitions that existed when you started the system up, and those that
 4017: you define yourself.
 4018: 
 4019: The examples also introduce the words @code{.} (dot), @code{."}
 4020: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
 4021: the stack and displays it. It's like @code{.s} except that it only
 4022: displays the top item of the stack and it is destructive; after it has
 4023: executed, the number is no longer on the stack. There is always one
 4024: space printed after the number, and no spaces before it. Dot-quote
 4025: defines a string (a sequence of characters) that will be printed when
 4026: the word is executed. The string can contain any printable characters
 4027: except @code{"}. A @code{"} has a special function; it is not a Forth
 4028: word but it acts as a delimiter (the way that delimiters work is
 4029: described in the next section). Finally, @code{dup} duplicates the value
 4030: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
 4031: 
 4032: We already know that the text interpreter searches through the
 4033: dictionary to locate names. If you've followed the examples earlier, you
 4034: will already have a definition called @code{add-two}. Lets try modifying
 4035: it by typing in a new definition:
 4036: 
 4037: @example
 4038: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two  ok
 4039: @end example
 4040: 
 4041: Forth recognised that we were defining a word that already exists, and
 4042: printed a message to warn us of that fact. Let's try out the new
 4043: definition:
 4044: 
 4045: @example
 4046: @kbd{9 add-two@key{RET}} 9 + 2 =11  ok
 4047: @end example
 4048: 
 4049: @noindent
 4050: All that we've actually done here, though, is to create a new
 4051: definition, with a particular name. The fact that there was already a
 4052: definition with the same name did not make any difference to the way
 4053: that the new definition was created (except that Forth printed a warning
 4054: message). The old definition of add-two still exists (try @code{demo}
 4055: again to see that this is true). Any new definition will use the new
 4056: definition of @code{add-two}, but old definitions continue to use the
 4057: version that already existed at the time that they were @code{compiled}.
 4058: 
 4059: Before you go on to the next section, try defining and redefining some
 4060: words of your own.
 4061: 
 4062: @comment ----------------------------------------------
 4063: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
 4064: @section How does that work?
 4065: @cindex parsing words
 4066: 
 4067: @c That's pretty deep (IMO way too deep) for an introduction. - anton
 4068: 
 4069: @c Is it a good idea to talk about the interpretation semantics of a
 4070: @c number? We don't have an xt to go along with it. - anton
 4071: 
 4072: @c Now that I have eliminated execution semantics, I wonder if it would not
 4073: @c be better to keep them (or add run-time semantics), to make it easier to
 4074: @c explain what compilation semantics usually does. - anton
 4075: 
 4076: @c nac-> I removed the term ``default compilation sematics'' from the
 4077: @c introductory chapter. Removing ``execution semantics'' was making
 4078: @c everything simpler to explain, then I think the use of this term made
 4079: @c everything more complex again. I replaced it with ``default
 4080: @c semantics'' (which is used elsewhere in the manual) by which I mean
 4081: @c ``a definition that has neither the immediate nor the compile-only
 4082: @c flag set''.
 4083: 
 4084: @c anton: I have eliminated default semantics (except in one place where it
 4085: @c means "default interpretation and compilation semantics"), because it
 4086: @c makes no sense in the presence of combined words.  I reverted to
 4087: @c "execution semantics" where necessary.
 4088: 
 4089: @c nac-> I reworded big chunks of the ``how does that work''
 4090: @c section (and, unusually for me, I think I even made it shorter!).  See
 4091: @c what you think -- I know I have not addressed your primary concern
 4092: @c that it is too heavy-going for an introduction. From what I understood
 4093: @c of your course notes it looks as though they might be a good framework. 
 4094: @c Things that I've tried to capture here are some things that came as a
 4095: @c great revelation here when I first understood them. Also, I like the
 4096: @c fact that a very simple code example shows up almost all of the issues
 4097: @c that you need to understand to see how Forth works. That's unique and
 4098: @c worthwhile to emphasise.
 4099: 
 4100: @c anton: I think it's a good idea to present the details, especially those
 4101: @c that you found to be a revelation, and probably the tutorial tries to be
 4102: @c too superficial and does not get some of the things across that make
 4103: @c Forth special.  I do believe that most of the time these things should
 4104: @c be discussed at the end of a section or in separate sections instead of
 4105: @c in the middle of a section (e.g., the stuff you added in "User-defined
 4106: @c defining words" leads in a completely different direction from the rest
 4107: @c of the section).
 4108: 
 4109: Now we're going to take another look at the definition of @code{add-two}
 4110: from the previous section. From our knowledge of the way that the text
 4111: interpreter works, we would have expected this result when we tried to
 4112: define @code{add-two}:
 4113: 
 4114: @example
 4115: @kbd{: add-two 2 + . ;@key{RET}}
 4116:   ^^^^^^^
 4117: Error: Undefined word
 4118: @end example
 4119: 
 4120: The reason that this didn't happen is bound up in the way that @code{:}
 4121: works. The word @code{:} does two special things. The first special
 4122: thing that it does prevents the text interpreter from ever seeing the
 4123: characters @code{add-two}. The text interpreter uses a variable called
 4124: @cindex modifying >IN
 4125: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
 4126: input line. When it encounters the word @code{:} it behaves in exactly
 4127: the same way as it does for any other word; it looks it up in the name
 4128: dictionary, finds its xt and executes it. When @code{:} executes, it
 4129: looks at the input buffer, finds the word @code{add-two} and advances the
 4130: value of @code{>IN} to point past it. It then does some other stuff
 4131: associated with creating the new definition (including creating an entry
 4132: for @code{add-two} in the name dictionary). When the execution of @code{:}
 4133: completes, control returns to the text interpreter, which is oblivious
 4134: to the fact that it has been tricked into ignoring part of the input
 4135: line.
 4136: 
 4137: @cindex parsing words
 4138: Words like @code{:} -- words that advance the value of @code{>IN} and so
 4139: prevent the text interpreter from acting on the whole of the input line
 4140: -- are called @dfn{parsing words}.
 4141: 
 4142: @cindex @code{state} - effect on the text interpreter
 4143: @cindex text interpreter - effect of state
 4144: The second special thing that @code{:} does is change the value of a
 4145: variable called @code{state}, which affects the way that the text
 4146: interpreter behaves. When Gforth starts up, @code{state} has the value
 4147: 0, and the text interpreter is said to be @dfn{interpreting}. During a
 4148: colon definition (started with @code{:}), @code{state} is set to -1 and
 4149: the text interpreter is said to be @dfn{compiling}.
 4150: 
 4151: In this example, the text interpreter is compiling when it processes the
 4152: string ``@code{2 + . ;}''. It still breaks the string down into
 4153: character sequences in the same way. However, instead of pushing the
 4154: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
 4155: into the definition of @code{add-two} that will make the number @code{2} get
 4156: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
 4157: the behaviours of @code{+} and @code{.} are also compiled into the
 4158: definition.
 4159: 
 4160: One category of words don't get compiled. These so-called @dfn{immediate
 4161: words} get executed (performed @i{now}) regardless of whether the text
 4162: interpreter is interpreting or compiling. The word @code{;} is an
 4163: immediate word. Rather than being compiled into the definition, it
 4164: executes. Its effect is to terminate the current definition, which
 4165: includes changing the value of @code{state} back to 0.
 4166: 
 4167: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
 4168: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
 4169: definition.
 4170: 
 4171: In Forth, every word or number can be described in terms of two
 4172: properties:
 4173: 
 4174: @itemize @bullet
 4175: @item
 4176: @cindex interpretation semantics
 4177: Its @dfn{interpretation semantics} describe how it will behave when the
 4178: text interpreter encounters it in @dfn{interpret} state. The
 4179: interpretation semantics of a word are represented by an @dfn{execution
 4180: token}.
 4181: @item
 4182: @cindex compilation semantics
 4183: Its @dfn{compilation semantics} describe how it will behave when the
 4184: text interpreter encounters it in @dfn{compile} state. The compilation
 4185: semantics of a word are represented in an implementation-dependent way;
 4186: Gforth uses a @dfn{compilation token}.
 4187: @end itemize
 4188: 
 4189: @noindent
 4190: Numbers are always treated in a fixed way:
 4191: 
 4192: @itemize @bullet
 4193: @item
 4194: When the number is @dfn{interpreted}, its behaviour is to push the
 4195: number onto the stack.
 4196: @item
 4197: When the number is @dfn{compiled}, a piece of code is appended to the
 4198: current definition that pushes the number when it runs. (In other words,
 4199: the compilation semantics of a number are to postpone its interpretation
 4200: semantics until the run-time of the definition that it is being compiled
 4201: into.)
 4202: @end itemize
 4203: 
 4204: Words don't behave in such a regular way, but most have @i{default
 4205: semantics} which means that they behave like this:
 4206: 
 4207: @itemize @bullet
 4208: @item
 4209: The @dfn{interpretation semantics} of the word are to do something useful.
 4210: @item
 4211: The @dfn{compilation semantics} of the word are to append its
 4212: @dfn{interpretation semantics} to the current definition (so that its
 4213: run-time behaviour is to do something useful).
 4214: @end itemize
 4215: 
 4216: @cindex immediate words
 4217: The actual behaviour of any particular word can be controlled by using
 4218: the words @code{immediate} and @code{compile-only} when the word is
 4219: defined. These words set flags in the name dictionary entry of the most
 4220: recently defined word, and these flags are retrieved by the text
 4221: interpreter when it finds the word in the name dictionary.
 4222: 
 4223: A word that is marked as @dfn{immediate} has compilation semantics that
 4224: are identical to its interpretation semantics. In other words, it
 4225: behaves like this:
 4226: 
 4227: @itemize @bullet
 4228: @item
 4229: The @dfn{interpretation semantics} of the word are to do something useful.
 4230: @item
 4231: The @dfn{compilation semantics} of the word are to do something useful
 4232: (and actually the same thing); i.e., it is executed during compilation.
 4233: @end itemize
 4234: 
 4235: Marking a word as @dfn{compile-only} prohibits the text interpreter from
 4236: performing the interpretation semantics of the word directly; an attempt
 4237: to do so will generate an error. It is never necessary to use
 4238: @code{compile-only} (and it is not even part of ANS Forth, though it is
 4239: provided by many implementations) but it is good etiquette to apply it
 4240: to a word that will not behave correctly (and might have unexpected
 4241: side-effects) in interpret state. For example, it is only legal to use
 4242: the conditional word @code{IF} within a definition. If you forget this
 4243: and try to use it elsewhere, the fact that (in Gforth) it is marked as
 4244: @code{compile-only} allows the text interpreter to generate a helpful
 4245: error message rather than subjecting you to the consequences of your
 4246: folly.
 4247: 
 4248: This example shows the difference between an immediate and a
 4249: non-immediate word:
 4250: 
 4251: @example
 4252: : show-state state @@ . ;
 4253: : show-state-now show-state ; immediate
 4254: : word1 show-state ;
 4255: : word2 show-state-now ;
 4256: @end example
 4257: 
 4258: The word @code{immediate} after the definition of @code{show-state-now}
 4259: makes that word an immediate word. These definitions introduce a new
 4260: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
 4261: variable, and leaves it on the stack. Therefore, the behaviour of
 4262: @code{show-state} is to print a number that represents the current value
 4263: of @code{state}.
 4264: 
 4265: When you execute @code{word1}, it prints the number 0, indicating that
 4266: the system is interpreting. When the text interpreter compiled the
 4267: definition of @code{word1}, it encountered @code{show-state} whose
 4268: compilation semantics are to append its interpretation semantics to the
 4269: current definition. When you execute @code{word1}, it performs the
 4270: interpretation semantics of @code{show-state}.  At the time that @code{word1}
 4271: (and therefore @code{show-state}) are executed, the system is
 4272: interpreting.
 4273: 
 4274: When you pressed @key{RET} after entering the definition of @code{word2},
 4275: you should have seen the number -1 printed, followed by ``@code{
 4276: ok}''. When the text interpreter compiled the definition of
 4277: @code{word2}, it encountered @code{show-state-now}, an immediate word,
 4278: whose compilation semantics are therefore to perform its interpretation
 4279: semantics. It is executed straight away (even before the text
 4280: interpreter has moved on to process another group of characters; the
 4281: @code{;} in this example). The effect of executing it are to display the
 4282: value of @code{state} @i{at the time that the definition of}
 4283: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
 4284: system is compiling at this time. If you execute @code{word2} it does
 4285: nothing at all.
 4286: 
 4287: @cindex @code{."}, how it works
 4288: Before leaving the subject of immediate words, consider the behaviour of
 4289: @code{."} in the definition of @code{greet}, in the previous
 4290: section. This word is both a parsing word and an immediate word. Notice
 4291: that there is a space between @code{."} and the start of the text
 4292: @code{Hello and welcome}, but that there is no space between the last
 4293: letter of @code{welcome} and the @code{"} character. The reason for this
 4294: is that @code{."} is a Forth word; it must have a space after it so that
 4295: the text interpreter can identify it. The @code{"} is not a Forth word;
 4296: it is a @dfn{delimiter}. The examples earlier show that, when the string
 4297: is displayed, there is neither a space before the @code{H} nor after the
 4298: @code{e}. Since @code{."} is an immediate word, it executes at the time
 4299: that @code{greet} is defined. When it executes, its behaviour is to
 4300: search forward in the input line looking for the delimiter. When it
 4301: finds the delimiter, it updates @code{>IN} to point past the
 4302: delimiter. It also compiles some magic code into the definition of
 4303: @code{greet}; the xt of a run-time routine that prints a text string. It
 4304: compiles the string @code{Hello and welcome} into memory so that it is
 4305: available to be printed later. When the text interpreter gains control,
 4306: the next word it finds in the input stream is @code{;} and so it
 4307: terminates the definition of @code{greet}.
 4308: 
 4309: 
 4310: @comment ----------------------------------------------
 4311: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
 4312: @section Forth is written in Forth
 4313: @cindex structure of Forth programs
 4314: 
 4315: When you start up a Forth compiler, a large number of definitions
 4316: already exist. In Forth, you develop a new application using bottom-up
 4317: programming techniques to create new definitions that are defined in
 4318: terms of existing definitions. As you create each definition you can
 4319: test and debug it interactively.
 4320: 
 4321: If you have tried out the examples in this section, you will probably
 4322: have typed them in by hand; when you leave Gforth, your definitions will
 4323: be lost. You can avoid this by using a text editor to enter Forth source
 4324: code into a file, and then loading code from the file using
 4325: @code{include} (@pxref{Forth source files}). A Forth source file is
 4326: processed by the text interpreter, just as though you had typed it in by
 4327: hand@footnote{Actually, there are some subtle differences -- see
 4328: @ref{The Text Interpreter}.}.
 4329: 
 4330: Gforth also supports the traditional Forth alternative to using text
 4331: files for program entry (@pxref{Blocks}).
 4332: 
 4333: In common with many, if not most, Forth compilers, most of Gforth is
 4334: actually written in Forth. All of the @file{.fs} files in the
 4335: installation directory@footnote{For example,
 4336: @file{/usr/local/share/gforth...}} are Forth source files, which you can
 4337: study to see examples of Forth programming.
 4338: 
 4339: Gforth maintains a history file that records every line that you type to
 4340: the text interpreter. This file is preserved between sessions, and is
 4341: used to provide a command-line recall facility. If you enter long
 4342: definitions by hand, you can use a text editor to paste them out of the
 4343: history file into a Forth source file for reuse at a later time
 4344: (for more information @pxref{Command-line editing}).
 4345: 
 4346: 
 4347: @comment ----------------------------------------------
 4348: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
 4349: @section Review - elements of a Forth system
 4350: @cindex elements of a Forth system
 4351: 
 4352: To summarise this chapter:
 4353: 
 4354: @itemize @bullet
 4355: @item
 4356: Forth programs use @dfn{factoring} to break a problem down into small
 4357: fragments called @dfn{words} or @dfn{definitions}.
 4358: @item
 4359: Forth program development is an interactive process.
 4360: @item
 4361: The main command loop that accepts input, and controls both
 4362: interpretation and compilation, is called the @dfn{text interpreter}
 4363: (also known as the @dfn{outer interpreter}).
 4364: @item
 4365: Forth has a very simple syntax, consisting of words and numbers
 4366: separated by spaces or carriage-return characters. Any additional syntax
 4367: is imposed by @dfn{parsing words}.
 4368: @item
 4369: Forth uses a stack to pass parameters between words. As a result, it
 4370: uses postfix notation.
 4371: @item
 4372: To use a word that has previously been defined, the text interpreter
 4373: searches for the word in the @dfn{name dictionary}.
 4374: @item
 4375: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
 4376: @item
 4377: The text interpreter uses the value of @code{state} to select between
 4378: the use of the @dfn{interpretation semantics} and the  @dfn{compilation
 4379: semantics} of a word that it encounters.
 4380: @item
 4381: The relationship between the @dfn{interpretation semantics} and
 4382: @dfn{compilation semantics} for a word
 4383: depend upon the way in which the word was defined (for example, whether
 4384: it is an @dfn{immediate} word).
 4385: @item
 4386: Forth definitions can be implemented in Forth (called @dfn{high-level
 4387: definitions}) or in some other way (usually a lower-level language and
 4388: as a result often called @dfn{low-level definitions}, @dfn{code
 4389: definitions} or @dfn{primitives}).
 4390: @item
 4391: Many Forth systems are implemented mainly in Forth.
 4392: @end itemize
 4393: 
 4394: 
 4395: @comment ----------------------------------------------
 4396: @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
 4397: @section Where To Go Next
 4398: @cindex where to go next
 4399: 
 4400: Amazing as it may seem, if you have read (and understood) this far, you
 4401: know almost all the fundamentals about the inner workings of a Forth
 4402: system. You certainly know enough to be able to read and understand the
 4403: rest of this manual and the ANS Forth document, to learn more about the
 4404: facilities that Forth in general and Gforth in particular provide. Even
 4405: scarier, you know almost enough to implement your own Forth system.
 4406: However, that's not a good idea just yet... better to try writing some
 4407: programs in Gforth.
 4408: 
 4409: Forth has such a rich vocabulary that it can be hard to know where to
 4410: start in learning it. This section suggests a few sets of words that are
 4411: enough to write small but useful programs. Use the word index in this
 4412: document to learn more about each word, then try it out and try to write
 4413: small definitions using it. Start by experimenting with these words:
 4414: 
 4415: @itemize @bullet
 4416: @item
 4417: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
 4418: @item
 4419: Comparison: @code{MIN MAX =}
 4420: @item
 4421: Logic: @code{AND OR XOR NOT}
 4422: @item
 4423: Stack manipulation: @code{DUP DROP SWAP OVER}
 4424: @item
 4425: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
 4426: @item
 4427: Input/Output: @code{. ." EMIT CR KEY}
 4428: @item
 4429: Defining words: @code{: ; CREATE}
 4430: @item
 4431: Memory allocation words: @code{ALLOT ,}
 4432: @item
 4433: Tools: @code{SEE WORDS .S MARKER}
 4434: @end itemize
 4435: 
 4436: When you have mastered those, go on to:
 4437: 
 4438: @itemize @bullet
 4439: @item
 4440: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
 4441: @item
 4442: Memory access: @code{@@ !}
 4443: @end itemize
 4444: 
 4445: When you have mastered these, there's nothing for it but to read through
 4446: the whole of this manual and find out what you've missed.
 4447: 
 4448: @comment ----------------------------------------------
 4449: @node Exercises,  , Where to go next, Introduction
 4450: @section Exercises
 4451: @cindex exercises
 4452: 
 4453: TODO: provide a set of programming excercises linked into the stuff done
 4454: already and into other sections of the manual. Provide solutions to all
 4455: the exercises in a .fs file in the distribution.
 4456: 
 4457: @c Get some inspiration from Starting Forth and Kelly&Spies.
 4458: 
 4459: @c excercises:
 4460: @c 1. take inches and convert to feet and inches.
 4461: @c 2. take temperature and convert from fahrenheight to celcius;
 4462: @c    may need to care about symmetric vs floored??
 4463: @c 3. take input line and do character substitution
 4464: @c    to encipher or decipher
 4465: @c 4. as above but work on a file for in and out
 4466: @c 5. take input line and convert to pig-latin 
 4467: @c
 4468: @c thing of sets of things to exercise then come up with
 4469: @c problems that need those things.
 4470: 
 4471: 
 4472: @c ******************************************************************
 4473: @node Words, Error messages, Introduction, Top
 4474: @chapter Forth Words
 4475: @cindex words
 4476: 
 4477: @menu
 4478: * Notation::                    
 4479: * Case insensitivity::          
 4480: * Comments::                    
 4481: * Boolean Flags::               
 4482: * Arithmetic::                  
 4483: * Stack Manipulation::          
 4484: * Memory::                      
 4485: * Control Structures::          
 4486: * Defining Words::              
 4487: * Interpretation and Compilation Semantics::  
 4488: * Tokens for Words::            
 4489: * Compiling words::             
 4490: * The Text Interpreter::        
 4491: * Word Lists::                  
 4492: * Environmental Queries::       
 4493: * Files::                       
 4494: * Blocks::                      
 4495: * Other I/O::                   
 4496: * Locals::                      
 4497: * Structures::                  
 4498: * Object-oriented Forth::       
 4499: * Programming Tools::           
 4500: * Assembler and Code Words::    
 4501: * Threading Words::             
 4502: * Passing Commands to the OS::  
 4503: * Keeping track of Time::       
 4504: * Miscellaneous Words::         
 4505: @end menu
 4506: 
 4507: @node Notation, Case insensitivity, Words, Words
 4508: @section Notation
 4509: @cindex notation of glossary entries
 4510: @cindex format of glossary entries
 4511: @cindex glossary notation format
 4512: @cindex word glossary entry format
 4513: 
 4514: The Forth words are described in this section in the glossary notation
 4515: that has become a de-facto standard for Forth texts:
 4516: 
 4517: @format
 4518: @i{word}     @i{Stack effect}   @i{wordset}   @i{pronunciation}
 4519: @end format
 4520: @i{Description}
 4521: 
 4522: @table @var
 4523: @item word
 4524: The name of the word.
 4525: 
 4526: @item Stack effect
 4527: @cindex stack effect
 4528: The stack effect is written in the notation @code{@i{before} --
 4529: @i{after}}, where @i{before} and @i{after} describe the top of
 4530: stack entries before and after the execution of the word. The rest of
 4531: the stack is not touched by the word. The top of stack is rightmost,
 4532: i.e., a stack sequence is written as it is typed in. Note that Gforth
 4533: uses a separate floating point stack, but a unified stack
 4534: notation. Also, return stack effects are not shown in @i{stack
 4535: effect}, but in @i{Description}. The name of a stack item describes
 4536: the type and/or the function of the item. See below for a discussion of
 4537: the types.
 4538: 
 4539: All words have two stack effects: A compile-time stack effect and a
 4540: run-time stack effect. The compile-time stack-effect of most words is
 4541: @i{ -- }. If the compile-time stack-effect of a word deviates from
 4542: this standard behaviour, or the word does other unusual things at
 4543: compile time, both stack effects are shown; otherwise only the run-time
 4544: stack effect is shown.
 4545: 
 4546: @cindex pronounciation of words
 4547: @item pronunciation
 4548: How the word is pronounced.
 4549: 
 4550: @cindex wordset
 4551: @cindex environment wordset
 4552: @item wordset
 4553: The ANS Forth standard is divided into several word sets. A standard
 4554: system need not support all of them. Therefore, in theory, the fewer
 4555: word sets your program uses the more portable it will be. However, we
 4556: suspect that most ANS Forth systems on personal machines will feature
 4557: all word sets. Words that are not defined in ANS Forth have
 4558: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
 4559: describes words that will work in future releases of Gforth;
 4560: @code{gforth-internal} words are more volatile. Environmental query
 4561: strings are also displayed like words; you can recognize them by the
 4562: @code{environment} in the word set field.
 4563: 
 4564: @item Description
 4565: A description of the behaviour of the word.
 4566: @end table
 4567: 
 4568: @cindex types of stack items
 4569: @cindex stack item types
 4570: The type of a stack item is specified by the character(s) the name
 4571: starts with:
 4572: 
 4573: @table @code
 4574: @item f
 4575: @cindex @code{f}, stack item type
 4576: Boolean flags, i.e. @code{false} or @code{true}.
 4577: @item c
 4578: @cindex @code{c}, stack item type
 4579: Char
 4580: @item w
 4581: @cindex @code{w}, stack item type
 4582: Cell, can contain an integer or an address
 4583: @item n
 4584: @cindex @code{n}, stack item type
 4585: signed integer
 4586: @item u
 4587: @cindex @code{u}, stack item type
 4588: unsigned integer
 4589: @item d
 4590: @cindex @code{d}, stack item type
 4591: double sized signed integer
 4592: @item ud
 4593: @cindex @code{ud}, stack item type
 4594: double sized unsigned integer
 4595: @item r
 4596: @cindex @code{r}, stack item type
 4597: Float (on the FP stack)
 4598: @item a-
 4599: @cindex @code{a_}, stack item type
 4600: Cell-aligned address
 4601: @item c-
 4602: @cindex @code{c_}, stack item type
 4603: Char-aligned address (note that a Char may have two bytes in Windows NT)
 4604: @item f-
 4605: @cindex @code{f_}, stack item type
 4606: Float-aligned address
 4607: @item df-
 4608: @cindex @code{df_}, stack item type
 4609: Address aligned for IEEE double precision float
 4610: @item sf-
 4611: @cindex @code{sf_}, stack item type
 4612: Address aligned for IEEE single precision float
 4613: @item xt
 4614: @cindex @code{xt}, stack item type
 4615: Execution token, same size as Cell
 4616: @item wid
 4617: @cindex @code{wid}, stack item type
 4618: Word list ID, same size as Cell
 4619: @item ior, wior
 4620: @cindex ior type description
 4621: @cindex wior type description
 4622: I/O result code, cell-sized.  In Gforth, you can @code{throw} iors.
 4623: @item f83name
 4624: @cindex @code{f83name}, stack item type
 4625: Pointer to a name structure
 4626: @item "
 4627: @cindex @code{"}, stack item type
 4628: string in the input stream (not on the stack). The terminating character
 4629: is a blank by default. If it is not a blank, it is shown in @code{<>}
 4630: quotes.
 4631: @end table
 4632: 
 4633: @comment ----------------------------------------------
 4634: @node Case insensitivity, Comments, Notation, Words
 4635: @section Case insensitivity
 4636: @cindex case sensitivity
 4637: @cindex upper and lower case
 4638: 
 4639: Gforth is case-insensitive; you can enter definitions and invoke
 4640: Standard words using upper, lower or mixed case (however,
 4641: @pxref{core-idef, Implementation-defined options, Implementation-defined
 4642: options}).
 4643: 
 4644: ANS Forth only @i{requires} implementations to recognise Standard words
 4645: when they are typed entirely in upper case. Therefore, a Standard
 4646: program must use upper case for all Standard words. You can use whatever
 4647: case you like for words that you define, but in a Standard program you
 4648: have to use the words in the same case that you defined them.
 4649: 
 4650: Gforth supports case sensitivity through @code{table}s (case-sensitive
 4651: wordlists, @pxref{Word Lists}).
 4652: 
 4653: Two people have asked how to convert Gforth to be case-sensitive; while
 4654: we think this is a bad idea, you can change all wordlists into tables
 4655: like this:
 4656: 
 4657: @example
 4658: ' table-find forth-wordlist wordlist-map @ !
 4659: @end example
 4660: 
 4661: Note that you now have to type the predefined words in the same case
 4662: that we defined them, which are varying.  You may want to convert them
 4663: to your favourite case before doing this operation (I won't explain how,
 4664: because if you are even contemplating doing this, you'd better have
 4665: enough knowledge of Forth systems to know this already).
 4666: 
 4667: @node Comments, Boolean Flags, Case insensitivity, Words
 4668: @section Comments
 4669: @cindex comments
 4670: 
 4671: Forth supports two styles of comment; the traditional @i{in-line} comment,
 4672: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
 4673: 
 4674: 
 4675: doc-(
 4676: doc-\
 4677: doc-\G
 4678: 
 4679: 
 4680: @node Boolean Flags, Arithmetic, Comments, Words
 4681: @section Boolean Flags
 4682: @cindex Boolean flags
 4683: 
 4684: A Boolean flag is cell-sized. A cell with all bits clear represents the
 4685: flag @code{false} and a flag with all bits set represents the flag
 4686: @code{true}. Words that check a flag (for example, @code{IF}) will treat
 4687: a cell that has @i{any} bit set as @code{true}.
 4688: @c on and off to Memory? 
 4689: @c true and false to "Bitwise operations" or "Numeric comparison"?
 4690: 
 4691: doc-true
 4692: doc-false
 4693: doc-on
 4694: doc-off
 4695: 
 4696: 
 4697: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
 4698: @section Arithmetic
 4699: @cindex arithmetic words
 4700: 
 4701: @cindex division with potentially negative operands
 4702: Forth arithmetic is not checked, i.e., you will not hear about integer
 4703: overflow on addition or multiplication, you may hear about division by
 4704: zero if you are lucky. The operator is written after the operands, but
 4705: the operands are still in the original order. I.e., the infix @code{2-1}
 4706: corresponds to @code{2 1 -}. Forth offers a variety of division
 4707: operators. If you perform division with potentially negative operands,
 4708: you do not want to use @code{/} or @code{/mod} with its undefined
 4709: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
 4710: former, @pxref{Mixed precision}).
 4711: @comment TODO discuss the different division forms and the std approach
 4712: 
 4713: @menu
 4714: * Single precision::            
 4715: * Double precision::            Double-cell integer arithmetic
 4716: * Bitwise operations::          
 4717: * Numeric comparison::          
 4718: * Mixed precision::             Operations with single and double-cell integers
 4719: * Floating Point::              
 4720: @end menu
 4721: 
 4722: @node Single precision, Double precision, Arithmetic, Arithmetic
 4723: @subsection Single precision
 4724: @cindex single precision arithmetic words
 4725: 
 4726: @c !! cell undefined
 4727: 
 4728: By default, numbers in Forth are single-precision integers that are one
 4729: cell in size. They can be signed or unsigned, depending upon how you
 4730: treat them. For the rules used by the text interpreter for recognising
 4731: single-precision integers see @ref{Number Conversion}.
 4732: 
 4733: These words are all defined for signed operands, but some of them also
 4734: work for unsigned numbers: @code{+}, @code{1+}, @code{-}, @code{1-},
 4735: @code{*}.
 4736: 
 4737: doc-+
 4738: doc-1+
 4739: doc--
 4740: doc-1-
 4741: doc-*
 4742: doc-/
 4743: doc-mod
 4744: doc-/mod
 4745: doc-negate
 4746: doc-abs
 4747: doc-min
 4748: doc-max
 4749: doc-floored
 4750: 
 4751: 
 4752: @node Double precision, Bitwise operations, Single precision, Arithmetic
 4753: @subsection Double precision
 4754: @cindex double precision arithmetic words
 4755: 
 4756: For the rules used by the text interpreter for
 4757: recognising double-precision integers, see @ref{Number Conversion}.
 4758: 
 4759: A double precision number is represented by a cell pair, with the most
 4760: significant cell at the TOS. It is trivial to convert an unsigned single
 4761: to a double: simply push a @code{0} onto the TOS. Since numbers are
 4762: represented by Gforth using 2's complement arithmetic, converting a
 4763: signed single to a (signed) double requires sign-extension across the
 4764: most significant cell. This can be achieved using @code{s>d}. The moral
 4765: of the story is that you cannot convert a number without knowing whether
 4766: it represents an unsigned or a signed number.
 4767: 
 4768: These words are all defined for signed operands, but some of them also
 4769: work for unsigned numbers: @code{d+}, @code{d-}.
 4770: 
 4771: doc-s>d
 4772: doc-d>s
 4773: doc-d+
 4774: doc-d-
 4775: doc-dnegate
 4776: doc-dabs
 4777: doc-dmin
 4778: doc-dmax
 4779: 
 4780: 
 4781: @node Bitwise operations, Numeric comparison, Double precision, Arithmetic
 4782: @subsection Bitwise operations
 4783: @cindex bitwise operation words
 4784: 
 4785: 
 4786: doc-and
 4787: doc-or
 4788: doc-xor
 4789: doc-invert
 4790: doc-lshift
 4791: doc-rshift
 4792: doc-2*
 4793: doc-d2*
 4794: doc-2/
 4795: doc-d2/
 4796: 
 4797: 
 4798: @node Numeric comparison, Mixed precision, Bitwise operations, Arithmetic
 4799: @subsection Numeric comparison
 4800: @cindex numeric comparison words
 4801: 
 4802: Note that the words that compare for equality (@code{= <> 0= 0<> d= d<>
 4803: d0= d0<>}) work for for both signed and unsigned numbers.
 4804: 
 4805: doc-<
 4806: doc-<=
 4807: doc-<>
 4808: doc-=
 4809: doc->
 4810: doc->=
 4811: 
 4812: doc-0<
 4813: doc-0<=
 4814: doc-0<>
 4815: doc-0=
 4816: doc-0>
 4817: doc-0>=
 4818: 
 4819: doc-u<
 4820: doc-u<=
 4821: @c u<> and u= exist but are the same as <> and =
 4822: @c doc-u<>
 4823: @c doc-u=
 4824: doc-u>
 4825: doc-u>=
 4826: 
 4827: doc-within
 4828: 
 4829: doc-d<
 4830: doc-d<=
 4831: doc-d<>
 4832: doc-d=
 4833: doc-d>
 4834: doc-d>=
 4835: 
 4836: doc-d0<
 4837: doc-d0<=
 4838: doc-d0<>
 4839: doc-d0=
 4840: doc-d0>
 4841: doc-d0>=
 4842: 
 4843: doc-du<
 4844: doc-du<=
 4845: @c du<> and du= exist but are the same as d<> and d=
 4846: @c doc-du<>
 4847: @c doc-du=
 4848: doc-du>
 4849: doc-du>=
 4850: 
 4851: 
 4852: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
 4853: @subsection Mixed precision
 4854: @cindex mixed precision arithmetic words
 4855: 
 4856: 
 4857: doc-m+
 4858: doc-*/
 4859: doc-*/mod
 4860: doc-m*
 4861: doc-um*
 4862: doc-m*/
 4863: doc-um/mod
 4864: doc-fm/mod
 4865: doc-sm/rem
 4866: 
 4867: 
 4868: @node Floating Point,  , Mixed precision, Arithmetic
 4869: @subsection Floating Point
 4870: @cindex floating point arithmetic words
 4871: 
 4872: For the rules used by the text interpreter for
 4873: recognising floating-point numbers see @ref{Number Conversion}.
 4874: 
 4875: Gforth has a separate floating point stack, but the documentation uses
 4876: the unified notation.@footnote{It's easy to generate the separate
 4877: notation from that by just separating the floating-point numbers out:
 4878: e.g. @code{( n r1 u r2 -- r3 )} becomes @code{( n u -- ) ( F: r1 r2 --
 4879: r3 )}.}
 4880: 
 4881: @cindex floating-point arithmetic, pitfalls
 4882: Floating point numbers have a number of unpleasant surprises for the
 4883: unwary (e.g., floating point addition is not associative) and even a few
 4884: for the wary. You should not use them unless you know what you are doing
 4885: or you don't care that the results you get are totally bogus. If you
 4886: want to learn about the problems of floating point numbers (and how to
 4887: avoid them), you might start with @cite{David Goldberg,
 4888: @uref{http://www.validgh.com/goldberg/paper.ps,What Every Computer
 4889: Scientist Should Know About Floating-Point Arithmetic}, ACM Computing
 4890: Surveys 23(1):5@minus{}48, March 1991}.
 4891: 
 4892: 
 4893: doc-d>f
 4894: doc-f>d
 4895: doc-f+
 4896: doc-f-
 4897: doc-f*
 4898: doc-f/
 4899: doc-fnegate
 4900: doc-fabs
 4901: doc-fmax
 4902: doc-fmin
 4903: doc-floor
 4904: doc-fround
 4905: doc-f**
 4906: doc-fsqrt
 4907: doc-fexp
 4908: doc-fexpm1
 4909: doc-fln
 4910: doc-flnp1
 4911: doc-flog
 4912: doc-falog
 4913: doc-f2*
 4914: doc-f2/
 4915: doc-1/f
 4916: doc-precision
 4917: doc-set-precision
 4918: 
 4919: @cindex angles in trigonometric operations
 4920: @cindex trigonometric operations
 4921: Angles in floating point operations are given in radians (a full circle
 4922: has 2 pi radians).
 4923: 
 4924: doc-fsin
 4925: doc-fcos
 4926: doc-fsincos
 4927: doc-ftan
 4928: doc-fasin
 4929: doc-facos
 4930: doc-fatan
 4931: doc-fatan2
 4932: doc-fsinh
 4933: doc-fcosh
 4934: doc-ftanh
 4935: doc-fasinh
 4936: doc-facosh
 4937: doc-fatanh
 4938: doc-pi
 4939: 
 4940: @cindex equality of floats
 4941: @cindex floating-point comparisons
 4942: One particular problem with floating-point arithmetic is that comparison
 4943: for equality often fails when you would expect it to succeed.  For this
 4944: reason approximate equality is often preferred (but you still have to
 4945: know what you are doing).  Also note that IEEE NaNs may compare
 4946: differently from what you might expect.  The comparison words are:
 4947: 
 4948: doc-f~rel
 4949: doc-f~abs
 4950: doc-f~
 4951: doc-f=
 4952: doc-f<>
 4953: 
 4954: doc-f<
 4955: doc-f<=
 4956: doc-f>
 4957: doc-f>=
 4958: 
 4959: doc-f0<
 4960: doc-f0<=
 4961: doc-f0<>
 4962: doc-f0=
 4963: doc-f0>
 4964: doc-f0>=
 4965: 
 4966: 
 4967: @node Stack Manipulation, Memory, Arithmetic, Words
 4968: @section Stack Manipulation
 4969: @cindex stack manipulation words
 4970: 
 4971: @cindex floating-point stack in the standard
 4972: Gforth maintains a number of separate stacks:
 4973: 
 4974: @cindex data stack
 4975: @cindex parameter stack
 4976: @itemize @bullet
 4977: @item
 4978: A data stack (also known as the @dfn{parameter stack}) -- for
 4979: characters, cells, addresses, and double cells.
 4980: 
 4981: @cindex floating-point stack
 4982: @item
 4983: A floating point stack -- for holding floating point (FP) numbers.
 4984: 
 4985: @cindex return stack
 4986: @item
 4987: A return stack -- for holding the return addresses of colon
 4988: definitions and other (non-FP) data.
 4989: 
 4990: @cindex locals stack
 4991: @item
 4992: A locals stack -- for holding local variables.
 4993: @end itemize
 4994: 
 4995: @menu
 4996: * Data stack::                  
 4997: * Floating point stack::        
 4998: * Return stack::                
 4999: * Locals stack::                
 5000: * Stack pointer manipulation::  
 5001: @end menu
 5002: 
 5003: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
 5004: @subsection Data stack
 5005: @cindex data stack manipulation words
 5006: @cindex stack manipulations words, data stack
 5007: 
 5008: 
 5009: doc-drop
 5010: doc-nip
 5011: doc-dup
 5012: doc-over
 5013: doc-tuck
 5014: doc-swap
 5015: doc-pick
 5016: doc-rot
 5017: doc--rot
 5018: doc-?dup
 5019: doc-roll
 5020: doc-2drop
 5021: doc-2nip
 5022: doc-2dup
 5023: doc-2over
 5024: doc-2tuck
 5025: doc-2swap
 5026: doc-2rot
 5027: 
 5028: 
 5029: @node Floating point stack, Return stack, Data stack, Stack Manipulation
 5030: @subsection Floating point stack
 5031: @cindex floating-point stack manipulation words
 5032: @cindex stack manipulation words, floating-point stack
 5033: 
 5034: Whilst every sane Forth has a separate floating-point stack, it is not
 5035: strictly required; an ANS Forth system could theoretically keep
 5036: floating-point numbers on the data stack. As an additional difficulty,
 5037: you don't know how many cells a floating-point number takes. It is
 5038: reportedly possible to write words in a way that they work also for a
 5039: unified stack model, but we do not recommend trying it. Instead, just
 5040: say that your program has an environmental dependency on a separate
 5041: floating-point stack.
 5042: 
 5043: doc-floating-stack
 5044: 
 5045: doc-fdrop
 5046: doc-fnip
 5047: doc-fdup
 5048: doc-fover
 5049: doc-ftuck
 5050: doc-fswap
 5051: doc-fpick
 5052: doc-frot
 5053: 
 5054: 
 5055: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
 5056: @subsection Return stack
 5057: @cindex return stack manipulation words
 5058: @cindex stack manipulation words, return stack
 5059: 
 5060: @cindex return stack and locals
 5061: @cindex locals and return stack
 5062: A Forth system is allowed to keep local variables on the
 5063: return stack. This is reasonable, as local variables usually eliminate
 5064: the need to use the return stack explicitly. So, if you want to produce
 5065: a standard compliant program and you are using local variables in a
 5066: word, forget about return stack manipulations in that word (refer to the
 5067: standard document for the exact rules).
 5068: 
 5069: doc->r
 5070: doc-r>
 5071: doc-r@
 5072: doc-rdrop
 5073: doc-2>r
 5074: doc-2r>
 5075: doc-2r@
 5076: doc-2rdrop
 5077: 
 5078: 
 5079: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
 5080: @subsection Locals stack
 5081: 
 5082: Gforth uses an extra locals stack.  It is described, along with the
 5083: reasons for its existence, in @ref{Locals implementation}.
 5084: 
 5085: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
 5086: @subsection Stack pointer manipulation
 5087: @cindex stack pointer manipulation words
 5088: 
 5089: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
 5090: doc-sp0
 5091: doc-sp@
 5092: doc-sp!
 5093: doc-fp0
 5094: doc-fp@
 5095: doc-fp!
 5096: doc-rp0
 5097: doc-rp@
 5098: doc-rp!
 5099: doc-lp0
 5100: doc-lp@
 5101: doc-lp!
 5102: 
 5103: 
 5104: @node Memory, Control Structures, Stack Manipulation, Words
 5105: @section Memory
 5106: @cindex memory words
 5107: 
 5108: @menu
 5109: * Memory model::                
 5110: * Dictionary allocation::       
 5111: * Heap Allocation::             
 5112: * Memory Access::               
 5113: * Address arithmetic::          
 5114: * Memory Blocks::               
 5115: @end menu
 5116: 
 5117: In addition to the standard Forth memory allocation words, there is also
 5118: a @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
 5119: garbage collector}.
 5120: 
 5121: @node Memory model, Dictionary allocation, Memory, Memory
 5122: @subsection ANS Forth and Gforth memory models
 5123: 
 5124: @c The ANS Forth description is a mess (e.g., is the heap part of
 5125: @c the dictionary?), so let's not stick to closely with it.
 5126: 
 5127: ANS Forth considers a Forth system as consisting of several address
 5128: spaces, of which only @dfn{data space} is managed and accessible with
 5129: the memory words.  Memory not necessarily in data space includes the
 5130: stacks, the code (called code space) and the headers (called name
 5131: space). In Gforth everything is in data space, but the code for the
 5132: primitives is usually read-only.
 5133: 
 5134: Data space is divided into a number of areas: The (data space portion of
 5135: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
 5136: refer to the search data structure embodied in word lists and headers,
 5137: because it is used for looking up names, just as you would in a
 5138: conventional dictionary.}, the heap, and a number of system-allocated
 5139: buffers.
 5140: 
 5141: @cindex address arithmetic restrictions, ANS vs. Gforth
 5142: @cindex contiguous regions, ANS vs. Gforth
 5143: In ANS Forth data space is also divided into contiguous regions.  You
 5144: can only use address arithmetic within a contiguous region, not between
 5145: them.  Usually each allocation gives you one contiguous region, but the
 5146: dictionary allocation words have additional rules (@pxref{Dictionary
 5147: allocation}).
 5148: 
 5149: Gforth provides one big address space, and address arithmetic can be
 5150: performed between any addresses. However, in the dictionary headers or
 5151: code are interleaved with data, so almost the only contiguous data space
 5152: regions there are those described by ANS Forth as contiguous; but you
 5153: can be sure that the dictionary is allocated towards increasing
 5154: addresses even between contiguous regions.  The memory order of
 5155: allocations in the heap is platform-dependent (and possibly different
 5156: from one run to the next).
 5157: 
 5158: 
 5159: @node Dictionary allocation, Heap Allocation, Memory model, Memory
 5160: @subsection Dictionary allocation
 5161: @cindex reserving data space
 5162: @cindex data space - reserving some
 5163: 
 5164: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
 5165: you want to deallocate X, you also deallocate everything
 5166: allocated after X.
 5167: 
 5168: @cindex contiguous regions in dictionary allocation
 5169: The allocations using the words below are contiguous and grow the region
 5170: towards increasing addresses.  Other words that allocate dictionary
 5171: memory of any kind (i.e., defining words including @code{:noname}) end
 5172: the contiguous region and start a new one.
 5173: 
 5174: In ANS Forth only @code{create}d words are guaranteed to produce an
 5175: address that is the start of the following contiguous region.  In
 5176: particular, the cell allocated by @code{variable} is not guaranteed to
 5177: be contiguous with following @code{allot}ed memory.
 5178: 
 5179: You can deallocate memory by using @code{allot} with a negative argument
 5180: (with some restrictions, see @code{allot}). For larger deallocations use
 5181: @code{marker}.
 5182: 
 5183: 
 5184: doc-here
 5185: doc-unused
 5186: doc-allot
 5187: doc-c,
 5188: doc-f,
 5189: doc-,
 5190: doc-2,
 5191: 
 5192: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
 5193: course you should allocate memory in an aligned way, too. I.e., before
 5194: allocating allocating a cell, @code{here} must be cell-aligned, etc.
 5195: The words below align @code{here} if it is not already.  Basically it is
 5196: only already aligned for a type, if the last allocation was a multiple
 5197: of the size of this type and if @code{here} was aligned for this type
 5198: before.
 5199: 
 5200: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
 5201: ANS Forth (@code{maxalign}ed in Gforth).
 5202: 
 5203: doc-align
 5204: doc-falign
 5205: doc-sfalign
 5206: doc-dfalign
 5207: doc-maxalign
 5208: doc-cfalign
 5209: 
 5210: 
 5211: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
 5212: @subsection Heap allocation
 5213: @cindex heap allocation
 5214: @cindex dynamic allocation of memory
 5215: @cindex memory-allocation word set
 5216: 
 5217: @cindex contiguous regions and heap allocation
 5218: Heap allocation supports deallocation of allocated memory in any
 5219: order. Dictionary allocation is not affected by it (i.e., it does not
 5220: end a contiguous region). In Gforth, these words are implemented using
 5221: the standard C library calls malloc(), free() and resize().
 5222: 
 5223: The memory region produced by one invocation of @code{allocate} or
 5224: @code{resize} is internally contiguous.  There is no contiguity between
 5225: such a region and any other region (including others allocated from the
 5226: heap).
 5227: 
 5228: doc-allocate
 5229: doc-free
 5230: doc-resize
 5231: 
 5232: 
 5233: @node Memory Access, Address arithmetic, Heap Allocation, Memory
 5234: @subsection Memory Access
 5235: @cindex memory access words
 5236: 
 5237: doc-@
 5238: doc-!
 5239: doc-+!
 5240: doc-c@
 5241: doc-c!
 5242: doc-2@
 5243: doc-2!
 5244: doc-f@
 5245: doc-f!
 5246: doc-sf@
 5247: doc-sf!
 5248: doc-df@
 5249: doc-df!
 5250: 
 5251: 
 5252: @node Address arithmetic, Memory Blocks, Memory Access, Memory
 5253: @subsection Address arithmetic
 5254: @cindex address arithmetic words
 5255: 
 5256: Address arithmetic is the foundation on which you can build data
 5257: structures like arrays, records (@pxref{Structures}) and objects
 5258: (@pxref{Object-oriented Forth}).
 5259: 
 5260: @cindex address unit
 5261: @cindex au (address unit)
 5262: ANS Forth does not specify the sizes of the data types. Instead, it
 5263: offers a number of words for computing sizes and doing address
 5264: arithmetic. Address arithmetic is performed in terms of address units
 5265: (aus); on most systems the address unit is one byte. Note that a
 5266: character may have more than one au, so @code{chars} is no noop (on
 5267: platforms where it is a noop, it compiles to nothing).
 5268: 
 5269: The basic address arithmetic words are @code{+} and @code{-}.  E.g., if
 5270: you have the address of a cell, perform @code{1 cells +}, and you will
 5271: have the address of the next cell.
 5272: 
 5273: @cindex contiguous regions and address arithmetic
 5274: In ANS Forth you can perform address arithmetic only within a contiguous
 5275: region, i.e., if you have an address into one region, you can only add
 5276: and subtract such that the result is still within the region; you can
 5277: only subtract or compare addresses from within the same contiguous
 5278: region.  Reasons: several contiguous regions can be arranged in memory
 5279: in any way; on segmented systems addresses may have unusual
 5280: representations, such that address arithmetic only works within a
 5281: region.  Gforth provides a few more guarantees (linear address space,
 5282: dictionary grows upwards), but in general I have found it easy to stay
 5283: within contiguous regions (exception: computing and comparing to the
 5284: address just beyond the end of an array).
 5285: 
 5286: @cindex alignment of addresses for types
 5287: ANS Forth also defines words for aligning addresses for specific
 5288: types. Many computers require that accesses to specific data types
 5289: must only occur at specific addresses; e.g., that cells may only be
 5290: accessed at addresses divisible by 4. Even if a machine allows unaligned
 5291: accesses, it can usually perform aligned accesses faster. 
 5292: 
 5293: For the performance-conscious: alignment operations are usually only
 5294: necessary during the definition of a data structure, not during the
 5295: (more frequent) accesses to it.
 5296: 
 5297: ANS Forth defines no words for character-aligning addresses. This is not
 5298: an oversight, but reflects the fact that addresses that are not
 5299: char-aligned have no use in the standard and therefore will not be
 5300: created.
 5301: 
 5302: @cindex @code{CREATE} and alignment
 5303: ANS Forth guarantees that addresses returned by @code{CREATE}d words
 5304: are cell-aligned; in addition, Gforth guarantees that these addresses
 5305: are aligned for all purposes.
 5306: 
 5307: Note that the ANS Forth word @code{char} has nothing to do with address
 5308: arithmetic.
 5309: 
 5310: 
 5311: doc-chars
 5312: doc-char+
 5313: doc-cells
 5314: doc-cell+
 5315: doc-cell
 5316: doc-aligned
 5317: doc-floats
 5318: doc-float+
 5319: doc-float
 5320: doc-faligned
 5321: doc-sfloats
 5322: doc-sfloat+
 5323: doc-sfaligned
 5324: doc-dfloats
 5325: doc-dfloat+
 5326: doc-dfaligned
 5327: doc-maxaligned
 5328: doc-cfaligned
 5329: doc-address-unit-bits
 5330: 
 5331: 
 5332: @node Memory Blocks,  , Address arithmetic, Memory
 5333: @subsection Memory Blocks
 5334: @cindex memory block words
 5335: @cindex character strings - moving and copying
 5336: 
 5337: Memory blocks often represent character strings; For ways of storing
 5338: character strings in memory see @ref{String Formats}.  For other
 5339: string-processing words see @ref{Displaying characters and strings}.
 5340: 
 5341: A few of these words work on address unit blocks.  In that case, you
 5342: usually have to insert @code{CHARS} before the word when working on
 5343: character strings.  Most words work on character blocks, and expect a
 5344: char-aligned address.
 5345: 
 5346: When copying characters between overlapping memory regions, use
 5347: @code{chars move} or choose carefully between @code{cmove} and
 5348: @code{cmove>}.
 5349: 
 5350: doc-move
 5351: doc-erase
 5352: doc-cmove
 5353: doc-cmove>
 5354: doc-fill
 5355: doc-blank
 5356: doc-compare
 5357: doc-search
 5358: doc--trailing
 5359: doc-/string
 5360: doc-bounds
 5361: 
 5362: @comment TODO examples
 5363: 
 5364: 
 5365: @node Control Structures, Defining Words, Memory, Words
 5366: @section Control Structures
 5367: @cindex control structures
 5368: 
 5369: Control structures in Forth cannot be used interpretively, only in a
 5370: colon definition@footnote{To be precise, they have no interpretation
 5371: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
 5372: not like this limitation, but have not seen a satisfying way around it
 5373: yet, although many schemes have been proposed.
 5374: 
 5375: @menu
 5376: * Selection::                   IF ... ELSE ... ENDIF
 5377: * Simple Loops::                BEGIN ...
 5378: * Counted Loops::               DO
 5379: * Arbitrary control structures::  
 5380: * Calls and returns::           
 5381: * Exception Handling::          
 5382: @end menu
 5383: 
 5384: @node Selection, Simple Loops, Control Structures, Control Structures
 5385: @subsection Selection
 5386: @cindex selection control structures
 5387: @cindex control structures for selection
 5388: 
 5389: @cindex @code{IF} control structure
 5390: @example
 5391: @i{flag}
 5392: IF
 5393:   @i{code}
 5394: ENDIF
 5395: @end example
 5396: @noindent
 5397: 
 5398: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
 5399: with any bit set represents truth) @i{code} is executed.
 5400: 
 5401: @example
 5402: @i{flag}
 5403: IF
 5404:   @i{code1}
 5405: ELSE
 5406:   @i{code2}
 5407: ENDIF
 5408: @end example
 5409: 
 5410: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
 5411: executed.
 5412: 
 5413: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
 5414: standard, and @code{ENDIF} is not, although it is quite popular. We
 5415: recommend using @code{ENDIF}, because it is less confusing for people
 5416: who also know other languages (and is not prone to reinforcing negative
 5417: prejudices against Forth in these people). Adding @code{ENDIF} to a
 5418: system that only supplies @code{THEN} is simple:
 5419: @example
 5420: : ENDIF   POSTPONE then ; immediate
 5421: @end example
 5422: 
 5423: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
 5424: (adv.)}  has the following meanings:
 5425: @quotation
 5426: ... 2b: following next after in order ... 3d: as a necessary consequence
 5427: (if you were there, then you saw them).
 5428: @end quotation
 5429: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
 5430: and many other programming languages has the meaning 3d.]
 5431: 
 5432: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
 5433: you can avoid using @code{?dup}. Using these alternatives is also more
 5434: efficient than using @code{?dup}. Definitions in ANS Forth
 5435: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
 5436: @file{compat/control.fs}.
 5437: 
 5438: @cindex @code{CASE} control structure
 5439: @example
 5440: @i{n}
 5441: CASE
 5442:   @i{n1} OF @i{code1} ENDOF
 5443:   @i{n2} OF @i{code2} ENDOF
 5444:   @dots{}
 5445:   ( n ) @i{default-code} ( n )
 5446: ENDCASE
 5447: @end example
 5448: 
 5449: Executes the first @i{codei}, where the @i{ni} is equal to @i{n}.  If no
 5450: @i{ni} matches, the optional @i{default-code} is executed. The optional
 5451: default case can be added by simply writing the code after the last
 5452: @code{ENDOF}. It may use @i{n}, which is on top of the stack, but must
 5453: not consume it.
 5454: 
 5455: @progstyle
 5456: To keep the code understandable, you should ensure that on all paths
 5457: through a selection construct the stack is changed in the same way
 5458: (wrt. number and types of stack items consumed and pushed).
 5459: 
 5460: @node Simple Loops, Counted Loops, Selection, Control Structures
 5461: @subsection Simple Loops
 5462: @cindex simple loops
 5463: @cindex loops without count 
 5464: 
 5465: @cindex @code{WHILE} loop
 5466: @example
 5467: BEGIN
 5468:   @i{code1}
 5469:   @i{flag}
 5470: WHILE
 5471:   @i{code2}
 5472: REPEAT
 5473: @end example
 5474: 
 5475: @i{code1} is executed and @i{flag} is computed. If it is true,
 5476: @i{code2} is executed and the loop is restarted; If @i{flag} is
 5477: false, execution continues after the @code{REPEAT}.
 5478: 
 5479: @cindex @code{UNTIL} loop
 5480: @example
 5481: BEGIN
 5482:   @i{code}
 5483:   @i{flag}
 5484: UNTIL
 5485: @end example
 5486: 
 5487: @i{code} is executed. The loop is restarted if @code{flag} is false.
 5488: 
 5489: @progstyle
 5490: To keep the code understandable, a complete iteration of the loop should
 5491: not change the number and types of the items on the stacks.
 5492: 
 5493: @cindex endless loop
 5494: @cindex loops, endless
 5495: @example
 5496: BEGIN
 5497:   @i{code}
 5498: AGAIN
 5499: @end example
 5500: 
 5501: This is an endless loop.
 5502: 
 5503: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
 5504: @subsection Counted Loops
 5505: @cindex counted loops
 5506: @cindex loops, counted
 5507: @cindex @code{DO} loops
 5508: 
 5509: The basic counted loop is:
 5510: @example
 5511: @i{limit} @i{start}
 5512: ?DO
 5513:   @i{body}
 5514: LOOP
 5515: @end example
 5516: 
 5517: This performs one iteration for every integer, starting from @i{start}
 5518: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
 5519: accessed with @code{i}. For example, the loop:
 5520: @example
 5521: 10 0 ?DO
 5522:   i .
 5523: LOOP
 5524: @end example
 5525: @noindent
 5526: prints @code{0 1 2 3 4 5 6 7 8 9}
 5527: 
 5528: The index of the innermost loop can be accessed with @code{i}, the index
 5529: of the next loop with @code{j}, and the index of the third loop with
 5530: @code{k}.
 5531: 
 5532: 
 5533: doc-i
 5534: doc-j
 5535: doc-k
 5536: 
 5537: 
 5538: The loop control data are kept on the return stack, so there are some
 5539: restrictions on mixing return stack accesses and counted loop words. In
 5540: particuler, if you put values on the return stack outside the loop, you
 5541: cannot read them inside the loop@footnote{well, not in a way that is
 5542: portable.}. If you put values on the return stack within a loop, you
 5543: have to remove them before the end of the loop and before accessing the
 5544: index of the loop.
 5545: 
 5546: There are several variations on the counted loop:
 5547: 
 5548: @itemize @bullet
 5549: @item
 5550: @code{LEAVE} leaves the innermost counted loop immediately; execution
 5551: continues after the associated @code{LOOP} or @code{NEXT}. For example:
 5552: 
 5553: @example
 5554: 10 0 ?DO  i DUP . 3 = IF LEAVE THEN LOOP
 5555: @end example
 5556: prints @code{0 1 2 3}
 5557: 
 5558: 
 5559: @item
 5560: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
 5561: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
 5562: return stack so @code{EXIT} can get to its return address. For example:
 5563: 
 5564: @example
 5565: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
 5566: @end example
 5567: prints @code{0 1 2 3}
 5568: 
 5569: 
 5570: @item
 5571: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
 5572: (and @code{LOOP} iterates until they become equal by wrap-around
 5573: arithmetic). This behaviour is usually not what you want. Therefore,
 5574: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
 5575: @code{?DO}), which do not enter the loop if @i{start} is greater than
 5576: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
 5577: unsigned loop parameters.
 5578: 
 5579: @item
 5580: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
 5581: the loop, independent of the loop parameters. Do not use @code{DO}, even
 5582: if you know that the loop is entered in any case. Such knowledge tends
 5583: to become invalid during maintenance of a program, and then the
 5584: @code{DO} will make trouble.
 5585: 
 5586: @item
 5587: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
 5588: index by @i{n} instead of by 1. The loop is terminated when the border
 5589: between @i{limit-1} and @i{limit} is crossed. E.g.:
 5590: 
 5591: @example
 5592: 4 0 +DO  i .  2 +LOOP
 5593: @end example
 5594: @noindent
 5595: prints @code{0 2}
 5596: 
 5597: @example
 5598: 4 1 +DO  i .  2 +LOOP
 5599: @end example
 5600: @noindent
 5601: prints @code{1 3}
 5602: 
 5603: @item
 5604: @cindex negative increment for counted loops
 5605: @cindex counted loops with negative increment
 5606: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
 5607: 
 5608: @example
 5609: -1 0 ?DO  i .  -1 +LOOP
 5610: @end example
 5611: @noindent
 5612: prints @code{0 -1}
 5613: 
 5614: @example
 5615: 0 0 ?DO  i .  -1 +LOOP
 5616: @end example
 5617: prints nothing.
 5618: 
 5619: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
 5620: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
 5621: index by @i{u} each iteration. The loop is terminated when the border
 5622: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
 5623: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
 5624: 
 5625: @example
 5626: -2 0 -DO  i .  1 -LOOP
 5627: @end example
 5628: @noindent
 5629: prints @code{0 -1}
 5630: 
 5631: @example
 5632: -1 0 -DO  i .  1 -LOOP
 5633: @end example
 5634: @noindent
 5635: prints @code{0}
 5636: 
 5637: @example
 5638: 0 0 -DO  i .  1 -LOOP
 5639: @end example
 5640: @noindent
 5641: prints nothing.
 5642: 
 5643: @end itemize
 5644: 
 5645: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
 5646: @code{-LOOP} are not defined in ANS Forth. However, an implementation
 5647: for these words that uses only standard words is provided in
 5648: @file{compat/loops.fs}.
 5649: 
 5650: 
 5651: @cindex @code{FOR} loops
 5652: Another counted loop is:
 5653: @example
 5654: @i{n}
 5655: FOR
 5656:   @i{body}
 5657: NEXT
 5658: @end example
 5659: This is the preferred loop of native code compiler writers who are too
 5660: lazy to optimize @code{?DO} loops properly. This loop structure is not
 5661: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
 5662: @code{i} produces values starting with @i{n} and ending with 0. Other
 5663: Forth systems may behave differently, even if they support @code{FOR}
 5664: loops. To avoid problems, don't use @code{FOR} loops.
 5665: 
 5666: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
 5667: @subsection Arbitrary control structures
 5668: @cindex control structures, user-defined
 5669: 
 5670: @cindex control-flow stack
 5671: ANS Forth permits and supports using control structures in a non-nested
 5672: way. Information about incomplete control structures is stored on the
 5673: control-flow stack. This stack may be implemented on the Forth data
 5674: stack, and this is what we have done in Gforth.
 5675: 
 5676: @cindex @code{orig}, control-flow stack item
 5677: @cindex @code{dest}, control-flow stack item
 5678: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
 5679: entry represents a backward branch target. A few words are the basis for
 5680: building any control structure possible (except control structures that
 5681: need storage, like calls, coroutines, and backtracking).
 5682: 
 5683: 
 5684: doc-if
 5685: doc-ahead
 5686: doc-then
 5687: doc-begin
 5688: doc-until
 5689: doc-again
 5690: doc-cs-pick
 5691: doc-cs-roll
 5692: 
 5693: 
 5694: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
 5695: manipulate the control-flow stack in a portable way. Without them, you
 5696: would need to know how many stack items are occupied by a control-flow
 5697: entry (many systems use one cell. In Gforth they currently take three,
 5698: but this may change in the future).
 5699: 
 5700: Some standard control structure words are built from these words:
 5701: 
 5702: 
 5703: doc-else
 5704: doc-while
 5705: doc-repeat
 5706: 
 5707: 
 5708: @noindent
 5709: Gforth adds some more control-structure words:
 5710: 
 5711: 
 5712: doc-endif
 5713: doc-?dup-if
 5714: doc-?dup-0=-if
 5715: 
 5716: 
 5717: @noindent
 5718: Counted loop words constitute a separate group of words:
 5719: 
 5720: 
 5721: doc-?do
 5722: doc-+do
 5723: doc-u+do
 5724: doc--do
 5725: doc-u-do
 5726: doc-do
 5727: doc-for
 5728: doc-loop
 5729: doc-+loop
 5730: doc--loop
 5731: doc-next
 5732: doc-leave
 5733: doc-?leave
 5734: doc-unloop
 5735: doc-done
 5736: 
 5737: 
 5738: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
 5739: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
 5740: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
 5741: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
 5742: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
 5743: resolved (by using one of the loop-ending words or @code{DONE}).
 5744: 
 5745: @noindent
 5746: Another group of control structure words are:
 5747: 
 5748: 
 5749: doc-case
 5750: doc-endcase
 5751: doc-of
 5752: doc-endof
 5753: 
 5754: 
 5755: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
 5756: @code{CS-ROLL}.
 5757: 
 5758: @subsubsection Programming Style
 5759: @cindex control structures programming style
 5760: @cindex programming style, arbitrary control structures
 5761: 
 5762: In order to ensure readability we recommend that you do not create
 5763: arbitrary control structures directly, but define new control structure
 5764: words for the control structure you want and use these words in your
 5765: program. For example, instead of writing:
 5766: 
 5767: @example
 5768: BEGIN
 5769:   ...
 5770: IF [ 1 CS-ROLL ]
 5771:   ...
 5772: AGAIN THEN
 5773: @end example
 5774: 
 5775: @noindent
 5776: we recommend defining control structure words, e.g.,
 5777: 
 5778: @example
 5779: : WHILE ( DEST -- ORIG DEST )
 5780:  POSTPONE IF
 5781:  1 CS-ROLL ; immediate
 5782: 
 5783: : REPEAT ( orig dest -- )
 5784:  POSTPONE AGAIN
 5785:  POSTPONE THEN ; immediate
 5786: @end example
 5787: 
 5788: @noindent
 5789: and then using these to create the control structure:
 5790: 
 5791: @example
 5792: BEGIN
 5793:   ...
 5794: WHILE
 5795:   ...
 5796: REPEAT
 5797: @end example
 5798: 
 5799: That's much easier to read, isn't it? Of course, @code{REPEAT} and
 5800: @code{WHILE} are predefined, so in this example it would not be
 5801: necessary to define them.
 5802: 
 5803: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
 5804: @subsection Calls and returns
 5805: @cindex calling a definition
 5806: @cindex returning from a definition
 5807: 
 5808: @cindex recursive definitions
 5809: A definition can be called simply be writing the name of the definition
 5810: to be called. Normally a definition is invisible during its own
 5811: definition. If you want to write a directly recursive definition, you
 5812: can use @code{recursive} to make the current definition visible, or
 5813: @code{recurse} to call the current definition directly.
 5814: 
 5815: 
 5816: doc-recursive
 5817: doc-recurse
 5818: 
 5819: 
 5820: @comment TODO add example of the two recursion methods
 5821: @quotation
 5822: @progstyle
 5823: I prefer using @code{recursive} to @code{recurse}, because calling the
 5824: definition by name is more descriptive (if the name is well-chosen) than
 5825: the somewhat cryptic @code{recurse}.  E.g., in a quicksort
 5826: implementation, it is much better to read (and think) ``now sort the
 5827: partitions'' than to read ``now do a recursive call''.
 5828: @end quotation
 5829: 
 5830: For mutual recursion, use @code{Defer}red words, like this:
 5831: 
 5832: @example
 5833: Defer foo
 5834: 
 5835: : bar ( ... -- ... )
 5836:  ... foo ... ;
 5837: 
 5838: :noname ( ... -- ... )
 5839:  ... bar ... ;
 5840: IS foo
 5841: @end example
 5842: 
 5843: Deferred words are discussed in more detail in @ref{Deferred words}.
 5844: 
 5845: The current definition returns control to the calling definition when
 5846: the end of the definition is reached or @code{EXIT} is encountered.
 5847: 
 5848: doc-exit
 5849: doc-;s
 5850: 
 5851: 
 5852: @node Exception Handling,  , Calls and returns, Control Structures
 5853: @subsection Exception Handling
 5854: @cindex exceptions
 5855: 
 5856: @c quit is a very bad idea for error handling, 
 5857: @c because it does not translate into a THROW
 5858: @c it also does not belong into this chapter
 5859: 
 5860: If a word detects an error condition that it cannot handle, it can
 5861: @code{throw} an exception.  In the simplest case, this will terminate
 5862: your program, and report an appropriate error.
 5863: 
 5864: doc-throw
 5865: 
 5866: @code{Throw} consumes a cell-sized error number on the stack. There are
 5867: some predefined error numbers in ANS Forth (see @file{errors.fs}).  In
 5868: Gforth (and most other systems) you can use the iors produced by various
 5869: words as error numbers (e.g., a typical use of @code{allocate} is
 5870: @code{allocate throw}).  Gforth also provides the word @code{exception}
 5871: to define your own error numbers (with decent error reporting); an ANS
 5872: Forth version of this word (but without the error messages) is available
 5873: in @code{compat/except.fs}.  And finally, you can use your own error
 5874: numbers (anything outside the range -4095..0), but won't get nice error
 5875: messages, only numbers.  For example, try:
 5876: 
 5877: @example
 5878: -10 throw                    \ ANS defined
 5879: -267 throw                   \ system defined
 5880: s" my error" exception throw \ user defined
 5881: 7 throw                      \ arbitrary number
 5882: @end example
 5883: 
 5884: doc---exception-exception
 5885: 
 5886: A common idiom to @code{THROW} a specific error if a flag is true is
 5887: this:
 5888: 
 5889: @example
 5890: @code{( flag ) 0<> @i{errno} and throw}
 5891: @end example
 5892: 
 5893: Your program can provide exception handlers to catch exceptions.  An
 5894: exception handler can be used to correct the problem, or to clean up
 5895: some data structures and just throw the exception to the next exception
 5896: handler.  Note that @code{throw} jumps to the dynamically innermost
 5897: exception handler.  The system's exception handler is outermost, and just
 5898: prints an error and restarts command-line interpretation (or, in batch
 5899: mode (i.e., while processing the shell command line), leaves Gforth).
 5900: 
 5901: The ANS Forth way to catch exceptions is @code{catch}:
 5902: 
 5903: doc-catch
 5904: 
 5905: The most common use of exception handlers is to clean up the state when
 5906: an error happens.  E.g.,
 5907: 
 5908: @example
 5909: base @ >r hex \ actually the hex should be inside foo, or we h
 5910: ['] foo catch ( nerror|0 )
 5911: r> base !
 5912: ( nerror|0 ) throw \ pass it on
 5913: @end example
 5914: 
 5915: A use of @code{catch} for handling the error @code{myerror} might look
 5916: like this:
 5917: 
 5918: @example
 5919: ['] foo catch
 5920: CASE
 5921:   myerror OF ... ( do something about it ) ENDOF
 5922:   dup throw \ default: pass other errors on, do nothing on non-errors
 5923: ENDCASE
 5924: @end example
 5925: 
 5926: Having to wrap the code into a separate word is often cumbersome,
 5927: therefore Gforth provides an alternative syntax:
 5928: 
 5929: @example
 5930: TRY
 5931:   @i{code1}
 5932: RECOVER     \ optional
 5933:   @i{code2} \ optional
 5934: ENDTRY
 5935: @end example
 5936: 
 5937: This performs @i{Code1}.  If @i{code1} completes normally, execution
 5938: continues after the @code{endtry}.  If @i{Code1} throws, the stacks are
 5939: reset to the state during @code{try}, the throw value is pushed on the
 5940: data stack, and execution constinues at @i{code2}, and finally falls
 5941: through the @code{endtry} into the following code. If there is no
 5942: @code{recover} clause, this works like an empty recover clause.
 5943: 
 5944: doc-try
 5945: doc-recover
 5946: doc-endtry
 5947: 
 5948: The cleanup example from above in this syntax:
 5949: 
 5950: @example
 5951: base @ >r TRY
 5952:   hex foo \ now the hex is placed correctly
 5953:   0       \ value for throw
 5954: ENDTRY
 5955: r> base ! throw
 5956: @end example
 5957: 
 5958: And here's the error handling example:
 5959: 
 5960: @example
 5961: TRY
 5962:   foo
 5963: RECOVER
 5964:   CASE
 5965:     myerror OF ... ( do something about it ) ENDOF
 5966:     throw \ pass other errors on
 5967:   ENDCASE
 5968: ENDTRY
 5969: @end example
 5970: 
 5971: @progstyle
 5972: As usual, you should ensure that the stack depth is statically known at
 5973: the end: either after the @code{throw} for passing on errors, or after
 5974: the @code{ENDTRY} (or, if you use @code{catch}, after the end of the
 5975: selection construct for handling the error).
 5976: 
 5977: There are two alternatives to @code{throw}: @code{Abort"} is conditional
 5978: and you can provide an error message.  @code{Abort} just produces an
 5979: ``Aborted'' error.
 5980: 
 5981: The problem with these words is that exception handlers cannot
 5982: differentiate between different @code{abort"}s; they just look like
 5983: @code{-2 throw} to them (the error message cannot be accessed by
 5984: standard programs).  Similar @code{abort} looks like @code{-1 throw} to
 5985: exception handlers.
 5986: 
 5987: doc-abort"
 5988: doc-abort
 5989: 
 5990: 
 5991: 
 5992: @c -------------------------------------------------------------
 5993: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
 5994: @section Defining Words
 5995: @cindex defining words
 5996: 
 5997: Defining words are used to extend Forth by creating new entries in the dictionary.
 5998: 
 5999: @menu
 6000: * CREATE::                      
 6001: * Variables::                   Variables and user variables
 6002: * Constants::                   
 6003: * Values::                      Initialised variables
 6004: * Colon Definitions::           
 6005: * Anonymous Definitions::       Definitions without names
 6006: * Supplying names::             Passing definition names as strings
 6007: * User-defined Defining Words::  
 6008: * Deferred words::              Allow forward references
 6009: * Aliases::                     
 6010: @end menu
 6011: 
 6012: @node CREATE, Variables, Defining Words, Defining Words
 6013: @subsection @code{CREATE}
 6014: @cindex simple defining words
 6015: @cindex defining words, simple
 6016: 
 6017: Defining words are used to create new entries in the dictionary. The
 6018: simplest defining word is @code{CREATE}. @code{CREATE} is used like
 6019: this:
 6020: 
 6021: @example
 6022: CREATE new-word1
 6023: @end example
 6024: 
 6025: @code{CREATE} is a parsing word, i.e., it takes an argument from the
 6026: input stream (@code{new-word1} in our example).  It generates a
 6027: dictionary entry for @code{new-word1}. When @code{new-word1} is
 6028: executed, all that it does is leave an address on the stack. The address
 6029: represents the value of the data space pointer (@code{HERE}) at the time
 6030: that @code{new-word1} was defined. Therefore, @code{CREATE} is a way of
 6031: associating a name with the address of a region of memory.
 6032: 
 6033: doc-create
 6034: 
 6035: Note that in ANS Forth guarantees only for @code{create} that its body
 6036: is in dictionary data space (i.e., where @code{here}, @code{allot}
 6037: etc. work, @pxref{Dictionary allocation}).  Also, in ANS Forth only
 6038: @code{create}d words can be modified with @code{does>}
 6039: (@pxref{User-defined Defining Words}).  And in ANS Forth @code{>body}
 6040: can only be applied to @code{create}d words.
 6041: 
 6042: By extending this example to reserve some memory in data space, we end
 6043: up with something like a @i{variable}. Here are two different ways to do
 6044: it:
 6045: 
 6046: @example
 6047: CREATE new-word2 1 cells allot  \ reserve 1 cell - initial value undefined
 6048: CREATE new-word3 4 ,            \ reserve 1 cell and initialise it (to 4)
 6049: @end example
 6050: 
 6051: The variable can be examined and modified using @code{@@} (``fetch'') and
 6052: @code{!} (``store'') like this:
 6053: 
 6054: @example
 6055: new-word2 @@ .      \ get address, fetch from it and display
 6056: 1234 new-word2 !   \ new value, get address, store to it
 6057: @end example
 6058: 
 6059: @cindex arrays
 6060: A similar mechanism can be used to create arrays. For example, an
 6061: 80-character text input buffer:
 6062: 
 6063: @example
 6064: CREATE text-buf 80 chars allot
 6065: 
 6066: text-buf 0 chars c@@ \ the 1st character (offset 0)
 6067: text-buf 3 chars c@@ \ the 4th character (offset 3)
 6068: @end example
 6069: 
 6070: You can build arbitrarily complex data structures by allocating
 6071: appropriate areas of memory. For further discussions of this, and to
 6072: learn about some Gforth tools that make it easier,
 6073: @xref{Structures}.
 6074: 
 6075: 
 6076: @node Variables, Constants, CREATE, Defining Words
 6077: @subsection Variables
 6078: @cindex variables
 6079: 
 6080: The previous section showed how a sequence of commands could be used to
 6081: generate a variable.  As a final refinement, the whole code sequence can
 6082: be wrapped up in a defining word (pre-empting the subject of the next
 6083: section), making it easier to create new variables:
 6084: 
 6085: @example
 6086: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
 6087: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
 6088: 
 6089: myvariableX foo \ variable foo starts off with an unknown value
 6090: myvariable0 joe \ whilst joe is initialised to 0
 6091: 
 6092: 45 3 * foo !   \ set foo to 135
 6093: 1234 joe !     \ set joe to 1234
 6094: 3 joe +!       \ increment joe by 3.. to 1237
 6095: @end example
 6096: 
 6097: Not surprisingly, there is no need to define @code{myvariable}, since
 6098: Forth already has a definition @code{Variable}. ANS Forth does not
 6099: guarantee that a @code{Variable} is initialised when it is created
 6100: (i.e., it may behave like @code{myvariableX}). In contrast, Gforth's
 6101: @code{Variable} initialises the variable to 0 (i.e., it behaves exactly
 6102: like @code{myvariable0}). Forth also provides @code{2Variable} and
 6103: @code{fvariable} for double and floating-point variables, respectively
 6104: -- they are initialised to 0. and 0e in Gforth. If you use a @code{Variable} to
 6105: store a boolean, you can use @code{on} and @code{off} to toggle its
 6106: state.
 6107: 
 6108: doc-variable
 6109: doc-2variable
 6110: doc-fvariable
 6111: 
 6112: @cindex user variables
 6113: @cindex user space
 6114: The defining word @code{User} behaves in the same way as @code{Variable}.
 6115: The difference is that it reserves space in @i{user (data) space} rather
 6116: than normal data space. In a Forth system that has a multi-tasker, each
 6117: task has its own set of user variables.
 6118: 
 6119: doc-user
 6120: @c doc-udp
 6121: @c doc-uallot
 6122: 
 6123: @comment TODO is that stuff about user variables strictly correct? Is it
 6124: @comment just terminal tasks that have user variables?
 6125: @comment should document tasker.fs (with some examples) elsewhere
 6126: @comment in this manual, then expand on user space and user variables.
 6127: 
 6128: @node Constants, Values, Variables, Defining Words
 6129: @subsection Constants
 6130: @cindex constants
 6131: 
 6132: @code{Constant} allows you to declare a fixed value and refer to it by
 6133: name. For example:
 6134: 
 6135: @example
 6136: 12 Constant INCHES-PER-FOOT
 6137: 3E+08 fconstant SPEED-O-LIGHT
 6138: @end example
 6139: 
 6140: A @code{Variable} can be both read and written, so its run-time
 6141: behaviour is to supply an address through which its current value can be
 6142: manipulated. In contrast, the value of a @code{Constant} cannot be
 6143: changed once it has been declared@footnote{Well, often it can be -- but
 6144: not in a Standard, portable way. It's safer to use a @code{Value} (read
 6145: on).} so it's not necessary to supply the address -- it is more
 6146: efficient to return the value of the constant directly. That's exactly
 6147: what happens; the run-time effect of a constant is to put its value on
 6148: the top of the stack (You can find one
 6149: way of implementing @code{Constant} in @ref{User-defined Defining Words}).
 6150: 
 6151: Forth also provides @code{2Constant} and @code{fconstant} for defining
 6152: double and floating-point constants, respectively.
 6153: 
 6154: doc-constant
 6155: doc-2constant
 6156: doc-fconstant
 6157: 
 6158: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
 6159: @c nac-> How could that not be true in an ANS Forth? You can't define a
 6160: @c constant, use it and then delete the definition of the constant..
 6161: 
 6162: @c anton->An ANS Forth system can compile a constant to a literal; On
 6163: @c decompilation you would see only the number, just as if it had been used
 6164: @c in the first place.  The word will stay, of course, but it will only be
 6165: @c used by the text interpreter (no run-time duties, except when it is 
 6166: @c POSTPONEd or somesuch).
 6167: 
 6168: @c nac:
 6169: @c I agree that it's rather deep, but IMO it is an important difference
 6170: @c relative to other programming languages.. often it's annoying: it
 6171: @c certainly changes my programming style relative to C.
 6172: 
 6173: @c anton: In what way?
 6174: 
 6175: Constants in Forth behave differently from their equivalents in other
 6176: programming languages. In other languages, a constant (such as an EQU in
 6177: assembler or a #define in C) only exists at compile-time; in the
 6178: executable program the constant has been translated into an absolute
 6179: number and, unless you are using a symbolic debugger, it's impossible to
 6180: know what abstract thing that number represents. In Forth a constant has
 6181: an entry in the header space and remains there after the code that uses
 6182: it has been defined. In fact, it must remain in the dictionary since it
 6183: has run-time duties to perform. For example:
 6184: 
 6185: @example
 6186: 12 Constant INCHES-PER-FOOT
 6187: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
 6188: @end example
 6189: 
 6190: @cindex in-lining of constants
 6191: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
 6192: associated with the constant @code{INCHES-PER-FOOT}. If you use
 6193: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
 6194: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
 6195: attempt to optimise constants by in-lining them where they are used. You
 6196: can force Gforth to in-line a constant like this:
 6197: 
 6198: @example
 6199: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
 6200: @end example
 6201: 
 6202: If you use @code{see} to decompile @i{this} version of
 6203: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
 6204: longer present. To understand how this works, read
 6205: @ref{Interpret/Compile states}, and @ref{Literals}.
 6206: 
 6207: In-lining constants in this way might improve execution time
 6208: fractionally, and can ensure that a constant is now only referenced at
 6209: compile-time. However, the definition of the constant still remains in
 6210: the dictionary. Some Forth compilers provide a mechanism for controlling
 6211: a second dictionary for holding transient words such that this second
 6212: dictionary can be deleted later in order to recover memory
 6213: space. However, there is no standard way of doing this.
 6214: 
 6215: 
 6216: @node Values, Colon Definitions, Constants, Defining Words
 6217: @subsection Values
 6218: @cindex values
 6219: 
 6220: A @code{Value} behaves like a @code{Constant}, but it can be changed.
 6221: @code{TO} is a parsing word that changes a @code{Values}.  In Gforth
 6222: (not in ANS Forth) you can access (and change) a @code{value} also with
 6223: @code{>body}.
 6224: 
 6225: Here are some
 6226: examples:
 6227: 
 6228: @example
 6229: 12 Value APPLES     \ Define APPLES with an initial value of 12
 6230: 34 TO APPLES        \ Change the value of APPLES. TO is a parsing word
 6231: 1 ' APPLES >body +! \ Increment APPLES.  Non-standard usage.
 6232: APPLES              \ puts 35 on the top of the stack.
 6233: @end example
 6234: 
 6235: doc-value
 6236: doc-to
 6237: 
 6238: 
 6239: 
 6240: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
 6241: @subsection Colon Definitions
 6242: @cindex colon definitions
 6243: 
 6244: @example
 6245: : name ( ... -- ... )
 6246:     word1 word2 word3 ;
 6247: @end example
 6248: 
 6249: @noindent
 6250: Creates a word called @code{name} that, upon execution, executes
 6251: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
 6252: 
 6253: The explanation above is somewhat superficial. For simple examples of
 6254: colon definitions see @ref{Your first definition}.  For an in-depth
 6255: discussion of some of the issues involved, @xref{Interpretation and
 6256: Compilation Semantics}.
 6257: 
 6258: doc-:
 6259: doc-;
 6260: 
 6261: 
 6262: @node Anonymous Definitions, Supplying names, Colon Definitions, Defining Words
 6263: @subsection Anonymous Definitions
 6264: @cindex colon definitions
 6265: @cindex defining words without name
 6266: 
 6267: Sometimes you want to define an @dfn{anonymous word}; a word without a
 6268: name. You can do this with:
 6269: 
 6270: doc-:noname
 6271: 
 6272: This leaves the execution token for the word on the stack after the
 6273: closing @code{;}. Here's an example in which a deferred word is
 6274: initialised with an @code{xt} from an anonymous colon definition:
 6275: 
 6276: @example
 6277: Defer deferred
 6278: :noname ( ... -- ... )
 6279:   ... ;
 6280: IS deferred
 6281: @end example
 6282: 
 6283: @noindent
 6284: Gforth provides an alternative way of doing this, using two separate
 6285: words:
 6286: 
 6287: doc-noname
 6288: @cindex execution token of last defined word
 6289: doc-lastxt
 6290: 
 6291: @noindent
 6292: The previous example can be rewritten using @code{noname} and
 6293: @code{lastxt}:
 6294: 
 6295: @example
 6296: Defer deferred
 6297: noname : ( ... -- ... )
 6298:   ... ;
 6299: lastxt IS deferred
 6300: @end example
 6301: 
 6302: @noindent
 6303: @code{noname} works with any defining word, not just @code{:}.
 6304: 
 6305: @code{lastxt} also works when the last word was not defined as
 6306: @code{noname}.  It does not work for combined words, though.  It also has
 6307: the useful property that is is valid as soon as the header for a
 6308: definition has been built. Thus:
 6309: 
 6310: @example
 6311: lastxt . : foo [ lastxt . ] ; ' foo .
 6312: @end example
 6313: 
 6314: @noindent
 6315: prints 3 numbers; the last two are the same.
 6316: 
 6317: @node Supplying names, User-defined Defining Words, Anonymous Definitions, Defining Words
 6318: @subsection Supplying the name of a defined word
 6319: @cindex names for defined words
 6320: @cindex defining words, name given in a string
 6321: 
 6322: By default, a defining word takes the name for the defined word from the
 6323: input stream. Sometimes you want to supply the name from a string. You
 6324: can do this with:
 6325: 
 6326: doc-nextname
 6327: 
 6328: For example:
 6329: 
 6330: @example
 6331: s" foo" nextname create
 6332: @end example
 6333: 
 6334: @noindent
 6335: is equivalent to:
 6336: 
 6337: @example
 6338: create foo
 6339: @end example
 6340: 
 6341: @noindent
 6342: @code{nextname} works with any defining word.
 6343: 
 6344: 
 6345: @node User-defined Defining Words, Deferred words, Supplying names, Defining Words
 6346: @subsection User-defined Defining Words
 6347: @cindex user-defined defining words
 6348: @cindex defining words, user-defined
 6349: 
 6350: You can create a new defining word by wrapping defining-time code around
 6351: an existing defining word and putting the sequence in a colon
 6352: definition. 
 6353: 
 6354: @c anton: This example is very complex and leads in a quite different
 6355: @c direction from the CREATE-DOES> stuff that follows.  It should probably
 6356: @c be done elsewhere, or as a subsubsection of this subsection (or as a
 6357: @c subsection of Defining Words)
 6358: 
 6359: For example, suppose that you have a word @code{stats} that
 6360: gathers statistics about colon definitions given the @i{xt} of the
 6361: definition, and you want every colon definition in your application to
 6362: make a call to @code{stats}. You can define and use a new version of
 6363: @code{:} like this:
 6364: 
 6365: @example
 6366: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
 6367:   ... ;  \ other code
 6368: 
 6369: : my: : lastxt postpone literal ['] stats compile, ;
 6370: 
 6371: my: foo + - ;
 6372: @end example
 6373: 
 6374: When @code{foo} is defined using @code{my:} these steps occur:
 6375: 
 6376: @itemize @bullet
 6377: @item
 6378: @code{my:} is executed.
 6379: @item
 6380: The @code{:} within the definition (the one between @code{my:} and
 6381: @code{lastxt}) is executed, and does just what it always does; it parses
 6382: the input stream for a name, builds a dictionary header for the name
 6383: @code{foo} and switches @code{state} from interpret to compile.
 6384: @item
 6385: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
 6386: being defined -- @code{foo} -- onto the stack.
 6387: @item
 6388: The code that was produced by @code{postpone literal} is executed; this
 6389: causes the value on the stack to be compiled as a literal in the code
 6390: area of @code{foo}.
 6391: @item
 6392: The code @code{['] stats} compiles a literal into the definition of
 6393: @code{my:}. When @code{compile,} is executed, that literal -- the
 6394: execution token for @code{stats} -- is layed down in the code area of
 6395: @code{foo} , following the literal@footnote{Strictly speaking, the
 6396: mechanism that @code{compile,} uses to convert an @i{xt} into something
 6397: in the code area is implementation-dependent. A threaded implementation
 6398: might spit out the execution token directly whilst another
 6399: implementation might spit out a native code sequence.}.
 6400: @item
 6401: At this point, the execution of @code{my:} is complete, and control
 6402: returns to the text interpreter. The text interpreter is in compile
 6403: state, so subsequent text @code{+ -} is compiled into the definition of
 6404: @code{foo} and the @code{;} terminates the definition as always.
 6405: @end itemize
 6406: 
 6407: You can use @code{see} to decompile a word that was defined using
 6408: @code{my:} and see how it is different from a normal @code{:}
 6409: definition. For example:
 6410: 
 6411: @example
 6412: : bar + - ;  \ like foo but using : rather than my:
 6413: see bar
 6414: : bar
 6415:   + - ;
 6416: see foo
 6417: : foo
 6418:   107645672 stats + - ;
 6419: 
 6420: \ use ' stats . to show that 107645672 is the xt for stats
 6421: @end example
 6422: 
 6423: You can use techniques like this to make new defining words in terms of
 6424: @i{any} existing defining word.
 6425: 
 6426: 
 6427: @cindex defining defining words
 6428: @cindex @code{CREATE} ... @code{DOES>}
 6429: If you want the words defined with your defining words to behave
 6430: differently from words defined with standard defining words, you can
 6431: write your defining word like this:
 6432: 
 6433: @example
 6434: : def-word ( "name" -- )
 6435:     CREATE @i{code1}
 6436: DOES> ( ... -- ... )
 6437:     @i{code2} ;
 6438: 
 6439: def-word name
 6440: @end example
 6441: 
 6442: @cindex child words
 6443: This fragment defines a @dfn{defining word} @code{def-word} and then
 6444: executes it.  When @code{def-word} executes, it @code{CREATE}s a new
 6445: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
 6446: is not executed at this time. The word @code{name} is sometimes called a
 6447: @dfn{child} of @code{def-word}.
 6448: 
 6449: When you execute @code{name}, the address of the body of @code{name} is
 6450: put on the data stack and @i{code2} is executed (the address of the body
 6451: of @code{name} is the address @code{HERE} returns immediately after the
 6452: @code{CREATE}, i.e., the address a @code{create}d word returns by
 6453: default).
 6454: 
 6455: @c anton:
 6456: @c www.dictionary.com says:
 6457: @c at·a·vism: 1.The reappearance of a characteristic in an organism after
 6458: @c several generations of absence, usually caused by the chance
 6459: @c recombination of genes.  2.An individual or a part that exhibits
 6460: @c atavism. Also called throwback.  3.The return of a trait or recurrence
 6461: @c of previous behavior after a period of absence.
 6462: @c
 6463: @c Doesn't seem to fit.
 6464: 
 6465: @c @cindex atavism in child words
 6466: You can use @code{def-word} to define a set of child words that behave
 6467: similarly; they all have a common run-time behaviour determined by
 6468: @i{code2}. Typically, the @i{code1} sequence builds a data area in the
 6469: body of the child word. The structure of the data is common to all
 6470: children of @code{def-word}, but the data values are specific -- and
 6471: private -- to each child word. When a child word is executed, the
 6472: address of its private data area is passed as a parameter on TOS to be
 6473: used and manipulated@footnote{It is legitimate both to read and write to
 6474: this data area.} by @i{code2}.
 6475: 
 6476: The two fragments of code that make up the defining words act (are
 6477: executed) at two completely separate times:
 6478: 
 6479: @itemize @bullet
 6480: @item
 6481: At @i{define time}, the defining word executes @i{code1} to generate a
 6482: child word
 6483: @item
 6484: At @i{child execution time}, when a child word is invoked, @i{code2}
 6485: is executed, using parameters (data) that are private and specific to
 6486: the child word.
 6487: @end itemize
 6488: 
 6489: Another way of understanding the behaviour of @code{def-word} and
 6490: @code{name} is to say that, if you make the following definitions:
 6491: @example
 6492: : def-word1 ( "name" -- )
 6493:     CREATE @i{code1} ;
 6494: 
 6495: : action1 ( ... -- ... )
 6496:     @i{code2} ;
 6497: 
 6498: def-word1 name1
 6499: @end example
 6500: 
 6501: @noindent
 6502: Then using @code{name1 action1} is equivalent to using @code{name}.
 6503: 
 6504: The classic example is that you can define @code{CONSTANT} in this way:
 6505: 
 6506: @example
 6507: : CONSTANT ( w "name" -- )
 6508:     CREATE ,
 6509: DOES> ( -- w )
 6510:     @@ ;
 6511: @end example
 6512: 
 6513: @comment There is a beautiful description of how this works and what
 6514: @comment it does in the Forthwrite 100th edition.. as well as an elegant
 6515: @comment commentary on the Counting Fruits problem.
 6516: 
 6517: When you create a constant with @code{5 CONSTANT five}, a set of
 6518: define-time actions take place; first a new word @code{five} is created,
 6519: then the value 5 is laid down in the body of @code{five} with
 6520: @code{,}. When @code{five} is executed, the address of the body is put on
 6521: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
 6522: no code of its own; it simply contains a data field and a pointer to the
 6523: code that follows @code{DOES>} in its defining word. That makes words
 6524: created in this way very compact.
 6525: 
 6526: The final example in this section is intended to remind you that space
 6527: reserved in @code{CREATE}d words is @i{data} space and therefore can be
 6528: both read and written by a Standard program@footnote{Exercise: use this
 6529: example as a starting point for your own implementation of @code{Value}
 6530: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
 6531: @code{[']}.}:
 6532: 
 6533: @example
 6534: : foo ( "name" -- )
 6535:     CREATE -1 ,
 6536: DOES> ( -- )
 6537:     @@ . ;
 6538: 
 6539: foo first-word
 6540: foo second-word
 6541: 
 6542: 123 ' first-word >BODY !
 6543: @end example
 6544: 
 6545: If @code{first-word} had been a @code{CREATE}d word, we could simply
 6546: have executed it to get the address of its data field. However, since it
 6547: was defined to have @code{DOES>} actions, its execution semantics are to
 6548: perform those @code{DOES>} actions. To get the address of its data field
 6549: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
 6550: translate the xt into the address of the data field.  When you execute
 6551: @code{first-word}, it will display @code{123}. When you execute
 6552: @code{second-word} it will display @code{-1}.
 6553: 
 6554: @cindex stack effect of @code{DOES>}-parts
 6555: @cindex @code{DOES>}-parts, stack effect
 6556: In the examples above the stack comment after the @code{DOES>} specifies
 6557: the stack effect of the defined words, not the stack effect of the
 6558: following code (the following code expects the address of the body on
 6559: the top of stack, which is not reflected in the stack comment). This is
 6560: the convention that I use and recommend (it clashes a bit with using
 6561: locals declarations for stack effect specification, though).
 6562: 
 6563: @menu
 6564: * CREATE..DOES> applications::  
 6565: * CREATE..DOES> details::       
 6566: * Advanced does> usage example::  
 6567: @end menu
 6568: 
 6569: @node CREATE..DOES> applications, CREATE..DOES> details, User-defined Defining Words, User-defined Defining Words
 6570: @subsubsection Applications of @code{CREATE..DOES>}
 6571: @cindex @code{CREATE} ... @code{DOES>}, applications
 6572: 
 6573: You may wonder how to use this feature. Here are some usage patterns:
 6574: 
 6575: @cindex factoring similar colon definitions
 6576: When you see a sequence of code occurring several times, and you can
 6577: identify a meaning, you will factor it out as a colon definition. When
 6578: you see similar colon definitions, you can factor them using
 6579: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
 6580: that look very similar:
 6581: @example
 6582: : ori, ( reg-target reg-source n -- )
 6583:     0 asm-reg-reg-imm ;
 6584: : andi, ( reg-target reg-source n -- )
 6585:     1 asm-reg-reg-imm ;
 6586: @end example
 6587: 
 6588: @noindent
 6589: This could be factored with:
 6590: @example
 6591: : reg-reg-imm ( op-code -- )
 6592:     CREATE ,
 6593: DOES> ( reg-target reg-source n -- )
 6594:     @@ asm-reg-reg-imm ;
 6595: 
 6596: 0 reg-reg-imm ori,
 6597: 1 reg-reg-imm andi,
 6598: @end example
 6599: 
 6600: @cindex currying
 6601: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
 6602: supply a part of the parameters for a word (known as @dfn{currying} in
 6603: the functional language community). E.g., @code{+} needs two
 6604: parameters. Creating versions of @code{+} with one parameter fixed can
 6605: be done like this:
 6606: 
 6607: @example
 6608: : curry+ ( n1 "name" -- )
 6609:     CREATE ,
 6610: DOES> ( n2 -- n1+n2 )
 6611:     @@ + ;
 6612: 
 6613:  3 curry+ 3+
 6614: -2 curry+ 2-
 6615: @end example
 6616: 
 6617: @node CREATE..DOES> details, Advanced does> usage example, CREATE..DOES> applications, User-defined Defining Words
 6618: @subsubsection The gory details of @code{CREATE..DOES>}
 6619: @cindex @code{CREATE} ... @code{DOES>}, details
 6620: 
 6621: doc-does>
 6622: 
 6623: @cindex @code{DOES>} in a separate definition
 6624: This means that you need not use @code{CREATE} and @code{DOES>} in the
 6625: same definition; you can put the @code{DOES>}-part in a separate
 6626: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
 6627: @example
 6628: : does1 
 6629: DOES> ( ... -- ... )
 6630:     ... ;
 6631: 
 6632: : does2
 6633: DOES> ( ... -- ... )
 6634:     ... ;
 6635: 
 6636: : def-word ( ... -- ... )
 6637:     create ...
 6638:     IF
 6639:        does1
 6640:     ELSE
 6641:        does2
 6642:     ENDIF ;
 6643: @end example
 6644: 
 6645: In this example, the selection of whether to use @code{does1} or
 6646: @code{does2} is made at definition-time; at the time that the child word is
 6647: @code{CREATE}d.
 6648: 
 6649: @cindex @code{DOES>} in interpretation state
 6650: In a standard program you can apply a @code{DOES>}-part only if the last
 6651: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
 6652: will override the behaviour of the last word defined in any case. In a
 6653: standard program, you can use @code{DOES>} only in a colon
 6654: definition. In Gforth, you can also use it in interpretation state, in a
 6655: kind of one-shot mode; for example:
 6656: @example
 6657: CREATE name ( ... -- ... )
 6658:   @i{initialization}
 6659: DOES>
 6660:   @i{code} ;
 6661: @end example
 6662: 
 6663: @noindent
 6664: is equivalent to the standard:
 6665: @example
 6666: :noname
 6667: DOES>
 6668:     @i{code} ;
 6669: CREATE name EXECUTE ( ... -- ... )
 6670:     @i{initialization}
 6671: @end example
 6672: 
 6673: doc->body
 6674: 
 6675: @node Advanced does> usage example,  , CREATE..DOES> details, User-defined Defining Words
 6676: @subsubsection Advanced does> usage example
 6677: 
 6678: The MIPS disassembler (@file{arch/mips/disasm.fs}) contains many words
 6679: for disassembling instructions, that follow a very repetetive scheme:
 6680: 
 6681: @example
 6682: :noname @var{disasm-operands} s" @var{inst-name}" type ;
 6683: @var{entry-num} cells @var{table} + !
 6684: @end example
 6685: 
 6686: Of course, this inspires the idea to factor out the commonalities to
 6687: allow a definition like
 6688: 
 6689: @example
 6690: @var{disasm-operands} @var{entry-num} @var{table} define-inst @var{inst-name}
 6691: @end example
 6692: 
 6693: The parameters @var{disasm-operands} and @var{table} are usually
 6694: correlated.  Moreover, before I wrote the disassembler, there already
 6695: existed code that defines instructions like this:
 6696: 
 6697: @example
 6698: @var{entry-num} @var{inst-format} @var{inst-name}
 6699: @end example
 6700: 
 6701: This code comes from the assembler and resides in
 6702: @file{arch/mips/insts.fs}.
 6703: 
 6704: So I had to define the @var{inst-format} words that performed the scheme
 6705: above when executed.  At first I chose to use run-time code-generation:
 6706: 
 6707: @example
 6708: : @var{inst-format} ( entry-num "name" -- ; compiled code: addr w -- )
 6709:   :noname Postpone @var{disasm-operands}
 6710:   name Postpone sliteral Postpone type Postpone ;
 6711:   swap cells @var{table} + ! ;
 6712: @end example
 6713: 
 6714: Note that this supplies the other two parameters of the scheme above.
 6715: 
 6716: An alternative would have been to write this using
 6717: @code{create}/@code{does>}:
 6718: 
 6719: @example
 6720: : @var{inst-format} ( entry-num "name" -- )
 6721:   here name string, ( entry-num c-addr ) \ parse and save "name"
 6722:   noname create , ( entry-num )
 6723:   lastxt swap cells @var{table} + !
 6724: does> ( addr w -- )
 6725:   \ disassemble instruction w at addr
 6726:   @@ >r 
 6727:   @var{disasm-operands}
 6728:   r> count type ;
 6729: @end example
 6730: 
 6731: Somehow the first solution is simpler, mainly because it's simpler to
 6732: shift a string from definition-time to use-time with @code{sliteral}
 6733: than with @code{string,} and friends.
 6734: 
 6735: I wrote a lot of words following this scheme and soon thought about
 6736: factoring out the commonalities among them.  Note that this uses a
 6737: two-level defining word, i.e., a word that defines ordinary defining
 6738: words.
 6739: 
 6740: This time a solution involving @code{postpone} and friends seemed more
 6741: difficult (try it as an exercise), so I decided to use a
 6742: @code{create}/@code{does>} word; since I was already at it, I also used
 6743: @code{create}/@code{does>} for the lower level (try using
 6744: @code{postpone} etc. as an exercise), resulting in the following
 6745: definition:
 6746: 
 6747: @example
 6748: : define-format ( disasm-xt table-xt -- )
 6749:     \ define an instruction format that uses disasm-xt for
 6750:     \ disassembling and enters the defined instructions into table
 6751:     \ table-xt
 6752:     create 2,
 6753: does> ( u "inst" -- )
 6754:     \ defines an anonymous word for disassembling instruction inst,
 6755:     \ and enters it as u-th entry into table-xt
 6756:     2@@ swap here name string, ( u table-xt disasm-xt c-addr ) \ remember string
 6757:     noname create 2,      \ define anonymous word
 6758:     execute lastxt swap ! \ enter xt of defined word into table-xt
 6759: does> ( addr w -- )
 6760:     \ disassemble instruction w at addr
 6761:     2@@ >r ( addr w disasm-xt R: c-addr )
 6762:     execute ( R: c-addr ) \ disassemble operands
 6763:     r> count type ; \ print name 
 6764: @end example
 6765: 
 6766: Note that the tables here (in contrast to above) do the @code{cells +}
 6767: by themselves (that's why you have to pass an xt).  This word is used in
 6768: the following way:
 6769: 
 6770: @example
 6771: ' @var{disasm-operands} ' @var{table} define-format @var{inst-format}
 6772: @end example
 6773: 
 6774: As shown above, the defined instruction format is then used like this:
 6775: 
 6776: @example
 6777: @var{entry-num} @var{inst-format} @var{inst-name}
 6778: @end example
 6779: 
 6780: In terms of currying, this kind of two-level defining word provides the
 6781: parameters in three stages: first @var{disasm-operands} and @var{table},
 6782: then @var{entry-num} and @var{inst-name}, finally @code{addr w}, i.e.,
 6783: the instruction to be disassembled.  
 6784: 
 6785: Of course this did not quite fit all the instruction format names used
 6786: in @file{insts.fs}, so I had to define a few wrappers that conditioned
 6787: the parameters into the right form.
 6788: 
 6789: If you have trouble following this section, don't worry.  First, this is
 6790: involved and takes time (and probably some playing around) to
 6791: understand; second, this is the first two-level
 6792: @code{create}/@code{does>} word I have written in seventeen years of
 6793: Forth; and if I did not have @file{insts.fs} to start with, I may well
 6794: have elected to use just a one-level defining word (with some repeating
 6795: of parameters when using the defining word). So it is not necessary to
 6796: understand this, but it may improve your understanding of Forth.
 6797: 
 6798: 
 6799: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
 6800: @subsection Deferred words
 6801: @cindex deferred words
 6802: 
 6803: The defining word @code{Defer} allows you to define a word by name
 6804: without defining its behaviour; the definition of its behaviour is
 6805: deferred. Here are two situation where this can be useful:
 6806: 
 6807: @itemize @bullet
 6808: @item
 6809: Where you want to allow the behaviour of a word to be altered later, and
 6810: for all precompiled references to the word to change when its behaviour
 6811: is changed.
 6812: @item
 6813: For mutual recursion; @xref{Calls and returns}.
 6814: @end itemize
 6815: 
 6816: In the following example, @code{foo} always invokes the version of
 6817: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
 6818: always invokes the version that prints ``@code{Hello}''. There is no way
 6819: of getting @code{foo} to use the later version without re-ordering the
 6820: source code and recompiling it.
 6821: 
 6822: @example
 6823: : greet ." Good morning" ;
 6824: : foo ... greet ... ;
 6825: : greet ." Hello" ;
 6826: : bar ... greet ... ;
 6827: @end example
 6828: 
 6829: This problem can be solved by defining @code{greet} as a @code{Defer}red
 6830: word. The behaviour of a @code{Defer}red word can be defined and
 6831: redefined at any time by using @code{IS} to associate the xt of a
 6832: previously-defined word with it. The previous example becomes:
 6833: 
 6834: @example
 6835: Defer greet ( -- )
 6836: : foo ... greet ... ;
 6837: : bar ... greet ... ;
 6838: : greet1 ( -- ) ." Good morning" ;
 6839: : greet2 ( -- ) ." Hello" ;
 6840: ' greet2 <IS> greet  \ make greet behave like greet2
 6841: @end example
 6842: 
 6843: @progstyle
 6844: You should write a stack comment for every deferred word, and put only
 6845: XTs into deferred words that conform to this stack effect.  Otherwise
 6846: it's too difficult to use the deferred word.
 6847: 
 6848: A deferred word can be used to improve the statistics-gathering example
 6849: from @ref{User-defined Defining Words}; rather than edit the
 6850: application's source code to change every @code{:} to a @code{my:}, do
 6851: this:
 6852: 
 6853: @example
 6854: : real: : ;     \ retain access to the original
 6855: defer :         \ redefine as a deferred word
 6856: ' my: <IS> :      \ use special version of :
 6857: \
 6858: \ load application here
 6859: \
 6860: ' real: <IS> :    \ go back to the original
 6861: @end example
 6862: 
 6863: 
 6864: One thing to note is that @code{<IS>} consumes its name when it is
 6865: executed.  If you want to specify the name at compile time, use
 6866: @code{[IS]}:
 6867: 
 6868: @example
 6869: : set-greet ( xt -- )
 6870:   [IS] greet ;
 6871: 
 6872: ' greet1 set-greet
 6873: @end example
 6874: 
 6875: A deferred word can only inherit execution semantics from the xt
 6876: (because that is all that an xt can represent -- for more discussion of
 6877: this @pxref{Tokens for Words}); by default it will have default
 6878: interpretation and compilation semantics deriving from this execution
 6879: semantics.  However, you can change the interpretation and compilation
 6880: semantics of the deferred word in the usual ways:
 6881: 
 6882: @example
 6883: : bar .... ; compile-only
 6884: Defer fred immediate
 6885: Defer jim
 6886: 
 6887: ' bar <IS> jim  \ jim has default semantics
 6888: ' bar <IS> fred \ fred is immediate
 6889: @end example
 6890: 
 6891: doc-defer
 6892: doc-<is>
 6893: doc-[is]
 6894: doc-is
 6895: @comment TODO document these: what's defers [is]
 6896: doc-what's
 6897: doc-defers
 6898: 
 6899: @c Use @code{words-deferred} to see a list of deferred words.
 6900: 
 6901: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
 6902: are provided in @file{compat/defer.fs}.
 6903: 
 6904: 
 6905: @node Aliases,  , Deferred words, Defining Words
 6906: @subsection Aliases
 6907: @cindex aliases
 6908: 
 6909: The defining word @code{Alias} allows you to define a word by name that
 6910: has the same behaviour as some other word. Here are two situation where
 6911: this can be useful:
 6912: 
 6913: @itemize @bullet
 6914: @item
 6915: When you want access to a word's definition from a different word list
 6916: (for an example of this, see the definition of the @code{Root} word list
 6917: in the Gforth source).
 6918: @item
 6919: When you want to create a synonym; a definition that can be known by
 6920: either of two names (for example, @code{THEN} and @code{ENDIF} are
 6921: aliases).
 6922: @end itemize
 6923: 
 6924: Like deferred words, an alias has default compilation and interpretation
 6925: semantics at the beginning (not the modifications of the other word),
 6926: but you can change them in the usual ways (@code{immediate},
 6927: @code{compile-only}). For example:
 6928: 
 6929: @example
 6930: : foo ... ; immediate
 6931: 
 6932: ' foo Alias bar \ bar is not an immediate word
 6933: ' foo Alias fooby immediate \ fooby is an immediate word
 6934: @end example
 6935: 
 6936: Words that are aliases have the same xt, different headers in the
 6937: dictionary, and consequently different name tokens (@pxref{Tokens for
 6938: Words}) and possibly different immediate flags.  An alias can only have
 6939: default or immediate compilation semantics; you can define aliases for
 6940: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
 6941: 
 6942: doc-alias
 6943: 
 6944: 
 6945: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
 6946: @section Interpretation and Compilation Semantics
 6947: @cindex semantics, interpretation and compilation
 6948: 
 6949: @c !! state and ' are used without explanation
 6950: @c example for immediate/compile-only? or is the tutorial enough
 6951: 
 6952: @cindex interpretation semantics
 6953: The @dfn{interpretation semantics} of a (named) word are what the text
 6954: interpreter does when it encounters the word in interpret state. It also
 6955: appears in some other contexts, e.g., the execution token returned by
 6956: @code{' @i{word}} identifies the interpretation semantics of @i{word}
 6957: (in other words, @code{' @i{word} execute} is equivalent to
 6958: interpret-state text interpretation of @code{@i{word}}).
 6959: 
 6960: @cindex compilation semantics
 6961: The @dfn{compilation semantics} of a (named) word are what the text
 6962: interpreter does when it encounters the word in compile state. It also
 6963: appears in other contexts, e.g, @code{POSTPONE @i{word}}
 6964: compiles@footnote{In standard terminology, ``appends to the current
 6965: definition''.} the compilation semantics of @i{word}.
 6966: 
 6967: @cindex execution semantics
 6968: The standard also talks about @dfn{execution semantics}. They are used
 6969: only for defining the interpretation and compilation semantics of many
 6970: words. By default, the interpretation semantics of a word are to
 6971: @code{execute} its execution semantics, and the compilation semantics of
 6972: a word are to @code{compile,} its execution semantics.@footnote{In
 6973: standard terminology: The default interpretation semantics are its
 6974: execution semantics; the default compilation semantics are to append its
 6975: execution semantics to the execution semantics of the current
 6976: definition.}
 6977: 
 6978: Unnamed words (@pxref{Anonymous Definitions}) cannot be encountered by
 6979: the text interpreter, ticked, or @code{postpone}d, so they have no
 6980: interpretation or compilation semantics.  Their behaviour is represented
 6981: by their XT (@pxref{Tokens for Words}), and we call it execution
 6982: semantics, too.
 6983: 
 6984: @comment TODO expand, make it co-operate with new sections on text interpreter.
 6985: 
 6986: @cindex immediate words
 6987: @cindex compile-only words
 6988: You can change the semantics of the most-recently defined word:
 6989: 
 6990: 
 6991: doc-immediate
 6992: doc-compile-only
 6993: doc-restrict
 6994: 
 6995: By convention, words with non-default compilation semantics (e.g.,
 6996: immediate words) often have names surrounded with brackets (e.g.,
 6997: @code{[']}, @pxref{Execution token}).
 6998: 
 6999: Note that ticking (@code{'}) a compile-only word gives an error
 7000: (``Interpreting a compile-only word'').
 7001: 
 7002: @menu
 7003: * Combined words::              
 7004: @end menu
 7005: 
 7006: 
 7007: @node Combined words,  , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
 7008: @subsection Combined Words
 7009: @cindex combined words
 7010: 
 7011: Gforth allows you to define @dfn{combined words} -- words that have an
 7012: arbitrary combination of interpretation and compilation semantics.
 7013: 
 7014: doc-interpret/compile:
 7015: 
 7016: This feature was introduced for implementing @code{TO} and @code{S"}. I
 7017: recommend that you do not define such words, as cute as they may be:
 7018: they make it hard to get at both parts of the word in some contexts.
 7019: E.g., assume you want to get an execution token for the compilation
 7020: part. Instead, define two words, one that embodies the interpretation
 7021: part, and one that embodies the compilation part.  Once you have done
 7022: that, you can define a combined word with @code{interpret/compile:} for
 7023: the convenience of your users.
 7024: 
 7025: You might try to use this feature to provide an optimizing
 7026: implementation of the default compilation semantics of a word. For
 7027: example, by defining:
 7028: @example
 7029: :noname
 7030:    foo bar ;
 7031: :noname
 7032:    POSTPONE foo POSTPONE bar ;
 7033: interpret/compile: opti-foobar
 7034: @end example
 7035: 
 7036: @noindent
 7037: as an optimizing version of:
 7038: 
 7039: @example
 7040: : foobar
 7041:     foo bar ;
 7042: @end example
 7043: 
 7044: Unfortunately, this does not work correctly with @code{[compile]},
 7045: because @code{[compile]} assumes that the compilation semantics of all
 7046: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
 7047: opti-foobar} would compile compilation semantics, whereas
 7048: @code{[compile] foobar} would compile interpretation semantics.
 7049: 
 7050: @cindex state-smart words (are a bad idea)
 7051: @anchor{state-smartness}
 7052: Some people try to use @dfn{state-smart} words to emulate the feature provided
 7053: by @code{interpret/compile:} (words are state-smart if they check
 7054: @code{STATE} during execution). E.g., they would try to code
 7055: @code{foobar} like this:
 7056: 
 7057: @example
 7058: : foobar
 7059:   STATE @@
 7060:   IF ( compilation state )
 7061:     POSTPONE foo POSTPONE bar
 7062:   ELSE
 7063:     foo bar
 7064:   ENDIF ; immediate
 7065: @end example
 7066: 
 7067: Although this works if @code{foobar} is only processed by the text
 7068: interpreter, it does not work in other contexts (like @code{'} or
 7069: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
 7070: for a state-smart word, not for the interpretation semantics of the
 7071: original @code{foobar}; when you execute this execution token (directly
 7072: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
 7073: state, the result will not be what you expected (i.e., it will not
 7074: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
 7075: write them@footnote{For a more detailed discussion of this topic, see
 7076: M. Anton Ertl,
 7077: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,@code{State}-smartness---Why
 7078: it is Evil and How to Exorcise it}}, EuroForth '98.}!
 7079: 
 7080: @cindex defining words with arbitrary semantics combinations
 7081: It is also possible to write defining words that define words with
 7082: arbitrary combinations of interpretation and compilation semantics. In
 7083: general, they look like this:
 7084: 
 7085: @example
 7086: : def-word
 7087:     create-interpret/compile
 7088:     @i{code1}
 7089: interpretation>
 7090:     @i{code2}
 7091: <interpretation
 7092: compilation>
 7093:     @i{code3}
 7094: <compilation ;
 7095: @end example
 7096: 
 7097: For a @i{word} defined with @code{def-word}, the interpretation
 7098: semantics are to push the address of the body of @i{word} and perform
 7099: @i{code2}, and the compilation semantics are to push the address of
 7100: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
 7101: can also be defined like this (except that the defined constants don't
 7102: behave correctly when @code{[compile]}d):
 7103: 
 7104: @example
 7105: : constant ( n "name" -- )
 7106:     create-interpret/compile
 7107:     ,
 7108: interpretation> ( -- n )
 7109:     @@
 7110: <interpretation
 7111: compilation> ( compilation. -- ; run-time. -- n )
 7112:     @@ postpone literal
 7113: <compilation ;
 7114: @end example
 7115: 
 7116: 
 7117: doc-create-interpret/compile
 7118: doc-interpretation>
 7119: doc-<interpretation
 7120: doc-compilation>
 7121: doc-<compilation
 7122: 
 7123: 
 7124: Words defined with @code{interpret/compile:} and
 7125: @code{create-interpret/compile} have an extended header structure that
 7126: differs from other words; however, unless you try to access them with
 7127: plain address arithmetic, you should not notice this. Words for
 7128: accessing the header structure usually know how to deal with this; e.g.,
 7129: @code{'} @i{word} @code{>body} also gives you the body of a word created
 7130: with @code{create-interpret/compile}.
 7131: 
 7132: 
 7133: @c -------------------------------------------------------------
 7134: @node Tokens for Words, Compiling words, Interpretation and Compilation Semantics, Words
 7135: @section Tokens for Words
 7136: @cindex tokens for words
 7137: 
 7138: This section describes the creation and use of tokens that represent
 7139: words.
 7140: 
 7141: @menu
 7142: * Execution token::             represents execution/interpretation semantics
 7143: * Compilation token::           represents compilation semantics
 7144: * Name token::                  represents named words
 7145: @end menu
 7146: 
 7147: @node Execution token, Compilation token, Tokens for Words, Tokens for Words
 7148: @subsection Execution token
 7149: 
 7150: @cindex xt
 7151: @cindex execution token
 7152: An @dfn{execution token} (@i{XT}) represents some behaviour of a word.
 7153: You can use @code{execute} to invoke this behaviour.
 7154: 
 7155: @cindex tick (')
 7156: You can use @code{'} to get an execution token that represents the
 7157: interpretation semantics of a named word:
 7158: 
 7159: @example
 7160: 5 ' .
 7161: execute
 7162: @end example
 7163: 
 7164: doc-'
 7165: 
 7166: @code{'} parses at run-time; there is also a word @code{[']} that parses
 7167: when it is compiled, and compiles the resulting XT:
 7168: 
 7169: @example
 7170: : foo ['] . execute ;
 7171: 5 foo
 7172: : bar ' execute ; \ by contrast,
 7173: 5 bar .           \ ' parses "." when bar executes
 7174: @end example
 7175: 
 7176: doc-[']
 7177: 
 7178: If you want the execution token of @i{word}, write @code{['] @i{word}}
 7179: in compiled code and @code{' @i{word}} in interpreted code.  Gforth's
 7180: @code{'} and @code{[']} behave somewhat unusually by complaining about
 7181: compile-only words (because these words have no interpretation
 7182: semantics).  You might get what you want by using @code{COMP' @i{word}
 7183: DROP} or @code{[COMP'] @i{word} DROP} (for details @pxref{Compilation
 7184: token}).
 7185: 
 7186: Another way to get an XT is @code{:noname} or @code{lastxt}
 7187: (@pxref{Anonymous Definitions}).  For anonymous words this gives an xt
 7188: for the only behaviour the word has (the execution semantics).  For
 7189: named words, @code{lastxt} produces an XT for the same behaviour it
 7190: would produce if the word was defined anonymously.
 7191: 
 7192: @example
 7193: :noname ." hello" ;
 7194: execute
 7195: @end example
 7196: 
 7197: An XT occupies one cell and can be manipulated like any other cell.
 7198: 
 7199: @cindex code field address
 7200: @cindex CFA
 7201: In ANS Forth the XT is just an abstract data type (i.e., defined by the
 7202: operations that produce or consume it).  For old hands: In Gforth, the
 7203: XT is implemented as a code field address (CFA).
 7204: 
 7205: doc-execute
 7206: doc-perform
 7207: 
 7208: @node Compilation token, Name token, Execution token, Tokens for Words
 7209: @subsection Compilation token
 7210: 
 7211: @cindex compilation token
 7212: @cindex CT (compilation token)
 7213: Gforth represents the compilation semantics of a named word by a
 7214: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
 7215: @i{xt} is an execution token. The compilation semantics represented by
 7216: the compilation token can be performed with @code{execute}, which
 7217: consumes the whole compilation token, with an additional stack effect
 7218: determined by the represented compilation semantics.
 7219: 
 7220: At present, the @i{w} part of a compilation token is an execution token,
 7221: and the @i{xt} part represents either @code{execute} or
 7222: @code{compile,}@footnote{Depending upon the compilation semantics of the
 7223: word. If the word has default compilation semantics, the @i{xt} will
 7224: represent @code{compile,}. Otherwise (e.g., for immediate words), the
 7225: @i{xt} will represent @code{execute}.}. However, don't rely on that
 7226: knowledge, unless necessary; future versions of Gforth may introduce
 7227: unusual compilation tokens (e.g., a compilation token that represents
 7228: the compilation semantics of a literal).
 7229: 
 7230: You can perform the compilation semantics represented by the compilation
 7231: token with @code{execute}.  You can compile the compilation semantics
 7232: with @code{postpone,}. I.e., @code{COMP' @i{word} postpone,} is
 7233: equivalent to @code{postpone @i{word}}.
 7234: 
 7235: doc-[comp']
 7236: doc-comp'
 7237: doc-postpone,
 7238: 
 7239: @node Name token,  , Compilation token, Tokens for Words
 7240: @subsection Name token
 7241: 
 7242: @cindex name token
 7243: @cindex name field address
 7244: @cindex NFA
 7245: Gforth represents named words by the @dfn{name token}, (@i{nt}). In
 7246: Gforth, the abstract data type @emph{name token} is implemented as a
 7247: name field address (NFA).
 7248: 
 7249: doc-find-name
 7250: doc-name>int
 7251: doc-name?int
 7252: doc-name>comp
 7253: doc-name>string
 7254: 
 7255: @c ----------------------------------------------------------
 7256: @node Compiling words, The Text Interpreter, Tokens for Words, Words
 7257: @section Compiling words
 7258: @cindex compiling words
 7259: @cindex macros
 7260: 
 7261: In contrast to most other languages, Forth has no strict boundary
 7262: between compilation and run-time.  E.g., you can run arbitrary code
 7263: between defining words (or for computing data used by defining words
 7264: like @code{constant}). Moreover, @code{Immediate} (@pxref{Interpretation
 7265: and Compilation Semantics} and @code{[}...@code{]} (see below) allow
 7266: running arbitrary code while compiling a colon definition (exception:
 7267: you must not allot dictionary space).
 7268: 
 7269: @menu
 7270: * Literals::                    Compiling data values
 7271: * Macros::                      Compiling words
 7272: @end menu
 7273: 
 7274: @node Literals, Macros, Compiling words, Compiling words
 7275: @subsection Literals
 7276: @cindex Literals
 7277: 
 7278: The simplest and most frequent example is to compute a literal during
 7279: compilation.  E.g., the following definition prints an array of strings,
 7280: one string per line:
 7281: 
 7282: @example
 7283: : .strings ( addr u -- ) \ gforth
 7284:     2* cells bounds U+DO
 7285: 	cr i 2@@ type
 7286:     2 cells +LOOP ;  
 7287: @end example
 7288: 
 7289: With a simple-minded compiler like Gforth's, this computes @code{2
 7290: cells} on every loop iteration.  You can compute this value once and for
 7291: all at compile time and compile it into the definition like this:
 7292: 
 7293: @example
 7294: : .strings ( addr u -- ) \ gforth
 7295:     2* cells bounds U+DO
 7296: 	cr i 2@@ type
 7297:     [ 2 cells ] literal +LOOP ;  
 7298: @end example
 7299: 
 7300: @code{[} switches the text interpreter to interpret state (you will get
 7301: an @code{ok} prompt if you type this example interactively and insert a
 7302: newline between @code{[} and @code{]}), so it performs the
 7303: interpretation semantics of @code{2 cells}; this computes a number.
 7304: @code{]} switches the text interpreter back into compile state.  It then
 7305: performs @code{Literal}'s compilation semantics, which are to compile
 7306: this number into the current word.  You can decompile the word with
 7307: @code{see .strings} to see the effect on the compiled code.
 7308: 
 7309: You can also optimize the @code{2* cells} into @code{[ 2 cells ] literal
 7310: *} in this way.
 7311: 
 7312: doc-[
 7313: doc-]
 7314: doc-literal
 7315: doc-]L
 7316: 
 7317: There are also words for compiling other data types than single cells as
 7318: literals:
 7319: 
 7320: doc-2literal
 7321: doc-fliteral
 7322: doc-sliteral
 7323: 
 7324: @cindex colon-sys, passing data across @code{:}
 7325: @cindex @code{:}, passing data across
 7326: You might be tempted to pass data from outside a colon definition to the
 7327: inside on the data stack.  This does not work, because @code{:} puhes a
 7328: colon-sys, making stuff below unaccessible.  E.g., this does not work:
 7329: 
 7330: @example
 7331: 5 : foo literal ; \ error: "unstructured"
 7332: @end example
 7333: 
 7334: Instead, you have to pass the value in some other way, e.g., through a
 7335: variable:
 7336: 
 7337: @example
 7338: variable temp
 7339: 5 temp !
 7340: : foo [ temp @@ ] literal ;
 7341: @end example
 7342: 
 7343: 
 7344: @node Macros,  , Literals, Compiling words
 7345: @subsection Macros
 7346: @cindex Macros
 7347: @cindex compiling compilation semantics
 7348: 
 7349: @code{Literal} and friends compile data values into the current
 7350: definition.  You can also write words that compile other words into the
 7351: current definition.  E.g.,
 7352: 
 7353: @example
 7354: : compile-+ ( -- ) \ compiled code: ( n1 n2 -- n )
 7355:   POSTPONE + ;
 7356: 
 7357: : foo ( n1 n2 -- n )
 7358:   [ compile-+ ] ;
 7359: 1 2 foo .
 7360: @end example
 7361: 
 7362: This is equivalent to @code{: foo + ;} (@code{see foo} to check this).
 7363: What happens in this example?  @code{Postpone} compiles the compilation
 7364: semantics of @code{+} into @code{compile-+}; later the text interpreter
 7365: executes @code{compile-+} and thus the compilation semantics of +, which
 7366: compile (the execution semantics of) @code{+} into
 7367: @code{foo}.@footnote{A recent RFI answer requires that compiling words
 7368: should only be executed in compile state, so this example is not
 7369: guaranteed to work on all standard systems, but on any decent system it
 7370: will work.}
 7371: 
 7372: doc-postpone
 7373: doc-[compile]
 7374: 
 7375: Compiling words like @code{compile-+} are usually immediate (or similar)
 7376: so you do not have to switch to interpret state to execute them;
 7377: mopifying the last example accordingly produces:
 7378: 
 7379: @example
 7380: : [compile-+] ( compilation: --; interpretation: -- )
 7381:   \ compiled code: ( n1 n2 -- n )
 7382:   POSTPONE + ; immediate
 7383: 
 7384: : foo ( n1 n2 -- n )
 7385:   [compile-+] ;
 7386: 1 2 foo .
 7387: @end example
 7388: 
 7389: Immediate compiling words are similar to macros in other languages (in
 7390: particular, Lisp).  The important differences to macros in, e.g., C are:
 7391: 
 7392: @itemize @bullet
 7393: 
 7394: @item
 7395: You use the same language for defining and processing macros, not a
 7396: separate preprocessing language and processor.
 7397: 
 7398: @item
 7399: Consequently, the full power of Forth is available in macro definitions.
 7400: E.g., you can perform arbitrarily complex computations, or generate
 7401: different code conditionally or in a loop (e.g., @pxref{Advanced macros
 7402: Tutorial}).  This power is very useful when writing a parser generators
 7403: or other code-generating software.
 7404: 
 7405: @item
 7406: Macros defined using @code{postpone} etc. deal with the language at a
 7407: higher level than strings; name binding happens at macro definition
 7408: time, so you can avoid the pitfalls of name collisions that can happen
 7409: in C macros.  Of course, Forth is a liberal language and also allows to
 7410: shoot yourself in the foot with text-interpreted macros like
 7411: 
 7412: @example
 7413: : [compile-+] s" +" evaluate ; immediate
 7414: @end example
 7415: 
 7416: Apart from binding the name at macro use time, using @code{evaluate}
 7417: also makes your definition @code{state}-smart (@pxref{state-smartness}).
 7418: @end itemize
 7419: 
 7420: You may want the macro to compile a number into a word.  The word to do
 7421: it is @code{literal}, but you have to @code{postpone} it, so its
 7422: compilation semantics take effect when the macro is executed, not when
 7423: it is compiled:
 7424: 
 7425: @example
 7426: : [compile-5] ( -- ) \ compiled code: ( -- n )
 7427:   5 POSTPONE literal ; immediate
 7428: 
 7429: : foo [compile-5] ;
 7430: foo .
 7431: @end example
 7432: 
 7433: You may want to pass parameters to a macro, that the macro should
 7434: compile into the current definition.  If the parameter is a number, then
 7435: you can use @code{postpone literal} (similar for other values).
 7436: 
 7437: If you want to pass a word that is to be compiled, the usual way is to
 7438: pass an execution token and @code{compile,} it:
 7439: 
 7440: @example
 7441: : twice1 ( xt -- ) \ compiled code: ... -- ...
 7442:   dup compile, compile, ;
 7443: 
 7444: : 2+ ( n1 -- n2 )
 7445:   [ ' 1+ twice1 ] ;
 7446: @end example
 7447: 
 7448: doc-compile,
 7449: 
 7450: An alternative available in Gforth, that allows you to pass compile-only
 7451: words as parameters is to use the compilation token (@pxref{Compilation
 7452: token}).  The same example in this technique:
 7453: 
 7454: @example
 7455: : twice ( ... ct -- ... ) \ compiled code: ... -- ...
 7456:   2dup 2>r execute 2r> execute ;
 7457: 
 7458: : 2+ ( n1 -- n2 )
 7459:   [ comp' 1+ twice ] ;
 7460: @end example
 7461: 
 7462: In the example above @code{2>r} and @code{2r>} ensure that @code{twice}
 7463: works even if the executed compilation semantics has an effect on the
 7464: data stack.
 7465: 
 7466: You can also define complete definitions with these words; this provides
 7467: an alternative to using @code{does>} (@pxref{User-defined Defining
 7468: Words}).  E.g., instead of
 7469: 
 7470: @example
 7471: : curry+ ( n1 "name" -- )
 7472:     CREATE ,
 7473: DOES> ( n2 -- n1+n2 )
 7474:     @@ + ;
 7475: @end example
 7476: 
 7477: you could define
 7478: 
 7479: @example
 7480: : curry+ ( n1 "name" -- )
 7481:   \ name execution: ( n2 -- n1+n2 )
 7482:   >r : r> POSTPONE literal POSTPONE + POSTPONE ; ;
 7483: 
 7484: -3 curry+ 3-
 7485: see 3-
 7486: @end example
 7487: 
 7488: The sequence @code{>r : r>} is necessary, because @code{:} puts a
 7489: colon-sys on the data stack that makes everything below it unaccessible.
 7490: 
 7491: This way of writing defining words is sometimes more, sometimes less
 7492: convenient than using @code{does>} (@pxref{Advanced does> usage
 7493: example}).  One advantage of this method is that it can be optimized
 7494: better, because the compiler knows that the value compiled with
 7495: @code{literal} is fixed, whereas the data associated with a
 7496: @code{create}d word can be changed.
 7497: 
 7498: @c ----------------------------------------------------------
 7499: @node The Text Interpreter, Word Lists, Compiling words, Words
 7500: @section  The Text Interpreter
 7501: @cindex interpreter - outer
 7502: @cindex text interpreter
 7503: @cindex outer interpreter
 7504: 
 7505: @c Should we really describe all these ugly details?  IMO the text
 7506: @c interpreter should be much cleaner, but that may not be possible within
 7507: @c ANS Forth. - anton
 7508: @c nac-> I wanted to explain how it works to show how you can exploit
 7509: @c it in your own programs. When I was writing a cross-compiler, figuring out
 7510: @c some of these gory details was very helpful to me. None of the textbooks
 7511: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
 7512: @c seems to positively avoid going into too much detail for some of
 7513: @c the internals.
 7514: 
 7515: @c anton: ok.  I wonder, though, if this is the right place; for some stuff
 7516: @c it is; for the ugly details, I would prefer another place.  I wonder
 7517: @c whether we should have a chapter before "Words" that describes some
 7518: @c basic concepts referred to in words, and a chapter after "Words" that
 7519: @c describes implementation details.
 7520: 
 7521: The text interpreter@footnote{This is an expanded version of the
 7522: material in @ref{Introducing the Text Interpreter}.} is an endless loop
 7523: that processes input from the current input device. It is also called
 7524: the outer interpreter, in contrast to the inner interpreter
 7525: (@pxref{Engine}) which executes the compiled Forth code on interpretive
 7526: implementations.
 7527: 
 7528: @cindex interpret state
 7529: @cindex compile state
 7530: The text interpreter operates in one of two states: @dfn{interpret
 7531: state} and @dfn{compile state}. The current state is defined by the
 7532: aptly-named variable @code{state}.
 7533: 
 7534: This section starts by describing how the text interpreter behaves when
 7535: it is in interpret state, processing input from the user input device --
 7536: the keyboard. This is the mode that a Forth system is in after it starts
 7537: up.
 7538: 
 7539: @cindex input buffer
 7540: @cindex terminal input buffer
 7541: The text interpreter works from an area of memory called the @dfn{input
 7542: buffer}@footnote{When the text interpreter is processing input from the
 7543: keyboard, this area of memory is called the @dfn{terminal input buffer}
 7544: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
 7545: @code{#TIB}.}, which stores your keyboard input when you press the
 7546: @key{RET} key. Starting at the beginning of the input buffer, it skips
 7547: leading spaces (called @dfn{delimiters}) then parses a string (a
 7548: sequence of non-space characters) until it reaches either a space
 7549: character or the end of the buffer. Having parsed a string, it makes two
 7550: attempts to process it:
 7551: 
 7552: @cindex dictionary
 7553: @itemize @bullet
 7554: @item
 7555: It looks for the string in a @dfn{dictionary} of definitions. If the
 7556: string is found, the string names a @dfn{definition} (also known as a
 7557: @dfn{word}) and the dictionary search returns information that allows
 7558: the text interpreter to perform the word's @dfn{interpretation
 7559: semantics}. In most cases, this simply means that the word will be
 7560: executed.
 7561: @item
 7562: If the string is not found in the dictionary, the text interpreter
 7563: attempts to treat it as a number, using the rules described in
 7564: @ref{Number Conversion}. If the string represents a legal number in the
 7565: current radix, the number is pushed onto a parameter stack (the data
 7566: stack for integers, the floating-point stack for floating-point
 7567: numbers).
 7568: @end itemize
 7569: 
 7570: If both attempts fail, or if the word is found in the dictionary but has
 7571: no interpretation semantics@footnote{This happens if the word was
 7572: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
 7573: remainder of the input buffer, issues an error message and waits for
 7574: more input. If one of the attempts succeeds, the text interpreter
 7575: repeats the parsing process until the whole of the input buffer has been
 7576: processed, at which point it prints the status message ``@code{ ok}''
 7577: and waits for more input.
 7578: 
 7579: @c anton: this should be in the input stream subsection (or below it)
 7580: 
 7581: @cindex parse area
 7582: The text interpreter keeps track of its position in the input buffer by
 7583: updating a variable called @code{>IN} (pronounced ``to-in''). The value
 7584: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
 7585: of the input buffer. The region from offset @code{>IN @@} to the end of
 7586: the input buffer is called the @dfn{parse area}@footnote{In other words,
 7587: the text interpreter processes the contents of the input buffer by
 7588: parsing strings from the parse area until the parse area is empty.}.
 7589: This example shows how @code{>IN} changes as the text interpreter parses
 7590: the input buffer:
 7591: 
 7592: @example
 7593: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
 7594:   CR ." ->" TYPE ." <-" ; IMMEDIATE 
 7595: 
 7596: 1 2 3 remaining + remaining . 
 7597: 
 7598: : foo 1 2 3 remaining SWAP remaining ;
 7599: @end example
 7600: 
 7601: @noindent
 7602: The result is:
 7603: 
 7604: @example
 7605: ->+ remaining .<-
 7606: ->.<-5  ok
 7607: 
 7608: ->SWAP remaining ;-<
 7609: ->;<-  ok
 7610: @end example
 7611: 
 7612: @cindex parsing words
 7613: The value of @code{>IN} can also be modified by a word in the input
 7614: buffer that is executed by the text interpreter.  This means that a word
 7615: can ``trick'' the text interpreter into either skipping a section of the
 7616: input buffer@footnote{This is how parsing words work.} or into parsing a
 7617: section twice. For example:
 7618: 
 7619: @example
 7620: : lat ." <<foo>>" ;
 7621: : flat ." <<bar>>" >IN DUP @@ 3 - SWAP ! ;
 7622: @end example
 7623: 
 7624: @noindent
 7625: When @code{flat} is executed, this output is produced@footnote{Exercise
 7626: for the reader: what would happen if the @code{3} were replaced with
 7627: @code{4}?}:
 7628: 
 7629: @example
 7630: <<bar>><<foo>>
 7631: @end example
 7632: 
 7633: This technique can be used to work around some of the interoperability
 7634: problems of parsing words.  Of course, it's better to avoid parsing
 7635: words where possible.
 7636: 
 7637: @noindent
 7638: Two important notes about the behaviour of the text interpreter:
 7639: 
 7640: @itemize @bullet
 7641: @item
 7642: It processes each input string to completion before parsing additional
 7643: characters from the input buffer.
 7644: @item
 7645: It treats the input buffer as a read-only region (and so must your code).
 7646: @end itemize
 7647: 
 7648: @noindent
 7649: When the text interpreter is in compile state, its behaviour changes in
 7650: these ways:
 7651: 
 7652: @itemize @bullet
 7653: @item
 7654: If a parsed string is found in the dictionary, the text interpreter will
 7655: perform the word's @dfn{compilation semantics}. In most cases, this
 7656: simply means that the execution semantics of the word will be appended
 7657: to the current definition.
 7658: @item
 7659: When a number is encountered, it is compiled into the current definition
 7660: (as a literal) rather than being pushed onto a parameter stack.
 7661: @item
 7662: If an error occurs, @code{state} is modified to put the text interpreter
 7663: back into interpret state.
 7664: @item
 7665: Each time a line is entered from the keyboard, Gforth prints
 7666: ``@code{ compiled}'' rather than `` @code{ok}''.
 7667: @end itemize
 7668: 
 7669: @cindex text interpreter - input sources
 7670: When the text interpreter is using an input device other than the
 7671: keyboard, its behaviour changes in these ways:
 7672: 
 7673: @itemize @bullet
 7674: @item
 7675: When the parse area is empty, the text interpreter attempts to refill
 7676: the input buffer from the input source. When the input source is
 7677: exhausted, the input source is set back to the previous input source.
 7678: @item
 7679: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
 7680: time the parse area is emptied.
 7681: @item
 7682: If an error occurs, the input source is set back to the user input
 7683: device.
 7684: @end itemize
 7685: 
 7686: You can read about this in more detail in @ref{Input Sources}.
 7687: 
 7688: doc->in
 7689: doc-source
 7690: 
 7691: doc-tib
 7692: doc-#tib
 7693: 
 7694: 
 7695: @menu
 7696: * Input Sources::               
 7697: * Number Conversion::           
 7698: * Interpret/Compile states::    
 7699: * Interpreter Directives::      
 7700: @end menu
 7701: 
 7702: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
 7703: @subsection Input Sources
 7704: @cindex input sources
 7705: @cindex text interpreter - input sources
 7706: 
 7707: By default, the text interpreter processes input from the user input
 7708: device (the keyboard) when Forth starts up. The text interpreter can
 7709: process input from any of these sources:
 7710: 
 7711: @itemize @bullet
 7712: @item
 7713: The user input device -- the keyboard.
 7714: @item
 7715: A file, using the words described in @ref{Forth source files}.
 7716: @item
 7717: A block, using the words described in @ref{Blocks}.
 7718: @item
 7719: A text string, using @code{evaluate}.
 7720: @end itemize
 7721: 
 7722: A program can identify the current input device from the values of
 7723: @code{source-id} and @code{blk}.
 7724: 
 7725: 
 7726: doc-source-id
 7727: doc-blk
 7728: 
 7729: doc-save-input
 7730: doc-restore-input
 7731: 
 7732: doc-evaluate
 7733: 
 7734: 
 7735: 
 7736: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
 7737: @subsection Number Conversion
 7738: @cindex number conversion
 7739: @cindex double-cell numbers, input format
 7740: @cindex input format for double-cell numbers
 7741: @cindex single-cell numbers, input format
 7742: @cindex input format for single-cell numbers
 7743: @cindex floating-point numbers, input format
 7744: @cindex input format for floating-point numbers
 7745: 
 7746: This section describes the rules that the text interpreter uses when it
 7747: tries to convert a string into a number.
 7748: 
 7749: Let <digit> represent any character that is a legal digit in the current
 7750: number base@footnote{For example, 0-9 when the number base is decimal or
 7751: 0-9, A-F when the number base is hexadecimal.}.
 7752: 
 7753: Let <decimal digit> represent any character in the range 0-9.
 7754: 
 7755: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
 7756: in the braces (@i{a} or @i{b} or neither).
 7757: 
 7758: Let * represent any number of instances of the previous character
 7759: (including none).
 7760: 
 7761: Let any other character represent itself.
 7762: 
 7763: @noindent
 7764: Now, the conversion rules are:
 7765: 
 7766: @itemize @bullet
 7767: @item
 7768: A string of the form <digit><digit>* is treated as a single-precision
 7769: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
 7770: @item
 7771: A string of the form -<digit><digit>* is treated as a single-precision
 7772: (cell-sized) negative integer, and is represented using 2's-complement
 7773: arithmetic. Examples are -45 -5681 -0
 7774: @item
 7775: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
 7776: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
 7777: (all three of these represent the same number).
 7778: @item
 7779: A string of the form -<digit><digit>*.<digit>* is treated as a
 7780: double-precision (double-cell-sized) negative integer, and is
 7781: represented using 2's-complement arithmetic. Examples are -3465. -3.465
 7782: -34.65 (all three of these represent the same number).
 7783: @item
 7784: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
 7785: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
 7786: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
 7787: number) +12.E-4
 7788: @end itemize
 7789: 
 7790: By default, the number base used for integer number conversion is given
 7791: by the contents of the variable @code{base}.  Note that a lot of
 7792: confusion can result from unexpected values of @code{base}.  If you
 7793: change @code{base} anywhere, make sure to save the old value and restore
 7794: it afterwards.  In general I recommend keeping @code{base} decimal, and
 7795: using the prefixes described below for the popular non-decimal bases.
 7796: 
 7797: doc-dpl
 7798: doc-base
 7799: doc-hex
 7800: doc-decimal
 7801: 
 7802: 
 7803: @cindex '-prefix for character strings
 7804: @cindex &-prefix for decimal numbers
 7805: @cindex %-prefix for binary numbers
 7806: @cindex $-prefix for hexadecimal numbers
 7807: Gforth allows you to override the value of @code{base} by using a
 7808: prefix@footnote{Some Forth implementations provide a similar scheme by
 7809: implementing @code{$} etc. as parsing words that process the subsequent
 7810: number in the input stream and push it onto the stack. For example, see
 7811: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
 7812: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
 7813: is required between the prefix and the number.} before the first digit
 7814: of an (integer) number. Four prefixes are supported:
 7815: 
 7816: @itemize @bullet
 7817: @item
 7818: @code{&} -- decimal
 7819: @item
 7820: @code{%} -- binary
 7821: @item
 7822: @code{$} -- hexadecimal
 7823: @item
 7824: @code{'} -- base @code{max-char+1}
 7825: @end itemize
 7826: 
 7827: Here are some examples, with the equivalent decimal number shown after
 7828: in braces:
 7829: 
 7830: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
 7831: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
 7832: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
 7833: &905 (905), $abc (2478), $ABC (2478).
 7834: 
 7835: @cindex number conversion - traps for the unwary
 7836: @noindent
 7837: Number conversion has a number of traps for the unwary:
 7838: 
 7839: @itemize @bullet
 7840: @item
 7841: You cannot determine the current number base using the code sequence
 7842: @code{base @@ .} -- the number base is always 10 in the current number
 7843: base. Instead, use something like @code{base @@ dec.}
 7844: @item
 7845: If the number base is set to a value greater than 14 (for example,
 7846: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
 7847: it to be intepreted as either a single-precision integer or a
 7848: floating-point number (Gforth treats it as an integer). The ambiguity
 7849: can be resolved by explicitly stating the sign of the mantissa and/or
 7850: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
 7851: ambiguity arises; either representation will be treated as a
 7852: floating-point number.
 7853: @item
 7854: There is a word @code{bin} but it does @i{not} set the number base!
 7855: It is used to specify file types.
 7856: @item
 7857: ANS Forth requires the @code{.} of a double-precision number to be the
 7858: final character in the string.  Gforth allows the @code{.} to be
 7859: anywhere after the first digit.
 7860: @item
 7861: The number conversion process does not check for overflow.
 7862: @item
 7863: In an ANS Forth program @code{base} is required to be decimal when
 7864: converting floating-point numbers.  In Gforth, number conversion to
 7865: floating-point numbers always uses base &10, irrespective of the value
 7866: of @code{base}.
 7867: @end itemize
 7868: 
 7869: You can read numbers into your programs with the words described in
 7870: @ref{Input}.
 7871: 
 7872: @node Interpret/Compile states, Interpreter Directives, Number Conversion, The Text Interpreter
 7873: @subsection Interpret/Compile states
 7874: @cindex Interpret/Compile states
 7875: 
 7876: A standard program is not permitted to change @code{state}
 7877: explicitly. However, it can change @code{state} implicitly, using the
 7878: words @code{[} and @code{]}. When @code{[} is executed it switches
 7879: @code{state} to interpret state, and therefore the text interpreter
 7880: starts interpreting. When @code{]} is executed it switches @code{state}
 7881: to compile state and therefore the text interpreter starts
 7882: compiling. The most common usage for these words is for switching into
 7883: interpret state and back from within a colon definition; this technique
 7884: can be used to compile a literal (for an example, @pxref{Literals}) or
 7885: for conditional compilation (for an example, @pxref{Interpreter
 7886: Directives}).
 7887: 
 7888: 
 7889: @c This is a bad example: It's non-standard, and it's not necessary.
 7890: @c However, I can't think of a good example for switching into compile
 7891: @c state when there is no current word (@code{state}-smart words are not a
 7892: @c good reason).  So maybe we should use an example for switching into
 7893: @c interpret @code{state} in a colon def. - anton
 7894: @c nac-> I agree. I started out by putting in the example, then realised
 7895: @c that it was non-ANS, so wrote more words around it. I hope this
 7896: @c re-written version is acceptable to you. I do want to keep the example
 7897: @c as it is helpful for showing what is and what is not portable, particularly
 7898: @c where it outlaws a style in common use.
 7899: 
 7900: @c anton: it's more important to show what's portable.  After we have done
 7901: @c that, we can also show what's not.  In any case, I have written a
 7902: @c section Compiling Words which also deals with [ ].
 7903: 
 7904: @code{[} and @code{]} also give you the ability to switch into compile
 7905: state and back, but we cannot think of any useful Standard application
 7906: for this ability. Pre-ANS Forth textbooks have examples like this:
 7907: 
 7908: @example
 7909: : AA ." this is A" ;
 7910: : BB ." this is B" ;
 7911: : CC ." this is C" ;
 7912: 
 7913: create table ] aa bb cc [
 7914: 
 7915: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
 7916:   cells table + @ execute ;
 7917: @end example
 7918: 
 7919: This example builds a jump table; @code{0 go} will display ``@code{this
 7920: is A}''. Using @code{[} and @code{]} in this example is equivalent to
 7921: defining @code{table} like this:
 7922: 
 7923: @example
 7924: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
 7925: @end example
 7926: 
 7927: The problem with this code is that the definition of @code{table} is not
 7928: portable -- it @i{compile}s execution tokens into code space. Whilst it
 7929: @i{may} work on systems where code space and data space co-incide, the
 7930: Standard only allows data space to be assigned for a @code{CREATE}d
 7931: word. In addition, the Standard only allows @code{@@} to access data
 7932: space, whilst this example is using it to access code space. The only
 7933: portable, Standard way to build this table is to build it in data space,
 7934: like this:
 7935: 
 7936: @example
 7937: create table ' aa , ' bb , ' cc ,
 7938: @end example
 7939: 
 7940: doc-state
 7941: 
 7942: 
 7943: @node Interpreter Directives,  , Interpret/Compile states, The Text Interpreter
 7944: @subsection Interpreter Directives
 7945: @cindex interpreter directives
 7946: @cindex conditional compilation
 7947: 
 7948: These words are usually used in interpret state; typically to control
 7949: which parts of a source file are processed by the text
 7950: interpreter. There are only a few ANS Forth Standard words, but Gforth
 7951: supplements these with a rich set of immediate control structure words
 7952: to compensate for the fact that the non-immediate versions can only be
 7953: used in compile state (@pxref{Control Structures}). Typical usages:
 7954: 
 7955: @example
 7956: FALSE Constant HAVE-ASSEMBLER
 7957: .
 7958: .
 7959: HAVE-ASSEMBLER [IF]
 7960: : ASSEMBLER-FEATURE
 7961:   ...
 7962: ;
 7963: [ENDIF]
 7964: .
 7965: .
 7966: : SEE
 7967:   ... \ general-purpose SEE code
 7968:   [ HAVE-ASSEMBLER [IF] ]
 7969:   ... \ assembler-specific SEE code
 7970:   [ [ENDIF] ]
 7971: ;
 7972: @end example
 7973: 
 7974: 
 7975: doc-[IF]
 7976: doc-[ELSE]
 7977: doc-[THEN]
 7978: doc-[ENDIF]
 7979: 
 7980: doc-[IFDEF]
 7981: doc-[IFUNDEF]
 7982: 
 7983: doc-[?DO]
 7984: doc-[DO]
 7985: doc-[FOR]
 7986: doc-[LOOP]
 7987: doc-[+LOOP]
 7988: doc-[NEXT]
 7989: 
 7990: doc-[BEGIN]
 7991: doc-[UNTIL]
 7992: doc-[AGAIN]
 7993: doc-[WHILE]
 7994: doc-[REPEAT]
 7995: 
 7996: 
 7997: @c -------------------------------------------------------------
 7998: @node Word Lists, Environmental Queries, The Text Interpreter, Words
 7999: @section Word Lists
 8000: @cindex word lists
 8001: @cindex header space
 8002: 
 8003: A wordlist is a list of named words; you can add new words and look up
 8004: words by name (and you can remove words in a restricted way with
 8005: markers).  Every named (and @code{reveal}ed) word is in one wordlist.
 8006: 
 8007: @cindex search order stack
 8008: The text interpreter searches the wordlists present in the search order
 8009: (a stack of wordlists), from the top to the bottom.  Within each
 8010: wordlist, the search starts conceptually at the newest word; i.e., if
 8011: two words in a wordlist have the same name, the newer word is found.
 8012: 
 8013: @cindex compilation word list
 8014: New words are added to the @dfn{compilation wordlist} (aka current
 8015: wordlist).
 8016: 
 8017: @cindex wid
 8018: A word list is identified by a cell-sized word list identifier (@i{wid})
 8019: in much the same way as a file is identified by a file handle. The
 8020: numerical value of the wid has no (portable) meaning, and might change
 8021: from session to session.
 8022: 
 8023: The ANS Forth ``Search order'' word set is intended to provide a set of
 8024: low-level tools that allow various different schemes to be
 8025: implemented. Gforth also provides @code{vocabulary}, a traditional Forth
 8026: word.  @file{compat/vocabulary.fs} provides an implementation in ANS
 8027: Forth.
 8028: 
 8029: @comment TODO: locals section refers to here, saying that every word list (aka
 8030: @comment vocabulary) has its own methods for searching etc. Need to document that.
 8031: @c anton: but better in a separate subsection on wordlist internals
 8032: 
 8033: @comment TODO: document markers, reveal, tables, mappedwordlist
 8034: 
 8035: @comment the gforthman- prefix is used to pick out the true definition of a
 8036: @comment word from the source files, rather than some alias.
 8037: 
 8038: doc-forth-wordlist
 8039: doc-definitions
 8040: doc-get-current
 8041: doc-set-current
 8042: doc-get-order
 8043: doc---gforthman-set-order
 8044: doc-wordlist
 8045: doc-table
 8046: doc->order
 8047: doc-previous
 8048: doc-also
 8049: doc---gforthman-forth
 8050: doc-only
 8051: doc---gforthman-order
 8052: 
 8053: doc-find
 8054: doc-search-wordlist
 8055: 
 8056: doc-words
 8057: doc-vlist
 8058: @c doc-words-deferred
 8059: 
 8060: @c doc-mappedwordlist @c map-structure undefined, implemantation-specific
 8061: doc-root
 8062: doc-vocabulary
 8063: doc-seal
 8064: doc-vocs
 8065: doc-current
 8066: doc-context
 8067: 
 8068: 
 8069: @menu
 8070: * Vocabularies::                
 8071: * Why use word lists?::         
 8072: * Word list example::           
 8073: @end menu
 8074: 
 8075: @node Vocabularies, Why use word lists?, Word Lists, Word Lists
 8076: @subsection Vocabularies
 8077: @cindex Vocabularies, detailed explanation
 8078: 
 8079: Here is an example of creating and using a new wordlist using ANS
 8080: Forth words:
 8081: 
 8082: @example
 8083: wordlist constant my-new-words-wordlist
 8084: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
 8085: 
 8086: \ add it to the search order
 8087: also my-new-words
 8088: 
 8089: \ alternatively, add it to the search order and make it
 8090: \ the compilation word list
 8091: also my-new-words definitions
 8092: \ type "order" to see the problem
 8093: @end example
 8094: 
 8095: The problem with this example is that @code{order} has no way to
 8096: associate the name @code{my-new-words} with the wid of the word list (in
 8097: Gforth, @code{order} and @code{vocs} will display @code{???}  for a wid
 8098: that has no associated name). There is no Standard way of associating a
 8099: name with a wid.
 8100: 
 8101: In Gforth, this example can be re-coded using @code{vocabulary}, which
 8102: associates a name with a wid:
 8103: 
 8104: @example
 8105: vocabulary my-new-words
 8106: 
 8107: \ add it to the search order
 8108: also my-new-words
 8109: 
 8110: \ alternatively, add it to the search order and make it
 8111: \ the compilation word list
 8112: my-new-words definitions
 8113: \ type "order" to see that the problem is solved
 8114: @end example
 8115: 
 8116: 
 8117: @node Why use word lists?, Word list example, Vocabularies, Word Lists
 8118: @subsection Why use word lists?
 8119: @cindex word lists - why use them?
 8120: 
 8121: Here are some reasons why people use wordlists:
 8122: 
 8123: @itemize @bullet
 8124: 
 8125: @c anton: Gforth's hashing implementation makes the search speed
 8126: @c independent from the number of words.  But it is linear with the number
 8127: @c of wordlists that have to be searched, so in effect using more wordlists
 8128: @c actually slows down compilation.
 8129: 
 8130: @c @item
 8131: @c To improve compilation speed by reducing the number of header space
 8132: @c entries that must be searched. This is achieved by creating a new
 8133: @c word list that contains all of the definitions that are used in the
 8134: @c definition of a Forth system but which would not usually be used by
 8135: @c programs running on that system. That word list would be on the search
 8136: @c list when the Forth system was compiled but would be removed from the
 8137: @c search list for normal operation. This can be a useful technique for
 8138: @c low-performance systems (for example, 8-bit processors in embedded
 8139: @c systems) but is unlikely to be necessary in high-performance desktop
 8140: @c systems.
 8141: 
 8142: @item
 8143: To prevent a set of words from being used outside the context in which
 8144: they are valid. Two classic examples of this are an integrated editor
 8145: (all of the edit commands are defined in a separate word list; the
 8146: search order is set to the editor word list when the editor is invoked;
 8147: the old search order is restored when the editor is terminated) and an
 8148: integrated assembler (the op-codes for the machine are defined in a
 8149: separate word list which is used when a @code{CODE} word is defined).
 8150: 
 8151: @item
 8152: To organize the words of an application or library into a user-visible
 8153: set (in @code{forth-wordlist} or some other common wordlist) and a set
 8154: of helper words used just for the implementation (hidden in a separate
 8155: wordlist).  This keeps @code{words}' output smaller, separates
 8156: implementation and interface, and reduces the chance of name conflicts
 8157: within the common wordlist.
 8158: 
 8159: @item
 8160: To prevent a name-space clash between multiple definitions with the same
 8161: name. For example, when building a cross-compiler you might have a word
 8162: @code{IF} that generates conditional code for your target system. By
 8163: placing this definition in a different word list you can control whether
 8164: the host system's @code{IF} or the target system's @code{IF} get used in
 8165: any particular context by controlling the order of the word lists on the
 8166: search order stack.
 8167: 
 8168: @end itemize
 8169: 
 8170: The downsides of using wordlists are:
 8171: 
 8172: @itemize
 8173: 
 8174: @item
 8175: Debugging becomes more cumbersome.
 8176: 
 8177: @item
 8178: Name conflicts worked around with wordlists are still there, and you
 8179: have to arrange the search order carefully to get the desired results;
 8180: if you forget to do that, you get hard-to-find errors (as in any case
 8181: where you read the code differently from the compiler; @code{see} can
 8182: help seeing which of several possible words the name resolves to in such
 8183: cases).  @code{See} displays just the name of the words, not what
 8184: wordlist they belong to, so it might be misleading.  Using unique names
 8185: is a better approach to avoid name conflicts.
 8186: 
 8187: @item
 8188: You have to explicitly undo any changes to the search order.  In many
 8189: cases it would be more convenient if this happened implicitly.  Gforth
 8190: currently does not provide such a feature, but it may do so in the
 8191: future.
 8192: @end itemize
 8193: 
 8194: 
 8195: @node Word list example,  , Why use word lists?, Word Lists
 8196: @subsection Word list example
 8197: @cindex word lists - example
 8198: 
 8199: The following example is from the
 8200: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
 8201: garbage collector} and uses wordlists to separate public words from
 8202: helper words:
 8203: 
 8204: @example
 8205: get-current ( wid )
 8206: vocabulary garbage-collector also garbage-collector definitions
 8207: ... \ define helper words
 8208: ( wid ) set-current \ restore original (i.e., public) compilation wordlist
 8209: ... \ define the public (i.e., API) words
 8210:     \ they can refer to the helper words
 8211: previous \ restore original search order (helper words become invisible)
 8212: @end example
 8213: 
 8214: @c -------------------------------------------------------------
 8215: @node Environmental Queries, Files, Word Lists, Words
 8216: @section Environmental Queries
 8217: @cindex environmental queries
 8218: 
 8219: ANS Forth introduced the idea of ``environmental queries'' as a way
 8220: for a program running on a system to determine certain characteristics of the system.
 8221: The Standard specifies a number of strings that might be recognised by a system.
 8222: 
 8223: The Standard requires that the header space used for environmental queries
 8224: be distinct from the header space used for definitions.
 8225: 
 8226: Typically, environmental queries are supported by creating a set of
 8227: definitions in a word list that is @i{only} used during environmental
 8228: queries; that is what Gforth does. There is no Standard way of adding
 8229: definitions to the set of recognised environmental queries, but any
 8230: implementation that supports the loading of optional word sets must have
 8231: some mechanism for doing this (after loading the word set, the
 8232: associated environmental query string must return @code{true}). In
 8233: Gforth, the word list used to honour environmental queries can be
 8234: manipulated just like any other word list.
 8235: 
 8236: 
 8237: doc-environment?
 8238: doc-environment-wordlist
 8239: 
 8240: doc-gforth
 8241: doc-os-class
 8242: 
 8243: 
 8244: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
 8245: returning two items on the stack, querying it using @code{environment?}
 8246: will return an additional item; the @code{true} flag that shows that the
 8247: string was recognised.
 8248: 
 8249: @comment TODO Document the standard strings or note where they are documented herein
 8250: 
 8251: Here are some examples of using environmental queries:
 8252: 
 8253: @example
 8254: s" address-unit-bits" environment? 0=
 8255: [IF]
 8256:      cr .( environmental attribute address-units-bits unknown... ) cr
 8257: [ELSE]
 8258:      drop \ ensure balanced stack effect
 8259: [THEN]
 8260: 
 8261: \ this might occur in the prelude of a standard program that uses THROW
 8262: s" exception" environment? [IF]
 8263:    0= [IF]
 8264:       : throw abort" exception thrown" ;
 8265:    [THEN]
 8266: [ELSE] \ we don't know, so make sure
 8267:    : throw abort" exception thrown" ;
 8268: [THEN]
 8269: 
 8270: s" gforth" environment? [IF] .( Gforth version ) TYPE
 8271:                         [ELSE] .( Not Gforth..) [THEN]
 8272: 
 8273: \ a program using v*
 8274: s" gforth" environment? [IF]
 8275:   s" 0.5.0" compare 0< [IF] \ v* is a primitive since 0.5.0
 8276:    : v* ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r )
 8277:      >r swap 2swap swap 0e r> 0 ?DO
 8278:        dup f@ over + 2swap dup f@ f* f+ over + 2swap
 8279:      LOOP
 8280:      2drop 2drop ; 
 8281:   [THEN]
 8282: [ELSE] \ 
 8283:   : v* ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r )
 8284:   ...
 8285: [THEN]
 8286: @end example
 8287: 
 8288: Here is an example of adding a definition to the environment word list:
 8289: 
 8290: @example
 8291: get-current environment-wordlist set-current
 8292: true constant block
 8293: true constant block-ext
 8294: set-current
 8295: @end example
 8296: 
 8297: You can see what definitions are in the environment word list like this:
 8298: 
 8299: @example
 8300: environment-wordlist >order words previous
 8301: @end example
 8302: 
 8303: 
 8304: @c -------------------------------------------------------------
 8305: @node Files, Blocks, Environmental Queries, Words
 8306: @section Files
 8307: @cindex files
 8308: @cindex I/O - file-handling
 8309: 
 8310: Gforth provides facilities for accessing files that are stored in the
 8311: host operating system's file-system. Files that are processed by Gforth
 8312: can be divided into two categories:
 8313: 
 8314: @itemize @bullet
 8315: @item
 8316: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
 8317: @item
 8318: Files that are processed by some other program (@dfn{general files}).
 8319: @end itemize
 8320: 
 8321: @menu
 8322: * Forth source files::          
 8323: * General files::               
 8324: * Search Paths::                
 8325: @end menu
 8326: 
 8327: @c -------------------------------------------------------------
 8328: @node Forth source files, General files, Files, Files
 8329: @subsection Forth source files
 8330: @cindex including files
 8331: @cindex Forth source files
 8332: 
 8333: The simplest way to interpret the contents of a file is to use one of
 8334: these two formats:
 8335: 
 8336: @example
 8337: include mysource.fs
 8338: s" mysource.fs" included
 8339: @end example
 8340: 
 8341: You usually want to include a file only if it is not included already
 8342: (by, say, another source file). In that case, you can use one of these
 8343: three formats:
 8344: 
 8345: @example
 8346: require mysource.fs
 8347: needs mysource.fs
 8348: s" mysource.fs" required
 8349: @end example
 8350: 
 8351: @cindex stack effect of included files
 8352: @cindex including files, stack effect
 8353: It is good practice to write your source files such that interpreting them
 8354: does not change the stack. Source files designed in this way can be used with
 8355: @code{required} and friends without complications. For example:
 8356: 
 8357: @example
 8358: 1024 require foo.fs drop
 8359: @end example
 8360: 
 8361: Here you want to pass the argument 1024 (e.g., a buffer size) to
 8362: @file{foo.fs}.  Interpreting @file{foo.fs} has the stack effect ( n -- n
 8363: ), which allows its use with @code{require}.  Of course with such
 8364: parameters to required files, you have to ensure that the first
 8365: @code{require} fits for all uses (i.e., @code{require} it early in the
 8366: master load file).
 8367: 
 8368: doc-include-file
 8369: doc-included
 8370: doc-included?
 8371: doc-include
 8372: doc-required
 8373: doc-require
 8374: doc-needs
 8375: @c doc-init-included-files @c internal
 8376: doc-sourcefilename
 8377: doc-sourceline#
 8378: 
 8379: A definition in ANS Forth for @code{required} is provided in
 8380: @file{compat/required.fs}.
 8381: 
 8382: @c -------------------------------------------------------------
 8383: @node General files, Search Paths, Forth source files, Files
 8384: @subsection General files
 8385: @cindex general files
 8386: @cindex file-handling
 8387: 
 8388: Files are opened/created by name and type. The following file access
 8389: methods (FAMs) are recognised:
 8390: 
 8391: @cindex fam (file access method)
 8392: doc-r/o
 8393: doc-r/w
 8394: doc-w/o
 8395: doc-bin
 8396: 
 8397: 
 8398: When a file is opened/created, it returns a file identifier,
 8399: @i{wfileid} that is used for all other file commands. All file
 8400: commands also return a status value, @i{wior}, that is 0 for a
 8401: successful operation and an implementation-defined non-zero value in the
 8402: case of an error.
 8403: 
 8404: 
 8405: doc-open-file
 8406: doc-create-file
 8407: 
 8408: doc-close-file
 8409: doc-delete-file
 8410: doc-rename-file
 8411: doc-read-file
 8412: doc-read-line
 8413: doc-write-file
 8414: doc-write-line
 8415: doc-emit-file
 8416: doc-flush-file
 8417: 
 8418: doc-file-status
 8419: doc-file-position
 8420: doc-reposition-file
 8421: doc-file-size
 8422: doc-resize-file
 8423: 
 8424: 
 8425: @c ---------------------------------------------------------
 8426: @node Search Paths,  , General files, Files
 8427: @subsection Search Paths
 8428: @cindex path for @code{included}
 8429: @cindex file search path
 8430: @cindex @code{include} search path
 8431: @cindex search path for files
 8432: 
 8433: If you specify an absolute filename (i.e., a filename starting with
 8434: @file{/} or @file{~}, or with @file{:} in the second position (as in
 8435: @samp{C:...})) for @code{included} and friends, that file is included
 8436: just as you would expect.
 8437: 
 8438: If the filename starts with @file{./}, this refers to the directory that
 8439: the present file was @code{included} from.  This allows files to include
 8440: other files relative to their own position (irrespective of the current
 8441: working directory or the absolute position).  This feature is essential
 8442: for libraries consisting of several files, where a file may include
 8443: other files from the library.  It corresponds to @code{#include "..."}
 8444: in C. If the current input source is not a file, @file{.} refers to the
 8445: directory of the innermost file being included, or, if there is no file
 8446: being included, to the current working directory.
 8447: 
 8448: For relative filenames (not starting with @file{./}), Gforth uses a
 8449: search path similar to Forth's search order (@pxref{Word Lists}). It
 8450: tries to find the given filename in the directories present in the path,
 8451: and includes the first one it finds. There are separate search paths for
 8452: Forth source files and general files.  If the search path contains the
 8453: directory @file{.}, this refers to the directory of the current file, or
 8454: the working directory, as if the file had been specified with @file{./}.
 8455: 
 8456: Use @file{~+} to refer to the current working directory (as in the
 8457: @code{bash}).
 8458: 
 8459: @c anton: fold the following subsubsections into this subsection?
 8460: 
 8461: @menu
 8462: * Source Search Paths::         
 8463: * General Search Paths::        
 8464: @end menu
 8465: 
 8466: @c ---------------------------------------------------------
 8467: @node Source Search Paths, General Search Paths, Search Paths, Search Paths
 8468: @subsubsection Source Search Paths
 8469: @cindex search path control, source files
 8470: 
 8471: The search path is initialized when you start Gforth (@pxref{Invoking
 8472: Gforth}). You can display it and change it using @code{fpath} in
 8473: combination with the general path handling words.
 8474: 
 8475: doc-fpath
 8476: @c the functionality of the following words is easily available through
 8477: @c   fpath and the general path words.  The may go away.
 8478: @c doc-.fpath
 8479: @c doc-fpath+
 8480: @c doc-fpath=
 8481: @c doc-open-fpath-file
 8482: 
 8483: @noindent
 8484: Here is an example of using @code{fpath} and @code{require}:
 8485: 
 8486: @example
 8487: fpath path= /usr/lib/forth/|./
 8488: require timer.fs
 8489: @end example
 8490: 
 8491: 
 8492: @c ---------------------------------------------------------
 8493: @node General Search Paths,  , Source Search Paths, Search Paths
 8494: @subsubsection General Search Paths
 8495: @cindex search path control, source files
 8496: 
 8497: Your application may need to search files in several directories, like
 8498: @code{included} does. To facilitate this, Gforth allows you to define
 8499: and use your own search paths, by providing generic equivalents of the
 8500: Forth search path words:
 8501: 
 8502: doc-open-path-file
 8503: doc-path-allot
 8504: doc-clear-path
 8505: doc-also-path
 8506: doc-.path
 8507: doc-path+
 8508: doc-path=
 8509: 
 8510: @c anton: better define a word for it, say "path-allot ( ucount -- path-addr )
 8511: 
 8512: Here's an example of creating an empty search path:
 8513: @c
 8514: @example
 8515: create mypath 500 path-allot \ maximum length 500 chars (is checked)
 8516: @end example
 8517: 
 8518: @c -------------------------------------------------------------
 8519: @node Blocks, Other I/O, Files, Words
 8520: @section Blocks
 8521: @cindex I/O - blocks
 8522: @cindex blocks
 8523: 
 8524: When you run Gforth on a modern desk-top computer, it runs under the
 8525: control of an operating system which provides certain services.  One of
 8526: these services is @var{file services}, which allows Forth source code
 8527: and data to be stored in files and read into Gforth (@pxref{Files}).
 8528: 
 8529: Traditionally, Forth has been an important programming language on
 8530: systems where it has interfaced directly to the underlying hardware with
 8531: no intervening operating system. Forth provides a mechanism, called
 8532: @dfn{blocks}, for accessing mass storage on such systems.
 8533: 
 8534: A block is a 1024-byte data area, which can be used to hold data or
 8535: Forth source code. No structure is imposed on the contents of the
 8536: block. A block is identified by its number; blocks are numbered
 8537: contiguously from 1 to an implementation-defined maximum.
 8538: 
 8539: A typical system that used blocks but no operating system might use a
 8540: single floppy-disk drive for mass storage, with the disks formatted to
 8541: provide 256-byte sectors. Blocks would be implemented by assigning the
 8542: first four sectors of the disk to block 1, the second four sectors to
 8543: block 2 and so on, up to the limit of the capacity of the disk. The disk
 8544: would not contain any file system information, just the set of blocks.
 8545: 
 8546: @cindex blocks file
 8547: On systems that do provide file services, blocks are typically
 8548: implemented by storing a sequence of blocks within a single @dfn{blocks
 8549: file}.  The size of the blocks file will be an exact multiple of 1024
 8550: bytes, corresponding to the number of blocks it contains. This is the
 8551: mechanism that Gforth uses.
 8552: 
 8553: @cindex @file{blocks.fb}
 8554: Only one blocks file can be open at a time. If you use block words without
 8555: having specified a blocks file, Gforth defaults to the blocks file
 8556: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
 8557: locate a blocks file (@pxref{Source Search Paths}).
 8558: 
 8559: @cindex block buffers
 8560: When you read and write blocks under program control, Gforth uses a
 8561: number of @dfn{block buffers} as intermediate storage. These buffers are
 8562: not used when you use @code{load} to interpret the contents of a block.
 8563: 
 8564: The behaviour of the block buffers is analagous to that of a cache.
 8565: Each block buffer has three states:
 8566: 
 8567: @itemize @bullet
 8568: @item
 8569: Unassigned
 8570: @item
 8571: Assigned-clean
 8572: @item
 8573: Assigned-dirty
 8574: @end itemize
 8575: 
 8576: Initially, all block buffers are @i{unassigned}. In order to access a
 8577: block, the block (specified by its block number) must be assigned to a
 8578: block buffer.
 8579: 
 8580: The assignment of a block to a block buffer is performed by @code{block}
 8581: or @code{buffer}. Use @code{block} when you wish to modify the existing
 8582: contents of a block. Use @code{buffer} when you don't care about the
 8583: existing contents of the block@footnote{The ANS Forth definition of
 8584: @code{buffer} is intended not to cause disk I/O; if the data associated
 8585: with the particular block is already stored in a block buffer due to an
 8586: earlier @code{block} command, @code{buffer} will return that block
 8587: buffer and the existing contents of the block will be
 8588: available. Otherwise, @code{buffer} will simply assign a new, empty
 8589: block buffer for the block.}.
 8590: 
 8591: Once a block has been assigned to a block buffer using @code{block} or
 8592: @code{buffer}, that block buffer becomes the @i{current block
 8593: buffer}. Data may only be manipulated (read or written) within the
 8594: current block buffer.
 8595: 
 8596: When the contents of the current block buffer has been modified it is
 8597: necessary, @emph{before calling @code{block} or @code{buffer} again}, to
 8598: either abandon the changes (by doing nothing) or mark the block as
 8599: changed (assigned-dirty), using @code{update}. Using @code{update} does
 8600: not change the blocks file; it simply changes a block buffer's state to
 8601: @i{assigned-dirty}.  The block will be written implicitly when it's
 8602: buffer is needed for another block, or explicitly by @code{flush} or
 8603: @code{save-buffers}.
 8604: 
 8605: word @code{Flush} writes all @i{assigned-dirty} blocks back to the
 8606: blocks file on disk. Leaving Gforth with @code{bye} also performs a
 8607: @code{flush}.
 8608: 
 8609: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
 8610: algorithm to assign a block buffer to a block. That means that any
 8611: particular block can only be assigned to one specific block buffer,
 8612: called (for the particular operation) the @i{victim buffer}. If the
 8613: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
 8614: the new block immediately. If it is @i{assigned-dirty} its current
 8615: contents are written back to the blocks file on disk before it is
 8616: allocated to the new block.
 8617: 
 8618: Although no structure is imposed on the contents of a block, it is
 8619: traditional to display the contents as 16 lines each of 64 characters.  A
 8620: block provides a single, continuous stream of input (for example, it
 8621: acts as a single parse area) -- there are no end-of-line characters
 8622: within a block, and no end-of-file character at the end of a
 8623: block. There are two consequences of this:
 8624: 
 8625: @itemize @bullet
 8626: @item
 8627: The last character of one line wraps straight into the first character
 8628: of the following line
 8629: @item
 8630: The word @code{\} -- comment to end of line -- requires special
 8631: treatment; in the context of a block it causes all characters until the
 8632: end of the current 64-character ``line'' to be ignored.
 8633: @end itemize
 8634: 
 8635: In Gforth, when you use @code{block} with a non-existent block number,
 8636: the current blocks file will be extended to the appropriate size and the
 8637: block buffer will be initialised with spaces.
 8638: 
 8639: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
 8640: for details) but doesn't encourage the use of blocks; the mechanism is
 8641: only provided for backward compatibility -- ANS Forth requires blocks to
 8642: be available when files are.
 8643: 
 8644: Common techniques that are used when working with blocks include:
 8645: 
 8646: @itemize @bullet
 8647: @item
 8648: A screen editor that allows you to edit blocks without leaving the Forth
 8649: environment.
 8650: @item
 8651: Shadow screens; where every code block has an associated block
 8652: containing comments (for example: code in odd block numbers, comments in
 8653: even block numbers). Typically, the block editor provides a convenient
 8654: mechanism to toggle between code and comments.
 8655: @item
 8656: Load blocks; a single block (typically block 1) contains a number of
 8657: @code{thru} commands which @code{load} the whole of the application.
 8658: @end itemize
 8659: 
 8660: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
 8661: integrated into a Forth programming environment.
 8662: 
 8663: @comment TODO what about errors on open-blocks?
 8664: 
 8665: doc-open-blocks
 8666: doc-use
 8667: doc-block-offset
 8668: doc-get-block-fid
 8669: doc-block-position
 8670: 
 8671: doc-list
 8672: doc-scr
 8673: 
 8674: doc---gforthman-block
 8675: doc-buffer
 8676: 
 8677: doc-empty-buffers
 8678: doc-empty-buffer
 8679: doc-update
 8680: doc-updated?
 8681: doc-save-buffers
 8682: doc-save-buffer
 8683: doc-flush
 8684: 
 8685: doc-load
 8686: doc-thru
 8687: doc-+load
 8688: doc-+thru
 8689: doc---gforthman--->
 8690: doc-block-included
 8691: 
 8692: 
 8693: @c -------------------------------------------------------------
 8694: @node Other I/O, Locals, Blocks, Words
 8695: @section Other I/O
 8696: @cindex I/O - keyboard and display
 8697: 
 8698: @menu
 8699: * Simple numeric output::       Predefined formats
 8700: * Formatted numeric output::    Formatted (pictured) output
 8701: * String Formats::              How Forth stores strings in memory
 8702: * Displaying characters and strings::  Other stuff
 8703: * Input::                       Input
 8704: @end menu
 8705: 
 8706: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
 8707: @subsection Simple numeric output
 8708: @cindex numeric output - simple/free-format
 8709: 
 8710: The simplest output functions are those that display numbers from the
 8711: data or floating-point stacks. Floating-point output is always displayed
 8712: using base 10. Numbers displayed from the data stack use the value stored
 8713: in @code{base}.
 8714: 
 8715: 
 8716: doc-.
 8717: doc-dec.
 8718: doc-hex.
 8719: doc-u.
 8720: doc-.r
 8721: doc-u.r
 8722: doc-d.
 8723: doc-ud.
 8724: doc-d.r
 8725: doc-ud.r
 8726: doc-f.
 8727: doc-fe.
 8728: doc-fs.
 8729: 
 8730: 
 8731: Examples of printing the number 1234.5678E23 in the different floating-point output
 8732: formats are shown below:
 8733: 
 8734: @example
 8735: f. 123456779999999000000000000.
 8736: fe. 123.456779999999E24
 8737: fs. 1.23456779999999E26
 8738: @end example
 8739: 
 8740: 
 8741: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
 8742: @subsection Formatted numeric output
 8743: @cindex formatted numeric output
 8744: @cindex pictured numeric output
 8745: @cindex numeric output - formatted
 8746: 
 8747: Forth traditionally uses a technique called @dfn{pictured numeric
 8748: output} for formatted printing of integers.  In this technique, digits
 8749: are extracted from the number (using the current output radix defined by
 8750: @code{base}), converted to ASCII codes and appended to a string that is
 8751: built in a scratch-pad area of memory (@pxref{core-idef,
 8752: Implementation-defined options, Implementation-defined
 8753: options}). Arbitrary characters can be appended to the string during the
 8754: extraction process. The completed string is specified by an address
 8755: and length and can be manipulated (@code{TYPE}ed, copied, modified)
 8756: under program control.
 8757: 
 8758: All of the integer output words described in the previous section
 8759: (@pxref{Simple numeric output}) are implemented in Gforth using pictured
 8760: numeric output.
 8761: 
 8762: Three important things to remember about pictured numeric output:
 8763: 
 8764: @itemize @bullet
 8765: @item
 8766: It always operates on double-precision numbers; to display a
 8767: single-precision number, convert it first (for ways of doing this
 8768: @pxref{Double precision}).
 8769: @item
 8770: It always treats the double-precision number as though it were
 8771: unsigned. The examples below show ways of printing signed numbers.
 8772: @item
 8773: The string is built up from right to left; least significant digit first.
 8774: @end itemize
 8775: 
 8776: 
 8777: doc-<#
 8778: doc-<<#
 8779: doc-#
 8780: doc-#s
 8781: doc-hold
 8782: doc-sign
 8783: doc-#>
 8784: doc-#>>
 8785: 
 8786: doc-represent
 8787: 
 8788: 
 8789: @noindent
 8790: Here are some examples of using pictured numeric output:
 8791: 
 8792: @example
 8793: : my-u. ( u -- )
 8794:   \ Simplest use of pns.. behaves like Standard u. 
 8795:   0              \ convert to unsigned double
 8796:   <<#            \ start conversion
 8797:   #s             \ convert all digits
 8798:   #>             \ complete conversion
 8799:   TYPE SPACE     \ display, with trailing space
 8800:   #>> ;          \ release hold area
 8801: 
 8802: : cents-only ( u -- )
 8803:   0              \ convert to unsigned double
 8804:   <<#            \ start conversion
 8805:   # #            \ convert two least-significant digits
 8806:   #>             \ complete conversion, discard other digits
 8807:   TYPE SPACE     \ display, with trailing space
 8808:   #>> ;          \ release hold area
 8809: 
 8810: : dollars-and-cents ( u -- )
 8811:   0              \ convert to unsigned double
 8812:   <<#            \ start conversion
 8813:   # #            \ convert two least-significant digits
 8814:   [char] . hold  \ insert decimal point
 8815:   #s             \ convert remaining digits
 8816:   [char] $ hold  \ append currency symbol
 8817:   #>             \ complete conversion
 8818:   TYPE SPACE     \ display, with trailing space
 8819:   #>> ;          \ release hold area
 8820: 
 8821: : my-. ( n -- )
 8822:   \ handling negatives.. behaves like Standard .
 8823:   s>d            \ convert to signed double
 8824:   swap over dabs \ leave sign byte followed by unsigned double
 8825:   <<#            \ start conversion
 8826:   #s             \ convert all digits
 8827:   rot sign       \ get at sign byte, append "-" if needed
 8828:   #>             \ complete conversion
 8829:   TYPE SPACE     \ display, with trailing space
 8830:   #>> ;          \ release hold area
 8831: 
 8832: : account. ( n -- )
 8833:   \ accountants don't like minus signs, they use parentheses
 8834:   \ for negative numbers
 8835:   s>d            \ convert to signed double
 8836:   swap over dabs \ leave sign byte followed by unsigned double
 8837:   <<#            \ start conversion
 8838:   2 pick         \ get copy of sign byte
 8839:   0< IF [char] ) hold THEN \ right-most character of output
 8840:   #s             \ convert all digits
 8841:   rot            \ get at sign byte
 8842:   0< IF [char] ( hold THEN
 8843:   #>             \ complete conversion
 8844:   TYPE SPACE     \ display, with trailing space
 8845:   #>> ;          \ release hold area
 8846: 
 8847: @end example
 8848: 
 8849: Here are some examples of using these words:
 8850: 
 8851: @example
 8852: 1 my-u. 1
 8853: hex -1 my-u. decimal FFFFFFFF
 8854: 1 cents-only 01
 8855: 1234 cents-only 34
 8856: 2 dollars-and-cents $0.02
 8857: 1234 dollars-and-cents $12.34
 8858: 123 my-. 123
 8859: -123 my. -123
 8860: 123 account. 123
 8861: -456 account. (456)
 8862: @end example
 8863: 
 8864: 
 8865: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
 8866: @subsection String Formats
 8867: @cindex strings - see character strings
 8868: @cindex character strings - formats
 8869: @cindex I/O - see character strings
 8870: @cindex counted strings
 8871: 
 8872: @c anton: this does not really belong here; maybe the memory section,
 8873: @c  or the principles chapter
 8874: 
 8875: Forth commonly uses two different methods for representing character
 8876: strings:
 8877: 
 8878: @itemize @bullet
 8879: @item
 8880: @cindex address of counted string
 8881: @cindex counted string
 8882: As a @dfn{counted string}, represented by a @i{c-addr}. The char
 8883: addressed by @i{c-addr} contains a character-count, @i{n}, of the
 8884: string and the string occupies the subsequent @i{n} char addresses in
 8885: memory.
 8886: @item
 8887: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
 8888: of the string in characters, and @i{c-addr} is the address of the
 8889: first byte of the string.
 8890: @end itemize
 8891: 
 8892: ANS Forth encourages the use of the second format when representing
 8893: strings.
 8894: 
 8895: 
 8896: doc-count
 8897: 
 8898: 
 8899: For words that move, copy and search for strings see @ref{Memory
 8900: Blocks}. For words that display characters and strings see
 8901: @ref{Displaying characters and strings}.
 8902: 
 8903: @node Displaying characters and strings, Input, String Formats, Other I/O
 8904: @subsection Displaying characters and strings
 8905: @cindex characters - compiling and displaying
 8906: @cindex character strings - compiling and displaying
 8907: 
 8908: This section starts with a glossary of Forth words and ends with a set
 8909: of examples.
 8910: 
 8911: 
 8912: doc-bl
 8913: doc-space
 8914: doc-spaces
 8915: doc-emit
 8916: doc-toupper
 8917: doc-."
 8918: doc-.(
 8919: doc-type
 8920: doc-typewhite
 8921: doc-cr
 8922: @cindex cursor control
 8923: doc-at-xy
 8924: doc-page
 8925: doc-s"
 8926: doc-c"
 8927: doc-char
 8928: doc-[char]
 8929: 
 8930: 
 8931: @noindent
 8932: As an example, consider the following text, stored in a file @file{test.fs}:
 8933: 
 8934: @example
 8935: .( text-1)
 8936: : my-word
 8937:   ." text-2" cr
 8938:   .( text-3)
 8939: ;
 8940: 
 8941: ." text-4"
 8942: 
 8943: : my-char
 8944:   [char] ALPHABET emit
 8945:   char emit
 8946: ;
 8947: @end example
 8948: 
 8949: When you load this code into Gforth, the following output is generated:
 8950: 
 8951: @example
 8952: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
 8953: @end example
 8954: 
 8955: @itemize @bullet
 8956: @item
 8957: Messages @code{text-1} and @code{text-3} are displayed because @code{.(} 
 8958: is an immediate word; it behaves in the same way whether it is used inside
 8959: or outside a colon definition.
 8960: @item
 8961: Message @code{text-4} is displayed because of Gforth's added interpretation
 8962: semantics for @code{."}.
 8963: @item
 8964: Message @code{text-2} is @i{not} displayed, because the text interpreter
 8965: performs the compilation semantics for @code{."} within the definition of
 8966: @code{my-word}.
 8967: @end itemize
 8968: 
 8969: Here are some examples of executing @code{my-word} and @code{my-char}:
 8970: 
 8971: @example
 8972: @kbd{my-word @key{RET}} text-2
 8973:  ok
 8974: @kbd{my-char fred @key{RET}} Af ok
 8975: @kbd{my-char jim @key{RET}} Aj ok
 8976: @end example
 8977: 
 8978: @itemize @bullet
 8979: @item
 8980: Message @code{text-2} is displayed because of the run-time behaviour of
 8981: @code{."}.
 8982: @item
 8983: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
 8984: on the stack at run-time. @code{emit} always displays the character
 8985: when @code{my-char} is executed.
 8986: @item
 8987: @code{char} parses a string at run-time and the second @code{emit} displays
 8988: the first character of the string.
 8989: @item
 8990: If you type @code{see my-char} you can see that @code{[char]} discarded
 8991: the text ``LPHABET'' and only compiled the display code for ``A'' into the
 8992: definition of @code{my-char}.
 8993: @end itemize
 8994: 
 8995: 
 8996: 
 8997: @node Input,  , Displaying characters and strings, Other I/O
 8998: @subsection Input
 8999: @cindex input
 9000: @cindex I/O - see input
 9001: @cindex parsing a string
 9002: 
 9003: For ways of storing character strings in memory see @ref{String Formats}.
 9004: 
 9005: @comment TODO examples for >number >float accept key key? pad parse word refill
 9006: @comment then index them
 9007: 
 9008: 
 9009: doc-key
 9010: doc-key?
 9011: doc-ekey
 9012: doc-ekey?
 9013: doc-ekey>char
 9014: doc->number
 9015: doc->float
 9016: doc-accept
 9017: doc-pad
 9018: @c anton: these belong in the input stream section
 9019: doc-parse
 9020: doc-word
 9021: doc-sword
 9022: doc-name
 9023: doc-refill
 9024: @comment obsolescent words..
 9025: doc-convert
 9026: doc-query
 9027: doc-expect
 9028: doc-span
 9029: 
 9030: 
 9031: @c -------------------------------------------------------------
 9032: @node Locals, Structures, Other I/O, Words
 9033: @section Locals
 9034: @cindex locals
 9035: 
 9036: Local variables can make Forth programming more enjoyable and Forth
 9037: programs easier to read. Unfortunately, the locals of ANS Forth are
 9038: laden with restrictions. Therefore, we provide not only the ANS Forth
 9039: locals wordset, but also our own, more powerful locals wordset (we
 9040: implemented the ANS Forth locals wordset through our locals wordset).
 9041: 
 9042: The ideas in this section have also been published in M. Anton Ertl,
 9043: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz,
 9044: Automatic Scoping of Local Variables}}, EuroForth '94.
 9045: 
 9046: @menu
 9047: * Gforth locals::               
 9048: * ANS Forth locals::            
 9049: @end menu
 9050: 
 9051: @node Gforth locals, ANS Forth locals, Locals, Locals
 9052: @subsection Gforth locals
 9053: @cindex Gforth locals
 9054: @cindex locals, Gforth style
 9055: 
 9056: Locals can be defined with
 9057: 
 9058: @example
 9059: @{ local1 local2 ... -- comment @}
 9060: @end example
 9061: or
 9062: @example
 9063: @{ local1 local2 ... @}
 9064: @end example
 9065: 
 9066: E.g.,
 9067: @example
 9068: : max @{ n1 n2 -- n3 @}
 9069:  n1 n2 > if
 9070:    n1
 9071:  else
 9072:    n2
 9073:  endif ;
 9074: @end example
 9075: 
 9076: The similarity of locals definitions with stack comments is intended. A
 9077: locals definition often replaces the stack comment of a word. The order
 9078: of the locals corresponds to the order in a stack comment and everything
 9079: after the @code{--} is really a comment.
 9080: 
 9081: This similarity has one disadvantage: It is too easy to confuse locals
 9082: declarations with stack comments, causing bugs and making them hard to
 9083: find. However, this problem can be avoided by appropriate coding
 9084: conventions: Do not use both notations in the same program. If you do,
 9085: they should be distinguished using additional means, e.g. by position.
 9086: 
 9087: @cindex types of locals
 9088: @cindex locals types
 9089: The name of the local may be preceded by a type specifier, e.g.,
 9090: @code{F:} for a floating point value:
 9091: 
 9092: @example
 9093: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
 9094: \ complex multiplication
 9095:  Ar Br f* Ai Bi f* f-
 9096:  Ar Bi f* Ai Br f* f+ ;
 9097: @end example
 9098: 
 9099: @cindex flavours of locals
 9100: @cindex locals flavours
 9101: @cindex value-flavoured locals
 9102: @cindex variable-flavoured locals
 9103: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
 9104: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
 9105: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
 9106: with @code{W:}, @code{D:} etc.) produces its value and can be changed
 9107: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
 9108: produces its address (which becomes invalid when the variable's scope is
 9109: left). E.g., the standard word @code{emit} can be defined in terms of
 9110: @code{type} like this:
 9111: 
 9112: @example
 9113: : emit @{ C^ char* -- @}
 9114:     char* 1 type ;
 9115: @end example
 9116: 
 9117: @cindex default type of locals
 9118: @cindex locals, default type
 9119: A local without type specifier is a @code{W:} local. Both flavours of
 9120: locals are initialized with values from the data or FP stack.
 9121: 
 9122: Currently there is no way to define locals with user-defined data
 9123: structures, but we are working on it.
 9124: 
 9125: Gforth allows defining locals everywhere in a colon definition. This
 9126: poses the following questions:
 9127: 
 9128: @menu
 9129: * Where are locals visible by name?::  
 9130: * How long do locals live?::    
 9131: * Locals programming style::    
 9132: * Locals implementation::       
 9133: @end menu
 9134: 
 9135: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
 9136: @subsubsection Where are locals visible by name?
 9137: @cindex locals visibility
 9138: @cindex visibility of locals
 9139: @cindex scope of locals
 9140: 
 9141: Basically, the answer is that locals are visible where you would expect
 9142: it in block-structured languages, and sometimes a little longer. If you
 9143: want to restrict the scope of a local, enclose its definition in
 9144: @code{SCOPE}...@code{ENDSCOPE}.
 9145: 
 9146: 
 9147: doc-scope
 9148: doc-endscope
 9149: 
 9150: 
 9151: These words behave like control structure words, so you can use them
 9152: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
 9153: arbitrary ways.
 9154: 
 9155: If you want a more exact answer to the visibility question, here's the
 9156: basic principle: A local is visible in all places that can only be
 9157: reached through the definition of the local@footnote{In compiler
 9158: construction terminology, all places dominated by the definition of the
 9159: local.}. In other words, it is not visible in places that can be reached
 9160: without going through the definition of the local. E.g., locals defined
 9161: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
 9162: defined in @code{BEGIN}...@code{UNTIL} are visible after the
 9163: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
 9164: 
 9165: The reasoning behind this solution is: We want to have the locals
 9166: visible as long as it is meaningful. The user can always make the
 9167: visibility shorter by using explicit scoping. In a place that can
 9168: only be reached through the definition of a local, the meaning of a
 9169: local name is clear. In other places it is not: How is the local
 9170: initialized at the control flow path that does not contain the
 9171: definition? Which local is meant, if the same name is defined twice in
 9172: two independent control flow paths?
 9173: 
 9174: This should be enough detail for nearly all users, so you can skip the
 9175: rest of this section. If you really must know all the gory details and
 9176: options, read on.
 9177: 
 9178: In order to implement this rule, the compiler has to know which places
 9179: are unreachable. It knows this automatically after @code{AHEAD},
 9180: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
 9181: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
 9182: compiler that the control flow never reaches that place. If
 9183: @code{UNREACHABLE} is not used where it could, the only consequence is
 9184: that the visibility of some locals is more limited than the rule above
 9185: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
 9186: lie to the compiler), buggy code will be produced.
 9187: 
 9188: 
 9189: doc-unreachable
 9190: 
 9191: 
 9192: Another problem with this rule is that at @code{BEGIN}, the compiler
 9193: does not know which locals will be visible on the incoming
 9194: back-edge. All problems discussed in the following are due to this
 9195: ignorance of the compiler (we discuss the problems using @code{BEGIN}
 9196: loops as examples; the discussion also applies to @code{?DO} and other
 9197: loops). Perhaps the most insidious example is:
 9198: @example
 9199: AHEAD
 9200: BEGIN
 9201:   x
 9202: [ 1 CS-ROLL ] THEN
 9203:   @{ x @}
 9204:   ...
 9205: UNTIL
 9206: @end example
 9207: 
 9208: This should be legal according to the visibility rule. The use of
 9209: @code{x} can only be reached through the definition; but that appears
 9210: textually below the use.
 9211: 
 9212: From this example it is clear that the visibility rules cannot be fully
 9213: implemented without major headaches. Our implementation treats common
 9214: cases as advertised and the exceptions are treated in a safe way: The
 9215: compiler makes a reasonable guess about the locals visible after a
 9216: @code{BEGIN}; if it is too pessimistic, the
 9217: user will get a spurious error about the local not being defined; if the
 9218: compiler is too optimistic, it will notice this later and issue a
 9219: warning. In the case above the compiler would complain about @code{x}
 9220: being undefined at its use. You can see from the obscure examples in
 9221: this section that it takes quite unusual control structures to get the
 9222: compiler into trouble, and even then it will often do fine.
 9223: 
 9224: If the @code{BEGIN} is reachable from above, the most optimistic guess
 9225: is that all locals visible before the @code{BEGIN} will also be
 9226: visible after the @code{BEGIN}. This guess is valid for all loops that
 9227: are entered only through the @code{BEGIN}, in particular, for normal
 9228: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
 9229: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
 9230: compiler. When the branch to the @code{BEGIN} is finally generated by
 9231: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
 9232: warns the user if it was too optimistic:
 9233: @example
 9234: IF
 9235:   @{ x @}
 9236: BEGIN
 9237:   \ x ? 
 9238: [ 1 cs-roll ] THEN
 9239:   ...
 9240: UNTIL
 9241: @end example
 9242: 
 9243: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
 9244: optimistically assumes that it lives until the @code{THEN}. It notices
 9245: this difference when it compiles the @code{UNTIL} and issues a
 9246: warning. The user can avoid the warning, and make sure that @code{x}
 9247: is not used in the wrong area by using explicit scoping:
 9248: @example
 9249: IF
 9250:   SCOPE
 9251:   @{ x @}
 9252:   ENDSCOPE
 9253: BEGIN
 9254: [ 1 cs-roll ] THEN
 9255:   ...
 9256: UNTIL
 9257: @end example
 9258: 
 9259: Since the guess is optimistic, there will be no spurious error messages
 9260: about undefined locals.
 9261: 
 9262: If the @code{BEGIN} is not reachable from above (e.g., after
 9263: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
 9264: optimistic guess, as the locals visible after the @code{BEGIN} may be
 9265: defined later. Therefore, the compiler assumes that no locals are
 9266: visible after the @code{BEGIN}. However, the user can use
 9267: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
 9268: visible at the BEGIN as at the point where the top control-flow stack
 9269: item was created.
 9270: 
 9271: 
 9272: doc-assume-live
 9273: 
 9274: 
 9275: @noindent
 9276: E.g.,
 9277: @example
 9278: @{ x @}
 9279: AHEAD
 9280: ASSUME-LIVE
 9281: BEGIN
 9282:   x
 9283: [ 1 CS-ROLL ] THEN
 9284:   ...
 9285: UNTIL
 9286: @end example
 9287: 
 9288: Other cases where the locals are defined before the @code{BEGIN} can be
 9289: handled by inserting an appropriate @code{CS-ROLL} before the
 9290: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
 9291: behind the @code{ASSUME-LIVE}).
 9292: 
 9293: Cases where locals are defined after the @code{BEGIN} (but should be
 9294: visible immediately after the @code{BEGIN}) can only be handled by
 9295: rearranging the loop. E.g., the ``most insidious'' example above can be
 9296: arranged into:
 9297: @example
 9298: BEGIN
 9299:   @{ x @}
 9300:   ... 0=
 9301: WHILE
 9302:   x
 9303: REPEAT
 9304: @end example
 9305: 
 9306: @node How long do locals live?, Locals programming style, Where are locals visible by name?, Gforth locals
 9307: @subsubsection How long do locals live?
 9308: @cindex locals lifetime
 9309: @cindex lifetime of locals
 9310: 
 9311: The right answer for the lifetime question would be: A local lives at
 9312: least as long as it can be accessed. For a value-flavoured local this
 9313: means: until the end of its visibility. However, a variable-flavoured
 9314: local could be accessed through its address far beyond its visibility
 9315: scope. Ultimately, this would mean that such locals would have to be
 9316: garbage collected. Since this entails un-Forth-like implementation
 9317: complexities, I adopted the same cowardly solution as some other
 9318: languages (e.g., C): The local lives only as long as it is visible;
 9319: afterwards its address is invalid (and programs that access it
 9320: afterwards are erroneous).
 9321: 
 9322: @node Locals programming style, Locals implementation, How long do locals live?, Gforth locals
 9323: @subsubsection Locals programming style
 9324: @cindex locals programming style
 9325: @cindex programming style, locals
 9326: 
 9327: The freedom to define locals anywhere has the potential to change
 9328: programming styles dramatically. In particular, the need to use the
 9329: return stack for intermediate storage vanishes. Moreover, all stack
 9330: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
 9331: determined arguments) can be eliminated: If the stack items are in the
 9332: wrong order, just write a locals definition for all of them; then
 9333: write the items in the order you want.
 9334: 
 9335: This seems a little far-fetched and eliminating stack manipulations is
 9336: unlikely to become a conscious programming objective. Still, the number
 9337: of stack manipulations will be reduced dramatically if local variables
 9338: are used liberally (e.g., compare @code{max} (@pxref{Gforth locals}) with
 9339: a traditional implementation of @code{max}).
 9340: 
 9341: This shows one potential benefit of locals: making Forth programs more
 9342: readable. Of course, this benefit will only be realized if the
 9343: programmers continue to honour the principle of factoring instead of
 9344: using the added latitude to make the words longer.
 9345: 
 9346: @cindex single-assignment style for locals
 9347: Using @code{TO} can and should be avoided.  Without @code{TO},
 9348: every value-flavoured local has only a single assignment and many
 9349: advantages of functional languages apply to Forth. I.e., programs are
 9350: easier to analyse, to optimize and to read: It is clear from the
 9351: definition what the local stands for, it does not turn into something
 9352: different later.
 9353: 
 9354: E.g., a definition using @code{TO} might look like this:
 9355: @example
 9356: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 9357:  u1 u2 min 0
 9358:  ?do
 9359:    addr1 c@@ addr2 c@@ -
 9360:    ?dup-if
 9361:      unloop exit
 9362:    then
 9363:    addr1 char+ TO addr1
 9364:    addr2 char+ TO addr2
 9365:  loop
 9366:  u1 u2 - ;
 9367: @end example
 9368: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
 9369: every loop iteration. @code{strcmp} is a typical example of the
 9370: readability problems of using @code{TO}. When you start reading
 9371: @code{strcmp}, you think that @code{addr1} refers to the start of the
 9372: string. Only near the end of the loop you realize that it is something
 9373: else.
 9374: 
 9375: This can be avoided by defining two locals at the start of the loop that
 9376: are initialized with the right value for the current iteration.
 9377: @example
 9378: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 9379:  addr1 addr2
 9380:  u1 u2 min 0 
 9381:  ?do @{ s1 s2 @}
 9382:    s1 c@@ s2 c@@ -
 9383:    ?dup-if
 9384:      unloop exit
 9385:    then
 9386:    s1 char+ s2 char+
 9387:  loop
 9388:  2drop
 9389:  u1 u2 - ;
 9390: @end example
 9391: Here it is clear from the start that @code{s1} has a different value
 9392: in every loop iteration.
 9393: 
 9394: @node Locals implementation,  , Locals programming style, Gforth locals
 9395: @subsubsection Locals implementation
 9396: @cindex locals implementation
 9397: @cindex implementation of locals
 9398: 
 9399: @cindex locals stack
 9400: Gforth uses an extra locals stack. The most compelling reason for
 9401: this is that the return stack is not float-aligned; using an extra stack
 9402: also eliminates the problems and restrictions of using the return stack
 9403: as locals stack. Like the other stacks, the locals stack grows toward
 9404: lower addresses. A few primitives allow an efficient implementation:
 9405: 
 9406: 
 9407: doc-@local#
 9408: doc-f@local#
 9409: doc-laddr#
 9410: doc-lp+!#
 9411: doc-lp!
 9412: doc->l
 9413: doc-f>l
 9414: 
 9415: 
 9416: In addition to these primitives, some specializations of these
 9417: primitives for commonly occurring inline arguments are provided for
 9418: efficiency reasons, e.g., @code{@@local0} as specialization of
 9419: @code{@@local#} for the inline argument 0. The following compiling words
 9420: compile the right specialized version, or the general version, as
 9421: appropriate:
 9422: 
 9423: 
 9424: doc-compile-@local
 9425: doc-compile-f@local
 9426: doc-compile-lp+!
 9427: 
 9428: 
 9429: Combinations of conditional branches and @code{lp+!#} like
 9430: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
 9431: is taken) are provided for efficiency and correctness in loops.
 9432: 
 9433: A special area in the dictionary space is reserved for keeping the
 9434: local variable names. @code{@{} switches the dictionary pointer to this
 9435: area and @code{@}} switches it back and generates the locals
 9436: initializing code. @code{W:} etc.@ are normal defining words. This
 9437: special area is cleared at the start of every colon definition.
 9438: 
 9439: @cindex word list for defining locals
 9440: A special feature of Gforth's dictionary is used to implement the
 9441: definition of locals without type specifiers: every word list (aka
 9442: vocabulary) has its own methods for searching
 9443: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
 9444: with a special search method: When it is searched for a word, it
 9445: actually creates that word using @code{W:}. @code{@{} changes the search
 9446: order to first search the word list containing @code{@}}, @code{W:} etc.,
 9447: and then the word list for defining locals without type specifiers.
 9448: 
 9449: The lifetime rules support a stack discipline within a colon
 9450: definition: The lifetime of a local is either nested with other locals
 9451: lifetimes or it does not overlap them.
 9452: 
 9453: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
 9454: pointer manipulation is generated. Between control structure words
 9455: locals definitions can push locals onto the locals stack. @code{AGAIN}
 9456: is the simplest of the other three control flow words. It has to
 9457: restore the locals stack depth of the corresponding @code{BEGIN}
 9458: before branching. The code looks like this:
 9459: @format
 9460: @code{lp+!#} current-locals-size @minus{} dest-locals-size
 9461: @code{branch} <begin>
 9462: @end format
 9463: 
 9464: @code{UNTIL} is a little more complicated: If it branches back, it
 9465: must adjust the stack just like @code{AGAIN}. But if it falls through,
 9466: the locals stack must not be changed. The compiler generates the
 9467: following code:
 9468: @format
 9469: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
 9470: @end format
 9471: The locals stack pointer is only adjusted if the branch is taken.
 9472: 
 9473: @code{THEN} can produce somewhat inefficient code:
 9474: @format
 9475: @code{lp+!#} current-locals-size @minus{} orig-locals-size
 9476: <orig target>:
 9477: @code{lp+!#} orig-locals-size @minus{} new-locals-size
 9478: @end format
 9479: The second @code{lp+!#} adjusts the locals stack pointer from the
 9480: level at the @i{orig} point to the level after the @code{THEN}. The
 9481: first @code{lp+!#} adjusts the locals stack pointer from the current
 9482: level to the level at the orig point, so the complete effect is an
 9483: adjustment from the current level to the right level after the
 9484: @code{THEN}.
 9485: 
 9486: @cindex locals information on the control-flow stack
 9487: @cindex control-flow stack items, locals information
 9488: In a conventional Forth implementation a dest control-flow stack entry
 9489: is just the target address and an orig entry is just the address to be
 9490: patched. Our locals implementation adds a word list to every orig or dest
 9491: item. It is the list of locals visible (or assumed visible) at the point
 9492: described by the entry. Our implementation also adds a tag to identify
 9493: the kind of entry, in particular to differentiate between live and dead
 9494: (reachable and unreachable) orig entries.
 9495: 
 9496: A few unusual operations have to be performed on locals word lists:
 9497: 
 9498: 
 9499: doc-common-list
 9500: doc-sub-list?
 9501: doc-list-size
 9502: 
 9503: 
 9504: Several features of our locals word list implementation make these
 9505: operations easy to implement: The locals word lists are organised as
 9506: linked lists; the tails of these lists are shared, if the lists
 9507: contain some of the same locals; and the address of a name is greater
 9508: than the address of the names behind it in the list.
 9509: 
 9510: Another important implementation detail is the variable
 9511: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
 9512: determine if they can be reached directly or only through the branch
 9513: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
 9514: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
 9515: definition, by @code{BEGIN} and usually by @code{THEN}.
 9516: 
 9517: Counted loops are similar to other loops in most respects, but
 9518: @code{LEAVE} requires special attention: It performs basically the same
 9519: service as @code{AHEAD}, but it does not create a control-flow stack
 9520: entry. Therefore the information has to be stored elsewhere;
 9521: traditionally, the information was stored in the target fields of the
 9522: branches created by the @code{LEAVE}s, by organizing these fields into a
 9523: linked list. Unfortunately, this clever trick does not provide enough
 9524: space for storing our extended control flow information. Therefore, we
 9525: introduce another stack, the leave stack. It contains the control-flow
 9526: stack entries for all unresolved @code{LEAVE}s.
 9527: 
 9528: Local names are kept until the end of the colon definition, even if
 9529: they are no longer visible in any control-flow path. In a few cases
 9530: this may lead to increased space needs for the locals name area, but
 9531: usually less than reclaiming this space would cost in code size.
 9532: 
 9533: 
 9534: @node ANS Forth locals,  , Gforth locals, Locals
 9535: @subsection ANS Forth locals
 9536: @cindex locals, ANS Forth style
 9537: 
 9538: The ANS Forth locals wordset does not define a syntax for locals, but
 9539: words that make it possible to define various syntaxes. One of the
 9540: possible syntaxes is a subset of the syntax we used in the Gforth locals
 9541: wordset, i.e.:
 9542: 
 9543: @example
 9544: @{ local1 local2 ... -- comment @}
 9545: @end example
 9546: @noindent
 9547: or
 9548: @example
 9549: @{ local1 local2 ... @}
 9550: @end example
 9551: 
 9552: The order of the locals corresponds to the order in a stack comment. The
 9553: restrictions are:
 9554: 
 9555: @itemize @bullet
 9556: @item
 9557: Locals can only be cell-sized values (no type specifiers are allowed).
 9558: @item
 9559: Locals can be defined only outside control structures.
 9560: @item
 9561: Locals can interfere with explicit usage of the return stack. For the
 9562: exact (and long) rules, see the standard. If you don't use return stack
 9563: accessing words in a definition using locals, you will be all right. The
 9564: purpose of this rule is to make locals implementation on the return
 9565: stack easier.
 9566: @item
 9567: The whole definition must be in one line.
 9568: @end itemize
 9569: 
 9570: Locals defined in ANS Forth behave like @code{VALUE}s
 9571: (@pxref{Values}). I.e., they are initialized from the stack. Using their
 9572: name produces their value. Their value can be changed using @code{TO}.
 9573: 
 9574: Since the syntax above is supported by Gforth directly, you need not do
 9575: anything to use it. If you want to port a program using this syntax to
 9576: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
 9577: syntax on the other system.
 9578: 
 9579: Note that a syntax shown in the standard, section A.13 looks
 9580: similar, but is quite different in having the order of locals
 9581: reversed. Beware!
 9582: 
 9583: The ANS Forth locals wordset itself consists of one word:
 9584: 
 9585: doc-(local)
 9586: 
 9587: The ANS Forth locals extension wordset defines a syntax using
 9588: @code{locals|}, but it is so awful that we strongly recommend not to use
 9589: it. We have implemented this syntax to make porting to Gforth easy, but
 9590: do not document it here. The problem with this syntax is that the locals
 9591: are defined in an order reversed with respect to the standard stack
 9592: comment notation, making programs harder to read, and easier to misread
 9593: and miswrite. The only merit of this syntax is that it is easy to
 9594: implement using the ANS Forth locals wordset.
 9595: 
 9596: 
 9597: @c ----------------------------------------------------------
 9598: @node Structures, Object-oriented Forth, Locals, Words
 9599: @section  Structures
 9600: @cindex structures
 9601: @cindex records
 9602: 
 9603: This section presents the structure package that comes with Gforth. A
 9604: version of the package implemented in ANS Forth is available in
 9605: @file{compat/struct.fs}. This package was inspired by a posting on
 9606: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
 9607: possibly John Hayes). A version of this section has been published in
 9608: M. Anton Ertl,
 9609: @uref{http://www.complang.tuwien.ac.at/forth/objects/structs.html, Yet
 9610: Another Forth Structures Package}, Forth Dimensions 19(3), pages
 9611: 13--16. Marcel Hendrix provided helpful comments.
 9612: 
 9613: @menu
 9614: * Why explicit structure support?::  
 9615: * Structure Usage::             
 9616: * Structure Naming Convention::  
 9617: * Structure Implementation::    
 9618: * Structure Glossary::          
 9619: @end menu
 9620: 
 9621: @node Why explicit structure support?, Structure Usage, Structures, Structures
 9622: @subsection Why explicit structure support?
 9623: 
 9624: @cindex address arithmetic for structures
 9625: @cindex structures using address arithmetic
 9626: If we want to use a structure containing several fields, we could simply
 9627: reserve memory for it, and access the fields using address arithmetic
 9628: (@pxref{Address arithmetic}). As an example, consider a structure with
 9629: the following fields
 9630: 
 9631: @table @code
 9632: @item a
 9633: is a float
 9634: @item b
 9635: is a cell
 9636: @item c
 9637: is a float
 9638: @end table
 9639: 
 9640: Given the (float-aligned) base address of the structure we get the
 9641: address of the field
 9642: 
 9643: @table @code
 9644: @item a
 9645: without doing anything further.
 9646: @item b
 9647: with @code{float+}
 9648: @item c
 9649: with @code{float+ cell+ faligned}
 9650: @end table
 9651: 
 9652: It is easy to see that this can become quite tiring. 
 9653: 
 9654: Moreover, it is not very readable, because seeing a
 9655: @code{cell+} tells us neither which kind of structure is
 9656: accessed nor what field is accessed; we have to somehow infer the kind
 9657: of structure, and then look up in the documentation, which field of
 9658: that structure corresponds to that offset.
 9659: 
 9660: Finally, this kind of address arithmetic also causes maintenance
 9661: troubles: If you add or delete a field somewhere in the middle of the
 9662: structure, you have to find and change all computations for the fields
 9663: afterwards.
 9664: 
 9665: So, instead of using @code{cell+} and friends directly, how
 9666: about storing the offsets in constants:
 9667: 
 9668: @example
 9669: 0 constant a-offset
 9670: 0 float+ constant b-offset
 9671: 0 float+ cell+ faligned c-offset
 9672: @end example
 9673: 
 9674: Now we can get the address of field @code{x} with @code{x-offset
 9675: +}. This is much better in all respects. Of course, you still
 9676: have to change all later offset definitions if you add a field. You can
 9677: fix this by declaring the offsets in the following way:
 9678: 
 9679: @example
 9680: 0 constant a-offset
 9681: a-offset float+ constant b-offset
 9682: b-offset cell+ faligned constant c-offset
 9683: @end example
 9684: 
 9685: Since we always use the offsets with @code{+}, we could use a defining
 9686: word @code{cfield} that includes the @code{+} in the action of the
 9687: defined word:
 9688: 
 9689: @example
 9690: : cfield ( n "name" -- )
 9691:     create ,
 9692: does> ( name execution: addr1 -- addr2 )
 9693:     @@ + ;
 9694: 
 9695: 0 cfield a
 9696: 0 a float+ cfield b
 9697: 0 b cell+ faligned cfield c
 9698: @end example
 9699: 
 9700: Instead of @code{x-offset +}, we now simply write @code{x}.
 9701: 
 9702: The structure field words now can be used quite nicely. However,
 9703: their definition is still a bit cumbersome: We have to repeat the
 9704: name, the information about size and alignment is distributed before
 9705: and after the field definitions etc.  The structure package presented
 9706: here addresses these problems.
 9707: 
 9708: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
 9709: @subsection Structure Usage
 9710: @cindex structure usage
 9711: 
 9712: @cindex @code{field} usage
 9713: @cindex @code{struct} usage
 9714: @cindex @code{end-struct} usage
 9715: You can define a structure for a (data-less) linked list with:
 9716: @example
 9717: struct
 9718:     cell% field list-next
 9719: end-struct list%
 9720: @end example
 9721: 
 9722: With the address of the list node on the stack, you can compute the
 9723: address of the field that contains the address of the next node with
 9724: @code{list-next}. E.g., you can determine the length of a list
 9725: with:
 9726: 
 9727: @example
 9728: : list-length ( list -- n )
 9729: \ "list" is a pointer to the first element of a linked list
 9730: \ "n" is the length of the list
 9731:     0 BEGIN ( list1 n1 )
 9732:         over
 9733:     WHILE ( list1 n1 )
 9734:         1+ swap list-next @@ swap
 9735:     REPEAT
 9736:     nip ;
 9737: @end example
 9738: 
 9739: You can reserve memory for a list node in the dictionary with
 9740: @code{list% %allot}, which leaves the address of the list node on the
 9741: stack. For the equivalent allocation on the heap you can use @code{list%
 9742: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
 9743: use @code{list% %allocate}). You can get the the size of a list
 9744: node with @code{list% %size} and its alignment with @code{list%
 9745: %alignment}.
 9746: 
 9747: Note that in ANS Forth the body of a @code{create}d word is
 9748: @code{aligned} but not necessarily @code{faligned};
 9749: therefore, if you do a:
 9750: 
 9751: @example
 9752: create @emph{name} foo% %allot drop
 9753: @end example
 9754: 
 9755: @noindent
 9756: then the memory alloted for @code{foo%} is guaranteed to start at the
 9757: body of @code{@emph{name}} only if @code{foo%} contains only character,
 9758: cell and double fields.  Therefore, if your structure contains floats,
 9759: better use
 9760: 
 9761: @example
 9762: foo% %allot constant @emph{name}
 9763: @end example
 9764: 
 9765: @cindex structures containing structures
 9766: You can include a structure @code{foo%} as a field of
 9767: another structure, like this:
 9768: @example
 9769: struct
 9770: ...
 9771:     foo% field ...
 9772: ...
 9773: end-struct ...
 9774: @end example
 9775: 
 9776: @cindex structure extension
 9777: @cindex extended records
 9778: Instead of starting with an empty structure, you can extend an
 9779: existing structure. E.g., a plain linked list without data, as defined
 9780: above, is hardly useful; You can extend it to a linked list of integers,
 9781: like this:@footnote{This feature is also known as @emph{extended
 9782: records}. It is the main innovation in the Oberon language; in other
 9783: words, adding this feature to Modula-2 led Wirth to create a new
 9784: language, write a new compiler etc.  Adding this feature to Forth just
 9785: required a few lines of code.}
 9786: 
 9787: @example
 9788: list%
 9789:     cell% field intlist-int
 9790: end-struct intlist%
 9791: @end example
 9792: 
 9793: @code{intlist%} is a structure with two fields:
 9794: @code{list-next} and @code{intlist-int}.
 9795: 
 9796: @cindex structures containing arrays
 9797: You can specify an array type containing @emph{n} elements of
 9798: type @code{foo%} like this:
 9799: 
 9800: @example
 9801: foo% @emph{n} *
 9802: @end example
 9803: 
 9804: You can use this array type in any place where you can use a normal
 9805: type, e.g., when defining a @code{field}, or with
 9806: @code{%allot}.
 9807: 
 9808: @cindex first field optimization
 9809: The first field is at the base address of a structure and the word for
 9810: this field (e.g., @code{list-next}) actually does not change the address
 9811: on the stack. You may be tempted to leave it away in the interest of
 9812: run-time and space efficiency. This is not necessary, because the
 9813: structure package optimizes this case: If you compile a first-field
 9814: words, no code is generated. So, in the interest of readability and
 9815: maintainability you should include the word for the field when accessing
 9816: the field.
 9817: 
 9818: 
 9819: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
 9820: @subsection Structure Naming Convention
 9821: @cindex structure naming convention
 9822: 
 9823: The field names that come to (my) mind are often quite generic, and,
 9824: if used, would cause frequent name clashes. E.g., many structures
 9825: probably contain a @code{counter} field. The structure names
 9826: that come to (my) mind are often also the logical choice for the names
 9827: of words that create such a structure.
 9828: 
 9829: Therefore, I have adopted the following naming conventions: 
 9830: 
 9831: @itemize @bullet
 9832: @cindex field naming convention
 9833: @item
 9834: The names of fields are of the form
 9835: @code{@emph{struct}-@emph{field}}, where
 9836: @code{@emph{struct}} is the basic name of the structure, and
 9837: @code{@emph{field}} is the basic name of the field. You can
 9838: think of field words as converting the (address of the)
 9839: structure into the (address of the) field.
 9840: 
 9841: @cindex structure naming convention
 9842: @item
 9843: The names of structures are of the form
 9844: @code{@emph{struct}%}, where
 9845: @code{@emph{struct}} is the basic name of the structure.
 9846: @end itemize
 9847: 
 9848: This naming convention does not work that well for fields of extended
 9849: structures; e.g., the integer list structure has a field
 9850: @code{intlist-int}, but has @code{list-next}, not
 9851: @code{intlist-next}.
 9852: 
 9853: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
 9854: @subsection Structure Implementation
 9855: @cindex structure implementation
 9856: @cindex implementation of structures
 9857: 
 9858: The central idea in the implementation is to pass the data about the
 9859: structure being built on the stack, not in some global
 9860: variable. Everything else falls into place naturally once this design
 9861: decision is made.
 9862: 
 9863: The type description on the stack is of the form @emph{align
 9864: size}. Keeping the size on the top-of-stack makes dealing with arrays
 9865: very simple.
 9866: 
 9867: @code{field} is a defining word that uses @code{Create}
 9868: and @code{DOES>}. The body of the field contains the offset
 9869: of the field, and the normal @code{DOES>} action is simply:
 9870: 
 9871: @example
 9872: @@ +
 9873: @end example
 9874: 
 9875: @noindent
 9876: i.e., add the offset to the address, giving the stack effect
 9877: @i{addr1 -- addr2} for a field.
 9878: 
 9879: @cindex first field optimization, implementation
 9880: This simple structure is slightly complicated by the optimization
 9881: for fields with offset 0, which requires a different
 9882: @code{DOES>}-part (because we cannot rely on there being
 9883: something on the stack if such a field is invoked during
 9884: compilation). Therefore, we put the different @code{DOES>}-parts
 9885: in separate words, and decide which one to invoke based on the
 9886: offset. For a zero offset, the field is basically a noop; it is
 9887: immediate, and therefore no code is generated when it is compiled.
 9888: 
 9889: @node Structure Glossary,  , Structure Implementation, Structures
 9890: @subsection Structure Glossary
 9891: @cindex structure glossary
 9892: 
 9893: 
 9894: doc-%align
 9895: doc-%alignment
 9896: doc-%alloc
 9897: doc-%allocate
 9898: doc-%allot
 9899: doc-cell%
 9900: doc-char%
 9901: doc-dfloat%
 9902: doc-double%
 9903: doc-end-struct
 9904: doc-field
 9905: doc-float%
 9906: doc-naligned
 9907: doc-sfloat%
 9908: doc-%size
 9909: doc-struct
 9910: 
 9911: 
 9912: @c -------------------------------------------------------------
 9913: @node Object-oriented Forth, Programming Tools, Structures, Words
 9914: @section Object-oriented Forth
 9915: 
 9916: Gforth comes with three packages for object-oriented programming:
 9917: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
 9918: is preloaded, so you have to @code{include} them before use. The most
 9919: important differences between these packages (and others) are discussed
 9920: in @ref{Comparison with other object models}. All packages are written
 9921: in ANS Forth and can be used with any other ANS Forth.
 9922: 
 9923: @menu
 9924: * Why object-oriented programming?::  
 9925: * Object-Oriented Terminology::  
 9926: * Objects::                     
 9927: * OOF::                         
 9928: * Mini-OOF::                    
 9929: * Comparison with other object models::  
 9930: @end menu
 9931: 
 9932: @c ----------------------------------------------------------------
 9933: @node Why object-oriented programming?, Object-Oriented Terminology, Object-oriented Forth, Object-oriented Forth
 9934: @subsection Why object-oriented programming?
 9935: @cindex object-oriented programming motivation
 9936: @cindex motivation for object-oriented programming
 9937: 
 9938: Often we have to deal with several data structures (@emph{objects}),
 9939: that have to be treated similarly in some respects, but differently in
 9940: others. Graphical objects are the textbook example: circles, triangles,
 9941: dinosaurs, icons, and others, and we may want to add more during program
 9942: development. We want to apply some operations to any graphical object,
 9943: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
 9944: has to do something different for every kind of object.
 9945: @comment TODO add some other operations eg perimeter, area
 9946: @comment and tie in to concrete examples later..
 9947: 
 9948: We could implement @code{draw} as a big @code{CASE}
 9949: control structure that executes the appropriate code depending on the
 9950: kind of object to be drawn. This would be not be very elegant, and,
 9951: moreover, we would have to change @code{draw} every time we add
 9952: a new kind of graphical object (say, a spaceship).
 9953: 
 9954: What we would rather do is: When defining spaceships, we would tell
 9955: the system: ``Here's how you @code{draw} a spaceship; you figure
 9956: out the rest''.
 9957: 
 9958: This is the problem that all systems solve that (rightfully) call
 9959: themselves object-oriented; the object-oriented packages presented here
 9960: solve this problem (and not much else).
 9961: @comment TODO ?list properties of oo systems.. oo vs o-based?
 9962: 
 9963: @c ------------------------------------------------------------------------
 9964: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
 9965: @subsection Object-Oriented Terminology
 9966: @cindex object-oriented terminology
 9967: @cindex terminology for object-oriented programming
 9968: 
 9969: This section is mainly for reference, so you don't have to understand
 9970: all of it right away.  The terminology is mainly Smalltalk-inspired.  In
 9971: short:
 9972: 
 9973: @table @emph
 9974: @cindex class
 9975: @item class
 9976: a data structure definition with some extras.
 9977: 
 9978: @cindex object
 9979: @item object
 9980: an instance of the data structure described by the class definition.
 9981: 
 9982: @cindex instance variables
 9983: @item instance variables
 9984: fields of the data structure.
 9985: 
 9986: @cindex selector
 9987: @cindex method selector
 9988: @cindex virtual function
 9989: @item selector
 9990: (or @emph{method selector}) a word (e.g.,
 9991: @code{draw}) that performs an operation on a variety of data
 9992: structures (classes). A selector describes @emph{what} operation to
 9993: perform. In C++ terminology: a (pure) virtual function.
 9994: 
 9995: @cindex method
 9996: @item method
 9997: the concrete definition that performs the operation
 9998: described by the selector for a specific class. A method specifies
 9999: @emph{how} the operation is performed for a specific class.
10000: 
10001: @cindex selector invocation
10002: @cindex message send
10003: @cindex invoking a selector
10004: @item selector invocation
10005: a call of a selector. One argument of the call (the TOS (top-of-stack))
10006: is used for determining which method is used. In Smalltalk terminology:
10007: a message (consisting of the selector and the other arguments) is sent
10008: to the object.
10009: 
10010: @cindex receiving object
10011: @item receiving object
10012: the object used for determining the method executed by a selector
10013: invocation. In the @file{objects.fs} model, it is the object that is on
10014: the TOS when the selector is invoked. (@emph{Receiving} comes from
10015: the Smalltalk @emph{message} terminology.)
10016: 
10017: @cindex child class
10018: @cindex parent class
10019: @cindex inheritance
10020: @item child class
10021: a class that has (@emph{inherits}) all properties (instance variables,
10022: selectors, methods) from a @emph{parent class}. In Smalltalk
10023: terminology: The subclass inherits from the superclass. In C++
10024: terminology: The derived class inherits from the base class.
10025: 
10026: @end table
10027: 
10028: @c If you wonder about the message sending terminology, it comes from
10029: @c a time when each object had it's own task and objects communicated via
10030: @c message passing; eventually the Smalltalk developers realized that
10031: @c they can do most things through simple (indirect) calls. They kept the
10032: @c terminology.
10033: 
10034: @c --------------------------------------------------------------
10035: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
10036: @subsection The @file{objects.fs} model
10037: @cindex objects
10038: @cindex object-oriented programming
10039: 
10040: @cindex @file{objects.fs}
10041: @cindex @file{oof.fs}
10042: 
10043: This section describes the @file{objects.fs} package. This material also
10044: has been published in M. Anton Ertl,
10045: @cite{@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html,
10046: Yet Another Forth Objects Package}}, Forth Dimensions 19(2), pages
10047: 37--43.
10048: @c McKewan's and Zsoter's packages
10049: 
10050: This section assumes that you have read @ref{Structures}.
10051: 
10052: The techniques on which this model is based have been used to implement
10053: the parser generator, Gray, and have also been used in Gforth for
10054: implementing the various flavours of word lists (hashed or not,
10055: case-sensitive or not, special-purpose word lists for locals etc.).
10056: 
10057: 
10058: @menu
10059: * Properties of the Objects model::  
10060: * Basic Objects Usage::         
10061: * The Objects base class::      
10062: * Creating objects::            
10063: * Object-Oriented Programming Style::  
10064: * Class Binding::               
10065: * Method conveniences::         
10066: * Classes and Scoping::         
10067: * Dividing classes::            
10068: * Object Interfaces::           
10069: * Objects Implementation::      
10070: * Objects Glossary::            
10071: @end menu
10072: 
10073: Marcel Hendrix provided helpful comments on this section.
10074: 
10075: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
10076: @subsubsection Properties of the @file{objects.fs} model
10077: @cindex @file{objects.fs} properties
10078: 
10079: @itemize @bullet
10080: @item
10081: It is straightforward to pass objects on the stack. Passing
10082: selectors on the stack is a little less convenient, but possible.
10083: 
10084: @item
10085: Objects are just data structures in memory, and are referenced by their
10086: address. You can create words for objects with normal defining words
10087: like @code{constant}. Likewise, there is no difference between instance
10088: variables that contain objects and those that contain other data.
10089: 
10090: @item
10091: Late binding is efficient and easy to use.
10092: 
10093: @item
10094: It avoids parsing, and thus avoids problems with state-smartness
10095: and reduced extensibility; for convenience there are a few parsing
10096: words, but they have non-parsing counterparts. There are also a few
10097: defining words that parse. This is hard to avoid, because all standard
10098: defining words parse (except @code{:noname}); however, such
10099: words are not as bad as many other parsing words, because they are not
10100: state-smart.
10101: 
10102: @item
10103: It does not try to incorporate everything. It does a few things and does
10104: them well (IMO). In particular, this model was not designed to support
10105: information hiding (although it has features that may help); you can use
10106: a separate package for achieving this.
10107: 
10108: @item
10109: It is layered; you don't have to learn and use all features to use this
10110: model. Only a few features are necessary (@pxref{Basic Objects Usage},
10111: @pxref{The Objects base class}, @pxref{Creating objects}.), the others
10112: are optional and independent of each other.
10113: 
10114: @item
10115: An implementation in ANS Forth is available.
10116: 
10117: @end itemize
10118: 
10119: 
10120: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
10121: @subsubsection Basic @file{objects.fs} Usage
10122: @cindex basic objects usage
10123: @cindex objects, basic usage
10124: 
10125: You can define a class for graphical objects like this:
10126: 
10127: @cindex @code{class} usage
10128: @cindex @code{end-class} usage
10129: @cindex @code{selector} usage
10130: @example
10131: object class \ "object" is the parent class
10132:   selector draw ( x y graphical -- )
10133: end-class graphical
10134: @end example
10135: 
10136: This code defines a class @code{graphical} with an
10137: operation @code{draw}.  We can perform the operation
10138: @code{draw} on any @code{graphical} object, e.g.:
10139: 
10140: @example
10141: 100 100 t-rex draw
10142: @end example
10143: 
10144: @noindent
10145: where @code{t-rex} is a word (say, a constant) that produces a
10146: graphical object.
10147: 
10148: @comment TODO add a 2nd operation eg perimeter.. and use for
10149: @comment a concrete example
10150: 
10151: @cindex abstract class
10152: How do we create a graphical object? With the present definitions,
10153: we cannot create a useful graphical object. The class
10154: @code{graphical} describes graphical objects in general, but not
10155: any concrete graphical object type (C++ users would call it an
10156: @emph{abstract class}); e.g., there is no method for the selector
10157: @code{draw} in the class @code{graphical}.
10158: 
10159: For concrete graphical objects, we define child classes of the
10160: class @code{graphical}, e.g.:
10161: 
10162: @cindex @code{overrides} usage
10163: @cindex @code{field} usage in class definition
10164: @example
10165: graphical class \ "graphical" is the parent class
10166:   cell% field circle-radius
10167: 
10168: :noname ( x y circle -- )
10169:   circle-radius @@ draw-circle ;
10170: overrides draw
10171: 
10172: :noname ( n-radius circle -- )
10173:   circle-radius ! ;
10174: overrides construct
10175: 
10176: end-class circle
10177: @end example
10178: 
10179: Here we define a class @code{circle} as a child of @code{graphical},
10180: with field @code{circle-radius} (which behaves just like a field
10181: (@pxref{Structures}); it defines (using @code{overrides}) new methods
10182: for the selectors @code{draw} and @code{construct} (@code{construct} is
10183: defined in @code{object}, the parent class of @code{graphical}).
10184: 
10185: Now we can create a circle on the heap (i.e.,
10186: @code{allocate}d memory) with:
10187: 
10188: @cindex @code{heap-new} usage
10189: @example
10190: 50 circle heap-new constant my-circle
10191: @end example
10192: 
10193: @noindent
10194: @code{heap-new} invokes @code{construct}, thus
10195: initializing the field @code{circle-radius} with 50. We can draw
10196: this new circle at (100,100) with:
10197: 
10198: @example
10199: 100 100 my-circle draw
10200: @end example
10201: 
10202: @cindex selector invocation, restrictions
10203: @cindex class definition, restrictions
10204: Note: You can only invoke a selector if the object on the TOS
10205: (the receiving object) belongs to the class where the selector was
10206: defined or one of its descendents; e.g., you can invoke
10207: @code{draw} only for objects belonging to @code{graphical}
10208: or its descendents (e.g., @code{circle}).  Immediately before
10209: @code{end-class}, the search order has to be the same as
10210: immediately after @code{class}.
10211: 
10212: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
10213: @subsubsection The @file{object.fs} base class
10214: @cindex @code{object} class
10215: 
10216: When you define a class, you have to specify a parent class.  So how do
10217: you start defining classes? There is one class available from the start:
10218: @code{object}. It is ancestor for all classes and so is the
10219: only class that has no parent. It has two selectors: @code{construct}
10220: and @code{print}.
10221: 
10222: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
10223: @subsubsection Creating objects
10224: @cindex creating objects
10225: @cindex object creation
10226: @cindex object allocation options
10227: 
10228: @cindex @code{heap-new} discussion
10229: @cindex @code{dict-new} discussion
10230: @cindex @code{construct} discussion
10231: You can create and initialize an object of a class on the heap with
10232: @code{heap-new} ( ... class -- object ) and in the dictionary
10233: (allocation with @code{allot}) with @code{dict-new} (
10234: ... class -- object ). Both words invoke @code{construct}, which
10235: consumes the stack items indicated by "..." above.
10236: 
10237: @cindex @code{init-object} discussion
10238: @cindex @code{class-inst-size} discussion
10239: If you want to allocate memory for an object yourself, you can get its
10240: alignment and size with @code{class-inst-size 2@@} ( class --
10241: align size ). Once you have memory for an object, you can initialize
10242: it with @code{init-object} ( ... class object -- );
10243: @code{construct} does only a part of the necessary work.
10244: 
10245: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
10246: @subsubsection Object-Oriented Programming Style
10247: @cindex object-oriented programming style
10248: @cindex programming style, object-oriented
10249: 
10250: This section is not exhaustive.
10251: 
10252: @cindex stack effects of selectors
10253: @cindex selectors and stack effects
10254: In general, it is a good idea to ensure that all methods for the
10255: same selector have the same stack effect: when you invoke a selector,
10256: you often have no idea which method will be invoked, so, unless all
10257: methods have the same stack effect, you will not know the stack effect
10258: of the selector invocation.
10259: 
10260: One exception to this rule is methods for the selector
10261: @code{construct}. We know which method is invoked, because we
10262: specify the class to be constructed at the same place. Actually, I
10263: defined @code{construct} as a selector only to give the users a
10264: convenient way to specify initialization. The way it is used, a
10265: mechanism different from selector invocation would be more natural
10266: (but probably would take more code and more space to explain).
10267: 
10268: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
10269: @subsubsection Class Binding
10270: @cindex class binding
10271: @cindex early binding
10272: 
10273: @cindex late binding
10274: Normal selector invocations determine the method at run-time depending
10275: on the class of the receiving object. This run-time selection is called
10276: @i{late binding}.
10277: 
10278: Sometimes it's preferable to invoke a different method. For example,
10279: you might want to use the simple method for @code{print}ing
10280: @code{object}s instead of the possibly long-winded @code{print} method
10281: of the receiver class. You can achieve this by replacing the invocation
10282: of @code{print} with:
10283: 
10284: @cindex @code{[bind]} usage
10285: @example
10286: [bind] object print
10287: @end example
10288: 
10289: @noindent
10290: in compiled code or:
10291: 
10292: @cindex @code{bind} usage
10293: @example
10294: bind object print
10295: @end example
10296: 
10297: @cindex class binding, alternative to
10298: @noindent
10299: in interpreted code. Alternatively, you can define the method with a
10300: name (e.g., @code{print-object}), and then invoke it through the
10301: name. Class binding is just a (often more convenient) way to achieve
10302: the same effect; it avoids name clutter and allows you to invoke
10303: methods directly without naming them first.
10304: 
10305: @cindex superclass binding
10306: @cindex parent class binding
10307: A frequent use of class binding is this: When we define a method
10308: for a selector, we often want the method to do what the selector does
10309: in the parent class, and a little more. There is a special word for
10310: this purpose: @code{[parent]}; @code{[parent]
10311: @emph{selector}} is equivalent to @code{[bind] @emph{parent
10312: selector}}, where @code{@emph{parent}} is the parent
10313: class of the current class. E.g., a method definition might look like:
10314: 
10315: @cindex @code{[parent]} usage
10316: @example
10317: :noname
10318:   dup [parent] foo \ do parent's foo on the receiving object
10319:   ... \ do some more
10320: ; overrides foo
10321: @end example
10322: 
10323: @cindex class binding as optimization
10324: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
10325: March 1997), Andrew McKewan presents class binding as an optimization
10326: technique. I recommend not using it for this purpose unless you are in
10327: an emergency. Late binding is pretty fast with this model anyway, so the
10328: benefit of using class binding is small; the cost of using class binding
10329: where it is not appropriate is reduced maintainability.
10330: 
10331: While we are at programming style questions: You should bind
10332: selectors only to ancestor classes of the receiving object. E.g., say,
10333: you know that the receiving object is of class @code{foo} or its
10334: descendents; then you should bind only to @code{foo} and its
10335: ancestors.
10336: 
10337: @node Method conveniences, Classes and Scoping, Class Binding, Objects
10338: @subsubsection Method conveniences
10339: @cindex method conveniences
10340: 
10341: In a method you usually access the receiving object pretty often.  If
10342: you define the method as a plain colon definition (e.g., with
10343: @code{:noname}), you may have to do a lot of stack
10344: gymnastics. To avoid this, you can define the method with @code{m:
10345: ... ;m}. E.g., you could define the method for
10346: @code{draw}ing a @code{circle} with
10347: 
10348: @cindex @code{this} usage
10349: @cindex @code{m:} usage
10350: @cindex @code{;m} usage
10351: @example
10352: m: ( x y circle -- )
10353:   ( x y ) this circle-radius @@ draw-circle ;m
10354: @end example
10355: 
10356: @cindex @code{exit} in @code{m: ... ;m}
10357: @cindex @code{exitm} discussion
10358: @cindex @code{catch} in @code{m: ... ;m}
10359: When this method is executed, the receiver object is removed from the
10360: stack; you can access it with @code{this} (admittedly, in this
10361: example the use of @code{m: ... ;m} offers no advantage). Note
10362: that I specify the stack effect for the whole method (i.e. including
10363: the receiver object), not just for the code between @code{m:}
10364: and @code{;m}. You cannot use @code{exit} in
10365: @code{m:...;m}; instead, use
10366: @code{exitm}.@footnote{Moreover, for any word that calls
10367: @code{catch} and was defined before loading
10368: @code{objects.fs}, you have to redefine it like I redefined
10369: @code{catch}: @code{: catch this >r catch r> to-this ;}}
10370: 
10371: @cindex @code{inst-var} usage
10372: You will frequently use sequences of the form @code{this
10373: @emph{field}} (in the example above: @code{this
10374: circle-radius}). If you use the field only in this way, you can
10375: define it with @code{inst-var} and eliminate the
10376: @code{this} before the field name. E.g., the @code{circle}
10377: class above could also be defined with:
10378: 
10379: @example
10380: graphical class
10381:   cell% inst-var radius
10382: 
10383: m: ( x y circle -- )
10384:   radius @@ draw-circle ;m
10385: overrides draw
10386: 
10387: m: ( n-radius circle -- )
10388:   radius ! ;m
10389: overrides construct
10390: 
10391: end-class circle
10392: @end example
10393: 
10394: @code{radius} can only be used in @code{circle} and its
10395: descendent classes and inside @code{m:...;m}.
10396: 
10397: @cindex @code{inst-value} usage
10398: You can also define fields with @code{inst-value}, which is
10399: to @code{inst-var} what @code{value} is to
10400: @code{variable}.  You can change the value of such a field with
10401: @code{[to-inst]}.  E.g., we could also define the class
10402: @code{circle} like this:
10403: 
10404: @example
10405: graphical class
10406:   inst-value radius
10407: 
10408: m: ( x y circle -- )
10409:   radius draw-circle ;m
10410: overrides draw
10411: 
10412: m: ( n-radius circle -- )
10413:   [to-inst] radius ;m
10414: overrides construct
10415: 
10416: end-class circle
10417: @end example
10418: 
10419: @c !! :m is easy to confuse with m:.  Another name would be better.
10420: 
10421: @c Finally, you can define named methods with @code{:m}.  One use of this
10422: @c feature is the definition of words that occur only in one class and are
10423: @c not intended to be overridden, but which still need method context
10424: @c (e.g., for accessing @code{inst-var}s).  Another use is for methods that
10425: @c would be bound frequently, if defined anonymously.
10426: 
10427: 
10428: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
10429: @subsubsection Classes and Scoping
10430: @cindex classes and scoping
10431: @cindex scoping and classes
10432: 
10433: Inheritance is frequent, unlike structure extension. This exacerbates
10434: the problem with the field name convention (@pxref{Structure Naming
10435: Convention}): One always has to remember in which class the field was
10436: originally defined; changing a part of the class structure would require
10437: changes for renaming in otherwise unaffected code.
10438: 
10439: @cindex @code{inst-var} visibility
10440: @cindex @code{inst-value} visibility
10441: To solve this problem, I added a scoping mechanism (which was not in my
10442: original charter): A field defined with @code{inst-var} (or
10443: @code{inst-value}) is visible only in the class where it is defined and in
10444: the descendent classes of this class.  Using such fields only makes
10445: sense in @code{m:}-defined methods in these classes anyway.
10446: 
10447: This scoping mechanism allows us to use the unadorned field name,
10448: because name clashes with unrelated words become much less likely.
10449: 
10450: @cindex @code{protected} discussion
10451: @cindex @code{private} discussion
10452: Once we have this mechanism, we can also use it for controlling the
10453: visibility of other words: All words defined after
10454: @code{protected} are visible only in the current class and its
10455: descendents. @code{public} restores the compilation
10456: (i.e. @code{current}) word list that was in effect before. If you
10457: have several @code{protected}s without an intervening
10458: @code{public} or @code{set-current}, @code{public}
10459: will restore the compilation word list in effect before the first of
10460: these @code{protected}s.
10461: 
10462: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
10463: @subsubsection Dividing classes
10464: @cindex Dividing classes
10465: @cindex @code{methods}...@code{end-methods}
10466: 
10467: You may want to do the definition of methods separate from the
10468: definition of the class, its selectors, fields, and instance variables,
10469: i.e., separate the implementation from the definition.  You can do this
10470: in the following way:
10471: 
10472: @example
10473: graphical class
10474:   inst-value radius
10475: end-class circle
10476: 
10477: ... \ do some other stuff
10478: 
10479: circle methods \ now we are ready
10480: 
10481: m: ( x y circle -- )
10482:   radius draw-circle ;m
10483: overrides draw
10484: 
10485: m: ( n-radius circle -- )
10486:   [to-inst] radius ;m
10487: overrides construct
10488: 
10489: end-methods
10490: @end example
10491: 
10492: You can use several @code{methods}...@code{end-methods} sections.  The
10493: only things you can do to the class in these sections are: defining
10494: methods, and overriding the class's selectors.  You must not define new
10495: selectors or fields.
10496: 
10497: Note that you often have to override a selector before using it.  In
10498: particular, you usually have to override @code{construct} with a new
10499: method before you can invoke @code{heap-new} and friends.  E.g., you
10500: must not create a circle before the @code{overrides construct} sequence
10501: in the example above.
10502: 
10503: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
10504: @subsubsection Object Interfaces
10505: @cindex object interfaces
10506: @cindex interfaces for objects
10507: 
10508: In this model you can only call selectors defined in the class of the
10509: receiving objects or in one of its ancestors. If you call a selector
10510: with a receiving object that is not in one of these classes, the
10511: result is undefined; if you are lucky, the program crashes
10512: immediately.
10513: 
10514: @cindex selectors common to hardly-related classes
10515: Now consider the case when you want to have a selector (or several)
10516: available in two classes: You would have to add the selector to a
10517: common ancestor class, in the worst case to @code{object}. You
10518: may not want to do this, e.g., because someone else is responsible for
10519: this ancestor class.
10520: 
10521: The solution for this problem is interfaces. An interface is a
10522: collection of selectors. If a class implements an interface, the
10523: selectors become available to the class and its descendents. A class
10524: can implement an unlimited number of interfaces. For the problem
10525: discussed above, we would define an interface for the selector(s), and
10526: both classes would implement the interface.
10527: 
10528: As an example, consider an interface @code{storage} for
10529: writing objects to disk and getting them back, and a class
10530: @code{foo} that implements it. The code would look like this:
10531: 
10532: @cindex @code{interface} usage
10533: @cindex @code{end-interface} usage
10534: @cindex @code{implementation} usage
10535: @example
10536: interface
10537:   selector write ( file object -- )
10538:   selector read1 ( file object -- )
10539: end-interface storage
10540: 
10541: bar class
10542:   storage implementation
10543: 
10544: ... overrides write
10545: ... overrides read1
10546: ...
10547: end-class foo
10548: @end example
10549: 
10550: @noindent
10551: (I would add a word @code{read} @i{( file -- object )} that uses
10552: @code{read1} internally, but that's beyond the point illustrated
10553: here.)
10554: 
10555: Note that you cannot use @code{protected} in an interface; and
10556: of course you cannot define fields.
10557: 
10558: In the Neon model, all selectors are available for all classes;
10559: therefore it does not need interfaces. The price you pay in this model
10560: is slower late binding, and therefore, added complexity to avoid late
10561: binding.
10562: 
10563: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
10564: @subsubsection @file{objects.fs} Implementation
10565: @cindex @file{objects.fs} implementation
10566: 
10567: @cindex @code{object-map} discussion
10568: An object is a piece of memory, like one of the data structures
10569: described with @code{struct...end-struct}. It has a field
10570: @code{object-map} that points to the method map for the object's
10571: class.
10572: 
10573: @cindex method map
10574: @cindex virtual function table
10575: The @emph{method map}@footnote{This is Self terminology; in C++
10576: terminology: virtual function table.} is an array that contains the
10577: execution tokens (@i{xt}s) of the methods for the object's class. Each
10578: selector contains an offset into a method map.
10579: 
10580: @cindex @code{selector} implementation, class
10581: @code{selector} is a defining word that uses
10582: @code{CREATE} and @code{DOES>}. The body of the
10583: selector contains the offset; the @code{DOES>} action for a
10584: class selector is, basically:
10585: 
10586: @example
10587: ( object addr ) @@ over object-map @@ + @@ execute
10588: @end example
10589: 
10590: Since @code{object-map} is the first field of the object, it
10591: does not generate any code. As you can see, calling a selector has a
10592: small, constant cost.
10593: 
10594: @cindex @code{current-interface} discussion
10595: @cindex class implementation and representation
10596: A class is basically a @code{struct} combined with a method
10597: map. During the class definition the alignment and size of the class
10598: are passed on the stack, just as with @code{struct}s, so
10599: @code{field} can also be used for defining class
10600: fields. However, passing more items on the stack would be
10601: inconvenient, so @code{class} builds a data structure in memory,
10602: which is accessed through the variable
10603: @code{current-interface}. After its definition is complete, the
10604: class is represented on the stack by a pointer (e.g., as parameter for
10605: a child class definition).
10606: 
10607: A new class starts off with the alignment and size of its parent,
10608: and a copy of the parent's method map. Defining new fields extends the
10609: size and alignment; likewise, defining new selectors extends the
10610: method map. @code{overrides} just stores a new @i{xt} in the method
10611: map at the offset given by the selector.
10612: 
10613: @cindex class binding, implementation
10614: Class binding just gets the @i{xt} at the offset given by the selector
10615: from the class's method map and @code{compile,}s (in the case of
10616: @code{[bind]}) it.
10617: 
10618: @cindex @code{this} implementation
10619: @cindex @code{catch} and @code{this}
10620: @cindex @code{this} and @code{catch}
10621: I implemented @code{this} as a @code{value}. At the
10622: start of an @code{m:...;m} method the old @code{this} is
10623: stored to the return stack and restored at the end; and the object on
10624: the TOS is stored @code{TO this}. This technique has one
10625: disadvantage: If the user does not leave the method via
10626: @code{;m}, but via @code{throw} or @code{exit},
10627: @code{this} is not restored (and @code{exit} may
10628: crash). To deal with the @code{throw} problem, I have redefined
10629: @code{catch} to save and restore @code{this}; the same
10630: should be done with any word that can catch an exception. As for
10631: @code{exit}, I simply forbid it (as a replacement, there is
10632: @code{exitm}).
10633: 
10634: @cindex @code{inst-var} implementation
10635: @code{inst-var} is just the same as @code{field}, with
10636: a different @code{DOES>} action:
10637: @example
10638: @@ this +
10639: @end example
10640: Similar for @code{inst-value}.
10641: 
10642: @cindex class scoping implementation
10643: Each class also has a word list that contains the words defined with
10644: @code{inst-var} and @code{inst-value}, and its protected
10645: words. It also has a pointer to its parent. @code{class} pushes
10646: the word lists of the class and all its ancestors onto the search order stack,
10647: and @code{end-class} drops them.
10648: 
10649: @cindex interface implementation
10650: An interface is like a class without fields, parent and protected
10651: words; i.e., it just has a method map. If a class implements an
10652: interface, its method map contains a pointer to the method map of the
10653: interface. The positive offsets in the map are reserved for class
10654: methods, therefore interface map pointers have negative
10655: offsets. Interfaces have offsets that are unique throughout the
10656: system, unlike class selectors, whose offsets are only unique for the
10657: classes where the selector is available (invokable).
10658: 
10659: This structure means that interface selectors have to perform one
10660: indirection more than class selectors to find their method. Their body
10661: contains the interface map pointer offset in the class method map, and
10662: the method offset in the interface method map. The
10663: @code{does>} action for an interface selector is, basically:
10664: 
10665: @example
10666: ( object selector-body )
10667: 2dup selector-interface @@ ( object selector-body object interface-offset )
10668: swap object-map @@ + @@ ( object selector-body map )
10669: swap selector-offset @@ + @@ execute
10670: @end example
10671: 
10672: where @code{object-map} and @code{selector-offset} are
10673: first fields and generate no code.
10674: 
10675: As a concrete example, consider the following code:
10676: 
10677: @example
10678: interface
10679:   selector if1sel1
10680:   selector if1sel2
10681: end-interface if1
10682: 
10683: object class
10684:   if1 implementation
10685:   selector cl1sel1
10686:   cell% inst-var cl1iv1
10687: 
10688: ' m1 overrides construct
10689: ' m2 overrides if1sel1
10690: ' m3 overrides if1sel2
10691: ' m4 overrides cl1sel2
10692: end-class cl1
10693: 
10694: create obj1 object dict-new drop
10695: create obj2 cl1    dict-new drop
10696: @end example
10697: 
10698: The data structure created by this code (including the data structure
10699: for @code{object}) is shown in the
10700: @uref{objects-implementation.eps,figure}, assuming a cell size of 4.
10701: @comment TODO add this diagram..
10702: 
10703: @node Objects Glossary,  , Objects Implementation, Objects
10704: @subsubsection @file{objects.fs} Glossary
10705: @cindex @file{objects.fs} Glossary
10706: 
10707: 
10708: doc---objects-bind
10709: doc---objects-<bind>
10710: doc---objects-bind'
10711: doc---objects-[bind]
10712: doc---objects-class
10713: doc---objects-class->map
10714: doc---objects-class-inst-size
10715: doc---objects-class-override!
10716: doc---objects-class-previous
10717: doc---objects-class>order
10718: doc---objects-construct
10719: doc---objects-current'
10720: doc---objects-[current]
10721: doc---objects-current-interface
10722: doc---objects-dict-new
10723: doc---objects-end-class
10724: doc---objects-end-class-noname
10725: doc---objects-end-interface
10726: doc---objects-end-interface-noname
10727: doc---objects-end-methods
10728: doc---objects-exitm
10729: doc---objects-heap-new
10730: doc---objects-implementation
10731: doc---objects-init-object
10732: doc---objects-inst-value
10733: doc---objects-inst-var
10734: doc---objects-interface
10735: doc---objects-m:
10736: doc---objects-:m
10737: doc---objects-;m
10738: doc---objects-method
10739: doc---objects-methods
10740: doc---objects-object
10741: doc---objects-overrides
10742: doc---objects-[parent]
10743: doc---objects-print
10744: doc---objects-protected
10745: doc---objects-public
10746: doc---objects-selector
10747: doc---objects-this
10748: doc---objects-<to-inst>
10749: doc---objects-[to-inst]
10750: doc---objects-to-this
10751: doc---objects-xt-new
10752: 
10753: 
10754: @c -------------------------------------------------------------
10755: @node OOF, Mini-OOF, Objects, Object-oriented Forth
10756: @subsection The @file{oof.fs} model
10757: @cindex oof
10758: @cindex object-oriented programming
10759: 
10760: @cindex @file{objects.fs}
10761: @cindex @file{oof.fs}
10762: 
10763: This section describes the @file{oof.fs} package.
10764: 
10765: The package described in this section has been used in bigFORTH since 1991, and
10766: used for two large applications: a chromatographic system used to
10767: create new medicaments, and a graphic user interface library (MINOS).
10768: 
10769: You can find a description (in German) of @file{oof.fs} in @cite{Object
10770: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
10771: 10(2), 1994.
10772: 
10773: @menu
10774: * Properties of the OOF model::  
10775: * Basic OOF Usage::             
10776: * The OOF base class::          
10777: * Class Declaration::           
10778: * Class Implementation::        
10779: @end menu
10780: 
10781: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
10782: @subsubsection Properties of the @file{oof.fs} model
10783: @cindex @file{oof.fs} properties
10784: 
10785: @itemize @bullet
10786: @item
10787: This model combines object oriented programming with information
10788: hiding. It helps you writing large application, where scoping is
10789: necessary, because it provides class-oriented scoping.
10790: 
10791: @item
10792: Named objects, object pointers, and object arrays can be created,
10793: selector invocation uses the ``object selector'' syntax. Selector invocation
10794: to objects and/or selectors on the stack is a bit less convenient, but
10795: possible.
10796: 
10797: @item
10798: Selector invocation and instance variable usage of the active object is
10799: straightforward, since both make use of the active object.
10800: 
10801: @item
10802: Late binding is efficient and easy to use.
10803: 
10804: @item
10805: State-smart objects parse selectors. However, extensibility is provided
10806: using a (parsing) selector @code{postpone} and a selector @code{'}.
10807: 
10808: @item
10809: An implementation in ANS Forth is available.
10810: 
10811: @end itemize
10812: 
10813: 
10814: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
10815: @subsubsection Basic @file{oof.fs} Usage
10816: @cindex @file{oof.fs} usage
10817: 
10818: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
10819: 
10820: You can define a class for graphical objects like this:
10821: 
10822: @cindex @code{class} usage
10823: @cindex @code{class;} usage
10824: @cindex @code{method} usage
10825: @example
10826: object class graphical \ "object" is the parent class
10827:   method draw ( x y graphical -- )
10828: class;
10829: @end example
10830: 
10831: This code defines a class @code{graphical} with an
10832: operation @code{draw}.  We can perform the operation
10833: @code{draw} on any @code{graphical} object, e.g.:
10834: 
10835: @example
10836: 100 100 t-rex draw
10837: @end example
10838: 
10839: @noindent
10840: where @code{t-rex} is an object or object pointer, created with e.g.
10841: @code{graphical : t-rex}.
10842: 
10843: @cindex abstract class
10844: How do we create a graphical object? With the present definitions,
10845: we cannot create a useful graphical object. The class
10846: @code{graphical} describes graphical objects in general, but not
10847: any concrete graphical object type (C++ users would call it an
10848: @emph{abstract class}); e.g., there is no method for the selector
10849: @code{draw} in the class @code{graphical}.
10850: 
10851: For concrete graphical objects, we define child classes of the
10852: class @code{graphical}, e.g.:
10853: 
10854: @example
10855: graphical class circle \ "graphical" is the parent class
10856:   cell var circle-radius
10857: how:
10858:   : draw ( x y -- )
10859:     circle-radius @@ draw-circle ;
10860: 
10861:   : init ( n-radius -- (
10862:     circle-radius ! ;
10863: class;
10864: @end example
10865: 
10866: Here we define a class @code{circle} as a child of @code{graphical},
10867: with a field @code{circle-radius}; it defines new methods for the
10868: selectors @code{draw} and @code{init} (@code{init} is defined in
10869: @code{object}, the parent class of @code{graphical}).
10870: 
10871: Now we can create a circle in the dictionary with:
10872: 
10873: @example
10874: 50 circle : my-circle
10875: @end example
10876: 
10877: @noindent
10878: @code{:} invokes @code{init}, thus initializing the field
10879: @code{circle-radius} with 50. We can draw this new circle at (100,100)
10880: with:
10881: 
10882: @example
10883: 100 100 my-circle draw
10884: @end example
10885: 
10886: @cindex selector invocation, restrictions
10887: @cindex class definition, restrictions
10888: Note: You can only invoke a selector if the receiving object belongs to
10889: the class where the selector was defined or one of its descendents;
10890: e.g., you can invoke @code{draw} only for objects belonging to
10891: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
10892: mechanism will check if you try to invoke a selector that is not
10893: defined in this class hierarchy, so you'll get an error at compilation
10894: time.
10895: 
10896: 
10897: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
10898: @subsubsection The @file{oof.fs} base class
10899: @cindex @file{oof.fs} base class
10900: 
10901: When you define a class, you have to specify a parent class.  So how do
10902: you start defining classes? There is one class available from the start:
10903: @code{object}. You have to use it as ancestor for all classes. It is the
10904: only class that has no parent. Classes are also objects, except that
10905: they don't have instance variables; class manipulation such as
10906: inheritance or changing definitions of a class is handled through
10907: selectors of the class @code{object}.
10908: 
10909: @code{object} provides a number of selectors:
10910: 
10911: @itemize @bullet
10912: @item
10913: @code{class} for subclassing, @code{definitions} to add definitions
10914: later on, and @code{class?} to get type informations (is the class a
10915: subclass of the class passed on the stack?).
10916: 
10917: doc---object-class
10918: doc---object-definitions
10919: doc---object-class?
10920: 
10921: 
10922: @item
10923: @code{init} and @code{dispose} as constructor and destructor of the
10924: object. @code{init} is invocated after the object's memory is allocated,
10925: while @code{dispose} also handles deallocation. Thus if you redefine
10926: @code{dispose}, you have to call the parent's dispose with @code{super
10927: dispose}, too.
10928: 
10929: doc---object-init
10930: doc---object-dispose
10931: 
10932: 
10933: @item
10934: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
10935: @code{[]} to create named and unnamed objects and object arrays or
10936: object pointers.
10937: 
10938: doc---object-new
10939: doc---object-new[]
10940: doc---object-:
10941: doc---object-ptr
10942: doc---object-asptr
10943: doc---object-[]
10944: 
10945: 
10946: @item
10947: @code{::} and @code{super} for explicit scoping. You should use explicit
10948: scoping only for super classes or classes with the same set of instance
10949: variables. Explicitly-scoped selectors use early binding.
10950: 
10951: doc---object-::
10952: doc---object-super
10953: 
10954: 
10955: @item
10956: @code{self} to get the address of the object
10957: 
10958: doc---object-self
10959: 
10960: 
10961: @item
10962: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
10963: pointers and instance defers.
10964: 
10965: doc---object-bind
10966: doc---object-bound
10967: doc---object-link
10968: doc---object-is
10969: 
10970: 
10971: @item
10972: @code{'} to obtain selector tokens, @code{send} to invocate selectors
10973: form the stack, and @code{postpone} to generate selector invocation code.
10974: 
10975: doc---object-'
10976: doc---object-postpone
10977: 
10978: 
10979: @item
10980: @code{with} and @code{endwith} to select the active object from the
10981: stack, and enable its scope. Using @code{with} and @code{endwith}
10982: also allows you to create code using selector @code{postpone} without being
10983: trapped by the state-smart objects.
10984: 
10985: doc---object-with
10986: doc---object-endwith
10987: 
10988: 
10989: @end itemize
10990: 
10991: @node Class Declaration, Class Implementation, The OOF base class, OOF
10992: @subsubsection Class Declaration
10993: @cindex class declaration
10994: 
10995: @itemize @bullet
10996: @item
10997: Instance variables
10998: 
10999: doc---oof-var
11000: 
11001: 
11002: @item
11003: Object pointers
11004: 
11005: doc---oof-ptr
11006: doc---oof-asptr
11007: 
11008: 
11009: @item
11010: Instance defers
11011: 
11012: doc---oof-defer
11013: 
11014: 
11015: @item
11016: Method selectors
11017: 
11018: doc---oof-early
11019: doc---oof-method
11020: 
11021: 
11022: @item
11023: Class-wide variables
11024: 
11025: doc---oof-static
11026: 
11027: 
11028: @item
11029: End declaration
11030: 
11031: doc---oof-how:
11032: doc---oof-class;
11033: 
11034: 
11035: @end itemize
11036: 
11037: @c -------------------------------------------------------------
11038: @node Class Implementation,  , Class Declaration, OOF
11039: @subsubsection Class Implementation
11040: @cindex class implementation
11041: 
11042: @c -------------------------------------------------------------
11043: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
11044: @subsection The @file{mini-oof.fs} model
11045: @cindex mini-oof
11046: 
11047: Gforth's third object oriented Forth package is a 12-liner. It uses a
11048: mixture of the @file{objects.fs} and the @file{oof.fs} syntax,
11049: and reduces to the bare minimum of features. This is based on a posting
11050: of Bernd Paysan in comp.lang.forth.
11051: 
11052: @menu
11053: * Basic Mini-OOF Usage::        
11054: * Mini-OOF Example::            
11055: * Mini-OOF Implementation::     
11056: @end menu
11057: 
11058: @c -------------------------------------------------------------
11059: @node Basic Mini-OOF Usage, Mini-OOF Example, Mini-OOF, Mini-OOF
11060: @subsubsection Basic @file{mini-oof.fs} Usage
11061: @cindex mini-oof usage
11062: 
11063: There is a base class (@code{class}, which allocates one cell for the
11064: object pointer) plus seven other words: to define a method, a variable,
11065: a class; to end a class, to resolve binding, to allocate an object and
11066: to compile a class method.
11067: @comment TODO better description of the last one
11068: 
11069: 
11070: doc-object
11071: doc-method
11072: doc-var
11073: doc-class
11074: doc-end-class
11075: doc-defines
11076: doc-new
11077: doc-::
11078: 
11079: 
11080: 
11081: @c -------------------------------------------------------------
11082: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
11083: @subsubsection Mini-OOF Example
11084: @cindex mini-oof example
11085: 
11086: A short example shows how to use this package. This example, in slightly
11087: extended form, is supplied as @file{moof-exm.fs}
11088: @comment TODO could flesh this out with some comments from the Forthwrite article
11089: 
11090: @example
11091: object class
11092:   method init
11093:   method draw
11094: end-class graphical
11095: @end example
11096: 
11097: This code defines a class @code{graphical} with an
11098: operation @code{draw}.  We can perform the operation
11099: @code{draw} on any @code{graphical} object, e.g.:
11100: 
11101: @example
11102: 100 100 t-rex draw
11103: @end example
11104: 
11105: where @code{t-rex} is an object or object pointer, created with e.g.
11106: @code{graphical new Constant t-rex}.
11107: 
11108: For concrete graphical objects, we define child classes of the
11109: class @code{graphical}, e.g.:
11110: 
11111: @example
11112: graphical class
11113:   cell var circle-radius
11114: end-class circle \ "graphical" is the parent class
11115: 
11116: :noname ( x y -- )
11117:   circle-radius @@ draw-circle ; circle defines draw
11118: :noname ( r -- )
11119:   circle-radius ! ; circle defines init
11120: @end example
11121: 
11122: There is no implicit init method, so we have to define one. The creation
11123: code of the object now has to call init explicitely.
11124: 
11125: @example
11126: circle new Constant my-circle
11127: 50 my-circle init
11128: @end example
11129: 
11130: It is also possible to add a function to create named objects with
11131: automatic call of @code{init}, given that all objects have @code{init}
11132: on the same place:
11133: 
11134: @example
11135: : new: ( .. o "name" -- )
11136:     new dup Constant init ;
11137: 80 circle new: large-circle
11138: @end example
11139: 
11140: We can draw this new circle at (100,100) with:
11141: 
11142: @example
11143: 100 100 my-circle draw
11144: @end example
11145: 
11146: @node Mini-OOF Implementation,  , Mini-OOF Example, Mini-OOF
11147: @subsubsection @file{mini-oof.fs} Implementation
11148: 
11149: Object-oriented systems with late binding typically use a
11150: ``vtable''-approach: the first variable in each object is a pointer to a
11151: table, which contains the methods as function pointers. The vtable
11152: may also contain other information.
11153: 
11154: So first, let's declare selectors:
11155: 
11156: @example
11157: : method ( m v "name" -- m' v ) Create  over , swap cell+ swap
11158:   DOES> ( ... o -- ... ) @@ over @@ + @@ execute ;
11159: @end example
11160: 
11161: During selector declaration, the number of selectors and instance
11162: variables is on the stack (in address units). @code{method} creates one
11163: selector and increments the selector number. To execute a selector, it
11164: takes the object, fetches the vtable pointer, adds the offset, and
11165: executes the method @i{xt} stored there. Each selector takes the object
11166: it is invoked with as top of stack parameter; it passes the parameters
11167: (including the object) unchanged to the appropriate method which should
11168: consume that object.
11169: 
11170: Now, we also have to declare instance variables
11171: 
11172: @example
11173: : var ( m v size "name" -- m v' ) Create  over , +
11174:   DOES> ( o -- addr ) @@ + ;
11175: @end example
11176: 
11177: As before, a word is created with the current offset. Instance
11178: variables can have different sizes (cells, floats, doubles, chars), so
11179: all we do is take the size and add it to the offset. If your machine
11180: has alignment restrictions, put the proper @code{aligned} or
11181: @code{faligned} before the variable, to adjust the variable
11182: offset. That's why it is on the top of stack.
11183: 
11184: We need a starting point (the base object) and some syntactic sugar:
11185: 
11186: @example
11187: Create object  1 cells , 2 cells ,
11188: : class ( class -- class selectors vars ) dup 2@@ ;
11189: @end example
11190: 
11191: For inheritance, the vtable of the parent object has to be
11192: copied when a new, derived class is declared. This gives all the
11193: methods of the parent class, which can be overridden, though.
11194: 
11195: @example
11196: : end-class  ( class selectors vars "name" -- )
11197:   Create  here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
11198:   cell+ dup cell+ r> rot @@ 2 cells /string move ;
11199: @end example
11200: 
11201: The first line creates the vtable, initialized with
11202: @code{noop}s. The second line is the inheritance mechanism, it
11203: copies the xts from the parent vtable.
11204: 
11205: We still have no way to define new methods, let's do that now:
11206: 
11207: @example
11208: : defines ( xt class "name" -- ) ' >body @@ + ! ;
11209: @end example
11210: 
11211: To allocate a new object, we need a word, too:
11212: 
11213: @example
11214: : new ( class -- o )  here over @@ allot swap over ! ;
11215: @end example
11216: 
11217: Sometimes derived classes want to access the method of the
11218: parent object. There are two ways to achieve this with Mini-OOF:
11219: first, you could use named words, and second, you could look up the
11220: vtable of the parent object.
11221: 
11222: @example
11223: : :: ( class "name" -- ) ' >body @@ + @@ compile, ;
11224: @end example
11225: 
11226: 
11227: Nothing can be more confusing than a good example, so here is
11228: one. First let's declare a text object (called
11229: @code{button}), that stores text and position:
11230: 
11231: @example
11232: object class
11233:   cell var text
11234:   cell var len
11235:   cell var x
11236:   cell var y
11237:   method init
11238:   method draw
11239: end-class button
11240: @end example
11241: 
11242: @noindent
11243: Now, implement the two methods, @code{draw} and @code{init}:
11244: 
11245: @example
11246: :noname ( o -- )
11247:  >r r@@ x @@ r@@ y @@ at-xy  r@@ text @@ r> len @@ type ;
11248:  button defines draw
11249: :noname ( addr u o -- )
11250:  >r 0 r@@ x ! 0 r@@ y ! r@@ len ! r> text ! ;
11251:  button defines init
11252: @end example
11253: 
11254: @noindent
11255: To demonstrate inheritance, we define a class @code{bold-button}, with no
11256: new data and no new selectors:
11257: 
11258: @example
11259: button class
11260: end-class bold-button
11261: 
11262: : bold   27 emit ." [1m" ;
11263: : normal 27 emit ." [0m" ;
11264: @end example
11265: 
11266: @noindent
11267: The class @code{bold-button} has a different draw method to
11268: @code{button}, but the new method is defined in terms of the draw method
11269: for @code{button}:
11270: 
11271: @example
11272: :noname bold [ button :: draw ] normal ; bold-button defines draw
11273: @end example
11274: 
11275: @noindent
11276: Finally, create two objects and apply selectors:
11277: 
11278: @example
11279: button new Constant foo
11280: s" thin foo" foo init
11281: page
11282: foo draw
11283: bold-button new Constant bar
11284: s" fat bar" bar init
11285: 1 bar y !
11286: bar draw
11287: @end example
11288: 
11289: 
11290: @node Comparison with other object models,  , Mini-OOF, Object-oriented Forth
11291: @subsection Comparison with other object models
11292: @cindex comparison of object models
11293: @cindex object models, comparison
11294: 
11295: Many object-oriented Forth extensions have been proposed (@cite{A survey
11296: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
11297: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
11298: relation of the object models described here to two well-known and two
11299: closely-related (by the use of method maps) models.  Andras Zsoter
11300: helped us with this section.
11301: 
11302: @cindex Neon model
11303: The most popular model currently seems to be the Neon model (see
11304: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
11305: 1997) by Andrew McKewan) but this model has a number of limitations
11306: @footnote{A longer version of this critique can be
11307: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
11308: Dimensions, May 1997) by Anton Ertl.}:
11309: 
11310: @itemize @bullet
11311: @item
11312: It uses a @code{@emph{selector object}} syntax, which makes it unnatural
11313: to pass objects on the stack.
11314: 
11315: @item
11316: It requires that the selector parses the input stream (at
11317: compile time); this leads to reduced extensibility and to bugs that are
11318: hard to find.
11319: 
11320: @item
11321: It allows using every selector on every object; this eliminates the
11322: need for interfaces, but makes it harder to create efficient
11323: implementations.
11324: @end itemize
11325: 
11326: @cindex Pountain's object-oriented model
11327: Another well-known publication is @cite{Object-Oriented Forth} (Academic
11328: Press, London, 1987) by Dick Pountain. However, it is not really about
11329: object-oriented programming, because it hardly deals with late
11330: binding. Instead, it focuses on features like information hiding and
11331: overloading that are characteristic of modular languages like Ada (83).
11332: 
11333: @cindex Zsoter's object-oriented model
11334: In @uref{http://www.forth.org/oopf.html, Does late binding have to be
11335: slow?} (Forth Dimensions 18(1) 1996, pages 31-35) Andras Zsoter
11336: describes a model that makes heavy use of an active object (like
11337: @code{this} in @file{objects.fs}): The active object is not only used
11338: for accessing all fields, but also specifies the receiving object of
11339: every selector invocation; you have to change the active object
11340: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
11341: changes more or less implicitly at @code{m: ... ;m}. Such a change at
11342: the method entry point is unnecessary with Zsoter's model, because the
11343: receiving object is the active object already. On the other hand, the
11344: explicit change is absolutely necessary in that model, because otherwise
11345: no one could ever change the active object. An ANS Forth implementation
11346: of this model is available through
11347: @uref{http://www.forth.org/oopf.html}.
11348: 
11349: @cindex @file{oof.fs}, differences to other models
11350: The @file{oof.fs} model combines information hiding and overloading
11351: resolution (by keeping names in various word lists) with object-oriented
11352: programming. It sets the active object implicitly on method entry, but
11353: also allows explicit changing (with @code{>o...o>} or with
11354: @code{with...endwith}). It uses parsing and state-smart objects and
11355: classes for resolving overloading and for early binding: the object or
11356: class parses the selector and determines the method from this. If the
11357: selector is not parsed by an object or class, it performs a call to the
11358: selector for the active object (late binding), like Zsoter's model.
11359: Fields are always accessed through the active object. The big
11360: disadvantage of this model is the parsing and the state-smartness, which
11361: reduces extensibility and increases the opportunities for subtle bugs;
11362: essentially, you are only safe if you never tick or @code{postpone} an
11363: object or class (Bernd disagrees, but I (Anton) am not convinced).
11364: 
11365: @cindex @file{mini-oof.fs}, differences to other models
11366: The @file{mini-oof.fs} model is quite similar to a very stripped-down
11367: version of the @file{objects.fs} model, but syntactically it is a
11368: mixture of the @file{objects.fs} and @file{oof.fs} models.
11369: 
11370: 
11371: @c -------------------------------------------------------------
11372: @node Programming Tools, Assembler and Code Words, Object-oriented Forth, Words
11373: @section Programming Tools
11374: @cindex programming tools
11375: 
11376: @c !! move this and assembler down below OO stuff.
11377: 
11378: @menu
11379: * Examining::                   
11380: * Forgetting words::            
11381: * Debugging::                   Simple and quick.
11382: * Assertions::                  Making your programs self-checking.
11383: * Singlestep Debugger::         Executing your program word by word.
11384: @end menu
11385: 
11386: @node Examining, Forgetting words, Programming Tools, Programming Tools
11387: @subsection Examining data and code
11388: @cindex examining data and code
11389: @cindex data examination
11390: @cindex code examination
11391: 
11392: The following words inspect the stack non-destructively:
11393: 
11394: doc-.s
11395: doc-f.s
11396: 
11397: There is a word @code{.r} but it does @i{not} display the return stack!
11398: It is used for formatted numeric output (@pxref{Simple numeric output}).
11399: 
11400: doc-depth
11401: doc-fdepth
11402: doc-clearstack
11403: 
11404: The following words inspect memory.
11405: 
11406: doc-?
11407: doc-dump
11408: 
11409: And finally, @code{see} allows to inspect code:
11410: 
11411: doc-see
11412: doc-xt-see
11413: 
11414: @node Forgetting words, Debugging, Examining, Programming Tools
11415: @subsection Forgetting words
11416: @cindex words, forgetting
11417: @cindex forgeting words
11418: 
11419: @c  anton: other, maybe better places for this subsection: Defining Words;
11420: @c  Dictionary allocation.  At least a reference should be there.
11421: 
11422: Forth allows you to forget words (and everything that was alloted in the
11423: dictonary after them) in a LIFO manner.
11424: 
11425: doc-marker
11426: 
11427: The most common use of this feature is during progam development: when
11428: you change a source file, forget all the words it defined and load it
11429: again (since you also forget everything defined after the source file
11430: was loaded, you have to reload that, too).  Note that effects like
11431: storing to variables and destroyed system words are not undone when you
11432: forget words.  With a system like Gforth, that is fast enough at
11433: starting up and compiling, I find it more convenient to exit and restart
11434: Gforth, as this gives me a clean slate.
11435: 
11436: Here's an example of using @code{marker} at the start of a source file
11437: that you are debugging; it ensures that you only ever have one copy of
11438: the file's definitions compiled at any time:
11439: 
11440: @example
11441: [IFDEF] my-code
11442:     my-code
11443: [ENDIF]
11444: 
11445: marker my-code
11446: init-included-files
11447: 
11448: \ .. definitions start here
11449: \ .
11450: \ .
11451: \ end
11452: @end example
11453: 
11454: 
11455: @node Debugging, Assertions, Forgetting words, Programming Tools
11456: @subsection Debugging
11457: @cindex debugging
11458: 
11459: Languages with a slow edit/compile/link/test development loop tend to
11460: require sophisticated tracing/stepping debuggers to facilate debugging.
11461: 
11462: A much better (faster) way in fast-compiling languages is to add
11463: printing code at well-selected places, let the program run, look at
11464: the output, see where things went wrong, add more printing code, etc.,
11465: until the bug is found.
11466: 
11467: The simple debugging aids provided in @file{debugs.fs}
11468: are meant to support this style of debugging.
11469: 
11470: The word @code{~~} prints debugging information (by default the source
11471: location and the stack contents). It is easy to insert. If you use Emacs
11472: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
11473: query-replace them with nothing). The deferred words
11474: @code{printdebugdata} and @code{printdebugline} control the output of
11475: @code{~~}. The default source location output format works well with
11476: Emacs' compilation mode, so you can step through the program at the
11477: source level using @kbd{C-x `} (the advantage over a stepping debugger
11478: is that you can step in any direction and you know where the crash has
11479: happened or where the strange data has occurred).
11480: 
11481: doc-~~
11482: doc-printdebugdata
11483: doc-printdebugline
11484: 
11485: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
11486: @subsection Assertions
11487: @cindex assertions
11488: 
11489: It is a good idea to make your programs self-checking, especially if you
11490: make an assumption that may become invalid during maintenance (for
11491: example, that a certain field of a data structure is never zero). Gforth
11492: supports @dfn{assertions} for this purpose. They are used like this:
11493: 
11494: @example
11495: assert( @i{flag} )
11496: @end example
11497: 
11498: The code between @code{assert(} and @code{)} should compute a flag, that
11499: should be true if everything is alright and false otherwise. It should
11500: not change anything else on the stack. The overall stack effect of the
11501: assertion is @code{( -- )}. E.g.
11502: 
11503: @example
11504: assert( 1 1 + 2 = ) \ what we learn in school
11505: assert( dup 0<> ) \ assert that the top of stack is not zero
11506: assert( false ) \ this code should not be reached
11507: @end example
11508: 
11509: The need for assertions is different at different times. During
11510: debugging, we want more checking, in production we sometimes care more
11511: for speed. Therefore, assertions can be turned off, i.e., the assertion
11512: becomes a comment. Depending on the importance of an assertion and the
11513: time it takes to check it, you may want to turn off some assertions and
11514: keep others turned on. Gforth provides several levels of assertions for
11515: this purpose:
11516: 
11517: 
11518: doc-assert0(
11519: doc-assert1(
11520: doc-assert2(
11521: doc-assert3(
11522: doc-assert(
11523: doc-)
11524: 
11525: 
11526: The variable @code{assert-level} specifies the highest assertions that
11527: are turned on. I.e., at the default @code{assert-level} of one,
11528: @code{assert0(} and @code{assert1(} assertions perform checking, while
11529: @code{assert2(} and @code{assert3(} assertions are treated as comments.
11530: 
11531: The value of @code{assert-level} is evaluated at compile-time, not at
11532: run-time. Therefore you cannot turn assertions on or off at run-time;
11533: you have to set the @code{assert-level} appropriately before compiling a
11534: piece of code. You can compile different pieces of code at different
11535: @code{assert-level}s (e.g., a trusted library at level 1 and
11536: newly-written code at level 3).
11537: 
11538: 
11539: doc-assert-level
11540: 
11541: 
11542: If an assertion fails, a message compatible with Emacs' compilation mode
11543: is produced and the execution is aborted (currently with @code{ABORT"}.
11544: If there is interest, we will introduce a special throw code. But if you
11545: intend to @code{catch} a specific condition, using @code{throw} is
11546: probably more appropriate than an assertion).
11547: 
11548: Definitions in ANS Forth for these assertion words are provided
11549: in @file{compat/assert.fs}.
11550: 
11551: 
11552: @node Singlestep Debugger,  , Assertions, Programming Tools
11553: @subsection Singlestep Debugger
11554: @cindex singlestep Debugger
11555: @cindex debugging Singlestep
11556: 
11557: When you create a new word there's often the need to check whether it
11558: behaves correctly or not. You can do this by typing @code{dbg
11559: badword}. A debug session might look like this:
11560: 
11561: @example
11562: : badword 0 DO i . LOOP ;  ok
11563: 2 dbg badword 
11564: : badword  
11565: Scanning code...
11566: 
11567: Nesting debugger ready!
11568: 
11569: 400D4738  8049BC4 0              -> [ 2 ] 00002 00000 
11570: 400D4740  8049F68 DO             -> [ 0 ] 
11571: 400D4744  804A0C8 i              -> [ 1 ] 00000 
11572: 400D4748 400C5E60 .              -> 0 [ 0 ] 
11573: 400D474C  8049D0C LOOP           -> [ 0 ] 
11574: 400D4744  804A0C8 i              -> [ 1 ] 00001 
11575: 400D4748 400C5E60 .              -> 1 [ 0 ] 
11576: 400D474C  8049D0C LOOP           -> [ 0 ] 
11577: 400D4758  804B384 ;              ->  ok
11578: @end example
11579: 
11580: Each line displayed is one step. You always have to hit return to
11581: execute the next word that is displayed. If you don't want to execute
11582: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
11583: an overview what keys are available:
11584: 
11585: @table @i
11586: 
11587: @item @key{RET}
11588: Next; Execute the next word.
11589: 
11590: @item n
11591: Nest; Single step through next word.
11592: 
11593: @item u
11594: Unnest; Stop debugging and execute rest of word. If we got to this word
11595: with nest, continue debugging with the calling word.
11596: 
11597: @item d
11598: Done; Stop debugging and execute rest.
11599: 
11600: @item s
11601: Stop; Abort immediately.
11602: 
11603: @end table
11604: 
11605: Debugging large application with this mechanism is very difficult, because
11606: you have to nest very deeply into the program before the interesting part
11607: begins. This takes a lot of time. 
11608: 
11609: To do it more directly put a @code{BREAK:} command into your source code.
11610: When program execution reaches @code{BREAK:} the single step debugger is
11611: invoked and you have all the features described above.
11612: 
11613: If you have more than one part to debug it is useful to know where the
11614: program has stopped at the moment. You can do this by the 
11615: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
11616: string is typed out when the ``breakpoint'' is reached.
11617: 
11618: 
11619: doc-dbg
11620: doc-break:
11621: doc-break"
11622: 
11623: 
11624: 
11625: @c -------------------------------------------------------------
11626: @node Assembler and Code Words, Threading Words, Programming Tools, Words
11627: @section Assembler and Code Words
11628: @cindex assembler
11629: @cindex code words
11630: 
11631: @menu
11632: * Code and ;code::              
11633: * Common Assembler::            Assembler Syntax
11634: * Common Disassembler::         
11635: * 386 Assembler::               Deviations and special cases
11636: * Alpha Assembler::             Deviations and special cases
11637: * MIPS assembler::              Deviations and special cases
11638: * Other assemblers::            How to write them
11639: @end menu
11640: 
11641: @node Code and ;code, Common Assembler, Assembler and Code Words, Assembler and Code Words
11642: @subsection @code{Code} and @code{;code}
11643: 
11644: Gforth provides some words for defining primitives (words written in
11645: machine code), and for defining the machine-code equivalent of
11646: @code{DOES>}-based defining words. However, the machine-independent
11647: nature of Gforth poses a few problems: First of all, Gforth runs on
11648: several architectures, so it can provide no standard assembler. What's
11649: worse is that the register allocation not only depends on the processor,
11650: but also on the @code{gcc} version and options used.
11651: 
11652: The words that Gforth offers encapsulate some system dependences (e.g.,
11653: the header structure), so a system-independent assembler may be used in
11654: Gforth. If you do not have an assembler, you can compile machine code
11655: directly with @code{,} and @code{c,}@footnote{This isn't portable,
11656: because these words emit stuff in @i{data} space; it works because
11657: Gforth has unified code/data spaces. Assembler isn't likely to be
11658: portable anyway.}.
11659: 
11660: 
11661: doc-assembler
11662: doc-init-asm
11663: doc-code
11664: doc-end-code
11665: doc-;code
11666: doc-flush-icache
11667: 
11668: 
11669: If @code{flush-icache} does not work correctly, @code{code} words
11670: etc. will not work (reliably), either.
11671: 
11672: The typical usage of these @code{code} words can be shown most easily by
11673: analogy to the equivalent high-level defining words:
11674: 
11675: @example
11676: : foo                              code foo
11677:    <high-level Forth words>              <assembler>
11678: ;                                  end-code
11679:                                 
11680: : bar                              : bar
11681:    <high-level Forth words>           <high-level Forth words>
11682:    CREATE                             CREATE
11683:       <high-level Forth words>           <high-level Forth words>
11684:    DOES>                              ;code
11685:       <high-level Forth words>           <assembler>
11686: ;                                  end-code
11687: @end example
11688: 
11689: @c anton: the following stuff is also in "Common Assembler", in less detail.
11690: 
11691: @cindex registers of the inner interpreter
11692: In the assembly code you will want to refer to the inner interpreter's
11693: registers (e.g., the data stack pointer) and you may want to use other
11694: registers for temporary storage. Unfortunately, the register allocation
11695: is installation-dependent.
11696: 
11697: In particular, @code{ip} (Forth instruction pointer) and @code{rp}
11698: (return stack pointer) are in different places in @code{gforth} and
11699: @code{gforth-fast}.  This means that you cannot write a @code{NEXT}
11700: routine that works on both versions; so for doing @code{NEXT}, I
11701: recomment jumping to @code{' noop >code-address}, which contains nothing
11702: but a @code{NEXT}.
11703: 
11704: For general accesses to the inner interpreter's registers, the easiest
11705: solution is to use explicit register declarations (@pxref{Explicit Reg
11706: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) for
11707: all of the inner interpreter's registers: You have to compile Gforth
11708: with @code{-DFORCE_REG} (configure option @code{--enable-force-reg}) and
11709: the appropriate declarations must be present in the @code{machine.h}
11710: file (see @code{mips.h} for an example; you can find a full list of all
11711: declarable register symbols with @code{grep register engine.c}). If you
11712: give explicit registers to all variables that are declared at the
11713: beginning of @code{engine()}, you should be able to use the other
11714: caller-saved registers for temporary storage. Alternatively, you can use
11715: the @code{gcc} option @code{-ffixed-REG} (@pxref{Code Gen Options, ,
11716: Options for Code Generation Conventions, gcc.info, GNU C Manual}) to
11717: reserve a register (however, this restriction on register allocation may
11718: slow Gforth significantly).
11719: 
11720: If this solution is not viable (e.g., because @code{gcc} does not allow
11721: you to explicitly declare all the registers you need), you have to find
11722: out by looking at the code where the inner interpreter's registers
11723: reside and which registers can be used for temporary storage. You can
11724: get an assembly listing of the engine's code with @code{make engine.s}.
11725: 
11726: In any case, it is good practice to abstract your assembly code from the
11727: actual register allocation. E.g., if the data stack pointer resides in
11728: register @code{$17}, create an alias for this register called @code{sp},
11729: and use that in your assembly code.
11730: 
11731: @cindex code words, portable
11732: Another option for implementing normal and defining words efficiently
11733: is to add the desired functionality to the source of Gforth. For normal
11734: words you just have to edit @file{primitives} (@pxref{Automatic
11735: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
11736: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
11737: @file{prims2x.fs}, and possibly @file{cross.fs}.
11738: 
11739: @node Common Assembler, Common Disassembler, Code and ;code, Assembler and Code Words
11740: @subsection Common Assembler
11741: 
11742: The assemblers in Gforth generally use a postfix syntax, i.e., the
11743: instruction name follows the operands.
11744: 
11745: The operands are passed in the usual order (the same that is used in the
11746: manual of the architecture).  Since they all are Forth words, they have
11747: to be separated by spaces; you can also use Forth words to compute the
11748: operands.
11749: 
11750: The instruction names usually end with a @code{,}.  This makes it easier
11751: to visually separate instructions if you put several of them on one
11752: line; it also avoids shadowing other Forth words (e.g., @code{and}).
11753: 
11754: Registers are usually specified by number; e.g., (decimal) @code{11}
11755: specifies registers R11 and F11 on the Alpha architecture (which one,
11756: depends on the instruction).  The usual names are also available, e.g.,
11757: @code{s2} for R11 on Alpha.
11758: 
11759: Control flow is specified similar to normal Forth code (@pxref{Arbitrary
11760: control structures}), with @code{if,}, @code{ahead,}, @code{then,},
11761: @code{begin,}, @code{until,}, @code{again,}, @code{cs-roll},
11762: @code{cs-pick}, @code{else,}, @code{while,}, and @code{repeat,}.  The
11763: conditions are specified in a way specific to each assembler.
11764: 
11765: Note that the register assignments of the Gforth engine can change
11766: between Gforth versions, or even between different compilations of the
11767: same Gforth version (e.g., if you use a different GCC version).  So if
11768: you want to refer to Gforth's registers (e.g., the stack pointer or
11769: TOS), I recommend defining your own words for refering to these
11770: registers, and using them later on; then you can easily adapt to a
11771: changed register assignment.  The stability of the register assignment
11772: is usually better if you build Gforth with @code{--enable-force-reg}.
11773: 
11774: In particular, the return stack pointer and the instruction pointer are
11775: in memory in @code{gforth}, and usually in registers in
11776: @code{gforth-fast}.  The most common use of these registers is to
11777: dispatch to the next word (the @code{next} routine).  A portable way to
11778: do this is to jump to @code{' noop >code-address} (of course, this is
11779: less efficient than integrating the @code{next} code and scheduling it
11780: well).
11781: 
11782: @node  Common Disassembler, 386 Assembler, Common Assembler, Assembler and Code Words
11783: @subsection Common Disassembler
11784: 
11785: You can disassemble a @code{code} word with @code{see}
11786: (@pxref{Debugging}).  You can disassemble a section of memory with
11787: 
11788: doc-disasm
11789: 
11790: The disassembler generally produces output that can be fed into the
11791: assembler (i.e., same syntax, etc.).  It also includes additional
11792: information in comments.  In particular, the address of the instruction
11793: is given in a comment before the instruction.
11794: 
11795: @code{See} may display more or less than the actual code of the word,
11796: because the recognition of the end of the code is unreliable.  You can
11797: use @code{disasm} if it did not display enough.  It may display more, if
11798: the code word is not immediately followed by a named word.  If you have
11799: something else there, you can follow the word with @code{align last @ ,}
11800: to ensure that the end is recognized.
11801: 
11802: @node 386 Assembler, Alpha Assembler, Common Disassembler, Assembler and Code Words
11803: @subsection 386 Assembler
11804: 
11805: The 386 assembler included in Gforth was written by Bernd Paysan, it's
11806: available under GPL, and originally part of bigFORTH.
11807: 
11808: The 386 disassembler included in Gforth was written by Andrew McKewan
11809: and is in the public domain.
11810: 
11811: The disassembler displays code in prefix Intel syntax.
11812: 
11813: The assembler uses a postfix syntax with reversed parameters.
11814: 
11815: The assembler includes all instruction of the Athlon, i.e. 486 core
11816: instructions, Pentium and PPro extensions, floating point, MMX, 3Dnow!,
11817: but not ISSE. It's an integrated 16- and 32-bit assembler. Default is 32
11818: bit, you can switch to 16 bit with .86 and back to 32 bit with .386.
11819: 
11820: There are several prefixes to switch between different operation sizes,
11821: @code{.b} for byte accesses, @code{.w} for word accesses, @code{.d} for
11822: double-word accesses. Addressing modes can be switched with @code{.wa}
11823: for 16 bit addresses, and @code{.da} for 32 bit addresses. You don't
11824: need a prefix for byte register names (@code{AL} et al).
11825: 
11826: For floating point operations, the prefixes are @code{.fs} (IEEE
11827: single), @code{.fl} (IEEE double), @code{.fx} (extended), @code{.fw}
11828: (word), @code{.fd} (double-word), and @code{.fq} (quad-word).
11829: 
11830: The MMX opcodes don't have size prefixes, they are spelled out like in
11831: the Intel assembler. Instead of move from and to memory, there are
11832: PLDQ/PLDD and PSTQ/PSTD.
11833: 
11834: The registers lack the 'e' prefix; even in 32 bit mode, eax is called
11835: ax.  Immediate values are indicated by postfixing them with @code{#},
11836: e.g., @code{3 #}.  Here are some examples of addressing modes:
11837: 
11838: @example
11839: 3 #          \ immediate
11840: 1000 #)      \ absolute
11841: ax           \ register
11842: 100 di d)    \ 100[edi]
11843: 4 bx cx di)  \ 4[ebx][ecx]
11844: di ax *4 i)  \ [edi][eax*4]
11845: 20 ax *4 i#) \ 20[eax*4]
11846: @end example
11847: 
11848: Some example of instructions are:
11849: 
11850: @example
11851: ax bx mov             \ move ebx,eax
11852: 3 # ax mov            \ mov eax,3
11853: 100 di ) ax mov       \ mov eax,100[edi]
11854: 4 bx cx di) ax mov    \ mov eax,4[ebx][ecx]
11855: .w ax bx mov          \ mov bx,ax
11856: @end example
11857: 
11858: The following forms are supported for binary instructions:
11859: 
11860: @example
11861: <reg> <reg> <inst>
11862: <n> # <reg> <inst>
11863: <mem> <reg> <inst>
11864: <reg> <mem> <inst>
11865: @end example
11866: 
11867: Immediate to memory is not supported.  The shift/rotate syntax is:
11868: 
11869: @example
11870: <reg/mem> 1 # shl \ shortens to shift without immediate
11871: <reg/mem> 4 # shl
11872: <reg/mem> cl shl
11873: @end example
11874: 
11875: Precede string instructions (@code{movs} etc.) with @code{.b} to get
11876: the byte version.
11877: 
11878: The control structure words @code{IF} @code{UNTIL} etc. must be preceded
11879: by one of these conditions: @code{vs vc u< u>= 0= 0<> u<= u> 0< 0>= ps
11880: pc < >= <= >}. (Note that most of these words shadow some Forth words
11881: when @code{assembler} is in front of @code{forth} in the search path,
11882: e.g., in @code{code} words).  Currently the control structure words use
11883: one stack item, so you have to use @code{roll} instead of @code{cs-roll}
11884: to shuffle them (you can also use @code{swap} etc.).
11885: 
11886: Here is an example of a @code{code} word (assumes that the stack pointer
11887: is in esi and the TOS is in ebx):
11888: 
11889: @example
11890: code my+ ( n1 n2 -- n )
11891:     4 si D) bx add
11892:     4 # si add
11893:     Next
11894: end-code
11895: @end example
11896: 
11897: @node Alpha Assembler, MIPS assembler, 386 Assembler, Assembler and Code Words
11898: @subsection Alpha Assembler
11899: 
11900: The Alpha assembler and disassembler were originally written by Bernd
11901: Thallner.
11902: 
11903: The register names @code{a0}--@code{a5} are not available to avoid
11904: shadowing hex numbers.
11905: 
11906: Immediate forms of arithmetic instructions are distinguished by a
11907: @code{#} just before the @code{,}, e.g., @code{and#,} (note: @code{lda,}
11908: does not count as arithmetic instruction).
11909: 
11910: You have to specify all operands to an instruction, even those that
11911: other assemblers consider optional, e.g., the destination register for
11912: @code{br,}, or the destination register and hint for @code{jmp,}.
11913: 
11914: You can specify conditions for @code{if,} by removing the first @code{b}
11915: and the trailing @code{,} from a branch with a corresponding name; e.g.,
11916: 
11917: @example
11918: 11 fgt if, \ if F11>0e
11919:   ...
11920: endif,
11921: @end example
11922: 
11923: @code{fbgt,} gives @code{fgt}.  
11924: 
11925: @node MIPS assembler, Other assemblers, Alpha Assembler, Assembler and Code Words
11926: @subsection MIPS assembler
11927: 
11928: The MIPS assembler was originally written by Christian Pirker.
11929: 
11930: Currently the assembler and disassembler only cover the MIPS-I
11931: architecture (R3000), and don't support FP instructions.
11932: 
11933: The register names @code{$a0}--@code{$a3} are not available to avoid
11934: shadowing hex numbers.
11935: 
11936: Because there is no way to distinguish registers from immediate values,
11937: you have to explicitly use the immediate forms of instructions, i.e.,
11938: @code{addiu,}, not just @code{addu,} (@command{as} does this
11939: implicitly).
11940: 
11941: If the architecture manual specifies several formats for the instruction
11942: (e.g., for @code{jalr,}), you usually have to use the one with more
11943: arguments (i.e., two for @code{jalr,}).  When in doubt, see
11944: @code{arch/mips/testasm.fs} for an example of correct use.
11945: 
11946: Branches and jumps in the MIPS architecture have a delay slot.  You have
11947: to fill it yourself (the simplest way is to use @code{nop,}), the
11948: assembler does not do it for you (unlike @command{as}).  Even
11949: @code{if,}, @code{ahead,}, @code{until,}, @code{again,}, @code{while,},
11950: @code{else,} and @code{repeat,} need a delay slot.  Since @code{begin,}
11951: and @code{then,} just specify branch targets, they are not affected.
11952: 
11953: Note that you must not put branches, jumps, or @code{li,} into the delay
11954: slot: @code{li,} may expand to several instructions, and control flow
11955: instructions may not be put into the branch delay slot in any case.
11956: 
11957: For branches the argument specifying the target is a relative address;
11958: You have to add the address of the delay slot to get the absolute
11959: address.
11960: 
11961: The MIPS architecture also has load delay slots and restrictions on
11962: using @code{mfhi,} and @code{mflo,}; you have to order the instructions
11963: yourself to satisfy these restrictions, the assembler does not do it for
11964: you.
11965: 
11966: You can specify the conditions for @code{if,} etc. by taking a
11967: conditional branch and leaving away the @code{b} at the start and the
11968: @code{,} at the end.  E.g.,
11969: 
11970: @example
11971: 4 5 eq if,
11972:   ... \ do something if $4 equals $5
11973: then,
11974: @end example
11975: 
11976: @node Other assemblers,  , MIPS assembler, Assembler and Code Words
11977: @subsection Other assemblers
11978: 
11979: If you want to contribute another assembler/disassembler, please contact
11980: us (@email{bug-gforth@@gnu.org}) to check if we have such an assembler
11981: already.  If you are writing them from scratch, please use a similar
11982: syntax style as the one we use (i.e., postfix, commas at the end of the
11983: instruction names, @pxref{Common Assembler}); make the output of the
11984: disassembler be valid input for the assembler, and keep the style
11985: similar to the style we used.
11986: 
11987: Hints on implementation: The most important part is to have a good test
11988: suite that contains all instructions.  Once you have that, the rest is
11989: easy.  For actual coding you can take a look at
11990: @file{arch/mips/disasm.fs} to get some ideas on how to use data for both
11991: the assembler and disassembler, avoiding redundancy and some potential
11992: bugs.  You can also look at that file (and @pxref{Advanced does> usage
11993: example}) to get ideas how to factor a disassembler.
11994: 
11995: Start with the disassembler, because it's easier to reuse data from the
11996: disassembler for the assembler than the other way round.
11997: 
11998: For the assembler, take a look at @file{arch/alpha/asm.fs}, which shows
11999: how simple it can be.
12000: 
12001: @c -------------------------------------------------------------
12002: @node Threading Words, Passing Commands to the OS, Assembler and Code Words, Words
12003: @section Threading Words
12004: @cindex threading words
12005: 
12006: @cindex code address
12007: These words provide access to code addresses and other threading stuff
12008: in Gforth (and, possibly, other interpretive Forths). It more or less
12009: abstracts away the differences between direct and indirect threading
12010: (and, for direct threading, the machine dependences). However, at
12011: present this wordset is still incomplete. It is also pretty low-level;
12012: some day it will hopefully be made unnecessary by an internals wordset
12013: that abstracts implementation details away completely.
12014: 
12015: The terminology used here stems from indirect threaded Forth systems; in
12016: such a system, the XT of a word is represented by the CFA (code field
12017: address) of a word; the CFA points to a cell that contains the code
12018: address.  The code address is the address of some machine code that
12019: performs the run-time action of invoking the word (e.g., the
12020: @code{dovar:} routine pushes the address of the body of the word (a
12021: variable) on the stack
12022: ).
12023: 
12024: @cindex code address
12025: @cindex code field address
12026: In an indirect threaded Forth, you can get the code address of @i{name}
12027: with @code{' @i{name} @@}; in Gforth you can get it with @code{' @i{name}
12028: >code-address}, independent of the threading method.
12029: 
12030: doc-threading-method
12031: doc->code-address
12032: doc-code-address!
12033: 
12034: @cindex @code{does>}-handler
12035: @cindex @code{does>}-code
12036: For a word defined with @code{DOES>}, the code address usually points to
12037: a jump instruction (the @dfn{does-handler}) that jumps to the dodoes
12038: routine (in Gforth on some platforms, it can also point to the dodoes
12039: routine itself).  What you are typically interested in, though, is
12040: whether a word is a @code{DOES>}-defined word, and what Forth code it
12041: executes; @code{>does-code} tells you that.
12042: 
12043: doc->does-code
12044: 
12045: To create a @code{DOES>}-defined word with the following basic words,
12046: you have to set up a @code{DOES>}-handler with @code{does-handler!};
12047: @code{/does-handler} aus behind you have to place your executable Forth
12048: code.  Finally you have to create a word and modify its behaviour with
12049: @code{does-handler!}.
12050: 
12051: doc-does-code!
12052: doc-does-handler!
12053: doc-/does-handler
12054: 
12055: The code addresses produced by various defining words are produced by
12056: the following words:
12057: 
12058: doc-docol:
12059: doc-docon:
12060: doc-dovar:
12061: doc-douser:
12062: doc-dodefer:
12063: doc-dofield:
12064: 
12065: @c -------------------------------------------------------------
12066: @node Passing Commands to the OS, Keeping track of Time, Threading Words, Words
12067: @section Passing Commands to the Operating System
12068: @cindex operating system - passing commands
12069: @cindex shell commands
12070: 
12071: Gforth allows you to pass an arbitrary string to the host operating
12072: system shell (if such a thing exists) for execution.
12073: 
12074: 
12075: doc-sh
12076: doc-system
12077: doc-$?
12078: doc-getenv
12079: 
12080: 
12081: @c -------------------------------------------------------------
12082: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
12083: @section Keeping track of Time
12084: @cindex time-related words
12085: 
12086: doc-ms
12087: doc-time&date
12088: doc-utime
12089: doc-cputime
12090: 
12091: 
12092: @c -------------------------------------------------------------
12093: @node Miscellaneous Words,  , Keeping track of Time, Words
12094: @section Miscellaneous Words
12095: @cindex miscellaneous words
12096: 
12097: @comment TODO find homes for these
12098: 
12099: These section lists the ANS Forth words that are not documented
12100: elsewhere in this manual. Ultimately, they all need proper homes.
12101: 
12102: doc-quit
12103: 
12104: The following ANS Forth words are not currently supported by Gforth 
12105: (@pxref{ANS conformance}):
12106: 
12107: @code{EDITOR} 
12108: @code{EMIT?} 
12109: @code{FORGET} 
12110: 
12111: @c ******************************************************************
12112: @node Error messages, Tools, Words, Top
12113: @chapter Error messages
12114: @cindex error messages
12115: @cindex backtrace
12116: 
12117: A typical Gforth error message looks like this:
12118: 
12119: @example
12120: in file included from \evaluated string/:-1
12121: in file included from ./yyy.fs:1
12122: ./xxx.fs:4: Invalid memory address
12123: bar
12124: ^^^
12125: Backtrace:
12126: $400E664C @@
12127: $400E6664 foo
12128: @end example
12129: 
12130: The message identifying the error is @code{Invalid memory address}.  The
12131: error happened when text-interpreting line 4 of the file
12132: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
12133: word on the line where the error happened, is pointed out (with
12134: @code{^^^}).
12135: 
12136: The file containing the error was included in line 1 of @file{./yyy.fs},
12137: and @file{yyy.fs} was included from a non-file (in this case, by giving
12138: @file{yyy.fs} as command-line parameter to Gforth).
12139: 
12140: At the end of the error message you find a return stack dump that can be
12141: interpreted as a backtrace (possibly empty). On top you find the top of
12142: the return stack when the @code{throw} happened, and at the bottom you
12143: find the return stack entry just above the return stack of the topmost
12144: text interpreter.
12145: 
12146: To the right of most return stack entries you see a guess for the word
12147: that pushed that return stack entry as its return address. This gives a
12148: backtrace. In our case we see that @code{bar} called @code{foo}, and
12149: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
12150: address} exception).
12151: 
12152: Note that the backtrace is not perfect: We don't know which return stack
12153: entries are return addresses (so we may get false positives); and in
12154: some cases (e.g., for @code{abort"}) we cannot determine from the return
12155: address the word that pushed the return address, so for some return
12156: addresses you see no names in the return stack dump.
12157: 
12158: @cindex @code{catch} and backtraces
12159: The return stack dump represents the return stack at the time when a
12160: specific @code{throw} was executed.  In programs that make use of
12161: @code{catch}, it is not necessarily clear which @code{throw} should be
12162: used for the return stack dump (e.g., consider one @code{throw} that
12163: indicates an error, which is caught, and during recovery another error
12164: happens; which @code{throw} should be used for the stack dump?).  Gforth
12165: presents the return stack dump for the first @code{throw} after the last
12166: executed (not returned-to) @code{catch}; this works well in the usual
12167: case.
12168: 
12169: @cindex @code{gforth-fast} and backtraces
12170: @cindex @code{gforth-fast}, difference from @code{gforth}
12171: @cindex backtraces with @code{gforth-fast}
12172: @cindex return stack dump with @code{gforth-fast}
12173: @code{Gforth} is able to do a return stack dump for throws generated
12174: from primitives (e.g., invalid memory address, stack empty etc.);
12175: @code{gforth-fast} is only able to do a return stack dump from a
12176: directly called @code{throw} (including @code{abort} etc.).  This is the
12177: only difference (apart from a speed factor of between 1.15 (K6-2) and
12178: 2 (21264)) between @code{gforth} and @code{gforth-fast}.  Given an
12179: exception caused by a primitive in @code{gforth-fast}, you will
12180: typically see no return stack dump at all; however, if the exception is
12181: caught by @code{catch} (e.g., for restoring some state), and then
12182: @code{throw}n again, the return stack dump will be for the first such
12183: @code{throw}.
12184: 
12185: @c ******************************************************************
12186: @node Tools, ANS conformance, Error messages, Top
12187: @chapter Tools
12188: 
12189: @menu
12190: * ANS Report::                  Report the words used, sorted by wordset.
12191: @end menu
12192: 
12193: See also @ref{Emacs and Gforth}.
12194: 
12195: @node ANS Report,  , Tools, Tools
12196: @section @file{ans-report.fs}: Report the words used, sorted by wordset
12197: @cindex @file{ans-report.fs}
12198: @cindex report the words used in your program
12199: @cindex words used in your program
12200: 
12201: If you want to label a Forth program as ANS Forth Program, you must
12202: document which wordsets the program uses; for extension wordsets, it is
12203: helpful to list the words the program requires from these wordsets
12204: (because Forth systems are allowed to provide only some words of them).
12205: 
12206: The @file{ans-report.fs} tool makes it easy for you to determine which
12207: words from which wordset and which non-ANS words your application
12208: uses. You simply have to include @file{ans-report.fs} before loading the
12209: program you want to check. After loading your program, you can get the
12210: report with @code{print-ans-report}. A typical use is to run this as
12211: batch job like this:
12212: @example
12213: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
12214: @end example
12215: 
12216: The output looks like this (for @file{compat/control.fs}):
12217: @example
12218: The program uses the following words
12219: from CORE :
12220: : POSTPONE THEN ; immediate ?dup IF 0= 
12221: from BLOCK-EXT :
12222: \ 
12223: from FILE :
12224: ( 
12225: @end example
12226: 
12227: @subsection Caveats
12228: 
12229: Note that @file{ans-report.fs} just checks which words are used, not whether
12230: they are used in an ANS Forth conforming way!
12231: 
12232: Some words are defined in several wordsets in the
12233: standard. @file{ans-report.fs} reports them for only one of the
12234: wordsets, and not necessarily the one you expect. It depends on usage
12235: which wordset is the right one to specify. E.g., if you only use the
12236: compilation semantics of @code{S"}, it is a Core word; if you also use
12237: its interpretation semantics, it is a File word.
12238: 
12239: @c ******************************************************************
12240: @node ANS conformance, Standard vs Extensions, Tools, Top
12241: @chapter ANS conformance
12242: @cindex ANS conformance of Gforth
12243: 
12244: To the best of our knowledge, Gforth is an
12245: 
12246: ANS Forth System
12247: @itemize @bullet
12248: @item providing the Core Extensions word set
12249: @item providing the Block word set
12250: @item providing the Block Extensions word set
12251: @item providing the Double-Number word set
12252: @item providing the Double-Number Extensions word set
12253: @item providing the Exception word set
12254: @item providing the Exception Extensions word set
12255: @item providing the Facility word set
12256: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
12257: @item providing the File Access word set
12258: @item providing the File Access Extensions word set
12259: @item providing the Floating-Point word set
12260: @item providing the Floating-Point Extensions word set
12261: @item providing the Locals word set
12262: @item providing the Locals Extensions word set
12263: @item providing the Memory-Allocation word set
12264: @item providing the Memory-Allocation Extensions word set (that one's easy)
12265: @item providing the Programming-Tools word set
12266: @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
12267: @item providing the Search-Order word set
12268: @item providing the Search-Order Extensions word set
12269: @item providing the String word set
12270: @item providing the String Extensions word set (another easy one)
12271: @end itemize
12272: 
12273: @cindex system documentation
12274: In addition, ANS Forth systems are required to document certain
12275: implementation choices. This chapter tries to meet these
12276: requirements. In many cases it gives a way to ask the system for the
12277: information instead of providing the information directly, in
12278: particular, if the information depends on the processor, the operating
12279: system or the installation options chosen, or if they are likely to
12280: change during the maintenance of Gforth.
12281: 
12282: @comment The framework for the rest has been taken from pfe.
12283: 
12284: @menu
12285: * The Core Words::              
12286: * The optional Block word set::  
12287: * The optional Double Number word set::  
12288: * The optional Exception word set::  
12289: * The optional Facility word set::  
12290: * The optional File-Access word set::  
12291: * The optional Floating-Point word set::  
12292: * The optional Locals word set::  
12293: * The optional Memory-Allocation word set::  
12294: * The optional Programming-Tools word set::  
12295: * The optional Search-Order word set::  
12296: @end menu
12297: 
12298: 
12299: @c =====================================================================
12300: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
12301: @comment  node-name,  next,  previous,  up
12302: @section The Core Words
12303: @c =====================================================================
12304: @cindex core words, system documentation
12305: @cindex system documentation, core words
12306: 
12307: @menu
12308: * core-idef::                   Implementation Defined Options                   
12309: * core-ambcond::                Ambiguous Conditions                
12310: * core-other::                  Other System Documentation                  
12311: @end menu
12312: 
12313: @c ---------------------------------------------------------------------
12314: @node core-idef, core-ambcond, The Core Words, The Core Words
12315: @subsection Implementation Defined Options
12316: @c ---------------------------------------------------------------------
12317: @cindex core words, implementation-defined options
12318: @cindex implementation-defined options, core words
12319: 
12320: 
12321: @table @i
12322: @item (Cell) aligned addresses:
12323: @cindex cell-aligned addresses
12324: @cindex aligned addresses
12325: processor-dependent. Gforth's alignment words perform natural alignment
12326: (e.g., an address aligned for a datum of size 8 is divisible by
12327: 8). Unaligned accesses usually result in a @code{-23 THROW}.
12328: 
12329: @item @code{EMIT} and non-graphic characters:
12330: @cindex @code{EMIT} and non-graphic characters
12331: @cindex non-graphic characters and @code{EMIT}
12332: The character is output using the C library function (actually, macro)
12333: @code{putc}.
12334: 
12335: @item character editing of @code{ACCEPT} and @code{EXPECT}:
12336: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
12337: @cindex editing in @code{ACCEPT} and @code{EXPECT}
12338: @cindex @code{ACCEPT}, editing
12339: @cindex @code{EXPECT}, editing
12340: This is modeled on the GNU readline library (@pxref{Readline
12341: Interaction, , Command Line Editing, readline, The GNU Readline
12342: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
12343: producing a full word completion every time you type it (instead of
12344: producing the common prefix of all completions). @xref{Command-line editing}.
12345: 
12346: @item character set:
12347: @cindex character set
12348: The character set of your computer and display device. Gforth is
12349: 8-bit-clean (but some other component in your system may make trouble).
12350: 
12351: @item Character-aligned address requirements:
12352: @cindex character-aligned address requirements
12353: installation-dependent. Currently a character is represented by a C
12354: @code{unsigned char}; in the future we might switch to @code{wchar_t}
12355: (Comments on that requested).
12356: 
12357: @item character-set extensions and matching of names:
12358: @cindex character-set extensions and matching of names
12359: @cindex case-sensitivity for name lookup
12360: @cindex name lookup, case-sensitivity
12361: @cindex locale and case-sensitivity
12362: Any character except the ASCII NUL character can be used in a
12363: name. Matching is case-insensitive (except in @code{TABLE}s). The
12364: matching is performed using the C library function @code{strncasecmp}, whose
12365: function is probably influenced by the locale. E.g., the @code{C} locale
12366: does not know about accents and umlauts, so they are matched
12367: case-sensitively in that locale. For portability reasons it is best to
12368: write programs such that they work in the @code{C} locale. Then one can
12369: use libraries written by a Polish programmer (who might use words
12370: containing ISO Latin-2 encoded characters) and by a French programmer
12371: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
12372: funny results for some of the words (which ones, depends on the font you
12373: are using)). Also, the locale you prefer may not be available in other
12374: operating systems. Hopefully, Unicode will solve these problems one day.
12375: 
12376: @item conditions under which control characters match a space delimiter:
12377: @cindex space delimiters
12378: @cindex control characters as delimiters
12379: If @code{WORD} is called with the space character as a delimiter, all
12380: white-space characters (as identified by the C macro @code{isspace()})
12381: are delimiters. @code{PARSE}, on the other hand, treats space like other
12382: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
12383: like @code{PARSE} otherwise. @code{Name}, which is used by the outer
12384: interpreter (aka text interpreter) by default, treats all white-space
12385: characters as delimiters.
12386: 
12387: @item format of the control-flow stack:
12388: @cindex control-flow stack, format
12389: The data stack is used as control-flow stack. The size of a control-flow
12390: stack item in cells is given by the constant @code{cs-item-size}. At the
12391: time of this writing, an item consists of a (pointer to a) locals list
12392: (third), an address in the code (second), and a tag for identifying the
12393: item (TOS). The following tags are used: @code{defstart},
12394: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
12395: @code{scopestart}.
12396: 
12397: @item conversion of digits > 35
12398: @cindex digits > 35
12399: The characters @code{[\]^_'} are the digits with the decimal value
12400: 36@minus{}41. There is no way to input many of the larger digits.
12401: 
12402: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
12403: @cindex @code{EXPECT}, display after end of input
12404: @cindex @code{ACCEPT}, display after end of input
12405: The cursor is moved to the end of the entered string. If the input is
12406: terminated using the @kbd{Return} key, a space is typed.
12407: 
12408: @item exception abort sequence of @code{ABORT"}:
12409: @cindex exception abort sequence of @code{ABORT"}
12410: @cindex @code{ABORT"}, exception abort sequence
12411: The error string is stored into the variable @code{"error} and a
12412: @code{-2 throw} is performed.
12413: 
12414: @item input line terminator:
12415: @cindex input line terminator
12416: @cindex line terminator on input
12417: @cindex newline character on input
12418: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
12419: lines. One of these characters is typically produced when you type the
12420: @kbd{Enter} or @kbd{Return} key.
12421: 
12422: @item maximum size of a counted string:
12423: @cindex maximum size of a counted string
12424: @cindex counted string, maximum size
12425: @code{s" /counted-string" environment? drop .}. Currently 255 characters
12426: on all platforms, but this may change.
12427: 
12428: @item maximum size of a parsed string:
12429: @cindex maximum size of a parsed string
12430: @cindex parsed string, maximum size
12431: Given by the constant @code{/line}. Currently 255 characters.
12432: 
12433: @item maximum size of a definition name, in characters:
12434: @cindex maximum size of a definition name, in characters
12435: @cindex name, maximum length
12436: 31
12437: 
12438: @item maximum string length for @code{ENVIRONMENT?}, in characters:
12439: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
12440: @cindex @code{ENVIRONMENT?} string length, maximum
12441: 31
12442: 
12443: @item method of selecting the user input device:
12444: @cindex user input device, method of selecting
12445: The user input device is the standard input. There is currently no way to
12446: change it from within Gforth. However, the input can typically be
12447: redirected in the command line that starts Gforth.
12448: 
12449: @item method of selecting the user output device:
12450: @cindex user output device, method of selecting
12451: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
12452: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
12453: output when the user output device is a terminal, otherwise the output
12454: is buffered.
12455: 
12456: @item methods of dictionary compilation:
12457: What are we expected to document here?
12458: 
12459: @item number of bits in one address unit:
12460: @cindex number of bits in one address unit
12461: @cindex address unit, size in bits
12462: @code{s" address-units-bits" environment? drop .}. 8 in all current
12463: platforms.
12464: 
12465: @item number representation and arithmetic:
12466: @cindex number representation and arithmetic
12467: Processor-dependent. Binary two's complement on all current platforms.
12468: 
12469: @item ranges for integer types:
12470: @cindex ranges for integer types
12471: @cindex integer types, ranges
12472: Installation-dependent. Make environmental queries for @code{MAX-N},
12473: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
12474: unsigned (and positive) types is 0. The lower bound for signed types on
12475: two's complement and one's complement machines machines can be computed
12476: by adding 1 to the upper bound.
12477: 
12478: @item read-only data space regions:
12479: @cindex read-only data space regions
12480: @cindex data-space, read-only regions
12481: The whole Forth data space is writable.
12482: 
12483: @item size of buffer at @code{WORD}:
12484: @cindex size of buffer at @code{WORD}
12485: @cindex @code{WORD} buffer size
12486: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12487: shared with the pictured numeric output string. If overwriting
12488: @code{PAD} is acceptable, it is as large as the remaining dictionary
12489: space, although only as much can be sensibly used as fits in a counted
12490: string.
12491: 
12492: @item size of one cell in address units:
12493: @cindex cell size
12494: @code{1 cells .}.
12495: 
12496: @item size of one character in address units:
12497: @cindex char size
12498: @code{1 chars .}. 1 on all current platforms.
12499: 
12500: @item size of the keyboard terminal buffer:
12501: @cindex size of the keyboard terminal buffer
12502: @cindex terminal buffer, size
12503: Varies. You can determine the size at a specific time using @code{lp@@
12504: tib - .}. It is shared with the locals stack and TIBs of files that
12505: include the current file. You can change the amount of space for TIBs
12506: and locals stack at Gforth startup with the command line option
12507: @code{-l}.
12508: 
12509: @item size of the pictured numeric output buffer:
12510: @cindex size of the pictured numeric output buffer
12511: @cindex pictured numeric output buffer, size
12512: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12513: shared with @code{WORD}.
12514: 
12515: @item size of the scratch area returned by @code{PAD}:
12516: @cindex size of the scratch area returned by @code{PAD}
12517: @cindex @code{PAD} size
12518: The remainder of dictionary space. @code{unused pad here - - .}.
12519: 
12520: @item system case-sensitivity characteristics:
12521: @cindex case-sensitivity characteristics
12522: Dictionary searches are case-insensitive (except in
12523: @code{TABLE}s). However, as explained above under @i{character-set
12524: extensions}, the matching for non-ASCII characters is determined by the
12525: locale you are using. In the default @code{C} locale all non-ASCII
12526: characters are matched case-sensitively.
12527: 
12528: @item system prompt:
12529: @cindex system prompt
12530: @cindex prompt
12531: @code{ ok} in interpret state, @code{ compiled} in compile state.
12532: 
12533: @item division rounding:
12534: @cindex division rounding
12535: installation dependent. @code{s" floored" environment? drop .}. We leave
12536: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
12537: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
12538: 
12539: @item values of @code{STATE} when true:
12540: @cindex @code{STATE} values
12541: -1.
12542: 
12543: @item values returned after arithmetic overflow:
12544: On two's complement machines, arithmetic is performed modulo
12545: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12546: arithmetic (with appropriate mapping for signed types). Division by zero
12547: typically results in a @code{-55 throw} (Floating-point unidentified
12548: fault) or @code{-10 throw} (divide by zero).
12549: 
12550: @item whether the current definition can be found after @t{DOES>}:
12551: @cindex @t{DOES>}, visibility of current definition
12552: No.
12553: 
12554: @end table
12555: 
12556: @c ---------------------------------------------------------------------
12557: @node core-ambcond, core-other, core-idef, The Core Words
12558: @subsection Ambiguous conditions
12559: @c ---------------------------------------------------------------------
12560: @cindex core words, ambiguous conditions
12561: @cindex ambiguous conditions, core words
12562: 
12563: @table @i
12564: 
12565: @item a name is neither a word nor a number:
12566: @cindex name not found
12567: @cindex undefined word
12568: @code{-13 throw} (Undefined word).
12569: 
12570: @item a definition name exceeds the maximum length allowed:
12571: @cindex word name too long
12572: @code{-19 throw} (Word name too long)
12573: 
12574: @item addressing a region not inside the various data spaces of the forth system:
12575: @cindex Invalid memory address
12576: The stacks, code space and header space are accessible. Machine code space is
12577: typically readable. Accessing other addresses gives results dependent on
12578: the operating system. On decent systems: @code{-9 throw} (Invalid memory
12579: address).
12580: 
12581: @item argument type incompatible with parameter:
12582: @cindex argument type mismatch
12583: This is usually not caught. Some words perform checks, e.g., the control
12584: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
12585: mismatch).
12586: 
12587: @item attempting to obtain the execution token of a word with undefined execution semantics:
12588: @cindex Interpreting a compile-only word, for @code{'} etc.
12589: @cindex execution token of words with undefined execution semantics
12590: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
12591: get an execution token for @code{compile-only-error} (which performs a
12592: @code{-14 throw} when executed).
12593: 
12594: @item dividing by zero:
12595: @cindex dividing by zero
12596: @cindex floating point unidentified fault, integer division
12597: On some platforms, this produces a @code{-10 throw} (Division by
12598: zero); on other systems, this typically results in a @code{-55 throw}
12599: (Floating-point unidentified fault).
12600: 
12601: @item insufficient data stack or return stack space:
12602: @cindex insufficient data stack or return stack space
12603: @cindex stack overflow
12604: @cindex address alignment exception, stack overflow
12605: @cindex Invalid memory address, stack overflow
12606: Depending on the operating system, the installation, and the invocation
12607: of Gforth, this is either checked by the memory management hardware, or
12608: it is not checked. If it is checked, you typically get a @code{-3 throw}
12609: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
12610: throw} (Invalid memory address) (depending on the platform and how you
12611: achieved the overflow) as soon as the overflow happens. If it is not
12612: checked, overflows typically result in mysterious illegal memory
12613: accesses, producing @code{-9 throw} (Invalid memory address) or
12614: @code{-23 throw} (Address alignment exception); they might also destroy
12615: the internal data structure of @code{ALLOCATE} and friends, resulting in
12616: various errors in these words.
12617: 
12618: @item insufficient space for loop control parameters:
12619: @cindex insufficient space for loop control parameters
12620: Like other return stack overflows.
12621: 
12622: @item insufficient space in the dictionary:
12623: @cindex insufficient space in the dictionary
12624: @cindex dictionary overflow
12625: If you try to allot (either directly with @code{allot}, or indirectly
12626: with @code{,}, @code{create} etc.) more memory than available in the
12627: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
12628: to access memory beyond the end of the dictionary, the results are
12629: similar to stack overflows.
12630: 
12631: @item interpreting a word with undefined interpretation semantics:
12632: @cindex interpreting a word with undefined interpretation semantics
12633: @cindex Interpreting a compile-only word
12634: For some words, we have defined interpretation semantics. For the
12635: others: @code{-14 throw} (Interpreting a compile-only word).
12636: 
12637: @item modifying the contents of the input buffer or a string literal:
12638: @cindex modifying the contents of the input buffer or a string literal
12639: These are located in writable memory and can be modified.
12640: 
12641: @item overflow of the pictured numeric output string:
12642: @cindex overflow of the pictured numeric output string
12643: @cindex pictured numeric output string, overflow
12644: @code{-17 throw} (Pictured numeric ouput string overflow).
12645: 
12646: @item parsed string overflow:
12647: @cindex parsed string overflow
12648: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
12649: 
12650: @item producing a result out of range:
12651: @cindex result out of range
12652: On two's complement machines, arithmetic is performed modulo
12653: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12654: arithmetic (with appropriate mapping for signed types). Division by zero
12655: typically results in a @code{-10 throw} (divide by zero) or @code{-55
12656: throw} (floating point unidentified fault). @code{convert} and
12657: @code{>number} currently overflow silently.
12658: 
12659: @item reading from an empty data or return stack:
12660: @cindex stack empty
12661: @cindex stack underflow
12662: @cindex return stack underflow
12663: The data stack is checked by the outer (aka text) interpreter after
12664: every word executed. If it has underflowed, a @code{-4 throw} (Stack
12665: underflow) is performed. Apart from that, stacks may be checked or not,
12666: depending on operating system, installation, and invocation. If they are
12667: caught by a check, they typically result in @code{-4 throw} (Stack
12668: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
12669: (Invalid memory address), depending on the platform and which stack
12670: underflows and by how much. Note that even if the system uses checking
12671: (through the MMU), your program may have to underflow by a significant
12672: number of stack items to trigger the reaction (the reason for this is
12673: that the MMU, and therefore the checking, works with a page-size
12674: granularity).  If there is no checking, the symptoms resulting from an
12675: underflow are similar to those from an overflow.  Unbalanced return
12676: stack errors can result in a variety of symptoms, including @code{-9 throw}
12677: (Invalid memory address) and Illegal Instruction (typically @code{-260
12678: throw}).
12679: 
12680: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
12681: @cindex unexpected end of the input buffer
12682: @cindex zero-length string as a name
12683: @cindex Attempt to use zero-length string as a name
12684: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
12685: use zero-length string as a name). Words like @code{'} probably will not
12686: find what they search. Note that it is possible to create zero-length
12687: names with @code{nextname} (should it not?).
12688: 
12689: @item @code{>IN} greater than input buffer:
12690: @cindex @code{>IN} greater than input buffer
12691: The next invocation of a parsing word returns a string with length 0.
12692: 
12693: @item @code{RECURSE} appears after @code{DOES>}:
12694: @cindex @code{RECURSE} appears after @code{DOES>}
12695: Compiles a recursive call to the defining word, not to the defined word.
12696: 
12697: @item argument input source different than current input source for @code{RESTORE-INPUT}:
12698: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
12699: @cindex argument type mismatch, @code{RESTORE-INPUT}
12700: @cindex @code{RESTORE-INPUT}, Argument type mismatch
12701: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
12702: the end of the file was reached), its source-id may be
12703: reused. Therefore, restoring an input source specification referencing a
12704: closed file may lead to unpredictable results instead of a @code{-12
12705: THROW}.
12706: 
12707: In the future, Gforth may be able to restore input source specifications
12708: from other than the current input source.
12709: 
12710: @item data space containing definitions gets de-allocated:
12711: @cindex data space containing definitions gets de-allocated
12712: Deallocation with @code{allot} is not checked. This typically results in
12713: memory access faults or execution of illegal instructions.
12714: 
12715: @item data space read/write with incorrect alignment:
12716: @cindex data space read/write with incorrect alignment
12717: @cindex alignment faults
12718: @cindex address alignment exception
12719: Processor-dependent. Typically results in a @code{-23 throw} (Address
12720: alignment exception). Under Linux-Intel on a 486 or later processor with
12721: alignment turned on, incorrect alignment results in a @code{-9 throw}
12722: (Invalid memory address). There are reportedly some processors with
12723: alignment restrictions that do not report violations.
12724: 
12725: @item data space pointer not properly aligned, @code{,}, @code{C,}:
12726: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
12727: Like other alignment errors.
12728: 
12729: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
12730: Like other stack underflows.
12731: 
12732: @item loop control parameters not available:
12733: @cindex loop control parameters not available
12734: Not checked. The counted loop words simply assume that the top of return
12735: stack items are loop control parameters and behave accordingly.
12736: 
12737: @item most recent definition does not have a name (@code{IMMEDIATE}):
12738: @cindex most recent definition does not have a name (@code{IMMEDIATE})
12739: @cindex last word was headerless
12740: @code{abort" last word was headerless"}.
12741: 
12742: @item name not defined by @code{VALUE} used by @code{TO}:
12743: @cindex name not defined by @code{VALUE} used by @code{TO}
12744: @cindex @code{TO} on non-@code{VALUE}s
12745: @cindex Invalid name argument, @code{TO}
12746: @code{-32 throw} (Invalid name argument) (unless name is a local or was
12747: defined by @code{CONSTANT}; in the latter case it just changes the constant).
12748: 
12749: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
12750: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
12751: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
12752: @code{-13 throw} (Undefined word)
12753: 
12754: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
12755: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
12756: Gforth behaves as if they were of the same type. I.e., you can predict
12757: the behaviour by interpreting all parameters as, e.g., signed.
12758: 
12759: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
12760: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
12761: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
12762: compilation semantics of @code{TO}.
12763: 
12764: @item String longer than a counted string returned by @code{WORD}:
12765: @cindex string longer than a counted string returned by @code{WORD}
12766: @cindex @code{WORD}, string overflow
12767: Not checked. The string will be ok, but the count will, of course,
12768: contain only the least significant bits of the length.
12769: 
12770: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
12771: @cindex @code{LSHIFT}, large shift counts
12772: @cindex @code{RSHIFT}, large shift counts
12773: Processor-dependent. Typical behaviours are returning 0 and using only
12774: the low bits of the shift count.
12775: 
12776: @item word not defined via @code{CREATE}:
12777: @cindex @code{>BODY} of non-@code{CREATE}d words
12778: @code{>BODY} produces the PFA of the word no matter how it was defined.
12779: 
12780: @cindex @code{DOES>} of non-@code{CREATE}d words
12781: @code{DOES>} changes the execution semantics of the last defined word no
12782: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
12783: @code{CREATE , DOES>}.
12784: 
12785: @item words improperly used outside @code{<#} and @code{#>}:
12786: Not checked. As usual, you can expect memory faults.
12787: 
12788: @end table
12789: 
12790: 
12791: @c ---------------------------------------------------------------------
12792: @node core-other,  , core-ambcond, The Core Words
12793: @subsection Other system documentation
12794: @c ---------------------------------------------------------------------
12795: @cindex other system documentation, core words
12796: @cindex core words, other system documentation
12797: 
12798: @table @i
12799: @item nonstandard words using @code{PAD}:
12800: @cindex @code{PAD} use by nonstandard words
12801: None.
12802: 
12803: @item operator's terminal facilities available:
12804: @cindex operator's terminal facilities available
12805: After processing the OS's command line, Gforth goes into interactive mode,
12806: and you can give commands to Gforth interactively. The actual facilities
12807: available depend on how you invoke Gforth.
12808: 
12809: @item program data space available:
12810: @cindex program data space available
12811: @cindex data space available
12812: @code{UNUSED .} gives the remaining dictionary space. The total
12813: dictionary space can be specified with the @code{-m} switch
12814: (@pxref{Invoking Gforth}) when Gforth starts up.
12815: 
12816: @item return stack space available:
12817: @cindex return stack space available
12818: You can compute the total return stack space in cells with
12819: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
12820: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
12821: 
12822: @item stack space available:
12823: @cindex stack space available
12824: You can compute the total data stack space in cells with
12825: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
12826: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
12827: 
12828: @item system dictionary space required, in address units:
12829: @cindex system dictionary space required, in address units
12830: Type @code{here forthstart - .} after startup. At the time of this
12831: writing, this gives 80080 (bytes) on a 32-bit system.
12832: @end table
12833: 
12834: 
12835: @c =====================================================================
12836: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
12837: @section The optional Block word set
12838: @c =====================================================================
12839: @cindex system documentation, block words
12840: @cindex block words, system documentation
12841: 
12842: @menu
12843: * block-idef::                  Implementation Defined Options
12844: * block-ambcond::               Ambiguous Conditions               
12845: * block-other::                 Other System Documentation                 
12846: @end menu
12847: 
12848: 
12849: @c ---------------------------------------------------------------------
12850: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
12851: @subsection Implementation Defined Options
12852: @c ---------------------------------------------------------------------
12853: @cindex implementation-defined options, block words
12854: @cindex block words, implementation-defined options
12855: 
12856: @table @i
12857: @item the format for display by @code{LIST}:
12858: @cindex @code{LIST} display format
12859: First the screen number is displayed, then 16 lines of 64 characters,
12860: each line preceded by the line number.
12861: 
12862: @item the length of a line affected by @code{\}:
12863: @cindex length of a line affected by @code{\}
12864: @cindex @code{\}, line length in blocks
12865: 64 characters.
12866: @end table
12867: 
12868: 
12869: @c ---------------------------------------------------------------------
12870: @node block-ambcond, block-other, block-idef, The optional Block word set
12871: @subsection Ambiguous conditions
12872: @c ---------------------------------------------------------------------
12873: @cindex block words, ambiguous conditions
12874: @cindex ambiguous conditions, block words
12875: 
12876: @table @i
12877: @item correct block read was not possible:
12878: @cindex block read not possible
12879: Typically results in a @code{throw} of some OS-derived value (between
12880: -512 and -2048). If the blocks file was just not long enough, blanks are
12881: supplied for the missing portion.
12882: 
12883: @item I/O exception in block transfer:
12884: @cindex I/O exception in block transfer
12885: @cindex block transfer, I/O exception
12886: Typically results in a @code{throw} of some OS-derived value (between
12887: -512 and -2048).
12888: 
12889: @item invalid block number:
12890: @cindex invalid block number
12891: @cindex block number invalid
12892: @code{-35 throw} (Invalid block number)
12893: 
12894: @item a program directly alters the contents of @code{BLK}:
12895: @cindex @code{BLK}, altering @code{BLK}
12896: The input stream is switched to that other block, at the same
12897: position. If the storing to @code{BLK} happens when interpreting
12898: non-block input, the system will get quite confused when the block ends.
12899: 
12900: @item no current block buffer for @code{UPDATE}:
12901: @cindex @code{UPDATE}, no current block buffer
12902: @code{UPDATE} has no effect.
12903: 
12904: @end table
12905: 
12906: @c ---------------------------------------------------------------------
12907: @node block-other,  , block-ambcond, The optional Block word set
12908: @subsection Other system documentation
12909: @c ---------------------------------------------------------------------
12910: @cindex other system documentation, block words
12911: @cindex block words, other system documentation
12912: 
12913: @table @i
12914: @item any restrictions a multiprogramming system places on the use of buffer addresses:
12915: No restrictions (yet).
12916: 
12917: @item the number of blocks available for source and data:
12918: depends on your disk space.
12919: 
12920: @end table
12921: 
12922: 
12923: @c =====================================================================
12924: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
12925: @section The optional Double Number word set
12926: @c =====================================================================
12927: @cindex system documentation, double words
12928: @cindex double words, system documentation
12929: 
12930: @menu
12931: * double-ambcond::              Ambiguous Conditions              
12932: @end menu
12933: 
12934: 
12935: @c ---------------------------------------------------------------------
12936: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
12937: @subsection Ambiguous conditions
12938: @c ---------------------------------------------------------------------
12939: @cindex double words, ambiguous conditions
12940: @cindex ambiguous conditions, double words
12941: 
12942: @table @i
12943: @item @i{d} outside of range of @i{n} in @code{D>S}:
12944: @cindex @code{D>S}, @i{d} out of range of @i{n} 
12945: The least significant cell of @i{d} is produced.
12946: 
12947: @end table
12948: 
12949: 
12950: @c =====================================================================
12951: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
12952: @section The optional Exception word set
12953: @c =====================================================================
12954: @cindex system documentation, exception words
12955: @cindex exception words, system documentation
12956: 
12957: @menu
12958: * exception-idef::              Implementation Defined Options              
12959: @end menu
12960: 
12961: 
12962: @c ---------------------------------------------------------------------
12963: @node exception-idef,  , The optional Exception word set, The optional Exception word set
12964: @subsection Implementation Defined Options
12965: @c ---------------------------------------------------------------------
12966: @cindex implementation-defined options, exception words
12967: @cindex exception words, implementation-defined options
12968: 
12969: @table @i
12970: @item @code{THROW}-codes used in the system:
12971: @cindex @code{THROW}-codes used in the system
12972: The codes -256@minus{}-511 are used for reporting signals. The mapping
12973: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
12974: codes -512@minus{}-2047 are used for OS errors (for file and memory
12975: allocation operations). The mapping from OS error numbers to throw codes
12976: is -512@minus{}@code{errno}. One side effect of this mapping is that
12977: undefined OS errors produce a message with a strange number; e.g.,
12978: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
12979: @end table
12980: 
12981: @c =====================================================================
12982: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
12983: @section The optional Facility word set
12984: @c =====================================================================
12985: @cindex system documentation, facility words
12986: @cindex facility words, system documentation
12987: 
12988: @menu
12989: * facility-idef::               Implementation Defined Options               
12990: * facility-ambcond::            Ambiguous Conditions            
12991: @end menu
12992: 
12993: 
12994: @c ---------------------------------------------------------------------
12995: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
12996: @subsection Implementation Defined Options
12997: @c ---------------------------------------------------------------------
12998: @cindex implementation-defined options, facility words
12999: @cindex facility words, implementation-defined options
13000: 
13001: @table @i
13002: @item encoding of keyboard events (@code{EKEY}):
13003: @cindex keyboard events, encoding in @code{EKEY}
13004: @cindex @code{EKEY}, encoding of keyboard events
13005: Keys corresponding to ASCII characters are encoded as ASCII characters.
13006: Other keys are encoded with the constants @code{k-left}, @code{k-right},
13007: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
13008: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
13009: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
13010: 
13011: 
13012: @item duration of a system clock tick:
13013: @cindex duration of a system clock tick
13014: @cindex clock tick duration
13015: System dependent. With respect to @code{MS}, the time is specified in
13016: microseconds. How well the OS and the hardware implement this, is
13017: another question.
13018: 
13019: @item repeatability to be expected from the execution of @code{MS}:
13020: @cindex repeatability to be expected from the execution of @code{MS}
13021: @cindex @code{MS}, repeatability to be expected
13022: System dependent. On Unix, a lot depends on load. If the system is
13023: lightly loaded, and the delay is short enough that Gforth does not get
13024: swapped out, the performance should be acceptable. Under MS-DOS and
13025: other single-tasking systems, it should be good.
13026: 
13027: @end table
13028: 
13029: 
13030: @c ---------------------------------------------------------------------
13031: @node facility-ambcond,  , facility-idef, The optional Facility word set
13032: @subsection Ambiguous conditions
13033: @c ---------------------------------------------------------------------
13034: @cindex facility words, ambiguous conditions
13035: @cindex ambiguous conditions, facility words
13036: 
13037: @table @i
13038: @item @code{AT-XY} can't be performed on user output device:
13039: @cindex @code{AT-XY} can't be performed on user output device
13040: Largely terminal dependent. No range checks are done on the arguments.
13041: No errors are reported. You may see some garbage appearing, you may see
13042: simply nothing happen.
13043: 
13044: @end table
13045: 
13046: 
13047: @c =====================================================================
13048: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
13049: @section The optional File-Access word set
13050: @c =====================================================================
13051: @cindex system documentation, file words
13052: @cindex file words, system documentation
13053: 
13054: @menu
13055: * file-idef::                   Implementation Defined Options
13056: * file-ambcond::                Ambiguous Conditions                
13057: @end menu
13058: 
13059: @c ---------------------------------------------------------------------
13060: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
13061: @subsection Implementation Defined Options
13062: @c ---------------------------------------------------------------------
13063: @cindex implementation-defined options, file words
13064: @cindex file words, implementation-defined options
13065: 
13066: @table @i
13067: @item file access methods used:
13068: @cindex file access methods used
13069: @code{R/O}, @code{R/W} and @code{BIN} work as you would
13070: expect. @code{W/O} translates into the C file opening mode @code{w} (or
13071: @code{wb}): The file is cleared, if it exists, and created, if it does
13072: not (with both @code{open-file} and @code{create-file}).  Under Unix
13073: @code{create-file} creates a file with 666 permissions modified by your
13074: umask.
13075: 
13076: @item file exceptions:
13077: @cindex file exceptions
13078: The file words do not raise exceptions (except, perhaps, memory access
13079: faults when you pass illegal addresses or file-ids).
13080: 
13081: @item file line terminator:
13082: @cindex file line terminator
13083: System-dependent. Gforth uses C's newline character as line
13084: terminator. What the actual character code(s) of this are is
13085: system-dependent.
13086: 
13087: @item file name format:
13088: @cindex file name format
13089: System dependent. Gforth just uses the file name format of your OS.
13090: 
13091: @item information returned by @code{FILE-STATUS}:
13092: @cindex @code{FILE-STATUS}, returned information
13093: @code{FILE-STATUS} returns the most powerful file access mode allowed
13094: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
13095: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
13096: along with the returned mode.
13097: 
13098: @item input file state after an exception when including source:
13099: @cindex exception when including source
13100: All files that are left via the exception are closed.
13101: 
13102: @item @i{ior} values and meaning:
13103: @cindex @i{ior} values and meaning
13104: @cindex @i{wior} values and meaning
13105: The @i{ior}s returned by the file and memory allocation words are
13106: intended as throw codes. They typically are in the range
13107: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
13108: @i{ior}s is -512@minus{}@i{errno}.
13109: 
13110: @item maximum depth of file input nesting:
13111: @cindex maximum depth of file input nesting
13112: @cindex file input nesting, maximum depth
13113: limited by the amount of return stack, locals/TIB stack, and the number
13114: of open files available. This should not give you troubles.
13115: 
13116: @item maximum size of input line:
13117: @cindex maximum size of input line
13118: @cindex input line size, maximum
13119: @code{/line}. Currently 255.
13120: 
13121: @item methods of mapping block ranges to files:
13122: @cindex mapping block ranges to files
13123: @cindex files containing blocks
13124: @cindex blocks in files
13125: By default, blocks are accessed in the file @file{blocks.fb} in the
13126: current working directory. The file can be switched with @code{USE}.
13127: 
13128: @item number of string buffers provided by @code{S"}:
13129: @cindex @code{S"}, number of string buffers
13130: 1
13131: 
13132: @item size of string buffer used by @code{S"}:
13133: @cindex @code{S"}, size of string buffer
13134: @code{/line}. currently 255.
13135: 
13136: @end table
13137: 
13138: @c ---------------------------------------------------------------------
13139: @node file-ambcond,  , file-idef, The optional File-Access word set
13140: @subsection Ambiguous conditions
13141: @c ---------------------------------------------------------------------
13142: @cindex file words, ambiguous conditions
13143: @cindex ambiguous conditions, file words
13144: 
13145: @table @i
13146: @item attempting to position a file outside its boundaries:
13147: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
13148: @code{REPOSITION-FILE} is performed as usual: Afterwards,
13149: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
13150: 
13151: @item attempting to read from file positions not yet written:
13152: @cindex reading from file positions not yet written
13153: End-of-file, i.e., zero characters are read and no error is reported.
13154: 
13155: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
13156: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid 
13157: An appropriate exception may be thrown, but a memory fault or other
13158: problem is more probable.
13159: 
13160: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
13161: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
13162: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
13163: The @i{ior} produced by the operation, that discovered the problem, is
13164: thrown.
13165: 
13166: @item named file cannot be opened (@code{INCLUDED}):
13167: @cindex @code{INCLUDED}, named file cannot be opened
13168: The @i{ior} produced by @code{open-file} is thrown.
13169: 
13170: @item requesting an unmapped block number:
13171: @cindex unmapped block numbers
13172: There are no unmapped legal block numbers. On some operating systems,
13173: writing a block with a large number may overflow the file system and
13174: have an error message as consequence.
13175: 
13176: @item using @code{source-id} when @code{blk} is non-zero:
13177: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
13178: @code{source-id} performs its function. Typically it will give the id of
13179: the source which loaded the block. (Better ideas?)
13180: 
13181: @end table
13182: 
13183: 
13184: @c =====================================================================
13185: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
13186: @section The optional Floating-Point word set
13187: @c =====================================================================
13188: @cindex system documentation, floating-point words
13189: @cindex floating-point words, system documentation
13190: 
13191: @menu
13192: * floating-idef::               Implementation Defined Options
13193: * floating-ambcond::            Ambiguous Conditions            
13194: @end menu
13195: 
13196: 
13197: @c ---------------------------------------------------------------------
13198: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
13199: @subsection Implementation Defined Options
13200: @c ---------------------------------------------------------------------
13201: @cindex implementation-defined options, floating-point words
13202: @cindex floating-point words, implementation-defined options
13203: 
13204: @table @i
13205: @item format and range of floating point numbers:
13206: @cindex format and range of floating point numbers
13207: @cindex floating point numbers, format and range
13208: System-dependent; the @code{double} type of C.
13209: 
13210: @item results of @code{REPRESENT} when @i{float} is out of range:
13211: @cindex  @code{REPRESENT}, results when @i{float} is out of range
13212: System dependent; @code{REPRESENT} is implemented using the C library
13213: function @code{ecvt()} and inherits its behaviour in this respect.
13214: 
13215: @item rounding or truncation of floating-point numbers:
13216: @cindex rounding of floating-point numbers
13217: @cindex truncation of floating-point numbers
13218: @cindex floating-point numbers, rounding or truncation
13219: System dependent; the rounding behaviour is inherited from the hosting C
13220: compiler. IEEE-FP-based (i.e., most) systems by default round to
13221: nearest, and break ties by rounding to even (i.e., such that the last
13222: bit of the mantissa is 0).
13223: 
13224: @item size of floating-point stack:
13225: @cindex floating-point stack size
13226: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
13227: the floating-point stack (in floats). You can specify this on startup
13228: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
13229: 
13230: @item width of floating-point stack:
13231: @cindex floating-point stack width 
13232: @code{1 floats}.
13233: 
13234: @end table
13235: 
13236: 
13237: @c ---------------------------------------------------------------------
13238: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
13239: @subsection Ambiguous conditions
13240: @c ---------------------------------------------------------------------
13241: @cindex floating-point words, ambiguous conditions
13242: @cindex ambiguous conditions, floating-point words
13243: 
13244: @table @i
13245: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
13246: @cindex @code{df@@} or @code{df!} used with an address that is not double-float  aligned
13247: System-dependent. Typically results in a @code{-23 THROW} like other
13248: alignment violations.
13249: 
13250: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
13251: @cindex @code{f@@} used with an address that is not float aligned
13252: @cindex @code{f!} used with an address that is not float aligned
13253: System-dependent. Typically results in a @code{-23 THROW} like other
13254: alignment violations.
13255: 
13256: @item floating-point result out of range:
13257: @cindex floating-point result out of range
13258: System-dependent. Can result in a @code{-43 throw} (floating point
13259: overflow), @code{-54 throw} (floating point underflow), @code{-41 throw}
13260: (floating point inexact result), @code{-55 THROW} (Floating-point
13261: unidentified fault), or can produce a special value representing, e.g.,
13262: Infinity.
13263: 
13264: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
13265: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned
13266: System-dependent. Typically results in an alignment fault like other
13267: alignment violations.
13268: 
13269: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
13270: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
13271: The floating-point number is converted into decimal nonetheless.
13272: 
13273: @item Both arguments are equal to zero (@code{FATAN2}):
13274: @cindex @code{FATAN2}, both arguments are equal to zero
13275: System-dependent. @code{FATAN2} is implemented using the C library
13276: function @code{atan2()}.
13277: 
13278: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
13279: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
13280: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
13281: because of small errors and the tan will be a very large (or very small)
13282: but finite number.
13283: 
13284: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
13285: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
13286: The result is rounded to the nearest float.
13287: 
13288: @item dividing by zero:
13289: @cindex dividing by zero, floating-point
13290: @cindex floating-point dividing by zero
13291: @cindex floating-point unidentified fault, FP divide-by-zero
13292: Platform-dependent; can produce an Infinity, NaN, @code{-42 throw}
13293: (floating point divide by zero) or @code{-55 throw} (Floating-point
13294: unidentified fault).
13295: 
13296: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
13297: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
13298: System dependent. On IEEE-FP based systems the number is converted into
13299: an infinity.
13300: 
13301: @item @i{float}<1 (@code{FACOSH}):
13302: @cindex @code{FACOSH}, @i{float}<1
13303: @cindex floating-point unidentified fault, @code{FACOSH}
13304: Platform-dependent; on IEEE-FP systems typically produces a NaN.
13305: 
13306: @item @i{float}=<-1 (@code{FLNP1}):
13307: @cindex @code{FLNP1}, @i{float}=<-1
13308: @cindex floating-point unidentified fault, @code{FLNP1}
13309: Platform-dependent; on IEEE-FP systems typically produces a NaN (or a
13310: negative infinity for @i{float}=-1).
13311: 
13312: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
13313: @cindex @code{FLN}, @i{float}=<0
13314: @cindex @code{FLOG}, @i{float}=<0
13315: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
13316: Platform-dependent; on IEEE-FP systems typically produces a NaN (or a
13317: negative infinity for @i{float}=0).
13318: 
13319: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
13320: @cindex @code{FASINH}, @i{float}<0
13321: @cindex @code{FSQRT}, @i{float}<0
13322: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
13323: Platform-dependent; for @code{fsqrt} this typically gives a NaN, for
13324: @code{fasinh} some platforms produce a NaN, others a number (bug in the
13325: C library?).
13326: 
13327: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
13328: @cindex @code{FACOS}, |@i{float}|>1
13329: @cindex @code{FASIN}, |@i{float}|>1
13330: @cindex @code{FATANH}, |@i{float}|>1
13331: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
13332: Platform-dependent; IEEE-FP systems typically produce a NaN.
13333: 
13334: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
13335: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
13336: @cindex floating-point unidentified fault, @code{F>D}
13337: Platform-dependent; typically, some double number is produced and no
13338: error is reported.
13339: 
13340: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
13341: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
13342: @code{Precision} characters of the numeric output area are used.  If
13343: @code{precision} is too high, these words will smash the data or code
13344: close to @code{here}.
13345: @end table
13346: 
13347: @c =====================================================================
13348: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
13349: @section The optional Locals word set
13350: @c =====================================================================
13351: @cindex system documentation, locals words
13352: @cindex locals words, system documentation
13353: 
13354: @menu
13355: * locals-idef::                 Implementation Defined Options                 
13356: * locals-ambcond::              Ambiguous Conditions              
13357: @end menu
13358: 
13359: 
13360: @c ---------------------------------------------------------------------
13361: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
13362: @subsection Implementation Defined Options
13363: @c ---------------------------------------------------------------------
13364: @cindex implementation-defined options, locals words
13365: @cindex locals words, implementation-defined options
13366: 
13367: @table @i
13368: @item maximum number of locals in a definition:
13369: @cindex maximum number of locals in a definition
13370: @cindex locals, maximum number in a definition
13371: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
13372: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
13373: characters. The number of locals in a definition is bounded by the size
13374: of locals-buffer, which contains the names of the locals.
13375: 
13376: @end table
13377: 
13378: 
13379: @c ---------------------------------------------------------------------
13380: @node locals-ambcond,  , locals-idef, The optional Locals word set
13381: @subsection Ambiguous conditions
13382: @c ---------------------------------------------------------------------
13383: @cindex locals words, ambiguous conditions
13384: @cindex ambiguous conditions, locals words
13385: 
13386: @table @i
13387: @item executing a named local in interpretation state:
13388: @cindex local in interpretation state
13389: @cindex Interpreting a compile-only word, for a local
13390: Locals have no interpretation semantics. If you try to perform the
13391: interpretation semantics, you will get a @code{-14 throw} somewhere
13392: (Interpreting a compile-only word). If you perform the compilation
13393: semantics, the locals access will be compiled (irrespective of state).
13394: 
13395: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
13396: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
13397: @cindex @code{TO} on non-@code{VALUE}s and non-locals
13398: @cindex Invalid name argument, @code{TO}
13399: @code{-32 throw} (Invalid name argument)
13400: 
13401: @end table
13402: 
13403: 
13404: @c =====================================================================
13405: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
13406: @section The optional Memory-Allocation word set
13407: @c =====================================================================
13408: @cindex system documentation, memory-allocation words
13409: @cindex memory-allocation words, system documentation
13410: 
13411: @menu
13412: * memory-idef::                 Implementation Defined Options                 
13413: @end menu
13414: 
13415: 
13416: @c ---------------------------------------------------------------------
13417: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
13418: @subsection Implementation Defined Options
13419: @c ---------------------------------------------------------------------
13420: @cindex implementation-defined options, memory-allocation words
13421: @cindex memory-allocation words, implementation-defined options
13422: 
13423: @table @i
13424: @item values and meaning of @i{ior}:
13425: @cindex  @i{ior} values and meaning
13426: The @i{ior}s returned by the file and memory allocation words are
13427: intended as throw codes. They typically are in the range
13428: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
13429: @i{ior}s is -512@minus{}@i{errno}.
13430: 
13431: @end table
13432: 
13433: @c =====================================================================
13434: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
13435: @section The optional Programming-Tools word set
13436: @c =====================================================================
13437: @cindex system documentation, programming-tools words
13438: @cindex programming-tools words, system documentation
13439: 
13440: @menu
13441: * programming-idef::            Implementation Defined Options            
13442: * programming-ambcond::         Ambiguous Conditions         
13443: @end menu
13444: 
13445: 
13446: @c ---------------------------------------------------------------------
13447: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
13448: @subsection Implementation Defined Options
13449: @c ---------------------------------------------------------------------
13450: @cindex implementation-defined options, programming-tools words
13451: @cindex programming-tools words, implementation-defined options
13452: 
13453: @table @i
13454: @item ending sequence for input following @code{;CODE} and @code{CODE}:
13455: @cindex @code{;CODE} ending sequence
13456: @cindex @code{CODE} ending sequence
13457: @code{END-CODE}
13458: 
13459: @item manner of processing input following @code{;CODE} and @code{CODE}:
13460: @cindex @code{;CODE}, processing input
13461: @cindex @code{CODE}, processing input
13462: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
13463: the input is processed by the text interpreter, (starting) in interpret
13464: state.
13465: 
13466: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
13467: @cindex @code{ASSEMBLER}, search order capability
13468: The ANS Forth search order word set.
13469: 
13470: @item source and format of display by @code{SEE}:
13471: @cindex @code{SEE}, source and format of output
13472: The source for @code{see} is the executable code used by the inner
13473: interpreter.  The current @code{see} tries to output Forth source code
13474: (and on some platforms, assembly code for primitives) as well as
13475: possible.
13476: 
13477: @end table
13478: 
13479: @c ---------------------------------------------------------------------
13480: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
13481: @subsection Ambiguous conditions
13482: @c ---------------------------------------------------------------------
13483: @cindex programming-tools words, ambiguous conditions
13484: @cindex ambiguous conditions, programming-tools words
13485: 
13486: @table @i
13487: 
13488: @item deleting the compilation word list (@code{FORGET}):
13489: @cindex @code{FORGET}, deleting the compilation word list
13490: Not implemented (yet).
13491: 
13492: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
13493: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
13494: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
13495: @cindex control-flow stack underflow
13496: This typically results in an @code{abort"} with a descriptive error
13497: message (may change into a @code{-22 throw} (Control structure mismatch)
13498: in the future). You may also get a memory access error. If you are
13499: unlucky, this ambiguous condition is not caught.
13500: 
13501: @item @i{name} can't be found (@code{FORGET}):
13502: @cindex @code{FORGET}, @i{name} can't be found
13503: Not implemented (yet).
13504: 
13505: @item @i{name} not defined via @code{CREATE}:
13506: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
13507: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
13508: the execution semantics of the last defined word no matter how it was
13509: defined.
13510: 
13511: @item @code{POSTPONE} applied to @code{[IF]}:
13512: @cindex @code{POSTPONE} applied to @code{[IF]}
13513: @cindex @code{[IF]} and @code{POSTPONE}
13514: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
13515: equivalent to @code{[IF]}.
13516: 
13517: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
13518: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
13519: Continue in the same state of conditional compilation in the next outer
13520: input source. Currently there is no warning to the user about this.
13521: 
13522: @item removing a needed definition (@code{FORGET}):
13523: @cindex @code{FORGET}, removing a needed definition
13524: Not implemented (yet).
13525: 
13526: @end table
13527: 
13528: 
13529: @c =====================================================================
13530: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
13531: @section The optional Search-Order word set
13532: @c =====================================================================
13533: @cindex system documentation, search-order words
13534: @cindex search-order words, system documentation
13535: 
13536: @menu
13537: * search-idef::                 Implementation Defined Options                 
13538: * search-ambcond::              Ambiguous Conditions              
13539: @end menu
13540: 
13541: 
13542: @c ---------------------------------------------------------------------
13543: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
13544: @subsection Implementation Defined Options
13545: @c ---------------------------------------------------------------------
13546: @cindex implementation-defined options, search-order words
13547: @cindex search-order words, implementation-defined options
13548: 
13549: @table @i
13550: @item maximum number of word lists in search order:
13551: @cindex maximum number of word lists in search order
13552: @cindex search order, maximum depth
13553: @code{s" wordlists" environment? drop .}. Currently 16.
13554: 
13555: @item minimum search order:
13556: @cindex minimum search order
13557: @cindex search order, minimum
13558: @code{root root}.
13559: 
13560: @end table
13561: 
13562: @c ---------------------------------------------------------------------
13563: @node search-ambcond,  , search-idef, The optional Search-Order word set
13564: @subsection Ambiguous conditions
13565: @c ---------------------------------------------------------------------
13566: @cindex search-order words, ambiguous conditions
13567: @cindex ambiguous conditions, search-order words
13568: 
13569: @table @i
13570: @item changing the compilation word list (during compilation):
13571: @cindex changing the compilation word list (during compilation)
13572: @cindex compilation word list, change before definition ends
13573: The word is entered into the word list that was the compilation word list
13574: at the start of the definition. Any changes to the name field (e.g.,
13575: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
13576: are applied to the latest defined word (as reported by @code{last} or
13577: @code{lastxt}), if possible, irrespective of the compilation word list.
13578: 
13579: @item search order empty (@code{previous}):
13580: @cindex @code{previous}, search order empty
13581: @cindex vocstack empty, @code{previous}
13582: @code{abort" Vocstack empty"}.
13583: 
13584: @item too many word lists in search order (@code{also}):
13585: @cindex @code{also}, too many word lists in search order
13586: @cindex vocstack full, @code{also}
13587: @code{abort" Vocstack full"}.
13588: 
13589: @end table
13590: 
13591: @c ***************************************************************
13592: @node Standard vs Extensions, Model, ANS conformance, Top
13593: @chapter Should I use Gforth extensions?
13594: @cindex Gforth extensions
13595: 
13596: As you read through the rest of this manual, you will see documentation
13597: for @i{Standard} words, and documentation for some appealing Gforth
13598: @i{extensions}. You might ask yourself the question: @i{``Should I
13599: restrict myself to the standard, or should I use the extensions?''}
13600: 
13601: The answer depends on the goals you have for the program you are working
13602: on:
13603: 
13604: @itemize @bullet
13605: 
13606: @item Is it just for yourself or do you want to share it with others?
13607: 
13608: @item
13609: If you want to share it, do the others all use Gforth?
13610: 
13611: @item
13612: If it is just for yourself, do you want to restrict yourself to Gforth?
13613: 
13614: @end itemize
13615: 
13616: If restricting the program to Gforth is ok, then there is no reason not
13617: to use extensions.  It is still a good idea to keep to the standard
13618: where it is easy, in case you want to reuse these parts in another
13619: program that you want to be portable.
13620: 
13621: If you want to be able to port the program to other Forth systems, there
13622: are the following points to consider:
13623: 
13624: @itemize @bullet
13625: 
13626: @item
13627: Most Forth systems that are being maintained support the ANS Forth
13628: standard.  So if your program complies with the standard, it will be
13629: portable among many systems.
13630: 
13631: @item
13632: A number of the Gforth extensions can be implemented in ANS Forth using
13633: public-domain files provided in the @file{compat/} directory. These are
13634: mentioned in the text in passing.  There is no reason not to use these
13635: extensions, your program will still be ANS Forth compliant; just include
13636: the appropriate compat files with your program.
13637: 
13638: @item
13639: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
13640: analyse your program and determine what non-Standard words it relies
13641: upon.  However, it does not check whether you use standard words in a
13642: non-standard way.
13643: 
13644: @item
13645: Some techniques are not standardized by ANS Forth, and are hard or
13646: impossible to implement in a standard way, but can be implemented in
13647: most Forth systems easily, and usually in similar ways (e.g., accessing
13648: word headers).  Forth has a rich historical precedent for programmers
13649: taking advantage of implementation-dependent features of their tools
13650: (for example, relying on a knowledge of the dictionary
13651: structure). Sometimes these techniques are necessary to extract every
13652: last bit of performance from the hardware, sometimes they are just a
13653: programming shorthand.
13654: 
13655: @item
13656: Does using a Gforth extension save more work than the porting this part
13657: to other Forth systems (if any) will cost?
13658: 
13659: @item
13660: Is the additional functionality worth the reduction in portability and
13661: the additional porting problems?
13662: 
13663: @end itemize
13664: 
13665: In order to perform these consideratios, you need to know what's
13666: standard and what's not.  This manual generally states if something is
13667: non-standard, but the authoritative source is the
13668: @uref{http://www.taygeta.com/forth/dpans.html,standard document}.
13669: Appendix A of the Standard (@var{Rationale}) provides a valuable insight
13670: into the thought processes of the technical committee.
13671: 
13672: Note also that portability between Forth systems is not the only
13673: portability issue; there is also the issue of portability between
13674: different platforms (processor/OS combinations).
13675: 
13676: @c ***************************************************************
13677: @node Model, Integrating Gforth, Standard vs Extensions, Top
13678: @chapter Model
13679: 
13680: This chapter has yet to be written. It will contain information, on
13681: which internal structures you can rely.
13682: 
13683: @c ***************************************************************
13684: @node Integrating Gforth, Emacs and Gforth, Model, Top
13685: @chapter Integrating Gforth into C programs
13686: 
13687: This is not yet implemented.
13688: 
13689: Several people like to use Forth as scripting language for applications
13690: that are otherwise written in C, C++, or some other language.
13691: 
13692: The Forth system ATLAST provides facilities for embedding it into
13693: applications; unfortunately it has several disadvantages: most
13694: importantly, it is not based on ANS Forth, and it is apparently dead
13695: (i.e., not developed further and not supported). The facilities
13696: provided by Gforth in this area are inspired by ATLAST's facilities, so
13697: making the switch should not be hard.
13698: 
13699: We also tried to design the interface such that it can easily be
13700: implemented by other Forth systems, so that we may one day arrive at a
13701: standardized interface. Such a standard interface would allow you to
13702: replace the Forth system without having to rewrite C code.
13703: 
13704: You embed the Gforth interpreter by linking with the library
13705: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
13706: global symbols in this library that belong to the interface, have the
13707: prefix @code{forth_}. (Global symbols that are used internally have the
13708: prefix @code{gforth_}).
13709: 
13710: You can include the declarations of Forth types and the functions and
13711: variables of the interface with @code{#include <forth.h>}.
13712: 
13713: Types.
13714: 
13715: Variables.
13716: 
13717: Data and FP Stack pointer. Area sizes.
13718: 
13719: functions.
13720: 
13721: forth_init(imagefile)
13722: forth_evaluate(string) exceptions?
13723: forth_goto(address) (or forth_execute(xt)?)
13724: forth_continue() (a corountining mechanism)
13725: 
13726: Adding primitives.
13727: 
13728: No checking.
13729: 
13730: Signals?
13731: 
13732: Accessing the Stacks
13733: 
13734: @c ******************************************************************
13735: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
13736: @chapter Emacs and Gforth
13737: @cindex Emacs and Gforth
13738: 
13739: @cindex @file{gforth.el}
13740: @cindex @file{forth.el}
13741: @cindex Rydqvist, Goran
13742: @cindex comment editing commands
13743: @cindex @code{\}, editing with Emacs
13744: @cindex debug tracer editing commands
13745: @cindex @code{~~}, removal with Emacs
13746: @cindex Forth mode in Emacs
13747: Gforth comes with @file{gforth.el}, an improved version of
13748: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
13749: improvements are:
13750: 
13751: @itemize @bullet
13752: @item
13753: A better (but still not perfect) handling of indentation.
13754: @item
13755: Comment paragraph filling (@kbd{M-q})
13756: @item
13757: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
13758: @item
13759: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
13760: @item
13761: Support of the @code{info-lookup} feature for looking up the
13762: documentation of a word.
13763: @end itemize
13764: 
13765: I left the stuff I do not use alone, even though some of it only makes
13766: sense for TILE. To get a description of these features, enter Forth mode
13767: and type @kbd{C-h m}.
13768: 
13769: @cindex source location of error or debugging output in Emacs
13770: @cindex error output, finding the source location in Emacs
13771: @cindex debugging output, finding the source location in Emacs
13772: In addition, Gforth supports Emacs quite well: The source code locations
13773: given in error messages, debugging output (from @code{~~}) and failed
13774: assertion messages are in the right format for Emacs' compilation mode
13775: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
13776: Manual}) so the source location corresponding to an error or other
13777: message is only a few keystrokes away (@kbd{C-x `} for the next error,
13778: @kbd{C-c C-c} for the error under the cursor).
13779: 
13780: @cindex @file{TAGS} file
13781: @cindex @file{etags.fs}
13782: @cindex viewing the source of a word in Emacs
13783: @cindex @code{require}, placement in files
13784: @cindex @code{include}, placement in files
13785: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
13786: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
13787: contains the definitions of all words defined afterwards. You can then
13788: find the source for a word using @kbd{M-.}. Note that emacs can use
13789: several tags files at the same time (e.g., one for the Gforth sources
13790: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
13791: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
13792: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
13793: @file{/usr/local/share/gforth/0.2.0/TAGS}).  To get the best behaviour
13794: with @file{etags.fs}, you should avoid putting definitions both before
13795: and after @code{require} etc., otherwise you will see the same file
13796: visited several times by commands like @code{tags-search}.
13797: 
13798: @cindex viewing the documentation of a word in Emacs
13799: @cindex context-sensitive help
13800: Moreover, for words documented in this manual, you can look up the
13801: glossary entry quickly by using @kbd{C-h TAB}
13802: (@code{info-lookup-symbol}, @pxref{Documentation, ,Documentation
13803: Commands, emacs, Emacs Manual}).  This feature requires Emacs 20.3 or
13804: later and does not work for words containing @code{:}.
13805: 
13806: 
13807: @cindex @file{.emacs}
13808: To get all these benefits, add the following lines to your @file{.emacs}
13809: file:
13810: 
13811: @example
13812: (autoload 'forth-mode "gforth.el")
13813: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
13814: @end example
13815: 
13816: @c ******************************************************************
13817: @node Image Files, Engine, Emacs and Gforth, Top
13818: @chapter Image Files
13819: @cindex image file
13820: @cindex @file{.fi} files
13821: @cindex precompiled Forth code
13822: @cindex dictionary in persistent form
13823: @cindex persistent form of dictionary
13824: 
13825: An image file is a file containing an image of the Forth dictionary,
13826: i.e., compiled Forth code and data residing in the dictionary.  By
13827: convention, we use the extension @code{.fi} for image files.
13828: 
13829: @menu
13830: * Image Licensing Issues::      Distribution terms for images.
13831: * Image File Background::       Why have image files?
13832: * Non-Relocatable Image Files::  don't always work.
13833: * Data-Relocatable Image Files::  are better.
13834: * Fully Relocatable Image Files::  better yet.
13835: * Stack and Dictionary Sizes::  Setting the default sizes for an image.
13836: * Running Image Files::         @code{gforth -i @i{file}} or @i{file}.
13837: * Modifying the Startup Sequence::  and turnkey applications.
13838: @end menu
13839: 
13840: @node Image Licensing Issues, Image File Background, Image Files, Image Files
13841: @section Image Licensing Issues
13842: @cindex license for images
13843: @cindex image license
13844: 
13845: An image created with @code{gforthmi} (@pxref{gforthmi}) or
13846: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
13847: original image; i.e., according to copyright law it is a derived work of
13848: the original image.
13849: 
13850: Since Gforth is distributed under the GNU GPL, the newly created image
13851: falls under the GNU GPL, too. In particular, this means that if you
13852: distribute the image, you have to make all of the sources for the image
13853: available, including those you wrote.  For details see @ref{License, ,
13854: GNU General Public License (Section 3)}.
13855: 
13856: If you create an image with @code{cross} (@pxref{cross.fs}), the image
13857: contains only code compiled from the sources you gave it; if none of
13858: these sources is under the GPL, the terms discussed above do not apply
13859: to the image. However, if your image needs an engine (a gforth binary)
13860: that is under the GPL, you should make sure that you distribute both in
13861: a way that is at most a @emph{mere aggregation}, if you don't want the
13862: terms of the GPL to apply to the image.
13863: 
13864: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
13865: @section Image File Background
13866: @cindex image file background
13867: 
13868: Gforth consists not only of primitives (in the engine), but also of
13869: definitions written in Forth. Since the Forth compiler itself belongs to
13870: those definitions, it is not possible to start the system with the
13871: engine and the Forth source alone. Therefore we provide the Forth
13872: code as an image file in nearly executable form. When Gforth starts up,
13873: a C routine loads the image file into memory, optionally relocates the
13874: addresses, then sets up the memory (stacks etc.) according to
13875: information in the image file, and (finally) starts executing Forth
13876: code.
13877: 
13878: The image file variants represent different compromises between the
13879: goals of making it easy to generate image files and making them
13880: portable.
13881: 
13882: @cindex relocation at run-time
13883: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
13884: run-time. This avoids many of the complications discussed below (image
13885: files are data relocatable without further ado), but costs performance
13886: (one addition per memory access).
13887: 
13888: @cindex relocation at load-time
13889: By contrast, the Gforth loader performs relocation at image load time. The
13890: loader also has to replace tokens that represent primitive calls with the
13891: appropriate code-field addresses (or code addresses in the case of
13892: direct threading).
13893: 
13894: There are three kinds of image files, with different degrees of
13895: relocatability: non-relocatable, data-relocatable, and fully relocatable
13896: image files.
13897: 
13898: @cindex image file loader
13899: @cindex relocating loader
13900: @cindex loader for image files
13901: These image file variants have several restrictions in common; they are
13902: caused by the design of the image file loader:
13903: 
13904: @itemize @bullet
13905: @item
13906: There is only one segment; in particular, this means, that an image file
13907: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
13908: them). The contents of the stacks are not represented, either.
13909: 
13910: @item
13911: The only kinds of relocation supported are: adding the same offset to
13912: all cells that represent data addresses; and replacing special tokens
13913: with code addresses or with pieces of machine code.
13914: 
13915: If any complex computations involving addresses are performed, the
13916: results cannot be represented in the image file. Several applications that
13917: use such computations come to mind:
13918: @itemize @minus
13919: @item
13920: Hashing addresses (or data structures which contain addresses) for table
13921: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
13922: purpose, you will have no problem, because the hash tables are
13923: recomputed automatically when the system is started. If you use your own
13924: hash tables, you will have to do something similar.
13925: 
13926: @item
13927: There's a cute implementation of doubly-linked lists that uses
13928: @code{XOR}ed addresses. You could represent such lists as singly-linked
13929: in the image file, and restore the doubly-linked representation on
13930: startup.@footnote{In my opinion, though, you should think thrice before
13931: using a doubly-linked list (whatever implementation).}
13932: 
13933: @item
13934: The code addresses of run-time routines like @code{docol:} cannot be
13935: represented in the image file (because their tokens would be replaced by
13936: machine code in direct threaded implementations). As a workaround,
13937: compute these addresses at run-time with @code{>code-address} from the
13938: executions tokens of appropriate words (see the definitions of
13939: @code{docol:} and friends in @file{kernel/getdoers.fs}).
13940: 
13941: @item
13942: On many architectures addresses are represented in machine code in some
13943: shifted or mangled form. You cannot put @code{CODE} words that contain
13944: absolute addresses in this form in a relocatable image file. Workarounds
13945: are representing the address in some relative form (e.g., relative to
13946: the CFA, which is present in some register), or loading the address from
13947: a place where it is stored in a non-mangled form.
13948: @end itemize
13949: @end itemize
13950: 
13951: @node  Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
13952: @section Non-Relocatable Image Files
13953: @cindex non-relocatable image files
13954: @cindex image file, non-relocatable
13955: 
13956: These files are simple memory dumps of the dictionary. They are specific
13957: to the executable (i.e., @file{gforth} file) they were created
13958: with. What's worse, they are specific to the place on which the
13959: dictionary resided when the image was created. Now, there is no
13960: guarantee that the dictionary will reside at the same place the next
13961: time you start Gforth, so there's no guarantee that a non-relocatable
13962: image will work the next time (Gforth will complain instead of crashing,
13963: though).
13964: 
13965: You can create a non-relocatable image file with
13966: 
13967: 
13968: doc-savesystem
13969: 
13970: 
13971: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
13972: @section Data-Relocatable Image Files
13973: @cindex data-relocatable image files
13974: @cindex image file, data-relocatable
13975: 
13976: These files contain relocatable data addresses, but fixed code addresses
13977: (instead of tokens). They are specific to the executable (i.e.,
13978: @file{gforth} file) they were created with. For direct threading on some
13979: architectures (e.g., the i386), data-relocatable images do not work. You
13980: get a data-relocatable image, if you use @file{gforthmi} with a
13981: Gforth binary that is not doubly indirect threaded (@pxref{Fully
13982: Relocatable Image Files}).
13983: 
13984: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
13985: @section Fully Relocatable Image Files
13986: @cindex fully relocatable image files
13987: @cindex image file, fully relocatable
13988: 
13989: @cindex @file{kern*.fi}, relocatability
13990: @cindex @file{gforth.fi}, relocatability
13991: These image files have relocatable data addresses, and tokens for code
13992: addresses. They can be used with different binaries (e.g., with and
13993: without debugging) on the same machine, and even across machines with
13994: the same data formats (byte order, cell size, floating point
13995: format). However, they are usually specific to the version of Gforth
13996: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
13997: are fully relocatable.
13998: 
13999: There are two ways to create a fully relocatable image file:
14000: 
14001: @menu
14002: * gforthmi::                    The normal way
14003: * cross.fs::                    The hard way
14004: @end menu
14005: 
14006: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
14007: @subsection @file{gforthmi}
14008: @cindex @file{comp-i.fs}
14009: @cindex @file{gforthmi}
14010: 
14011: You will usually use @file{gforthmi}. If you want to create an
14012: image @i{file} that contains everything you would load by invoking
14013: Gforth with @code{gforth @i{options}}, you simply say:
14014: @example
14015: gforthmi @i{file} @i{options}
14016: @end example
14017: 
14018: E.g., if you want to create an image @file{asm.fi} that has the file
14019: @file{asm.fs} loaded in addition to the usual stuff, you could do it
14020: like this:
14021: 
14022: @example
14023: gforthmi asm.fi asm.fs
14024: @end example
14025: 
14026: @file{gforthmi} is implemented as a sh script and works like this: It
14027: produces two non-relocatable images for different addresses and then
14028: compares them. Its output reflects this: first you see the output (if
14029: any) of the two Gforth invocations that produce the non-relocatable image
14030: files, then you see the output of the comparing program: It displays the
14031: offset used for data addresses and the offset used for code addresses;
14032: moreover, for each cell that cannot be represented correctly in the
14033: image files, it displays a line like this:
14034: 
14035: @example
14036:      78DC         BFFFFA50         BFFFFA40
14037: @end example
14038: 
14039: This means that at offset $78dc from @code{forthstart}, one input image
14040: contains $bffffa50, and the other contains $bffffa40. Since these cells
14041: cannot be represented correctly in the output image, you should examine
14042: these places in the dictionary and verify that these cells are dead
14043: (i.e., not read before they are written).
14044: 
14045: @cindex --application, @code{gforthmi} option
14046: If you insert the option @code{--application} in front of the image file
14047: name, you will get an image that uses the @code{--appl-image} option
14048: instead of the @code{--image-file} option (@pxref{Invoking
14049: Gforth}). When you execute such an image on Unix (by typing the image
14050: name as command), the Gforth engine will pass all options to the image
14051: instead of trying to interpret them as engine options.
14052: 
14053: If you type @file{gforthmi} with no arguments, it prints some usage
14054: instructions.
14055: 
14056: @cindex @code{savesystem} during @file{gforthmi}
14057: @cindex @code{bye} during @file{gforthmi}
14058: @cindex doubly indirect threaded code
14059: @cindex environment variables
14060: @cindex @code{GFORTHD} -- environment variable
14061: @cindex @code{GFORTH} -- environment variable
14062: @cindex @code{gforth-ditc}
14063: There are a few wrinkles: After processing the passed @i{options}, the
14064: words @code{savesystem} and @code{bye} must be visible. A special doubly
14065: indirect threaded version of the @file{gforth} executable is used for
14066: creating the non-relocatable images; you can pass the exact filename of
14067: this executable through the environment variable @code{GFORTHD}
14068: (default: @file{gforth-ditc}); if you pass a version that is not doubly
14069: indirect threaded, you will not get a fully relocatable image, but a
14070: data-relocatable image (because there is no code address offset). The
14071: normal @file{gforth} executable is used for creating the relocatable
14072: image; you can pass the exact filename of this executable through the
14073: environment variable @code{GFORTH}.
14074: 
14075: @node cross.fs,  , gforthmi, Fully Relocatable Image Files
14076: @subsection @file{cross.fs}
14077: @cindex @file{cross.fs}
14078: @cindex cross-compiler
14079: @cindex metacompiler
14080: @cindex target compiler
14081: 
14082: You can also use @code{cross}, a batch compiler that accepts a Forth-like
14083: programming language (@pxref{Cross Compiler}).
14084: 
14085: @code{cross} allows you to create image files for machines with
14086: different data sizes and data formats than the one used for generating
14087: the image file. You can also use it to create an application image that
14088: does not contain a Forth compiler. These features are bought with
14089: restrictions and inconveniences in programming. E.g., addresses have to
14090: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
14091: order to make the code relocatable.
14092: 
14093: 
14094: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
14095: @section Stack and Dictionary Sizes
14096: @cindex image file, stack and dictionary sizes
14097: @cindex dictionary size default
14098: @cindex stack size default
14099: 
14100: If you invoke Gforth with a command line flag for the size
14101: (@pxref{Invoking Gforth}), the size you specify is stored in the
14102: dictionary. If you save the dictionary with @code{savesystem} or create
14103: an image with @file{gforthmi}, this size will become the default
14104: for the resulting image file. E.g., the following will create a
14105: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
14106: 
14107: @example
14108: gforthmi gforth.fi -m 1M
14109: @end example
14110: 
14111: In other words, if you want to set the default size for the dictionary
14112: and the stacks of an image, just invoke @file{gforthmi} with the
14113: appropriate options when creating the image.
14114: 
14115: @cindex stack size, cache-friendly
14116: Note: For cache-friendly behaviour (i.e., good performance), you should
14117: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
14118: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
14119: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
14120: 
14121: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
14122: @section Running Image Files
14123: @cindex running image files
14124: @cindex invoking image files
14125: @cindex image file invocation
14126: 
14127: @cindex -i, invoke image file
14128: @cindex --image file, invoke image file
14129: You can invoke Gforth with an image file @i{image} instead of the
14130: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
14131: @example
14132: gforth -i @i{image}
14133: @end example
14134: 
14135: @cindex executable image file
14136: @cindex image file, executable
14137: If your operating system supports starting scripts with a line of the
14138: form @code{#! ...}, you just have to type the image file name to start
14139: Gforth with this image file (note that the file extension @code{.fi} is
14140: just a convention). I.e., to run Gforth with the image file @i{image},
14141: you can just type @i{image} instead of @code{gforth -i @i{image}}.
14142: This works because every @code{.fi} file starts with a line of this
14143: format:
14144: 
14145: @example
14146: #! /usr/local/bin/gforth-0.4.0 -i
14147: @end example
14148: 
14149: The file and pathname for the Gforth engine specified on this line is
14150: the specific Gforth executable that it was built against; i.e. the value
14151: of the environment variable @code{GFORTH} at the time that
14152: @file{gforthmi} was executed.
14153: 
14154: You can make use of the same shell capability to make a Forth source
14155: file into an executable. For example, if you place this text in a file:
14156: 
14157: @example
14158: #! /usr/local/bin/gforth
14159: 
14160: ." Hello, world" CR
14161: bye
14162: @end example
14163: 
14164: @noindent
14165: and then make the file executable (chmod +x in Unix), you can run it
14166: directly from the command line. The sequence @code{#!} is used in two
14167: ways; firstly, it is recognised as a ``magic sequence'' by the operating
14168: system@footnote{The Unix kernel actually recognises two types of files:
14169: executable files and files of data, where the data is processed by an
14170: interpreter that is specified on the ``interpreter line'' -- the first
14171: line of the file, starting with the sequence #!. There may be a small
14172: limit (e.g., 32) on the number of characters that may be specified on
14173: the interpreter line.} secondly it is treated as a comment character by
14174: Gforth. Because of the second usage, a space is required between
14175: @code{#!} and the path to the executable (moreover, some Unixes
14176: require the sequence @code{#! /}).
14177: 
14178: The disadvantage of this latter technique, compared with using
14179: @file{gforthmi}, is that it is slightly slower; the Forth source code is
14180: compiled on-the-fly, each time the program is invoked.
14181: 
14182: doc-#!
14183: 
14184: 
14185: @node Modifying the Startup Sequence,  , Running Image Files, Image Files
14186: @section Modifying the Startup Sequence
14187: @cindex startup sequence for image file
14188: @cindex image file initialization sequence
14189: @cindex initialization sequence of image file
14190: 
14191: You can add your own initialization to the startup sequence through the
14192: deferred word @code{'cold}. @code{'cold} is invoked just before the
14193: image-specific command line processing (i.e., loading files and
14194: evaluating (@code{-e}) strings) starts.
14195: 
14196: A sequence for adding your initialization usually looks like this:
14197: 
14198: @example
14199: :noname
14200:     Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
14201:     ... \ your stuff
14202: ; IS 'cold
14203: @end example
14204: 
14205: @cindex turnkey image files
14206: @cindex image file, turnkey applications
14207: You can make a turnkey image by letting @code{'cold} execute a word
14208: (your turnkey application) that never returns; instead, it exits Gforth
14209: via @code{bye} or @code{throw}.
14210: 
14211: @cindex command-line arguments, access
14212: @cindex arguments on the command line, access
14213: You can access the (image-specific) command-line arguments through the
14214: variables @code{argc} and @code{argv}. @code{arg} provides convenient
14215: access to @code{argv}.
14216: 
14217: If @code{'cold} exits normally, Gforth processes the command-line
14218: arguments as files to be loaded and strings to be evaluated.  Therefore,
14219: @code{'cold} should remove the arguments it has used in this case.
14220: 
14221: 
14222: 
14223: doc-'cold
14224: doc-argc
14225: doc-argv
14226: doc-arg
14227: 
14228: 
14229: 
14230: @c ******************************************************************
14231: @node Engine, Binding to System Library, Image Files, Top
14232: @chapter Engine
14233: @cindex engine
14234: @cindex virtual machine
14235: 
14236: Reading this chapter is not necessary for programming with Gforth. It
14237: may be helpful for finding your way in the Gforth sources.
14238: 
14239: The ideas in this section have also been published in Bernd Paysan,
14240: @cite{ANS fig/GNU/??? Forth} (in German), Forth-Tagung '93 and M. Anton
14241: Ertl, @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z, A
14242: Portable Forth Engine}}, EuroForth '93.
14243: 
14244: @menu
14245: * Portability::                 
14246: * Threading::                   
14247: * Primitives::                  
14248: * Performance::                 
14249: @end menu
14250: 
14251: @node Portability, Threading, Engine, Engine
14252: @section Portability
14253: @cindex engine portability
14254: 
14255: An important goal of the Gforth Project is availability across a wide
14256: range of personal machines. fig-Forth, and, to a lesser extent, F83,
14257: achieved this goal by manually coding the engine in assembly language
14258: for several then-popular processors. This approach is very
14259: labor-intensive and the results are short-lived due to progress in
14260: computer architecture.
14261: 
14262: @cindex C, using C for the engine
14263: Others have avoided this problem by coding in C, e.g., Mitch Bradley
14264: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
14265: particularly popular for UNIX-based Forths due to the large variety of
14266: architectures of UNIX machines. Unfortunately an implementation in C
14267: does not mix well with the goals of efficiency and with using
14268: traditional techniques: Indirect or direct threading cannot be expressed
14269: in C, and switch threading, the fastest technique available in C, is
14270: significantly slower. Another problem with C is that it is very
14271: cumbersome to express double integer arithmetic.
14272: 
14273: @cindex GNU C for the engine
14274: @cindex long long
14275: Fortunately, there is a portable language that does not have these
14276: limitations: GNU C, the version of C processed by the GNU C compiler
14277: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
14278: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
14279: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
14280: threading possible, its @code{long long} type (@pxref{Long Long, ,
14281: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
14282: double numbers@footnote{Unfortunately, long longs are not implemented
14283: properly on all machines (e.g., on alpha-osf1, long longs are only 64
14284: bits, the same size as longs (and pointers), but they should be twice as
14285: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
14286: C Manual}). So, we had to implement doubles in C after all. Still, on
14287: most machines we can use long longs and achieve better performance than
14288: with the emulation package.}. GNU C is available for free on all
14289: important (and many unimportant) UNIX machines, VMS, 80386s running
14290: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
14291: on all these machines.
14292: 
14293: Writing in a portable language has the reputation of producing code that
14294: is slower than assembly. For our Forth engine we repeatedly looked at
14295: the code produced by the compiler and eliminated most compiler-induced
14296: inefficiencies by appropriate changes in the source code.
14297: 
14298: @cindex explicit register declarations
14299: @cindex --enable-force-reg, configuration flag
14300: @cindex -DFORCE_REG
14301: However, register allocation cannot be portably influenced by the
14302: programmer, leading to some inefficiencies on register-starved
14303: machines. We use explicit register declarations (@pxref{Explicit Reg
14304: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
14305: improve the speed on some machines. They are turned on by using the
14306: configuration flag @code{--enable-force-reg} (@code{gcc} switch
14307: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
14308: machine, but also on the compiler version: On some machines some
14309: compiler versions produce incorrect code when certain explicit register
14310: declarations are used. So by default @code{-DFORCE_REG} is not used.
14311: 
14312: @node Threading, Primitives, Portability, Engine
14313: @section Threading
14314: @cindex inner interpreter implementation
14315: @cindex threaded code implementation
14316: 
14317: @cindex labels as values
14318: GNU C's labels as values extension (available since @code{gcc-2.0},
14319: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
14320: makes it possible to take the address of @i{label} by writing
14321: @code{&&@i{label}}.  This address can then be used in a statement like
14322: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
14323: @code{goto x}.
14324: 
14325: @cindex @code{NEXT}, indirect threaded
14326: @cindex indirect threaded inner interpreter
14327: @cindex inner interpreter, indirect threaded
14328: With this feature an indirect threaded @code{NEXT} looks like:
14329: @example
14330: cfa = *ip++;
14331: ca = *cfa;
14332: goto *ca;
14333: @end example
14334: @cindex instruction pointer
14335: For those unfamiliar with the names: @code{ip} is the Forth instruction
14336: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
14337: execution token and points to the code field of the next word to be
14338: executed; The @code{ca} (code address) fetched from there points to some
14339: executable code, e.g., a primitive or the colon definition handler
14340: @code{docol}.
14341: 
14342: @cindex @code{NEXT}, direct threaded
14343: @cindex direct threaded inner interpreter
14344: @cindex inner interpreter, direct threaded
14345: Direct threading is even simpler:
14346: @example
14347: ca = *ip++;
14348: goto *ca;
14349: @end example
14350: 
14351: Of course we have packaged the whole thing neatly in macros called
14352: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
14353: 
14354: @menu
14355: * Scheduling::                  
14356: * Direct or Indirect Threaded?::  
14357: * DOES>::                       
14358: @end menu
14359: 
14360: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
14361: @subsection Scheduling
14362: @cindex inner interpreter optimization
14363: 
14364: There is a little complication: Pipelined and superscalar processors,
14365: i.e., RISC and some modern CISC machines can process independent
14366: instructions while waiting for the results of an instruction. The
14367: compiler usually reorders (schedules) the instructions in a way that
14368: achieves good usage of these delay slots. However, on our first tries
14369: the compiler did not do well on scheduling primitives. E.g., for
14370: @code{+} implemented as
14371: @example
14372: n=sp[0]+sp[1];
14373: sp++;
14374: sp[0]=n;
14375: NEXT;
14376: @end example
14377: the @code{NEXT} comes strictly after the other code, i.e., there is
14378: nearly no scheduling. After a little thought the problem becomes clear:
14379: The compiler cannot know that @code{sp} and @code{ip} point to different
14380: addresses (and the version of @code{gcc} we used would not know it even
14381: if it was possible), so it could not move the load of the cfa above the
14382: store to the TOS. Indeed the pointers could be the same, if code on or
14383: very near the top of stack were executed. In the interest of speed we
14384: chose to forbid this probably unused ``feature'' and helped the compiler
14385: in scheduling: @code{NEXT} is divided into several parts:
14386: @code{NEXT_P0}, @code{NEXT_P1} and @code{NEXT_P2}). @code{+} now looks
14387: like:
14388: @example
14389: NEXT_P0;
14390: n=sp[0]+sp[1];
14391: sp++;
14392: NEXT_P1;
14393: sp[0]=n;
14394: NEXT_P2;
14395: @end example
14396: 
14397: There are various schemes that distribute the different operations of
14398: NEXT between these parts in several ways; in general, different schemes
14399: perform best on different processors.  We use a scheme for most
14400: architectures that performs well for most processors of this
14401: architecture; in the furture we may switch to benchmarking and chosing
14402: the scheme on installation time.
14403: 
14404: 
14405: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
14406: @subsection Direct or Indirect Threaded?
14407: @cindex threading, direct or indirect?
14408: 
14409: @cindex -DDIRECT_THREADED
14410: Both! After packaging the nasty details in macro definitions we
14411: realized that we could switch between direct and indirect threading by
14412: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
14413: defining a few machine-specific macros for the direct-threading case.
14414: On the Forth level we also offer access words that hide the
14415: differences between the threading methods (@pxref{Threading Words}).
14416: 
14417: Indirect threading is implemented completely machine-independently.
14418: Direct threading needs routines for creating jumps to the executable
14419: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
14420: machine-dependent, but they do not amount to many source lines. Therefore,
14421: even porting direct threading to a new machine requires little effort.
14422: 
14423: @cindex --enable-indirect-threaded, configuration flag
14424: @cindex --enable-direct-threaded, configuration flag
14425: The default threading method is machine-dependent. You can enforce a
14426: specific threading method when building Gforth with the configuration
14427: flag @code{--enable-direct-threaded} or
14428: @code{--enable-indirect-threaded}. Note that direct threading is not
14429: supported on all machines.
14430: 
14431: @node DOES>,  , Direct or Indirect Threaded?, Threading
14432: @subsection DOES>
14433: @cindex @code{DOES>} implementation