File:  [gforth] / gforth / doc / gforth.ds
Revision 1.106: download - view: text, annotated - select for diffs
Tue Jan 21 10:24:44 2003 UTC (21 years, 2 months ago) by anton
Branches: MAIN
CVS tags: HEAD
loadfilename#>str is now safer
documented the interaction of markers and ~~ and assertions

    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
  173:  --- The Detailed Node Listing ---
  174: 
  175: Gforth Environment
  176: 
  177: * Invoking Gforth::             Getting in
  178: * Leaving Gforth::              Getting out
  179: * Command-line editing::        
  180: * Environment variables::       that affect how Gforth starts up
  181: * Gforth Files::                What gets installed and where
  182: * Startup speed::               When 35ms is not fast enough ...
  183: 
  184: Forth Tutorial
  185: 
  186: * Starting Gforth Tutorial::    
  187: * Syntax Tutorial::             
  188: * Crash Course Tutorial::       
  189: * Stack Tutorial::              
  190: * Arithmetics Tutorial::        
  191: * Stack Manipulation Tutorial::  
  192: * Using files for Forth code Tutorial::  
  193: * Comments Tutorial::           
  194: * Colon Definitions Tutorial::  
  195: * Decompilation Tutorial::      
  196: * Stack-Effect Comments Tutorial::  
  197: * Types Tutorial::              
  198: * Factoring Tutorial::          
  199: * Designing the stack effect Tutorial::  
  200: * Local Variables Tutorial::    
  201: * Conditional execution Tutorial::  
  202: * Flags and Comparisons Tutorial::  
  203: * General Loops Tutorial::      
  204: * Counted loops Tutorial::      
  205: * Recursion Tutorial::          
  206: * Leaving definitions or loops Tutorial::  
  207: * Return Stack Tutorial::       
  208: * Memory Tutorial::             
  209: * Characters and Strings Tutorial::  
  210: * Alignment Tutorial::          
  211: * Files Tutorial::              
  212: * Interpretation and Compilation Semantics and Immediacy Tutorial::  
  213: * Execution Tokens Tutorial::   
  214: * Exceptions Tutorial::         
  215: * Defining Words Tutorial::     
  216: * Arrays and Records Tutorial::  
  217: * POSTPONE Tutorial::           
  218: * Literal Tutorial::            
  219: * Advanced macros Tutorial::    
  220: * Compilation Tokens Tutorial::  
  221: * Wordlists and Search Order Tutorial::  
  222: 
  223: An Introduction to ANS Forth
  224: 
  225: * Introducing the Text Interpreter::  
  226: * Stacks and Postfix notation::  
  227: * Your first definition::       
  228: * How does that work?::         
  229: * Forth is written in Forth::   
  230: * Review - elements of a Forth system::  
  231: * Where to go next::            
  232: * Exercises::                   
  233: 
  234: Forth Words
  235: 
  236: * Notation::                    
  237: * Case insensitivity::          
  238: * Comments::                    
  239: * Boolean Flags::               
  240: * Arithmetic::                  
  241: * Stack Manipulation::          
  242: * Memory::                      
  243: * Control Structures::          
  244: * Defining Words::              
  245: * Interpretation and Compilation Semantics::  
  246: * Tokens for Words::            
  247: * Compiling words::             
  248: * The Text Interpreter::        
  249: * Word Lists::                  
  250: * Environmental Queries::       
  251: * Files::                       
  252: * Blocks::                      
  253: * Other I/O::                   
  254: * Locals::                      
  255: * Structures::                  
  256: * Object-oriented Forth::       
  257: * Programming Tools::           
  258: * Assembler and Code Words::    
  259: * Threading Words::             
  260: * Passing Commands to the OS::  
  261: * Keeping track of Time::       
  262: * Miscellaneous Words::         
  263: 
  264: Arithmetic
  265: 
  266: * Single precision::            
  267: * Double precision::            Double-cell integer arithmetic
  268: * Bitwise operations::          
  269: * Numeric comparison::          
  270: * Mixed precision::             Operations with single and double-cell integers
  271: * Floating Point::              
  272: 
  273: Stack Manipulation
  274: 
  275: * Data stack::                  
  276: * Floating point stack::        
  277: * Return stack::                
  278: * Locals stack::                
  279: * Stack pointer manipulation::  
  280: 
  281: Memory
  282: 
  283: * Memory model::                
  284: * Dictionary allocation::       
  285: * Heap Allocation::             
  286: * Memory Access::               
  287: * Address arithmetic::          
  288: * Memory Blocks::               
  289: 
  290: Control Structures
  291: 
  292: * Selection::                   IF ... ELSE ... ENDIF
  293: * Simple Loops::                BEGIN ...
  294: * Counted Loops::               DO
  295: * Arbitrary control structures::  
  296: * Calls and returns::           
  297: * Exception Handling::          
  298: 
  299: Defining Words
  300: 
  301: * CREATE::                      
  302: * Variables::                   Variables and user variables
  303: * Constants::                   
  304: * Values::                      Initialised variables
  305: * Colon Definitions::           
  306: * Anonymous Definitions::       Definitions without names
  307: * Supplying names::             Passing definition names as strings
  308: * User-defined Defining Words::  
  309: * Deferred words::              Allow forward references
  310: * Aliases::                     
  311: 
  312: User-defined Defining Words
  313: 
  314: * CREATE..DOES> applications::  
  315: * CREATE..DOES> details::       
  316: * Advanced does> usage example::  
  317: * @code{Const-does>}::          
  318: 
  319: Interpretation and Compilation Semantics
  320: 
  321: * Combined words::              
  322: 
  323: Tokens for Words
  324: 
  325: * Execution token::             represents execution/interpretation semantics
  326: * Compilation token::           represents compilation semantics
  327: * Name token::                  represents named words
  328: 
  329: Compiling words
  330: 
  331: * Literals::                    Compiling data values
  332: * Macros::                      Compiling words
  333: 
  334: The Text Interpreter
  335: 
  336: * Input Sources::               
  337: * Number Conversion::           
  338: * Interpret/Compile states::    
  339: * Interpreter Directives::      
  340: 
  341: Word Lists
  342: 
  343: * Vocabularies::                
  344: * Why use word lists?::         
  345: * Word list example::           
  346: 
  347: Files
  348: 
  349: * Forth source files::          
  350: * General files::               
  351: * Search Paths::                
  352: 
  353: Search Paths
  354: 
  355: * Source Search Paths::         
  356: * General Search Paths::        
  357: 
  358: Other I/O
  359: 
  360: * Simple numeric output::       Predefined formats
  361: * Formatted numeric output::    Formatted (pictured) output
  362: * String Formats::              How Forth stores strings in memory
  363: * Displaying characters and strings::  Other stuff
  364: * Input::                       Input
  365: 
  366: Locals
  367: 
  368: * Gforth locals::               
  369: * ANS Forth locals::            
  370: 
  371: Gforth locals
  372: 
  373: * Where are locals visible by name?::  
  374: * How long do locals live?::    
  375: * Locals programming style::    
  376: * Locals implementation::       
  377: 
  378: Structures
  379: 
  380: * Why explicit structure support?::  
  381: * Structure Usage::             
  382: * Structure Naming Convention::  
  383: * Structure Implementation::    
  384: * Structure Glossary::          
  385: 
  386: Object-oriented Forth
  387: 
  388: * Why object-oriented programming?::  
  389: * Object-Oriented Terminology::  
  390: * Objects::                     
  391: * OOF::                         
  392: * Mini-OOF::                    
  393: * Comparison with other object models::  
  394: 
  395: The @file{objects.fs} model
  396: 
  397: * Properties of the Objects model::  
  398: * Basic Objects Usage::         
  399: * The Objects base class::      
  400: * Creating objects::            
  401: * Object-Oriented Programming Style::  
  402: * Class Binding::               
  403: * Method conveniences::         
  404: * Classes and Scoping::         
  405: * Dividing classes::            
  406: * Object Interfaces::           
  407: * Objects Implementation::      
  408: * Objects Glossary::            
  409: 
  410: The @file{oof.fs} model
  411: 
  412: * Properties of the OOF model::  
  413: * Basic OOF Usage::             
  414: * The OOF base class::          
  415: * Class Declaration::           
  416: * Class Implementation::        
  417: 
  418: The @file{mini-oof.fs} model
  419: 
  420: * Basic Mini-OOF Usage::        
  421: * Mini-OOF Example::            
  422: * Mini-OOF Implementation::     
  423: 
  424: Programming Tools
  425: 
  426: * Examining::                   
  427: * Forgetting words::            
  428: * Debugging::                   Simple and quick.
  429: * Assertions::                  Making your programs self-checking.
  430: * Singlestep Debugger::         Executing your program word by word.
  431: 
  432: Assembler and Code Words
  433: 
  434: * Code and ;code::              
  435: * Common Assembler::            Assembler Syntax
  436: * Common Disassembler::         
  437: * 386 Assembler::               Deviations and special cases
  438: * Alpha Assembler::             Deviations and special cases
  439: * MIPS assembler::              Deviations and special cases
  440: * Other assemblers::            How to write them
  441: 
  442: Tools
  443: 
  444: * ANS Report::                  Report the words used, sorted by wordset.
  445: 
  446: ANS conformance
  447: 
  448: * The Core Words::              
  449: * The optional Block word set::  
  450: * The optional Double Number word set::  
  451: * The optional Exception word set::  
  452: * The optional Facility word set::  
  453: * The optional File-Access word set::  
  454: * The optional Floating-Point word set::  
  455: * The optional Locals word set::  
  456: * The optional Memory-Allocation word set::  
  457: * The optional Programming-Tools word set::  
  458: * The optional Search-Order word set::  
  459: 
  460: The Core Words
  461: 
  462: * core-idef::                   Implementation Defined Options                   
  463: * core-ambcond::                Ambiguous Conditions                
  464: * core-other::                  Other System Documentation                  
  465: 
  466: The optional Block word set
  467: 
  468: * block-idef::                  Implementation Defined Options
  469: * block-ambcond::               Ambiguous Conditions               
  470: * block-other::                 Other System Documentation                 
  471: 
  472: The optional Double Number word set
  473: 
  474: * double-ambcond::              Ambiguous Conditions              
  475: 
  476: The optional Exception word set
  477: 
  478: * exception-idef::              Implementation Defined Options              
  479: 
  480: The optional Facility word set
  481: 
  482: * facility-idef::               Implementation Defined Options               
  483: * facility-ambcond::            Ambiguous Conditions            
  484: 
  485: The optional File-Access word set
  486: 
  487: * file-idef::                   Implementation Defined Options
  488: * file-ambcond::                Ambiguous Conditions                
  489: 
  490: The optional Floating-Point word set
  491: 
  492: * floating-idef::               Implementation Defined Options
  493: * floating-ambcond::            Ambiguous Conditions            
  494: 
  495: The optional Locals word set
  496: 
  497: * locals-idef::                 Implementation Defined Options                 
  498: * locals-ambcond::              Ambiguous Conditions              
  499: 
  500: The optional Memory-Allocation word set
  501: 
  502: * memory-idef::                 Implementation Defined Options                 
  503: 
  504: The optional Programming-Tools word set
  505: 
  506: * programming-idef::            Implementation Defined Options            
  507: * programming-ambcond::         Ambiguous Conditions         
  508: 
  509: The optional Search-Order word set
  510: 
  511: * search-idef::                 Implementation Defined Options                 
  512: * search-ambcond::              Ambiguous Conditions              
  513: 
  514: Image Files
  515: 
  516: * Image Licensing Issues::      Distribution terms for images.
  517: * Image File Background::       Why have image files?
  518: * Non-Relocatable Image Files::  don't always work.
  519: * Data-Relocatable Image Files::  are better.
  520: * Fully Relocatable Image Files::  better yet.
  521: * Stack and Dictionary Sizes::  Setting the default sizes for an image.
  522: * Running Image Files::         @code{gforth -i @i{file}} or @i{file}.
  523: * Modifying the Startup Sequence::  and turnkey applications.
  524: 
  525: Fully Relocatable Image Files
  526: 
  527: * gforthmi::                    The normal way
  528: * cross.fs::                    The hard way
  529: 
  530: Engine
  531: 
  532: * Portability::                 
  533: * Threading::                   
  534: * Primitives::                  
  535: * Performance::                 
  536: 
  537: Threading
  538: 
  539: * Scheduling::                  
  540: * Direct or Indirect Threaded?::  
  541: * DOES>::                       
  542: 
  543: Primitives
  544: 
  545: * Automatic Generation::        
  546: * TOS Optimization::            
  547: * Produced code::               
  548: 
  549: Cross Compiler
  550: 
  551: * Using the Cross Compiler::    
  552: * How the Cross Compiler Works::  
  553: 
  554: @end detailmenu
  555: @end menu
  556: 
  557: @node License, Goals, Top, Top
  558: @unnumbered GNU GENERAL PUBLIC LICENSE
  559: @center Version 2, June 1991
  560: 
  561: @display
  562: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
  563: 59 Temple Place, Suite 330, Boston, MA 02111, USA
  564: 
  565: Everyone is permitted to copy and distribute verbatim copies
  566: of this license document, but changing it is not allowed.
  567: @end display
  568: 
  569: @unnumberedsec Preamble
  570: 
  571:   The licenses for most software are designed to take away your
  572: freedom to share and change it.  By contrast, the GNU General Public
  573: License is intended to guarantee your freedom to share and change free
  574: software---to make sure the software is free for all its users.  This
  575: General Public License applies to most of the Free Software
  576: Foundation's software and to any other program whose authors commit to
  577: using it.  (Some other Free Software Foundation software is covered by
  578: the GNU Library General Public License instead.)  You can apply it to
  579: your programs, too.
  580: 
  581:   When we speak of free software, we are referring to freedom, not
  582: price.  Our General Public Licenses are designed to make sure that you
  583: have the freedom to distribute copies of free software (and charge for
  584: this service if you wish), that you receive source code or can get it
  585: if you want it, that you can change the software or use pieces of it
  586: in new free programs; and that you know you can do these things.
  587: 
  588:   To protect your rights, we need to make restrictions that forbid
  589: anyone to deny you these rights or to ask you to surrender the rights.
  590: These restrictions translate to certain responsibilities for you if you
  591: distribute copies of the software, or if you modify it.
  592: 
  593:   For example, if you distribute copies of such a program, whether
  594: gratis or for a fee, you must give the recipients all the rights that
  595: you have.  You must make sure that they, too, receive or can get the
  596: source code.  And you must show them these terms so they know their
  597: rights.
  598: 
  599:   We protect your rights with two steps: (1) copyright the software, and
  600: (2) offer you this license which gives you legal permission to copy,
  601: distribute and/or modify the software.
  602: 
  603:   Also, for each author's protection and ours, we want to make certain
  604: that everyone understands that there is no warranty for this free
  605: software.  If the software is modified by someone else and passed on, we
  606: want its recipients to know that what they have is not the original, so
  607: that any problems introduced by others will not reflect on the original
  608: authors' reputations.
  609: 
  610:   Finally, any free program is threatened constantly by software
  611: patents.  We wish to avoid the danger that redistributors of a free
  612: program will individually obtain patent licenses, in effect making the
  613: program proprietary.  To prevent this, we have made it clear that any
  614: patent must be licensed for everyone's free use or not licensed at all.
  615: 
  616:   The precise terms and conditions for copying, distribution and
  617: modification follow.
  618: 
  619: @iftex
  620: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  621: @end iftex
  622: @ifnottex
  623: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  624: @end ifnottex
  625: 
  626: @enumerate 0
  627: @item
  628: This License applies to any program or other work which contains
  629: a notice placed by the copyright holder saying it may be distributed
  630: under the terms of this General Public License.  The ``Program'', below,
  631: refers to any such program or work, and a ``work based on the Program''
  632: means either the Program or any derivative work under copyright law:
  633: that is to say, a work containing the Program or a portion of it,
  634: either verbatim or with modifications and/or translated into another
  635: language.  (Hereinafter, translation is included without limitation in
  636: the term ``modification''.)  Each licensee is addressed as ``you''.
  637: 
  638: Activities other than copying, distribution and modification are not
  639: covered by this License; they are outside its scope.  The act of
  640: running the Program is not restricted, and the output from the Program
  641: is covered only if its contents constitute a work based on the
  642: Program (independent of having been made by running the Program).
  643: Whether that is true depends on what the Program does.
  644: 
  645: @item
  646: You may copy and distribute verbatim copies of the Program's
  647: source code as you receive it, in any medium, provided that you
  648: conspicuously and appropriately publish on each copy an appropriate
  649: copyright notice and disclaimer of warranty; keep intact all the
  650: notices that refer to this License and to the absence of any warranty;
  651: and give any other recipients of the Program a copy of this License
  652: along with the Program.
  653: 
  654: You may charge a fee for the physical act of transferring a copy, and
  655: you may at your option offer warranty protection in exchange for a fee.
  656: 
  657: @item
  658: You may modify your copy or copies of the Program or any portion
  659: of it, thus forming a work based on the Program, and copy and
  660: distribute such modifications or work under the terms of Section 1
  661: above, provided that you also meet all of these conditions:
  662: 
  663: @enumerate a
  664: @item
  665: You must cause the modified files to carry prominent notices
  666: stating that you changed the files and the date of any change.
  667: 
  668: @item
  669: You must cause any work that you distribute or publish, that in
  670: whole or in part contains or is derived from the Program or any
  671: part thereof, to be licensed as a whole at no charge to all third
  672: parties under the terms of this License.
  673: 
  674: @item
  675: If the modified program normally reads commands interactively
  676: when run, you must cause it, when started running for such
  677: interactive use in the most ordinary way, to print or display an
  678: announcement including an appropriate copyright notice and a
  679: notice that there is no warranty (or else, saying that you provide
  680: a warranty) and that users may redistribute the program under
  681: these conditions, and telling the user how to view a copy of this
  682: License.  (Exception: if the Program itself is interactive but
  683: does not normally print such an announcement, your work based on
  684: the Program is not required to print an announcement.)
  685: @end enumerate
  686: 
  687: These requirements apply to the modified work as a whole.  If
  688: identifiable sections of that work are not derived from the Program,
  689: and can be reasonably considered independent and separate works in
  690: themselves, then this License, and its terms, do not apply to those
  691: sections when you distribute them as separate works.  But when you
  692: distribute the same sections as part of a whole which is a work based
  693: on the Program, the distribution of the whole must be on the terms of
  694: this License, whose permissions for other licensees extend to the
  695: entire whole, and thus to each and every part regardless of who wrote it.
  696: 
  697: Thus, it is not the intent of this section to claim rights or contest
  698: your rights to work written entirely by you; rather, the intent is to
  699: exercise the right to control the distribution of derivative or
  700: collective works based on the Program.
  701: 
  702: In addition, mere aggregation of another work not based on the Program
  703: with the Program (or with a work based on the Program) on a volume of
  704: a storage or distribution medium does not bring the other work under
  705: the scope of this License.
  706: 
  707: @item
  708: You may copy and distribute the Program (or a work based on it,
  709: under Section 2) in object code or executable form under the terms of
  710: Sections 1 and 2 above provided that you also do one of the following:
  711: 
  712: @enumerate a
  713: @item
  714: Accompany it with the complete corresponding machine-readable
  715: source code, which must be distributed under the terms of Sections
  716: 1 and 2 above on a medium customarily used for software interchange; or,
  717: 
  718: @item
  719: Accompany it with a written offer, valid for at least three
  720: years, to give any third party, for a charge no more than your
  721: cost of physically performing source distribution, a complete
  722: machine-readable copy of the corresponding source code, to be
  723: distributed under the terms of Sections 1 and 2 above on a medium
  724: customarily used for software interchange; or,
  725: 
  726: @item
  727: Accompany it with the information you received as to the offer
  728: to distribute corresponding source code.  (This alternative is
  729: allowed only for noncommercial distribution and only if you
  730: received the program in object code or executable form with such
  731: an offer, in accord with Subsection b above.)
  732: @end enumerate
  733: 
  734: The source code for a work means the preferred form of the work for
  735: making modifications to it.  For an executable work, complete source
  736: code means all the source code for all modules it contains, plus any
  737: associated interface definition files, plus the scripts used to
  738: control compilation and installation of the executable.  However, as a
  739: special exception, the source code distributed need not include
  740: anything that is normally distributed (in either source or binary
  741: form) with the major components (compiler, kernel, and so on) of the
  742: operating system on which the executable runs, unless that component
  743: itself accompanies the executable.
  744: 
  745: If distribution of executable or object code is made by offering
  746: access to copy from a designated place, then offering equivalent
  747: access to copy the source code from the same place counts as
  748: distribution of the source code, even though third parties are not
  749: compelled to copy the source along with the object code.
  750: 
  751: @item
  752: You may not copy, modify, sublicense, or distribute the Program
  753: except as expressly provided under this License.  Any attempt
  754: otherwise to copy, modify, sublicense or distribute the Program is
  755: void, and will automatically terminate your rights under this License.
  756: However, parties who have received copies, or rights, from you under
  757: this License will not have their licenses terminated so long as such
  758: parties remain in full compliance.
  759: 
  760: @item
  761: You are not required to accept this License, since you have not
  762: signed it.  However, nothing else grants you permission to modify or
  763: distribute the Program or its derivative works.  These actions are
  764: prohibited by law if you do not accept this License.  Therefore, by
  765: modifying or distributing the Program (or any work based on the
  766: Program), you indicate your acceptance of this License to do so, and
  767: all its terms and conditions for copying, distributing or modifying
  768: the Program or works based on it.
  769: 
  770: @item
  771: Each time you redistribute the Program (or any work based on the
  772: Program), the recipient automatically receives a license from the
  773: original licensor to copy, distribute or modify the Program subject to
  774: these terms and conditions.  You may not impose any further
  775: restrictions on the recipients' exercise of the rights granted herein.
  776: You are not responsible for enforcing compliance by third parties to
  777: this License.
  778: 
  779: @item
  780: If, as a consequence of a court judgment or allegation of patent
  781: infringement or for any other reason (not limited to patent issues),
  782: conditions are imposed on you (whether by court order, agreement or
  783: otherwise) that contradict the conditions of this License, they do not
  784: excuse you from the conditions of this License.  If you cannot
  785: distribute so as to satisfy simultaneously your obligations under this
  786: License and any other pertinent obligations, then as a consequence you
  787: may not distribute the Program at all.  For example, if a patent
  788: license would not permit royalty-free redistribution of the Program by
  789: all those who receive copies directly or indirectly through you, then
  790: the only way you could satisfy both it and this License would be to
  791: refrain entirely from distribution of the Program.
  792: 
  793: If any portion of this section is held invalid or unenforceable under
  794: any particular circumstance, the balance of the section is intended to
  795: apply and the section as a whole is intended to apply in other
  796: circumstances.
  797: 
  798: It is not the purpose of this section to induce you to infringe any
  799: patents or other property right claims or to contest validity of any
  800: such claims; this section has the sole purpose of protecting the
  801: integrity of the free software distribution system, which is
  802: implemented by public license practices.  Many people have made
  803: generous contributions to the wide range of software distributed
  804: through that system in reliance on consistent application of that
  805: system; it is up to the author/donor to decide if he or she is willing
  806: to distribute software through any other system and a licensee cannot
  807: impose that choice.
  808: 
  809: This section is intended to make thoroughly clear what is believed to
  810: be a consequence of the rest of this License.
  811: 
  812: @item
  813: If the distribution and/or use of the Program is restricted in
  814: certain countries either by patents or by copyrighted interfaces, the
  815: original copyright holder who places the Program under this License
  816: may add an explicit geographical distribution limitation excluding
  817: those countries, so that distribution is permitted only in or among
  818: countries not thus excluded.  In such case, this License incorporates
  819: the limitation as if written in the body of this License.
  820: 
  821: @item
  822: The Free Software Foundation may publish revised and/or new versions
  823: of the General Public License from time to time.  Such new versions will
  824: be similar in spirit to the present version, but may differ in detail to
  825: address new problems or concerns.
  826: 
  827: Each version is given a distinguishing version number.  If the Program
  828: specifies a version number of this License which applies to it and ``any
  829: later version'', you have the option of following the terms and conditions
  830: either of that version or of any later version published by the Free
  831: Software Foundation.  If the Program does not specify a version number of
  832: this License, you may choose any version ever published by the Free Software
  833: Foundation.
  834: 
  835: @item
  836: If you wish to incorporate parts of the Program into other free
  837: programs whose distribution conditions are different, write to the author
  838: to ask for permission.  For software which is copyrighted by the Free
  839: Software Foundation, write to the Free Software Foundation; we sometimes
  840: make exceptions for this.  Our decision will be guided by the two goals
  841: of preserving the free status of all derivatives of our free software and
  842: of promoting the sharing and reuse of software generally.
  843: 
  844: @iftex
  845: @heading NO WARRANTY
  846: @end iftex
  847: @ifnottex
  848: @center NO WARRANTY
  849: @end ifnottex
  850: 
  851: @item
  852: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
  853: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
  854: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
  855: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
  856: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  857: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
  858: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
  859: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
  860: REPAIR OR CORRECTION.
  861: 
  862: @item
  863: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
  864: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
  865: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
  866: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
  867: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
  868: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
  869: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
  870: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
  871: POSSIBILITY OF SUCH DAMAGES.
  872: @end enumerate
  873: 
  874: @iftex
  875: @heading END OF TERMS AND CONDITIONS
  876: @end iftex
  877: @ifnottex
  878: @center END OF TERMS AND CONDITIONS
  879: @end ifnottex
  880: 
  881: @page
  882: @unnumberedsec How to Apply These Terms to Your New Programs
  883: 
  884:   If you develop a new program, and you want it to be of the greatest
  885: possible use to the public, the best way to achieve this is to make it
  886: free software which everyone can redistribute and change under these terms.
  887: 
  888:   To do so, attach the following notices to the program.  It is safest
  889: to attach them to the start of each source file to most effectively
  890: convey the exclusion of warranty; and each file should have at least
  891: the ``copyright'' line and a pointer to where the full notice is found.
  892: 
  893: @smallexample
  894: @var{one line to give the program's name and a brief idea of what it does.}
  895: Copyright (C) 19@var{yy}  @var{name of author}
  896: 
  897: This program is free software; you can redistribute it and/or modify 
  898: it under the terms of the GNU General Public License as published by 
  899: the Free Software Foundation; either version 2 of the License, or 
  900: (at your option) any later version.
  901: 
  902: This program is distributed in the hope that it will be useful,
  903: but WITHOUT ANY WARRANTY; without even the implied warranty of
  904: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  905: GNU General Public License for more details.
  906: 
  907: You should have received a copy of the GNU General Public License
  908: along with this program; if not, write to the Free Software
  909: Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111, USA.
  910: @end smallexample
  911: 
  912: Also add information on how to contact you by electronic and paper mail.
  913: 
  914: If the program is interactive, make it output a short notice like this
  915: when it starts in an interactive mode:
  916: 
  917: @smallexample
  918: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
  919: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
  920: type `show w'.  
  921: This is free software, and you are welcome to redistribute it 
  922: under certain conditions; type `show c' for details.
  923: @end smallexample
  924: 
  925: The hypothetical commands @samp{show w} and @samp{show c} should show
  926: the appropriate parts of the General Public License.  Of course, the
  927: commands you use may be called something other than @samp{show w} and
  928: @samp{show c}; they could even be mouse-clicks or menu items---whatever
  929: suits your program.
  930: 
  931: You should also get your employer (if you work as a programmer) or your
  932: school, if any, to sign a ``copyright disclaimer'' for the program, if
  933: necessary.  Here is a sample; alter the names:
  934: 
  935: @smallexample
  936: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
  937: `Gnomovision' (which makes passes at compilers) written by James Hacker.
  938: 
  939: @var{signature of Ty Coon}, 1 April 1989
  940: Ty Coon, President of Vice
  941: @end smallexample
  942: 
  943: This General Public License does not permit incorporating your program into
  944: proprietary programs.  If your program is a subroutine library, you may
  945: consider it more useful to permit linking proprietary applications with the
  946: library.  If this is what you want to do, use the GNU Library General
  947: Public License instead of this License.
  948: 
  949: @iftex
  950: @unnumbered Preface
  951: @cindex Preface
  952: This manual documents Gforth. Some introductory material is provided for
  953: readers who are unfamiliar with Forth or who are migrating to Gforth
  954: from other Forth compilers. However, this manual is primarily a
  955: reference manual.
  956: @end iftex
  957: 
  958: @comment TODO much more blurb here.
  959: 
  960: @c ******************************************************************
  961: @node Goals, Gforth Environment, License, Top
  962: @comment node-name,     next,           previous, up
  963: @chapter Goals of Gforth
  964: @cindex goals of the Gforth project
  965: The goal of the Gforth Project is to develop a standard model for
  966: ANS Forth. This can be split into several subgoals:
  967: 
  968: @itemize @bullet
  969: @item
  970: Gforth should conform to the ANS Forth Standard.
  971: @item
  972: It should be a model, i.e. it should define all the
  973: implementation-dependent things.
  974: @item
  975: It should become standard, i.e. widely accepted and used. This goal
  976: is the most difficult one.
  977: @end itemize
  978: 
  979: To achieve these goals Gforth should be
  980: @itemize @bullet
  981: @item
  982: Similar to previous models (fig-Forth, F83)
  983: @item
  984: Powerful. It should provide for all the things that are considered
  985: necessary today and even some that are not yet considered necessary.
  986: @item
  987: Efficient. It should not get the reputation of being exceptionally
  988: slow.
  989: @item
  990: Free.
  991: @item
  992: Available on many machines/easy to port.
  993: @end itemize
  994: 
  995: Have we achieved these goals? Gforth conforms to the ANS Forth
  996: standard. It may be considered a model, but we have not yet documented
  997: which parts of the model are stable and which parts we are likely to
  998: change. It certainly has not yet become a de facto standard, but it
  999: appears to be quite popular. It has some similarities to and some
 1000: differences from previous models. It has some powerful features, but not
 1001: yet everything that we envisioned. We certainly have achieved our
 1002: execution speed goals (@pxref{Performance})@footnote{However, in 1998
 1003: the bar was raised when the major commercial Forth vendors switched to
 1004: native code compilers.}.  It is free and available on many machines.
 1005: 
 1006: @c ******************************************************************
 1007: @node Gforth Environment, Tutorial, Goals, Top
 1008: @chapter Gforth Environment
 1009: @cindex Gforth environment
 1010: 
 1011: Note: ultimately, the Gforth man page will be auto-generated from the
 1012: material in this chapter.
 1013: 
 1014: @menu
 1015: * Invoking Gforth::             Getting in
 1016: * Leaving Gforth::              Getting out
 1017: * Command-line editing::        
 1018: * Environment variables::       that affect how Gforth starts up
 1019: * Gforth Files::                What gets installed and where
 1020: * Startup speed::               When 35ms is not fast enough ...
 1021: @end menu
 1022: 
 1023: For related information about the creation of images see @ref{Image Files}.
 1024: 
 1025: @comment ----------------------------------------------
 1026: @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
 1027: @section Invoking Gforth
 1028: @cindex invoking Gforth
 1029: @cindex running Gforth
 1030: @cindex command-line options
 1031: @cindex options on the command line
 1032: @cindex flags on the command line
 1033: 
 1034: Gforth is made up of two parts; an executable ``engine'' (named
 1035: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
 1036: will usually just say @code{gforth} -- this automatically loads the
 1037: default image file @file{gforth.fi}. In many other cases the default
 1038: Gforth image will be invoked like this:
 1039: @example
 1040: gforth [file | -e forth-code] ...
 1041: @end example
 1042: @noindent
 1043: This interprets the contents of the files and the Forth code in the order they
 1044: are given.
 1045: 
 1046: In addition to the @file{gforth} engine, there is also an engine called
 1047: @file{gforth-fast}, which is faster, but gives less informative error
 1048: messages (@pxref{Error messages}) and may catch fewer stack underflows.
 1049: You should use it for debugged, performance-critical programs.
 1050: 
 1051: In general, the command line looks like this:
 1052: 
 1053: @example
 1054: gforth[-fast] [engine options] [image options]
 1055: @end example
 1056: 
 1057: The engine options must come before the rest of the command
 1058: line. They are:
 1059: 
 1060: @table @code
 1061: @cindex -i, command-line option
 1062: @cindex --image-file, command-line option
 1063: @item --image-file @i{file}
 1064: @itemx -i @i{file}
 1065: Loads the Forth image @i{file} instead of the default
 1066: @file{gforth.fi} (@pxref{Image Files}).
 1067: 
 1068: @cindex --appl-image, command-line option
 1069: @item --appl-image @i{file}
 1070: Loads the image @i{file} and leaves all further command-line arguments
 1071: to the image (instead of processing them as engine options).  This is
 1072: useful for building executable application images on Unix, built with
 1073: @code{gforthmi --application ...}.
 1074: 
 1075: @cindex --path, command-line option
 1076: @cindex -p, command-line option
 1077: @item --path @i{path}
 1078: @itemx -p @i{path}
 1079: Uses @i{path} for searching the image file and Forth source code files
 1080: instead of the default in the environment variable @code{GFORTHPATH} or
 1081: the path specified at installation time (e.g.,
 1082: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
 1083: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
 1084: 
 1085: @cindex --dictionary-size, command-line option
 1086: @cindex -m, command-line option
 1087: @cindex @i{size} parameters for command-line options
 1088: @cindex size of the dictionary and the stacks
 1089: @item --dictionary-size @i{size}
 1090: @itemx -m @i{size}
 1091: Allocate @i{size} space for the Forth dictionary space instead of
 1092: using the default specified in the image (typically 256K). The
 1093: @i{size} specification for this and subsequent options consists of
 1094: an integer and a unit (e.g.,
 1095: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
 1096: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
 1097: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
 1098: @code{e} is used.
 1099: 
 1100: @cindex --data-stack-size, command-line option
 1101: @cindex -d, command-line option
 1102: @item --data-stack-size @i{size}
 1103: @itemx -d @i{size}
 1104: Allocate @i{size} space for the data stack instead of using the
 1105: default specified in the image (typically 16K).
 1106: 
 1107: @cindex --return-stack-size, command-line option
 1108: @cindex -r, command-line option
 1109: @item --return-stack-size @i{size}
 1110: @itemx -r @i{size}
 1111: Allocate @i{size} space for the return stack instead of using the
 1112: default specified in the image (typically 15K).
 1113: 
 1114: @cindex --fp-stack-size, command-line option
 1115: @cindex -f, command-line option
 1116: @item --fp-stack-size @i{size}
 1117: @itemx -f @i{size}
 1118: Allocate @i{size} space for the floating point stack instead of
 1119: using the default specified in the image (typically 15.5K). In this case
 1120: the unit specifier @code{e} refers to floating point numbers.
 1121: 
 1122: @cindex --locals-stack-size, command-line option
 1123: @cindex -l, command-line option
 1124: @item --locals-stack-size @i{size}
 1125: @itemx -l @i{size}
 1126: Allocate @i{size} space for the locals stack instead of using the
 1127: default specified in the image (typically 14.5K).
 1128: 
 1129: @cindex -h, command-line option
 1130: @cindex --help, command-line option
 1131: @item --help
 1132: @itemx -h
 1133: Print a message about the command-line options
 1134: 
 1135: @cindex -v, command-line option
 1136: @cindex --version, command-line option
 1137: @item --version
 1138: @itemx -v
 1139: Print version and exit
 1140: 
 1141: @cindex --debug, command-line option
 1142: @item --debug
 1143: Print some information useful for debugging on startup.
 1144: 
 1145: @cindex --offset-image, command-line option
 1146: @item --offset-image
 1147: Start the dictionary at a slightly different position than would be used
 1148: otherwise (useful for creating data-relocatable images,
 1149: @pxref{Data-Relocatable Image Files}).
 1150: 
 1151: @cindex --no-offset-im, command-line option
 1152: @item --no-offset-im
 1153: Start the dictionary at the normal position.
 1154: 
 1155: @cindex --clear-dictionary, command-line option
 1156: @item --clear-dictionary
 1157: Initialize all bytes in the dictionary to 0 before loading the image
 1158: (@pxref{Data-Relocatable Image Files}).
 1159: 
 1160: @cindex --die-on-signal, command-line-option
 1161: @item --die-on-signal
 1162: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
 1163: or the segmentation violation SIGSEGV) by translating it into a Forth
 1164: @code{THROW}. With this option, Gforth exits if it receives such a
 1165: signal. This option is useful when the engine and/or the image might be
 1166: severely broken (such that it causes another signal before recovering
 1167: from the first); this option avoids endless loops in such cases.
 1168: @end table
 1169: 
 1170: @cindex loading files at startup
 1171: @cindex executing code on startup
 1172: @cindex batch processing with Gforth
 1173: As explained above, the image-specific command-line arguments for the
 1174: default image @file{gforth.fi} consist of a sequence of filenames and
 1175: @code{-e @var{forth-code}} options that are interpreted in the sequence
 1176: in which they are given. The @code{-e @var{forth-code}} or
 1177: @code{--evaluate @var{forth-code}} option evaluates the Forth
 1178: code. This option takes only one argument; if you want to evaluate more
 1179: Forth words, you have to quote them or use @code{-e} several times. To exit
 1180: after processing the command line (instead of entering interactive mode)
 1181: append @code{-e bye} to the command line.
 1182: 
 1183: @cindex versions, invoking other versions of Gforth
 1184: If you have several versions of Gforth installed, @code{gforth} will
 1185: invoke the version that was installed last. @code{gforth-@i{version}}
 1186: invokes a specific version. If your environment contains the variable
 1187: @code{GFORTHPATH}, you may want to override it by using the
 1188: @code{--path} option.
 1189: 
 1190: Not yet implemented:
 1191: On startup the system first executes the system initialization file
 1192: (unless the option @code{--no-init-file} is given; note that the system
 1193: resulting from using this option may not be ANS Forth conformant). Then
 1194: the user initialization file @file{.gforth.fs} is executed, unless the
 1195: option @code{--no-rc} is given; this file is searched for in @file{.},
 1196: then in @file{~}, then in the normal path (see above).
 1197: 
 1198: 
 1199: 
 1200: @comment ----------------------------------------------
 1201: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
 1202: @section Leaving Gforth
 1203: @cindex Gforth - leaving
 1204: @cindex leaving Gforth
 1205: 
 1206: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
 1207: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
 1208: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
 1209: data are discarded.  For ways of saving the state of the system before
 1210: leaving Gforth see @ref{Image Files}.
 1211: 
 1212: doc-bye
 1213: 
 1214: 
 1215: @comment ----------------------------------------------
 1216: @node Command-line editing, Environment variables, Leaving Gforth, Gforth Environment
 1217: @section Command-line editing
 1218: @cindex command-line editing
 1219: 
 1220: Gforth maintains a history file that records every line that you type to
 1221: the text interpreter. This file is preserved between sessions, and is
 1222: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
 1223: repeatedly you can recall successively older commands from this (or
 1224: previous) session(s). The full list of command-line editing facilities is:
 1225: 
 1226: @itemize @bullet
 1227: @item
 1228: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
 1229: commands from the history buffer.
 1230: @item
 1231: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
 1232: from the history buffer.
 1233: @item
 1234: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
 1235: @item
 1236: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
 1237: @item
 1238: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
 1239: closing up the line.
 1240: @item
 1241: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
 1242: @item
 1243: @kbd{Ctrl-a} to move the cursor to the start of the line.
 1244: @item
 1245: @kbd{Ctrl-e} to move the cursor to the end of the line.
 1246: @item
 1247: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
 1248: line.
 1249: @item
 1250: @key{TAB} to step through all possible full-word completions of the word
 1251: currently being typed.
 1252: @item
 1253: @kbd{Ctrl-d} on an empty line line to terminate Gforth (gracefully,
 1254: using @code{bye}). 
 1255: @item
 1256: @kbd{Ctrl-x} (or @code{Ctrl-d} on a non-empty line) to delete the
 1257: character under the cursor.
 1258: @end itemize
 1259: 
 1260: When editing, displayable characters are inserted to the left of the
 1261: cursor position; the line is always in ``insert'' (as opposed to
 1262: ``overstrike'') mode.
 1263: 
 1264: @cindex history file
 1265: @cindex @file{.gforth-history}
 1266: On Unix systems, the history file is @file{~/.gforth-history} by
 1267: default@footnote{i.e. it is stored in the user's home directory.}. You
 1268: can find out the name and location of your history file using:
 1269: 
 1270: @example 
 1271: history-file type \ Unix-class systems
 1272: 
 1273: history-file type \ Other systems
 1274: history-dir  type
 1275: @end example
 1276: 
 1277: If you enter long definitions by hand, you can use a text editor to
 1278: paste them out of the history file into a Forth source file for reuse at
 1279: a later time.
 1280: 
 1281: Gforth never trims the size of the history file, so you should do this
 1282: periodically, if necessary.
 1283: 
 1284: @comment this is all defined in history.fs
 1285: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
 1286: @comment chosen?
 1287: 
 1288: 
 1289: @comment ----------------------------------------------
 1290: @node Environment variables, Gforth Files, Command-line editing, Gforth Environment
 1291: @section Environment variables
 1292: @cindex environment variables
 1293: 
 1294: Gforth uses these environment variables:
 1295: 
 1296: @itemize @bullet
 1297: @item
 1298: @cindex @code{GFORTHHIST} -- environment variable
 1299: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
 1300: open/create the history file, @file{.gforth-history}. Default:
 1301: @code{$HOME}.
 1302: 
 1303: @item
 1304: @cindex @code{GFORTHPATH} -- environment variable
 1305: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
 1306: for Forth source-code files.
 1307: 
 1308: @item
 1309: @cindex @code{GFORTH} -- environment variable
 1310: @code{GFORTH} -- used by @file{gforthmi}, @xref{gforthmi}.
 1311: 
 1312: @item
 1313: @cindex @code{GFORTHD} -- environment variable
 1314: @code{GFORTHD} -- used by @file{gforthmi}, @xref{gforthmi}.
 1315: 
 1316: @item
 1317: @cindex @code{TMP}, @code{TEMP} - environment variable
 1318: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
 1319: location for the history file.
 1320: @end itemize
 1321: 
 1322: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
 1323: @comment mentioning these.
 1324: 
 1325: All the Gforth environment variables default to sensible values if they
 1326: are not set.
 1327: 
 1328: 
 1329: @comment ----------------------------------------------
 1330: @node Gforth Files, Startup speed, Environment variables, Gforth Environment
 1331: @section Gforth files
 1332: @cindex Gforth files
 1333: 
 1334: When you install Gforth on a Unix system, it installs files in these
 1335: locations by default:
 1336: 
 1337: @itemize @bullet
 1338: @item
 1339: @file{/usr/local/bin/gforth}
 1340: @item
 1341: @file{/usr/local/bin/gforthmi}
 1342: @item
 1343: @file{/usr/local/man/man1/gforth.1} - man page.
 1344: @item
 1345: @file{/usr/local/info} - the Info version of this manual.
 1346: @item
 1347: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
 1348: @item
 1349: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
 1350: @item
 1351: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
 1352: @item
 1353: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
 1354: @end itemize
 1355: 
 1356: You can select different places for installation by using
 1357: @code{configure} options (listed with @code{configure --help}).
 1358: 
 1359: @comment ----------------------------------------------
 1360: @node Startup speed,  , Gforth Files, Gforth Environment
 1361: @section Startup speed
 1362: @cindex Startup speed
 1363: @cindex speed, startup
 1364: 
 1365: If Gforth is used for CGI scripts or in shell scripts, its startup
 1366: speed may become a problem.  On a 300MHz 21064a under Linux-2.2.13 with
 1367: glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
 1368: system time.
 1369: 
 1370: If startup speed is a problem, you may consider the following ways to
 1371: improve it; or you may consider ways to reduce the number of startups
 1372: (for example, by using Fast-CGI).
 1373: 
 1374: The first step to improve startup speed is to statically link Gforth, by
 1375: building it with @code{XLDFLAGS=-static}.  This requires more memory for
 1376: the code and will therefore slow down the first invocation, but
 1377: subsequent invocations avoid the dynamic linking overhead.  Another
 1378: disadvantage is that Gforth won't profit from library upgrades.  As a
 1379: result, @code{gforth-static -e bye} takes about 17.1ms user and
 1380: 8.2ms system time.
 1381: 
 1382: The next step to improve startup speed is to use a non-relocatable image
 1383: (@pxref{Non-Relocatable Image Files}).  You can create this image with
 1384: @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
 1385: @code{gforth -i gforthnr.fi ...}.  This avoids the relocation overhead
 1386: and a part of the copy-on-write overhead.  The disadvantage is that the
 1387: non-relocatable image does not work if the OS gives Gforth a different
 1388: address for the dictionary, for whatever reason; so you better provide a
 1389: fallback on a relocatable image.  @code{gforth-static -i gforthnr.fi -e
 1390: bye} takes about 15.3ms user and 7.5ms system time.
 1391: 
 1392: The final step is to disable dictionary hashing in Gforth.  Gforth
 1393: builds the hash table on startup, which takes much of the startup
 1394: overhead. You can do this by commenting out the @code{include hash.fs}
 1395: in @file{startup.fs} and everything that requires @file{hash.fs} (at the
 1396: moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
 1397: The disadvantages are that functionality like @code{table} and
 1398: @code{ekey} is missing and that text interpretation (e.g., compiling)
 1399: now takes much longer. So, you should only use this method if there is
 1400: no significant text interpretation to perform (the script should be
 1401: compiled into the image, amongst other things).  @code{gforth-static -i
 1402: gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
 1403: 
 1404: @c ******************************************************************
 1405: @node Tutorial, Introduction, Gforth Environment, Top
 1406: @chapter Forth Tutorial
 1407: @cindex Tutorial
 1408: @cindex Forth Tutorial
 1409: 
 1410: @c Topics from nac's Introduction that could be mentioned:
 1411: @c press <ret> after each line
 1412: @c Prompt
 1413: @c numbers vs. words in dictionary on text interpretation
 1414: @c what happens on redefinition
 1415: @c parsing words (in particular, defining words)
 1416: 
 1417: The difference of this chapter from the Introduction
 1418: (@pxref{Introduction}) is that this tutorial is more fast-paced, should
 1419: be used while sitting in front of a computer, and covers much more
 1420: material, but does not explain how the Forth system works.
 1421: 
 1422: This tutorial can be used with any ANS-compliant Forth; any
 1423: Gforth-specific features are marked as such and you can skip them if you
 1424: work with another Forth.  This tutorial does not explain all features of
 1425: Forth, just enough to get you started and give you some ideas about the
 1426: facilities available in Forth.  Read the rest of the manual and the
 1427: standard when you are through this.
 1428: 
 1429: The intended way to use this tutorial is that you work through it while
 1430: sitting in front of the console, take a look at the examples and predict
 1431: what they will do, then try them out; if the outcome is not as expected,
 1432: find out why (e.g., by trying out variations of the example), so you
 1433: understand what's going on.  There are also some assignments that you
 1434: should solve.
 1435: 
 1436: This tutorial assumes that you have programmed before and know what,
 1437: e.g., a loop is.
 1438: 
 1439: @c !! explain compat library
 1440: 
 1441: @menu
 1442: * Starting Gforth Tutorial::    
 1443: * Syntax Tutorial::             
 1444: * Crash Course Tutorial::       
 1445: * Stack Tutorial::              
 1446: * Arithmetics Tutorial::        
 1447: * Stack Manipulation Tutorial::  
 1448: * Using files for Forth code Tutorial::  
 1449: * Comments Tutorial::           
 1450: * Colon Definitions Tutorial::  
 1451: * Decompilation Tutorial::      
 1452: * Stack-Effect Comments Tutorial::  
 1453: * Types Tutorial::              
 1454: * Factoring Tutorial::          
 1455: * Designing the stack effect Tutorial::  
 1456: * Local Variables Tutorial::    
 1457: * Conditional execution Tutorial::  
 1458: * Flags and Comparisons Tutorial::  
 1459: * General Loops Tutorial::      
 1460: * Counted loops Tutorial::      
 1461: * Recursion Tutorial::          
 1462: * Leaving definitions or loops Tutorial::  
 1463: * Return Stack Tutorial::       
 1464: * Memory Tutorial::             
 1465: * Characters and Strings Tutorial::  
 1466: * Alignment Tutorial::          
 1467: * Files Tutorial::              
 1468: * Interpretation and Compilation Semantics and Immediacy Tutorial::  
 1469: * Execution Tokens Tutorial::   
 1470: * Exceptions Tutorial::         
 1471: * Defining Words Tutorial::     
 1472: * Arrays and Records Tutorial::  
 1473: * POSTPONE Tutorial::           
 1474: * Literal Tutorial::            
 1475: * Advanced macros Tutorial::    
 1476: * Compilation Tokens Tutorial::  
 1477: * Wordlists and Search Order Tutorial::  
 1478: @end menu
 1479: 
 1480: @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
 1481: @section Starting Gforth
 1482: @cindex starting Gforth tutorial
 1483: You can start Gforth by typing its name:
 1484: 
 1485: @example
 1486: gforth
 1487: @end example
 1488: 
 1489: That puts you into interactive mode; you can leave Gforth by typing
 1490: @code{bye}.  While in Gforth, you can edit the command line and access
 1491: the command line history with cursor keys, similar to bash.
 1492: 
 1493: 
 1494: @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
 1495: @section Syntax
 1496: @cindex syntax tutorial
 1497: 
 1498: A @dfn{word} is a sequence of arbitrary characters (expcept white
 1499: space).  Words are separated by white space.  E.g., each of the
 1500: following lines contains exactly one word:
 1501: 
 1502: @example
 1503: word
 1504: !@@#$%^&*()
 1505: 1234567890
 1506: 5!a
 1507: @end example
 1508: 
 1509: A frequent beginner's error is to leave away necessary white space,
 1510: resulting in an error like @samp{Undefined word}; so if you see such an
 1511: error, check if you have put spaces wherever necessary.
 1512: 
 1513: @example
 1514: ." hello, world" \ correct
 1515: ."hello, world"  \ gives an "Undefined word" error
 1516: @end example
 1517: 
 1518: Gforth and most other Forth systems ignore differences in case (they are
 1519: case-insensitive), i.e., @samp{word} is the same as @samp{Word}.  If
 1520: your system is case-sensitive, you may have to type all the examples
 1521: given here in upper case.
 1522: 
 1523: 
 1524: @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
 1525: @section Crash Course
 1526: 
 1527: Type
 1528: 
 1529: @example
 1530: 0 0 !
 1531: here execute
 1532: ' catch >body 20 erase abort
 1533: ' (quit) >body 20 erase
 1534: @end example
 1535: 
 1536: The last two examples are guaranteed to destroy parts of Gforth (and
 1537: most other systems), so you better leave Gforth afterwards (if it has
 1538: not finished by itself).  On some systems you may have to kill gforth
 1539: from outside (e.g., in Unix with @code{kill}).
 1540: 
 1541: Now that you know how to produce crashes (and that there's not much to
 1542: them), let's learn how to produce meaningful programs.
 1543: 
 1544: 
 1545: @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
 1546: @section Stack
 1547: @cindex stack tutorial
 1548: 
 1549: The most obvious feature of Forth is the stack.  When you type in a
 1550: number, it is pushed on the stack.  You can display the content of the
 1551: stack with @code{.s}.
 1552: 
 1553: @example
 1554: 1 2 .s
 1555: 3 .s
 1556: @end example
 1557: 
 1558: @code{.s} displays the top-of-stack to the right, i.e., the numbers
 1559: appear in @code{.s} output as they appeared in the input.
 1560: 
 1561: You can print the top of stack element with @code{.}.
 1562: 
 1563: @example
 1564: 1 2 3 . . .
 1565: @end example
 1566: 
 1567: In general, words consume their stack arguments (@code{.s} is an
 1568: exception).
 1569: 
 1570: @assignment
 1571: What does the stack contain after @code{5 6 7 .}?
 1572: @endassignment
 1573: 
 1574: 
 1575: @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
 1576: @section Arithmetics
 1577: @cindex arithmetics tutorial
 1578: 
 1579: The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
 1580: operate on the top two stack items:
 1581: 
 1582: @example
 1583: 2 2 .s
 1584: + .s
 1585: .
 1586: 2 1 - .
 1587: 7 3 mod .
 1588: @end example
 1589: 
 1590: The operands of @code{-}, @code{/}, and @code{mod} are in the same order
 1591: as in the corresponding infix expression (this is generally the case in
 1592: Forth).
 1593: 
 1594: Parentheses are superfluous (and not available), because the order of
 1595: the words unambiguously determines the order of evaluation and the
 1596: operands:
 1597: 
 1598: @example
 1599: 3 4 + 5 * .
 1600: 3 4 5 * + .
 1601: @end example
 1602: 
 1603: @assignment
 1604: What are the infix expressions corresponding to the Forth code above?
 1605: Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
 1606: known as Postfix or RPN (Reverse Polish Notation).}.
 1607: @endassignment
 1608: 
 1609: To change the sign, use @code{negate}:
 1610: 
 1611: @example
 1612: 2 negate .
 1613: @end example
 1614: 
 1615: @assignment
 1616: Convert -(-3)*4-5 to Forth.
 1617: @endassignment
 1618: 
 1619: @code{/mod} performs both @code{/} and @code{mod}.
 1620: 
 1621: @example
 1622: 7 3 /mod . .
 1623: @end example
 1624: 
 1625: Reference: @ref{Arithmetic}.
 1626: 
 1627: 
 1628: @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
 1629: @section Stack Manipulation
 1630: @cindex stack manipulation tutorial
 1631: 
 1632: Stack manipulation words rearrange the data on the stack.
 1633: 
 1634: @example
 1635: 1 .s drop .s
 1636: 1 .s dup .s drop drop .s
 1637: 1 2 .s over .s drop drop drop
 1638: 1 2 .s swap .s drop drop
 1639: 1 2 3 .s rot .s drop drop drop
 1640: @end example
 1641: 
 1642: These are the most important stack manipulation words.  There are also
 1643: variants that manipulate twice as many stack items:
 1644: 
 1645: @example
 1646: 1 2 3 4 .s 2swap .s 2drop 2drop
 1647: @end example
 1648: 
 1649: Two more stack manipulation words are:
 1650: 
 1651: @example
 1652: 1 2 .s nip .s drop
 1653: 1 2 .s tuck .s 2drop drop
 1654: @end example
 1655: 
 1656: @assignment
 1657: Replace @code{nip} and @code{tuck} with combinations of other stack
 1658: manipulation words.
 1659: 
 1660: @example
 1661: Given:          How do you get:
 1662: 1 2 3           3 2 1           
 1663: 1 2 3           1 2 3 2                 
 1664: 1 2 3           1 2 3 3                 
 1665: 1 2 3           1 3 3           
 1666: 1 2 3           2 1 3           
 1667: 1 2 3 4         4 3 2 1         
 1668: 1 2 3           1 2 3 1 2 3             
 1669: 1 2 3 4         1 2 3 4 1 2             
 1670: 1 2 3
 1671: 1 2 3           1 2 3 4                 
 1672: 1 2 3           1 3             
 1673: @end example
 1674: @endassignment
 1675: 
 1676: @example
 1677: 5 dup * .
 1678: @end example
 1679: 
 1680: @assignment
 1681: Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
 1682: Write a piece of Forth code that expects two numbers on the stack
 1683: (@var{a} and @var{b}, with @var{b} on top) and computes
 1684: @code{(a-b)(a+1)}.
 1685: @endassignment
 1686: 
 1687: Reference: @ref{Stack Manipulation}.
 1688: 
 1689: 
 1690: @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
 1691: @section Using files for Forth code
 1692: @cindex loading Forth code, tutorial
 1693: @cindex files containing Forth code, tutorial
 1694: 
 1695: While working at the Forth command line is convenient for one-line
 1696: examples and short one-off code, you probably want to store your source
 1697: code in files for convenient editing and persistence.  You can use your
 1698: favourite editor (Gforth includes Emacs support, @pxref{Emacs and
 1699: Gforth}) to create @var{file.fs} and use
 1700: 
 1701: @example
 1702: s" @var{file.fs}" included
 1703: @end example
 1704: 
 1705: to load it into your Forth system.  The file name extension I use for
 1706: Forth files is @samp{.fs}.
 1707: 
 1708: You can easily start Gforth with some files loaded like this:
 1709: 
 1710: @example
 1711: gforth @var{file1.fs} @var{file2.fs}
 1712: @end example
 1713: 
 1714: If an error occurs during loading these files, Gforth terminates,
 1715: whereas an error during @code{INCLUDED} within Gforth usually gives you
 1716: a Gforth command line.  Starting the Forth system every time gives you a
 1717: clean start every time, without interference from the results of earlier
 1718: tries.
 1719: 
 1720: I often put all the tests in a file, then load the code and run the
 1721: tests with
 1722: 
 1723: @example
 1724: gforth @var{code.fs} @var{tests.fs} -e bye
 1725: @end example
 1726: 
 1727: (often by performing this command with @kbd{C-x C-e} in Emacs).  The
 1728: @code{-e bye} ensures that Gforth terminates afterwards so that I can
 1729: restart this command without ado.
 1730: 
 1731: The advantage of this approach is that the tests can be repeated easily
 1732: every time the program ist changed, making it easy to catch bugs
 1733: introduced by the change.
 1734: 
 1735: Reference: @ref{Forth source files}.
 1736: 
 1737: 
 1738: @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
 1739: @section Comments
 1740: @cindex comments tutorial
 1741: 
 1742: @example
 1743: \ That's a comment; it ends at the end of the line
 1744: ( Another comment; it ends here: )  .s
 1745: @end example
 1746: 
 1747: @code{\} and @code{(} are ordinary Forth words and therefore have to be
 1748: separated with white space from the following text.
 1749: 
 1750: @example
 1751: \This gives an "Undefined word" error
 1752: @end example
 1753: 
 1754: The first @code{)} ends a comment started with @code{(}, so you cannot
 1755: nest @code{(}-comments; and you cannot comment out text containing a
 1756: @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
 1757: avoid @code{)} in word names.}.
 1758: 
 1759: I use @code{\}-comments for descriptive text and for commenting out code
 1760: of one or more line; I use @code{(}-comments for describing the stack
 1761: effect, the stack contents, or for commenting out sub-line pieces of
 1762: code.
 1763: 
 1764: The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
 1765: these uses by commenting out a region with @kbd{C-x \}, uncommenting a
 1766: region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
 1767: with @kbd{M-q}.
 1768: 
 1769: Reference: @ref{Comments}.
 1770: 
 1771: 
 1772: @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
 1773: @section Colon Definitions
 1774: @cindex colon definitions, tutorial
 1775: @cindex definitions, tutorial
 1776: @cindex procedures, tutorial
 1777: @cindex functions, tutorial
 1778: 
 1779: are similar to procedures and functions in other programming languages.
 1780: 
 1781: @example
 1782: : squared ( n -- n^2 )
 1783:    dup * ;
 1784: 5 squared .
 1785: 7 squared .
 1786: @end example
 1787: 
 1788: @code{:} starts the colon definition; its name is @code{squared}.  The
 1789: following comment describes its stack effect.  The words @code{dup *}
 1790: are not executed, but compiled into the definition.  @code{;} ends the
 1791: colon definition.
 1792: 
 1793: The newly-defined word can be used like any other word, including using
 1794: it in other definitions:
 1795: 
 1796: @example
 1797: : cubed ( n -- n^3 )
 1798:    dup squared * ;
 1799: -5 cubed .
 1800: : fourth-power ( n -- n^4 )
 1801:    squared squared ;
 1802: 3 fourth-power .
 1803: @end example
 1804: 
 1805: @assignment
 1806: Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
 1807: @code{/mod} in terms of other Forth words, and check if they work (hint:
 1808: test your tests on the originals first).  Don't let the
 1809: @samp{redefined}-Messages spook you, they are just warnings.
 1810: @endassignment
 1811: 
 1812: Reference: @ref{Colon Definitions}.
 1813: 
 1814: 
 1815: @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
 1816: @section Decompilation
 1817: @cindex decompilation tutorial
 1818: @cindex see tutorial
 1819: 
 1820: You can decompile colon definitions with @code{see}:
 1821: 
 1822: @example
 1823: see squared
 1824: see cubed
 1825: @end example
 1826: 
 1827: In Gforth @code{see} shows you a reconstruction of the source code from
 1828: the executable code.  Informations that were present in the source, but
 1829: not in the executable code, are lost (e.g., comments).
 1830: 
 1831: You can also decompile the predefined words:
 1832: 
 1833: @example
 1834: see .
 1835: see +
 1836: @end example
 1837: 
 1838: 
 1839: @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
 1840: @section Stack-Effect Comments
 1841: @cindex stack-effect comments, tutorial
 1842: @cindex --, tutorial
 1843: By convention the comment after the name of a definition describes the
 1844: stack effect: The part in from of the @samp{--} describes the state of
 1845: the stack before the execution of the definition, i.e., the parameters
 1846: that are passed into the colon definition; the part behind the @samp{--}
 1847: is the state of the stack after the execution of the definition, i.e.,
 1848: the results of the definition.  The stack comment only shows the top
 1849: stack items that the definition accesses and/or changes.
 1850: 
 1851: You should put a correct stack effect on every definition, even if it is
 1852: just @code{( -- )}.  You should also add some descriptive comment to
 1853: more complicated words (I usually do this in the lines following
 1854: @code{:}).  If you don't do this, your code becomes unreadable (because
 1855: you have to work through every definition before you can undertsand
 1856: any).
 1857: 
 1858: @assignment
 1859: The stack effect of @code{swap} can be written like this: @code{x1 x2 --
 1860: x2 x1}.  Describe the stack effect of @code{-}, @code{drop}, @code{dup},
 1861: @code{over}, @code{rot}, @code{nip}, and @code{tuck}.  Hint: When you
 1862: are done, you can compare your stack effects to those in this manual
 1863: (@pxref{Word Index}).
 1864: @endassignment
 1865: 
 1866: Sometimes programmers put comments at various places in colon
 1867: definitions that describe the contents of the stack at that place (stack
 1868: comments); i.e., they are like the first part of a stack-effect
 1869: comment. E.g.,
 1870: 
 1871: @example
 1872: : cubed ( n -- n^3 )
 1873:    dup squared  ( n n^2 ) * ;
 1874: @end example
 1875: 
 1876: In this case the stack comment is pretty superfluous, because the word
 1877: is simple enough.  If you think it would be a good idea to add such a
 1878: comment to increase readability, you should also consider factoring the
 1879: word into several simpler words (@pxref{Factoring Tutorial,,
 1880: Factoring}), which typically eliminates the need for the stack comment;
 1881: however, if you decide not to refactor it, then having such a comment is
 1882: better than not having it.
 1883: 
 1884: The names of the stack items in stack-effect and stack comments in the
 1885: standard, in this manual, and in many programs specify the type through
 1886: a type prefix, similar to Fortran and Hungarian notation.  The most
 1887: frequent prefixes are:
 1888: 
 1889: @table @code
 1890: @item n
 1891: signed integer
 1892: @item u
 1893: unsigned integer
 1894: @item c
 1895: character
 1896: @item f
 1897: Boolean flags, i.e. @code{false} or @code{true}.
 1898: @item a-addr,a-
 1899: Cell-aligned address
 1900: @item c-addr,c-
 1901: Char-aligned address (note that a Char may have two bytes in Windows NT)
 1902: @item xt
 1903: Execution token, same size as Cell
 1904: @item w,x
 1905: Cell, can contain an integer or an address.  It usually takes 32, 64 or
 1906: 16 bits (depending on your platform and Forth system). A cell is more
 1907: commonly known as machine word, but the term @emph{word} already means
 1908: something different in Forth.
 1909: @item d
 1910: signed double-cell integer
 1911: @item ud
 1912: unsigned double-cell integer
 1913: @item r
 1914: Float (on the FP stack)
 1915: @end table
 1916: 
 1917: You can find a more complete list in @ref{Notation}.
 1918: 
 1919: @assignment
 1920: Write stack-effect comments for all definitions you have written up to
 1921: now.
 1922: @endassignment
 1923: 
 1924: 
 1925: @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
 1926: @section Types
 1927: @cindex types tutorial
 1928: 
 1929: In Forth the names of the operations are not overloaded; so similar
 1930: operations on different types need different names; e.g., @code{+} adds
 1931: integers, and you have to use @code{f+} to add floating-point numbers.
 1932: The following prefixes are often used for related operations on
 1933: different types:
 1934: 
 1935: @table @code
 1936: @item (none)
 1937: signed integer
 1938: @item u
 1939: unsigned integer
 1940: @item c
 1941: character
 1942: @item d
 1943: signed double-cell integer
 1944: @item ud, du
 1945: unsigned double-cell integer
 1946: @item 2
 1947: two cells (not-necessarily double-cell numbers)
 1948: @item m, um
 1949: mixed single-cell and double-cell operations
 1950: @item f
 1951: floating-point (note that in stack comments @samp{f} represents flags,
 1952: and @samp{r} represents FP numbers).
 1953: @end table
 1954: 
 1955: If there are no differences between the signed and the unsigned variant
 1956: (e.g., for @code{+}), there is only the prefix-less variant.
 1957: 
 1958: Forth does not perform type checking, neither at compile time, nor at
 1959: run time.  If you use the wrong oeration, the data are interpreted
 1960: incorrectly:
 1961: 
 1962: @example
 1963: -1 u.
 1964: @end example
 1965: 
 1966: If you have only experience with type-checked languages until now, and
 1967: have heard how important type-checking is, don't panic!  In my
 1968: experience (and that of other Forthers), type errors in Forth code are
 1969: usually easy to find (once you get used to it), the increased vigilance
 1970: of the programmer tends to catch some harder errors in addition to most
 1971: type errors, and you never have to work around the type system, so in
 1972: most situations the lack of type-checking seems to be a win (projects to
 1973: add type checking to Forth have not caught on).
 1974: 
 1975: 
 1976: @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
 1977: @section Factoring
 1978: @cindex factoring tutorial
 1979: 
 1980: If you try to write longer definitions, you will soon find it hard to
 1981: keep track of the stack contents.  Therefore, good Forth programmers
 1982: tend to write only short definitions (e.g., three lines).  The art of
 1983: finding meaningful short definitions is known as factoring (as in
 1984: factoring polynomials).
 1985: 
 1986: Well-factored programs offer additional advantages: smaller, more
 1987: general words, are easier to test and debug and can be reused more and
 1988: better than larger, specialized words.
 1989: 
 1990: So, if you run into difficulties with stack management, when writing
 1991: code, try to define meaningful factors for the word, and define the word
 1992: in terms of those.  Even if a factor contains only two words, it is
 1993: often helpful.
 1994: 
 1995: Good factoring is not easy, and it takes some practice to get the knack
 1996: for it; but even experienced Forth programmers often don't find the
 1997: right solution right away, but only when rewriting the program.  So, if
 1998: you don't come up with a good solution immediately, keep trying, don't
 1999: despair.
 2000: 
 2001: @c example !!
 2002: 
 2003: 
 2004: @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
 2005: @section Designing the stack effect
 2006: @cindex Stack effect design, tutorial
 2007: @cindex design of stack effects, tutorial
 2008: 
 2009: In other languages you can use an arbitrary order of parameters for a
 2010: function; and since there is only one result, you don't have to deal with
 2011: the order of results, either.
 2012: 
 2013: In Forth (and other stack-based languages, e.g., Postscript) the
 2014: parameter and result order of a definition is important and should be
 2015: designed well.  The general guideline is to design the stack effect such
 2016: that the word is simple to use in most cases, even if that complicates
 2017: the implementation of the word.  Some concrete rules are:
 2018: 
 2019: @itemize @bullet
 2020: 
 2021: @item
 2022: Words consume all of their parameters (e.g., @code{.}).
 2023: 
 2024: @item
 2025: If there is a convention on the order of parameters (e.g., from
 2026: mathematics or another programming language), stick with it (e.g.,
 2027: @code{-}).
 2028: 
 2029: @item
 2030: If one parameter usually requires only a short computation (e.g., it is
 2031: a constant), pass it on the top of the stack.  Conversely, parameters
 2032: that usually require a long sequence of code to compute should be passed
 2033: as the bottom (i.e., first) parameter.  This makes the code easier to
 2034: read, because reader does not need to keep track of the bottom item
 2035: through a long sequence of code (or, alternatively, through stack
 2036: manipulations). E.g., @code{!} (store, @pxref{Memory}) expects the
 2037: address on top of the stack because it is usually simpler to compute
 2038: than the stored value (often the address is just a variable).
 2039: 
 2040: @item
 2041: Similarly, results that are usually consumed quickly should be returned
 2042: on the top of stack, whereas a result that is often used in long
 2043: computations should be passed as bottom result.  E.g., the file words
 2044: like @code{open-file} return the error code on the top of stack, because
 2045: it is usually consumed quickly by @code{throw}; moreover, the error code
 2046: has to be checked before doing anything with the other results.
 2047: 
 2048: @end itemize
 2049: 
 2050: These rules are just general guidelines, don't lose sight of the overall
 2051: goal to make the words easy to use.  E.g., if the convention rule
 2052: conflicts with the computation-length rule, you might decide in favour
 2053: of the convention if the word will be used rarely, and in favour of the
 2054: computation-length rule if the word will be used frequently (because
 2055: with frequent use the cost of breaking the computation-length rule would
 2056: be quite high, and frequent use makes it easier to remember an
 2057: unconventional order).
 2058: 
 2059: @c example !! structure package
 2060: 
 2061: 
 2062: @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
 2063: @section Local Variables
 2064: @cindex local variables, tutorial
 2065: 
 2066: You can define local variables (@emph{locals}) in a colon definition:
 2067: 
 2068: @example
 2069: : swap @{ a b -- b a @}
 2070:   b a ;
 2071: 1 2 swap .s 2drop
 2072: @end example
 2073: 
 2074: (If your Forth system does not support this syntax, include
 2075: @file{compat/anslocals.fs} first).
 2076: 
 2077: In this example @code{@{ a b -- b a @}} is the locals definition; it
 2078: takes two cells from the stack, puts the top of stack in @code{b} and
 2079: the next stack element in @code{a}.  @code{--} starts a comment ending
 2080: with @code{@}}.  After the locals definition, using the name of the
 2081: local will push its value on the stack.  You can leave the comment
 2082: part (@code{-- b a}) away:
 2083: 
 2084: @example
 2085: : swap ( x1 x2 -- x2 x1 )
 2086:   @{ a b @} b a ;
 2087: @end example
 2088: 
 2089: In Gforth you can have several locals definitions, anywhere in a colon
 2090: definition; in contrast, in a standard program you can have only one
 2091: locals definition per colon definition, and that locals definition must
 2092: be outside any controll structure.
 2093: 
 2094: With locals you can write slightly longer definitions without running
 2095: into stack trouble.  However, I recommend trying to write colon
 2096: definitions without locals for exercise purposes to help you gain the
 2097: essential factoring skills.
 2098: 
 2099: @assignment
 2100: Rewrite your definitions until now with locals
 2101: @endassignment
 2102: 
 2103: Reference: @ref{Locals}.
 2104: 
 2105: 
 2106: @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
 2107: @section Conditional execution
 2108: @cindex conditionals, tutorial
 2109: @cindex if, tutorial
 2110: 
 2111: In Forth you can use control structures only inside colon definitions.
 2112: An @code{if}-structure looks like this:
 2113: 
 2114: @example
 2115: : abs ( n1 -- +n2 )
 2116:     dup 0 < if
 2117:         negate
 2118:     endif ;
 2119: 5 abs .
 2120: -5 abs .
 2121: @end example
 2122: 
 2123: @code{if} takes a flag from the stack.  If the flag is non-zero (true),
 2124: the following code is performed, otherwise execution continues after the
 2125: @code{endif} (or @code{else}).  @code{<} compares the top two stack
 2126: elements and prioduces a flag:
 2127: 
 2128: @example
 2129: 1 2 < .
 2130: 2 1 < .
 2131: 1 1 < .
 2132: @end example
 2133: 
 2134: Actually the standard name for @code{endif} is @code{then}.  This
 2135: tutorial presents the examples using @code{endif}, because this is often
 2136: less confusing for people familiar with other programming languages
 2137: where @code{then} has a different meaning.  If your system does not have
 2138: @code{endif}, define it with
 2139: 
 2140: @example
 2141: : endif postpone then ; immediate
 2142: @end example
 2143: 
 2144: You can optionally use an @code{else}-part:
 2145: 
 2146: @example
 2147: : min ( n1 n2 -- n )
 2148:   2dup < if
 2149:     drop
 2150:   else
 2151:     nip
 2152:   endif ;
 2153: 2 3 min .
 2154: 3 2 min .
 2155: @end example
 2156: 
 2157: @assignment
 2158: Write @code{min} without @code{else}-part (hint: what's the definition
 2159: of @code{nip}?).
 2160: @endassignment
 2161: 
 2162: Reference: @ref{Selection}.
 2163: 
 2164: 
 2165: @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
 2166: @section Flags and Comparisons
 2167: @cindex flags tutorial
 2168: @cindex comparison tutorial
 2169: 
 2170: In a false-flag all bits are clear (0 when interpreted as integer).  In
 2171: a canonical true-flag all bits are set (-1 as a twos-complement signed
 2172: integer); in many contexts (e.g., @code{if}) any non-zero value is
 2173: treated as true flag.
 2174: 
 2175: @example
 2176: false .
 2177: true .
 2178: true hex u. decimal
 2179: @end example
 2180: 
 2181: Comparison words produce canonical flags:
 2182: 
 2183: @example
 2184: 1 1 = .
 2185: 1 0= .
 2186: 0 1 < .
 2187: 0 0 < .
 2188: -1 1 u< . \ type error, u< interprets -1 as large unsigned number
 2189: -1 1 < .
 2190: @end example
 2191: 
 2192: Gforth supports all combinations of the prefixes @code{0 u d d0 du f f0}
 2193: (or none) and the comparisons @code{= <> < > <= >=}.  Only a part of
 2194: these combinations are standard (for details see the standard,
 2195: @ref{Numeric comparison}, @ref{Floating Point} or @ref{Word Index}).
 2196: 
 2197: You can use @code{and or xor invert} can be used as operations on
 2198: canonical flags.  Actually they are bitwise operations:
 2199: 
 2200: @example
 2201: 1 2 and .
 2202: 1 2 or .
 2203: 1 3 xor .
 2204: 1 invert .
 2205: @end example
 2206: 
 2207: You can convert a zero/non-zero flag into a canonical flag with
 2208: @code{0<>} (and complement it on the way with @code{0=}).
 2209: 
 2210: @example
 2211: 1 0= .
 2212: 1 0<> .
 2213: @end example
 2214: 
 2215: You can use the all-bits-set feature of canonical flags and the bitwise
 2216: operation of the Boolean operations to avoid @code{if}s:
 2217: 
 2218: @example
 2219: : foo ( n1 -- n2 )
 2220:   0= if
 2221:     14
 2222:   else
 2223:     0
 2224:   endif ;
 2225: 0 foo .
 2226: 1 foo .
 2227: 
 2228: : foo ( n1 -- n2 )
 2229:   0= 14 and ;
 2230: 0 foo .
 2231: 1 foo .
 2232: @end example
 2233: 
 2234: @assignment
 2235: Write @code{min} without @code{if}.
 2236: @endassignment
 2237: 
 2238: For reference, see @ref{Boolean Flags}, @ref{Numeric comparison}, and
 2239: @ref{Bitwise operations}.
 2240: 
 2241: 
 2242: @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
 2243: @section General Loops
 2244: @cindex loops, indefinite, tutorial
 2245: 
 2246: The endless loop is the most simple one:
 2247: 
 2248: @example
 2249: : endless ( -- )
 2250:   0 begin
 2251:     dup . 1+
 2252:   again ;
 2253: endless
 2254: @end example
 2255: 
 2256: Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth).  @code{begin}
 2257: does nothing at run-time, @code{again} jumps back to @code{begin}.
 2258: 
 2259: A loop with one exit at any place looks like this:
 2260: 
 2261: @example
 2262: : log2 ( +n1 -- n2 )
 2263: \ logarithmus dualis of n1>0, rounded down to the next integer
 2264:   assert( dup 0> )
 2265:   2/ 0 begin
 2266:     over 0> while
 2267:       1+ swap 2/ swap
 2268:   repeat
 2269:   nip ;
 2270: 7 log2 .
 2271: 8 log2 .
 2272: @end example
 2273: 
 2274: At run-time @code{while} consumes a flag; if it is 0, execution
 2275: continues behind the @code{repeat}; if the flag is non-zero, execution
 2276: continues behind the @code{while}.  @code{Repeat} jumps back to
 2277: @code{begin}, just like @code{again}.
 2278: 
 2279: In Forth there are many combinations/abbreviations, like @code{1+}.
 2280: However, @code{2/} is not one of them; it shifts its argument right by
 2281: one bit (arithmetic shift right):
 2282: 
 2283: @example
 2284: -5 2 / .
 2285: -5 2/ .
 2286: @end example
 2287: 
 2288: @code{assert(} is no standard word, but you can get it on systems other
 2289: then Gforth by including @file{compat/assert.fs}.  You can see what it
 2290: does by trying
 2291: 
 2292: @example
 2293: 0 log2 .
 2294: @end example
 2295: 
 2296: Here's a loop with an exit at the end:
 2297: 
 2298: @example
 2299: : log2 ( +n1 -- n2 )
 2300: \ logarithmus dualis of n1>0, rounded down to the next integer
 2301:   assert( dup 0 > )
 2302:   -1 begin
 2303:     1+ swap 2/ swap
 2304:     over 0 <=
 2305:   until
 2306:   nip ;
 2307: @end example
 2308: 
 2309: @code{Until} consumes a flag; if it is non-zero, execution continues at
 2310: the @code{begin}, otherwise after the @code{until}.
 2311: 
 2312: @assignment
 2313: Write a definition for computing the greatest common divisor.
 2314: @endassignment
 2315: 
 2316: Reference: @ref{Simple Loops}.
 2317: 
 2318: 
 2319: @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
 2320: @section Counted loops
 2321: @cindex loops, counted, tutorial
 2322: 
 2323: @example
 2324: : ^ ( n1 u -- n )
 2325: \ n = the uth power of u1
 2326:   1 swap 0 u+do
 2327:     over *
 2328:   loop
 2329:   nip ;
 2330: 3 2 ^ .
 2331: 4 3 ^ .
 2332: @end example
 2333: 
 2334: @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
 2335: have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
 2336: performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
 2337: times (or not at all, if @code{u3-u4<0}).
 2338: 
 2339: You can see the stack effect design rules at work in the stack effect of
 2340: the loop start words: Since the start value of the loop is more
 2341: frequently constant than the end value, the start value is passed on
 2342: the top-of-stack.
 2343: 
 2344: You can access the counter of a counted loop with @code{i}:
 2345: 
 2346: @example
 2347: : fac ( u -- u! )
 2348:   1 swap 1+ 1 u+do
 2349:     i *
 2350:   loop ;
 2351: 5 fac .
 2352: 7 fac .
 2353: @end example
 2354: 
 2355: There is also @code{+do}, which expects signed numbers (important for
 2356: deciding whether to enter the loop).
 2357: 
 2358: @assignment
 2359: Write a definition for computing the nth Fibonacci number.
 2360: @endassignment
 2361: 
 2362: You can also use increments other than 1:
 2363: 
 2364: @example
 2365: : up2 ( n1 n2 -- )
 2366:   +do
 2367:     i .
 2368:   2 +loop ;
 2369: 10 0 up2
 2370: 
 2371: : down2 ( n1 n2 -- )
 2372:   -do
 2373:     i .
 2374:   2 -loop ;
 2375: 0 10 down2
 2376: @end example
 2377: 
 2378: Reference: @ref{Counted Loops}.
 2379: 
 2380: 
 2381: @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
 2382: @section Recursion
 2383: @cindex recursion tutorial
 2384: 
 2385: Usually the name of a definition is not visible in the definition; but
 2386: earlier definitions are usually visible:
 2387: 
 2388: @example
 2389: 1 0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
 2390: : / ( n1 n2 -- n )
 2391:   dup 0= if
 2392:     -10 throw \ report division by zero
 2393:   endif
 2394:   /           \ old version
 2395: ;
 2396: 1 0 /
 2397: @end example
 2398: 
 2399: For recursive definitions you can use @code{recursive} (non-standard) or
 2400: @code{recurse}:
 2401: 
 2402: @example
 2403: : fac1 ( n -- n! ) recursive
 2404:  dup 0> if
 2405:    dup 1- fac1 *
 2406:  else
 2407:    drop 1
 2408:  endif ;
 2409: 7 fac1 .
 2410: 
 2411: : fac2 ( n -- n! )
 2412:  dup 0> if
 2413:    dup 1- recurse *
 2414:  else
 2415:    drop 1
 2416:  endif ;
 2417: 8 fac2 .
 2418: @end example
 2419: 
 2420: @assignment
 2421: Write a recursive definition for computing the nth Fibonacci number.
 2422: @endassignment
 2423: 
 2424: Reference (including indirect recursion): @xref{Calls and returns}.
 2425: 
 2426: 
 2427: @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
 2428: @section Leaving definitions or loops
 2429: @cindex leaving definitions, tutorial
 2430: @cindex leaving loops, tutorial
 2431: 
 2432: @code{EXIT} exits the current definition right away.  For every counted
 2433: loop that is left in this way, an @code{UNLOOP} has to be performed
 2434: before the @code{EXIT}:
 2435: 
 2436: @c !! real examples
 2437: @example
 2438: : ...
 2439:  ... u+do
 2440:    ... if
 2441:      ... unloop exit
 2442:    endif
 2443:    ...
 2444:  loop
 2445:  ... ;
 2446: @end example
 2447: 
 2448: @code{LEAVE} leaves the innermost counted loop right away:
 2449: 
 2450: @example
 2451: : ...
 2452:  ... u+do
 2453:    ... if
 2454:      ... leave
 2455:    endif
 2456:    ...
 2457:  loop
 2458:  ... ;
 2459: @end example
 2460: 
 2461: @c !! example
 2462: 
 2463: Reference: @ref{Calls and returns}, @ref{Counted Loops}.
 2464: 
 2465: 
 2466: @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
 2467: @section Return Stack
 2468: @cindex return stack tutorial
 2469: 
 2470: In addition to the data stack Forth also has a second stack, the return
 2471: stack; most Forth systems store the return addresses of procedure calls
 2472: there (thus its name).  Programmers can also use this stack:
 2473: 
 2474: @example
 2475: : foo ( n1 n2 -- )
 2476:  .s
 2477:  >r .s
 2478:  r@@ .
 2479:  >r .s
 2480:  r@@ .
 2481:  r> .
 2482:  r@@ .
 2483:  r> . ;
 2484: 1 2 foo
 2485: @end example
 2486: 
 2487: @code{>r} takes an element from the data stack and pushes it onto the
 2488: return stack; conversely, @code{r>} moves an elementm from the return to
 2489: the data stack; @code{r@@} pushes a copy of the top of the return stack
 2490: on the return stack.
 2491: 
 2492: Forth programmers usually use the return stack for storing data
 2493: temporarily, if using the data stack alone would be too complex, and
 2494: factoring and locals are not an option:
 2495: 
 2496: @example
 2497: : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
 2498:  rot >r rot r> ;
 2499: @end example
 2500: 
 2501: The return address of the definition and the loop control parameters of
 2502: counted loops usually reside on the return stack, so you have to take
 2503: all items, that you have pushed on the return stack in a colon
 2504: definition or counted loop, from the return stack before the definition
 2505: or loop ends.  You cannot access items that you pushed on the return
 2506: stack outside some definition or loop within the definition of loop.
 2507: 
 2508: If you miscount the return stack items, this usually ends in a crash:
 2509: 
 2510: @example
 2511: : crash ( n -- )
 2512:   >r ;
 2513: 5 crash
 2514: @end example
 2515: 
 2516: You cannot mix using locals and using the return stack (according to the
 2517: standard; Gforth has no problem).  However, they solve the same
 2518: problems, so this shouldn't be an issue.
 2519: 
 2520: @assignment
 2521: Can you rewrite any of the definitions you wrote until now in a better
 2522: way using the return stack?
 2523: @endassignment
 2524: 
 2525: Reference: @ref{Return stack}.
 2526: 
 2527: 
 2528: @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
 2529: @section Memory
 2530: @cindex memory access/allocation tutorial
 2531: 
 2532: You can create a global variable @code{v} with
 2533: 
 2534: @example
 2535: variable v ( -- addr )
 2536: @end example
 2537: 
 2538: @code{v} pushes the address of a cell in memory on the stack.  This cell
 2539: was reserved by @code{variable}.  You can use @code{!} (store) to store
 2540: values into this cell and @code{@@} (fetch) to load the value from the
 2541: stack into memory:
 2542: 
 2543: @example
 2544: v .
 2545: 5 v ! .s
 2546: v @@ .
 2547: @end example
 2548: 
 2549: You can see a raw dump of memory with @code{dump}:
 2550: 
 2551: @example
 2552: v 1 cells .s dump
 2553: @end example
 2554: 
 2555: @code{Cells ( n1 -- n2 )} gives you the number of bytes (or, more
 2556: generally, address units (aus)) that @code{n1 cells} occupy.  You can
 2557: also reserve more memory:
 2558: 
 2559: @example
 2560: create v2 20 cells allot
 2561: v2 20 cells dump
 2562: @end example
 2563: 
 2564: creates a word @code{v2} and reserves 20 uninitialized cells; the
 2565: address pushed by @code{v2} points to the start of these 20 cells.  You
 2566: can use address arithmetic to access these cells:
 2567: 
 2568: @example
 2569: 3 v2 5 cells + !
 2570: v2 20 cells dump
 2571: @end example
 2572: 
 2573: You can reserve and initialize memory with @code{,}:
 2574: 
 2575: @example
 2576: create v3
 2577:   5 , 4 , 3 , 2 , 1 ,
 2578: v3 @@ .
 2579: v3 cell+ @@ .
 2580: v3 2 cells + @@ .
 2581: v3 5 cells dump
 2582: @end example
 2583: 
 2584: @assignment
 2585: Write a definition @code{vsum ( addr u -- n )} that computes the sum of
 2586: @code{u} cells, with the first of these cells at @code{addr}, the next
 2587: one at @code{addr cell+} etc.
 2588: @endassignment
 2589: 
 2590: You can also reserve memory without creating a new word:
 2591: 
 2592: @example
 2593: here 10 cells allot .
 2594: here .
 2595: @end example
 2596: 
 2597: @code{Here} pushes the start address of the memory area.  You should
 2598: store it somewhere, or you will have a hard time finding the memory area
 2599: again.
 2600: 
 2601: @code{Allot} manages dictionary memory.  The dictionary memory contains
 2602: the system's data structures for words etc. on Gforth and most other
 2603: Forth systems.  It is managed like a stack: You can free the memory that
 2604: you have just @code{allot}ed with
 2605: 
 2606: @example
 2607: -10 cells allot
 2608: here .
 2609: @end example
 2610: 
 2611: Note that you cannot do this if you have created a new word in the
 2612: meantime (because then your @code{allot}ed memory is no longer on the
 2613: top of the dictionary ``stack'').
 2614: 
 2615: Alternatively, you can use @code{allocate} and @code{free} which allow
 2616: freeing memory in any order:
 2617: 
 2618: @example
 2619: 10 cells allocate throw .s
 2620: 20 cells allocate throw .s
 2621: swap
 2622: free throw
 2623: free throw
 2624: @end example
 2625: 
 2626: The @code{throw}s deal with errors (e.g., out of memory).
 2627: 
 2628: And there is also a
 2629: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
 2630: garbage collector}, which eliminates the need to @code{free} memory
 2631: explicitly.
 2632: 
 2633: Reference: @ref{Memory}.
 2634: 
 2635: 
 2636: @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
 2637: @section Characters and Strings
 2638: @cindex strings tutorial
 2639: @cindex characters tutorial
 2640: 
 2641: On the stack characters take up a cell, like numbers.  In memory they
 2642: have their own size (one 8-bit byte on most systems), and therefore
 2643: require their own words for memory access:
 2644: 
 2645: @example
 2646: create v4 
 2647:   104 c, 97 c, 108 c, 108 c, 111 c,
 2648: v4 4 chars + c@@ .
 2649: v4 5 chars dump
 2650: @end example
 2651: 
 2652: The preferred representation of strings on the stack is @code{addr
 2653: u-count}, where @code{addr} is the address of the first character and
 2654: @code{u-count} is the number of characters in the string.
 2655: 
 2656: @example
 2657: v4 5 type
 2658: @end example
 2659: 
 2660: You get a string constant with
 2661: 
 2662: @example
 2663: s" hello, world" .s
 2664: type
 2665: @end example
 2666: 
 2667: Make sure you have a space between @code{s"} and the string; @code{s"}
 2668: is a normal Forth word and must be delimited with white space (try what
 2669: happens when you remove the space).
 2670: 
 2671: However, this interpretive use of @code{s"} is quite restricted: the
 2672: string exists only until the next call of @code{s"} (some Forth systems
 2673: keep more than one of these strings, but usually they still have a
 2674: limited lifetime).
 2675: 
 2676: @example
 2677: s" hello," s" world" .s
 2678: type
 2679: type
 2680: @end example
 2681: 
 2682: You can also use @code{s"} in a definition, and the resulting
 2683: strings then live forever (well, for as long as the definition):
 2684: 
 2685: @example
 2686: : foo s" hello," s" world" ;
 2687: foo .s
 2688: type
 2689: type
 2690: @end example
 2691: 
 2692: @assignment
 2693: @code{Emit ( c -- )} types @code{c} as character (not a number).
 2694: Implement @code{type ( addr u -- )}.
 2695: @endassignment
 2696: 
 2697: Reference: @ref{Memory Blocks}.
 2698: 
 2699: 
 2700: @node Alignment Tutorial, Files Tutorial, Characters and Strings Tutorial, Tutorial
 2701: @section Alignment
 2702: @cindex alignment tutorial
 2703: @cindex memory alignment tutorial
 2704: 
 2705: On many processors cells have to be aligned in memory, if you want to
 2706: access them with @code{@@} and @code{!} (and even if the processor does
 2707: not require alignment, access to aligned cells is faster).
 2708: 
 2709: @code{Create} aligns @code{here} (i.e., the place where the next
 2710: allocation will occur, and that the @code{create}d word points to).
 2711: Likewise, the memory produced by @code{allocate} starts at an aligned
 2712: address.  Adding a number of @code{cells} to an aligned address produces
 2713: another aligned address.
 2714: 
 2715: However, address arithmetic involving @code{char+} and @code{chars} can
 2716: create an address that is not cell-aligned.  @code{Aligned ( addr --
 2717: a-addr )} produces the next aligned address:
 2718: 
 2719: @example
 2720: v3 char+ aligned .s @@ .
 2721: v3 char+ .s @@ .
 2722: @end example
 2723: 
 2724: Similarly, @code{align} advances @code{here} to the next aligned
 2725: address:
 2726: 
 2727: @example
 2728: create v5 97 c,
 2729: here .
 2730: align here .
 2731: 1000 ,
 2732: @end example
 2733: 
 2734: Note that you should use aligned addresses even if your processor does
 2735: not require them, if you want your program to be portable.
 2736: 
 2737: Reference: @ref{Address arithmetic}.
 2738: 
 2739: 
 2740: @node Files Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Alignment Tutorial, Tutorial
 2741: @section Files
 2742: @cindex files tutorial
 2743: 
 2744: This section gives a short introduction into how to use files inside
 2745: Forth. It's broken up into five easy steps:
 2746: 
 2747: @enumerate 1
 2748: @item Opened an ASCII text file for input
 2749: @item Opened a file for output
 2750: @item Read input file until string matched (or some other condition matched)
 2751: @item Wrote some lines from input ( modified or not) to output
 2752: @item Closed the files.
 2753: @end enumerate
 2754: 
 2755: @subsection Open file for input
 2756: 
 2757: @example
 2758: s" foo.in"  r/o open-file throw Value fd-in
 2759: @end example
 2760: 
 2761: @subsection Create file for output
 2762: 
 2763: @example
 2764: s" foo.out" w/o create-file throw Value fd-out
 2765: @end example
 2766: 
 2767: The available file modes are r/o for read-only access, r/w for
 2768: read-write access, and w/o for write-only access. You could open both
 2769: files with r/w, too, if you like. All file words return error codes; for
 2770: most applications, it's best to pass there error codes with @code{throw}
 2771: to the outer error handler.
 2772: 
 2773: If you want words for opening and assigning, define them as follows:
 2774: 
 2775: @example
 2776: 0 Value fd-in
 2777: 0 Value fd-out
 2778: : open-input ( addr u -- )  r/o open-file throw to fd-in ;
 2779: : open-output ( addr u -- )  w/o create-file throw to fd-out ;
 2780: @end example
 2781: 
 2782: Usage example:
 2783: 
 2784: @example
 2785: s" foo.in" open-input
 2786: s" foo.out" open-output
 2787: @end example
 2788: 
 2789: @subsection Scan file for a particular line
 2790: 
 2791: @example
 2792: 256 Constant max-line
 2793: Create line-buffer  max-line 2 + allot
 2794: 
 2795: : scan-file ( addr u -- )
 2796:   begin
 2797:       line-buffer max-line fd-in read-line throw
 2798:   while
 2799:          >r 2dup line-buffer r> compare 0=
 2800:      until
 2801:   else
 2802:      drop
 2803:   then
 2804:   2drop ;
 2805: @end example
 2806: 
 2807: @code{read-line ( addr u1 fd -- u2 flag ior )} reads up to u1 bytes into
 2808: the buffer at addr, and returns the number of bytes read, a flag that is
 2809: false when the end of file is reached, and an error code.
 2810: 
 2811: @code{compare ( addr1 u1 addr2 u2 -- n )} compares two strings and
 2812: returns zero if both strings are equal. It returns a positive number if
 2813: the first string is lexically greater, a negative if the second string
 2814: is lexically greater.
 2815: 
 2816: We haven't seen this loop here; it has two exits. Since the @code{while}
 2817: exits with the number of bytes read on the stack, we have to clean up
 2818: that separately; that's after the @code{else}.
 2819: 
 2820: Usage example:
 2821: 
 2822: @example
 2823: s" The text I search is here" scan-file
 2824: @end example
 2825: 
 2826: @subsection Copy input to output
 2827: 
 2828: @example
 2829: : copy-file ( -- )
 2830:   begin
 2831:       line-buffer max-line fd-in read-line throw
 2832:   while
 2833:       line-buffer swap fd-out write-file throw
 2834:   repeat ;
 2835: @end example
 2836: 
 2837: @subsection Close files
 2838: 
 2839: @example
 2840: fd-in close-file throw
 2841: fd-out close-file throw
 2842: @end example
 2843: 
 2844: Likewise, you can put that into definitions, too:
 2845: 
 2846: @example
 2847: : close-input ( -- )  fd-in close-file throw ;
 2848: : close-output ( -- )  fd-out close-file throw ;
 2849: @end example
 2850: 
 2851: @assignment
 2852: How could you modify @code{copy-file} so that it copies until a second line is
 2853: matched? Can you write a program that extracts a section of a text file,
 2854: given the line that starts and the line that terminates that section?
 2855: @endassignment
 2856: 
 2857: @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Files Tutorial, Tutorial
 2858: @section Interpretation and Compilation Semantics and Immediacy
 2859: @cindex semantics tutorial
 2860: @cindex interpretation semantics tutorial
 2861: @cindex compilation semantics tutorial
 2862: @cindex immediate, tutorial
 2863: 
 2864: When a word is compiled, it behaves differently from being interpreted.
 2865: E.g., consider @code{+}:
 2866: 
 2867: @example
 2868: 1 2 + .
 2869: : foo + ;
 2870: @end example
 2871: 
 2872: These two behaviours are known as compilation and interpretation
 2873: semantics.  For normal words (e.g., @code{+}), the compilation semantics
 2874: is to append the interpretation semantics to the currently defined word
 2875: (@code{foo} in the example above).  I.e., when @code{foo} is executed
 2876: later, the interpretation semantics of @code{+} (i.e., adding two
 2877: numbers) will be performed.
 2878: 
 2879: However, there are words with non-default compilation semantics, e.g.,
 2880: the control-flow words like @code{if}.  You can use @code{immediate} to
 2881: change the compilation semantics of the last defined word to be equal to
 2882: the interpretation semantics:
 2883: 
 2884: @example
 2885: : [FOO] ( -- )
 2886:  5 . ; immediate
 2887: 
 2888: [FOO]
 2889: : bar ( -- )
 2890:   [FOO] ;
 2891: bar
 2892: see bar
 2893: @end example
 2894: 
 2895: Two conventions to mark words with non-default compilation semnatics are
 2896: names with brackets (more frequently used) and to write them all in
 2897: upper case (less frequently used).
 2898: 
 2899: In Gforth (and many other systems) you can also remove the
 2900: interpretation semantics with @code{compile-only} (the compilation
 2901: semantics is derived from the original interpretation semantics):
 2902: 
 2903: @example
 2904: : flip ( -- )
 2905:  6 . ; compile-only \ but not immediate
 2906: flip
 2907: 
 2908: : flop ( -- )
 2909:  flip ;
 2910: flop
 2911: @end example
 2912: 
 2913: In this example the interpretation semantics of @code{flop} is equal to
 2914: the original interpretation semantics of @code{flip}.
 2915: 
 2916: The text interpreter has two states: in interpret state, it performs the
 2917: interpretation semantics of words it encounters; in compile state, it
 2918: performs the compilation semantics of these words.
 2919: 
 2920: Among other things, @code{:} switches into compile state, and @code{;}
 2921: switches back to interpret state.  They contain the factors @code{]}
 2922: (switch to compile state) and @code{[} (switch to interpret state), that
 2923: do nothing but switch the state.
 2924: 
 2925: @example
 2926: : xxx ( -- )
 2927:   [ 5 . ]
 2928: ;
 2929: 
 2930: xxx
 2931: see xxx
 2932: @end example
 2933: 
 2934: These brackets are also the source of the naming convention mentioned
 2935: above.
 2936: 
 2937: Reference: @ref{Interpretation and Compilation Semantics}.
 2938: 
 2939: 
 2940: @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
 2941: @section Execution Tokens
 2942: @cindex execution tokens tutorial
 2943: @cindex XT tutorial
 2944: 
 2945: @code{' word} gives you the execution token (XT) of a word.  The XT is a
 2946: cell representing the interpretation semantics of a word.  You can
 2947: execute this semantics with @code{execute}:
 2948: 
 2949: @example
 2950: ' + .s
 2951: 1 2 rot execute .
 2952: @end example
 2953: 
 2954: The XT is similar to a function pointer in C.  However, parameter
 2955: passing through the stack makes it a little more flexible:
 2956: 
 2957: @example
 2958: : map-array ( ... addr u xt -- ... )
 2959: \ executes xt ( ... x -- ... ) for every element of the array starting
 2960: \ at addr and containing u elements
 2961:   @{ xt @}
 2962:   cells over + swap ?do
 2963:     i @@ xt execute
 2964:   1 cells +loop ;
 2965: 
 2966: create a 3 , 4 , 2 , -1 , 4 ,
 2967: a 5 ' . map-array .s
 2968: 0 a 5 ' + map-array .
 2969: s" max-n" environment? drop .s
 2970: a 5 ' min map-array .
 2971: @end example
 2972: 
 2973: You can use map-array with the XTs of words that consume one element
 2974: more than they produce.  In theory you can also use it with other XTs,
 2975: but the stack effect then depends on the size of the array, which is
 2976: hard to understand.
 2977: 
 2978: Since XTs are cell-sized, you can store them in memory and manipulate
 2979: them on the stack like other cells.  You can also compile the XT into a
 2980: word with @code{compile,}:
 2981: 
 2982: @example
 2983: : foo1 ( n1 n2 -- n )
 2984:    [ ' + compile, ] ;
 2985: see foo
 2986: @end example
 2987: 
 2988: This is non-standard, because @code{compile,} has no compilation
 2989: semantics in the standard, but it works in good Forth systems.  For the
 2990: broken ones, use
 2991: 
 2992: @example
 2993: : [compile,] compile, ; immediate
 2994: 
 2995: : foo1 ( n1 n2 -- n )
 2996:    [ ' + ] [compile,] ;
 2997: see foo
 2998: @end example
 2999: 
 3000: @code{'} is a word with default compilation semantics; it parses the
 3001: next word when its interpretation semantics are executed, not during
 3002: compilation:
 3003: 
 3004: @example
 3005: : foo ( -- xt )
 3006:   ' ;
 3007: see foo
 3008: : bar ( ... "word" -- ... )
 3009:   ' execute ;
 3010: see bar
 3011: 1 2 bar + .
 3012: @end example
 3013: 
 3014: You often want to parse a word during compilation and compile its XT so
 3015: it will be pushed on the stack at run-time.  @code{[']} does this:
 3016: 
 3017: @example
 3018: : xt-+ ( -- xt )
 3019:   ['] + ;
 3020: see xt-+
 3021: 1 2 xt-+ execute .
 3022: @end example
 3023: 
 3024: Many programmers tend to see @code{'} and the word it parses as one
 3025: unit, and expect it to behave like @code{[']} when compiled, and are
 3026: confused by the actual behaviour.  If you are, just remember that the
 3027: Forth system just takes @code{'} as one unit and has no idea that it is
 3028: a parsing word (attempts to convenience programmers in this issue have
 3029: usually resulted in even worse pitfalls, see
 3030: @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,
 3031: @code{State}-smartness---Why it is evil and How to Exorcise it}).
 3032: 
 3033: Note that the state of the interpreter does not come into play when
 3034: creating and executing XTs.  I.e., even when you execute @code{'} in
 3035: compile state, it still gives you the interpretation semantics.  And
 3036: whatever that state is, @code{execute} performs the semantics
 3037: represented by the XT (i.e., for XTs produced with @code{'} the
 3038: interpretation semantics).
 3039: 
 3040: Reference: @ref{Tokens for Words}.
 3041: 
 3042: 
 3043: @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
 3044: @section Exceptions
 3045: @cindex exceptions tutorial
 3046: 
 3047: @code{throw ( n -- )} causes an exception unless n is zero.
 3048: 
 3049: @example
 3050: 100 throw .s
 3051: 0 throw .s
 3052: @end example
 3053: 
 3054: @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
 3055: it catches exceptions and pushes the number of the exception on the
 3056: stack (or 0, if the xt executed without exception).  If there was an
 3057: exception, the stacks have the same depth as when entering @code{catch}:
 3058: 
 3059: @example
 3060: .s
 3061: 3 0 ' / catch .s
 3062: 3 2 ' / catch .s
 3063: @end example
 3064: 
 3065: @assignment
 3066: Try the same with @code{execute} instead of @code{catch}.
 3067: @endassignment
 3068: 
 3069: @code{Throw} always jumps to the dynamically next enclosing
 3070: @code{catch}, even if it has to leave several call levels to achieve
 3071: this:
 3072: 
 3073: @example
 3074: : foo 100 throw ;
 3075: : foo1 foo ." after foo" ;
 3076: : bar ['] foo1 catch ;
 3077: bar .
 3078: @end example
 3079: 
 3080: It is often important to restore a value upon leaving a definition, even
 3081: if the definition is left through an exception.  You can ensure this
 3082: like this:
 3083: 
 3084: @example
 3085: : ...
 3086:    save-x
 3087:    ['] word-changing-x catch ( ... n )
 3088:    restore-x
 3089:    ( ... n ) throw ;
 3090: @end example
 3091: 
 3092: Gforth provides an alternative syntax in addition to @code{catch}:
 3093: @code{try ... recover ... endtry}.  If the code between @code{try} and
 3094: @code{recover} has an exception, the stack depths are restored, the
 3095: exception number is pushed on the stack, and the code between
 3096: @code{recover} and @code{endtry} is performed.  E.g., the definition for
 3097: @code{catch} is
 3098: 
 3099: @example
 3100: : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
 3101:   try
 3102:     execute 0
 3103:   recover
 3104:     nip
 3105:   endtry ;
 3106: @end example
 3107: 
 3108: The equivalent to the restoration code above is
 3109: 
 3110: @example
 3111: : ...
 3112:   save-x
 3113:   try
 3114:     word-changing-x 0
 3115:   recover endtry
 3116:   restore-x
 3117:   throw ;
 3118: @end example
 3119: 
 3120: This works if @code{word-changing-x} does not change the stack depth,
 3121: otherwise you should add some code between @code{recover} and
 3122: @code{endtry} to balance the stack.
 3123: 
 3124: Reference: @ref{Exception Handling}.
 3125: 
 3126: 
 3127: @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
 3128: @section Defining Words
 3129: @cindex defining words tutorial
 3130: @cindex does> tutorial
 3131: @cindex create...does> tutorial
 3132: 
 3133: @c before semantics?
 3134: 
 3135: @code{:}, @code{create}, and @code{variable} are definition words: They
 3136: define other words.  @code{Constant} is another definition word:
 3137: 
 3138: @example
 3139: 5 constant foo
 3140: foo .
 3141: @end example
 3142: 
 3143: You can also use the prefixes @code{2} (double-cell) and @code{f}
 3144: (floating point) with @code{variable} and @code{constant}.
 3145: 
 3146: You can also define your own defining words.  E.g.:
 3147: 
 3148: @example
 3149: : variable ( "name" -- )
 3150:   create 0 , ;
 3151: @end example
 3152: 
 3153: You can also define defining words that create words that do something
 3154: other than just producing their address:
 3155: 
 3156: @example
 3157: : constant ( n "name" -- )
 3158:   create ,
 3159: does> ( -- n )
 3160:   ( addr ) @@ ;
 3161: 
 3162: 5 constant foo
 3163: foo .
 3164: @end example
 3165: 
 3166: The definition of @code{constant} above ends at the @code{does>}; i.e.,
 3167: @code{does>} replaces @code{;}, but it also does something else: It
 3168: changes the last defined word such that it pushes the address of the
 3169: body of the word and then performs the code after the @code{does>}
 3170: whenever it is called.
 3171: 
 3172: In the example above, @code{constant} uses @code{,} to store 5 into the
 3173: body of @code{foo}.  When @code{foo} executes, it pushes the address of
 3174: the body onto the stack, then (in the code after the @code{does>})
 3175: fetches the 5 from there.
 3176: 
 3177: The stack comment near the @code{does>} reflects the stack effect of the
 3178: defined word, not the stack effect of the code after the @code{does>}
 3179: (the difference is that the code expects the address of the body that
 3180: the stack comment does not show).
 3181: 
 3182: You can use these definition words to do factoring in cases that involve
 3183: (other) definition words.  E.g., a field offset is always added to an
 3184: address.  Instead of defining
 3185: 
 3186: @example
 3187: 2 cells constant offset-field1
 3188: @end example
 3189: 
 3190: and using this like
 3191: 
 3192: @example
 3193: ( addr ) offset-field1 +
 3194: @end example
 3195: 
 3196: you can define a definition word
 3197: 
 3198: @example
 3199: : simple-field ( n "name" -- )
 3200:   create ,
 3201: does> ( n1 -- n1+n )
 3202:   ( addr ) @@ + ;
 3203: @end example
 3204: 
 3205: Definition and use of field offsets now look like this:
 3206: 
 3207: @example
 3208: 2 cells simple-field field1
 3209: create mystruct 4 cells allot
 3210: mystruct .s field1 .s drop
 3211: @end example
 3212: 
 3213: If you want to do something with the word without performing the code
 3214: after the @code{does>}, you can access the body of a @code{create}d word
 3215: with @code{>body ( xt -- addr )}:
 3216: 
 3217: @example
 3218: : value ( n "name" -- )
 3219:   create ,
 3220: does> ( -- n1 )
 3221:   @@ ;
 3222: : to ( n "name" -- )
 3223:   ' >body ! ;
 3224: 
 3225: 5 value foo
 3226: foo .
 3227: 7 to foo
 3228: foo .
 3229: @end example
 3230: 
 3231: @assignment
 3232: Define @code{defer ( "name" -- )}, which creates a word that stores an
 3233: XT (at the start the XT of @code{abort}), and upon execution
 3234: @code{execute}s the XT.  Define @code{is ( xt "name" -- )} that stores
 3235: @code{xt} into @code{name}, a word defined with @code{defer}.  Indirect
 3236: recursion is one application of @code{defer}.
 3237: @endassignment
 3238: 
 3239: Reference: @ref{User-defined Defining Words}.
 3240: 
 3241: 
 3242: @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
 3243: @section Arrays and Records
 3244: @cindex arrays tutorial
 3245: @cindex records tutorial
 3246: @cindex structs tutorial
 3247: 
 3248: Forth has no standard words for defining data structures such as arrays
 3249: and records (structs in C terminology), but you can build them yourself
 3250: based on address arithmetic.  You can also define words for defining
 3251: arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
 3252: 
 3253: One of the first projects a Forth newcomer sets out upon when learning
 3254: about defining words is an array defining word (possibly for
 3255: n-dimensional arrays).  Go ahead and do it, I did it, too; you will
 3256: learn something from it.  However, don't be disappointed when you later
 3257: learn that you have little use for these words (inappropriate use would
 3258: be even worse).  I have not yet found a set of useful array words yet;
 3259: the needs are just too diverse, and named, global arrays (the result of
 3260: naive use of defining words) are often not flexible enough (e.g.,
 3261: consider how to pass them as parameters).  Another such project is a set
 3262: of words to help dealing with strings.
 3263: 
 3264: On the other hand, there is a useful set of record words, and it has
 3265: been defined in @file{compat/struct.fs}; these words are predefined in
 3266: Gforth.  They are explained in depth elsewhere in this manual (see
 3267: @pxref{Structures}).  The @code{simple-field} example above is
 3268: simplified variant of fields in this package.
 3269: 
 3270: 
 3271: @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
 3272: @section @code{POSTPONE}
 3273: @cindex postpone tutorial
 3274: 
 3275: You can compile the compilation semantics (instead of compiling the
 3276: interpretation semantics) of a word with @code{POSTPONE}:
 3277: 
 3278: @example
 3279: : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
 3280:  POSTPONE + ; immediate
 3281: : foo ( n1 n2 -- n )
 3282:  MY-+ ;
 3283: 1 2 foo .
 3284: see foo
 3285: @end example
 3286: 
 3287: During the definition of @code{foo} the text interpreter performs the
 3288: compilation semantics of @code{MY-+}, which performs the compilation
 3289: semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
 3290: 
 3291: This example also displays separate stack comments for the compilation
 3292: semantics and for the stack effect of the compiled code.  For words with
 3293: default compilation semantics these stack effects are usually not
 3294: displayed; the stack effect of the compilation semantics is always
 3295: @code{( -- )} for these words, the stack effect for the compiled code is
 3296: the stack effect of the interpretation semantics.
 3297: 
 3298: Note that the state of the interpreter does not come into play when
 3299: performing the compilation semantics in this way.  You can also perform
 3300: it interpretively, e.g.:
 3301: 
 3302: @example
 3303: : foo2 ( n1 n2 -- n )
 3304:  [ MY-+ ] ;
 3305: 1 2 foo .
 3306: see foo
 3307: @end example
 3308: 
 3309: However, there are some broken Forth systems where this does not always
 3310: work, and therefore this practice was been declared non-standard in
 3311: 1999.
 3312: @c !! repair.fs
 3313: 
 3314: Here is another example for using @code{POSTPONE}:
 3315: 
 3316: @example
 3317: : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
 3318:  POSTPONE negate POSTPONE + ; immediate compile-only
 3319: : bar ( n1 n2 -- n )
 3320:   MY-- ;
 3321: 2 1 bar .
 3322: see bar
 3323: @end example
 3324: 
 3325: You can define @code{ENDIF} in this way:
 3326: 
 3327: @example
 3328: : ENDIF ( Compilation: orig -- )
 3329:   POSTPONE then ; immediate
 3330: @end example
 3331: 
 3332: @assignment
 3333: Write @code{MY-2DUP} that has compilation semantics equivalent to
 3334: @code{2dup}, but compiles @code{over over}.
 3335: @endassignment
 3336: 
 3337: @c !! @xref{Macros} for reference
 3338: 
 3339: 
 3340: @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
 3341: @section @code{Literal}
 3342: @cindex literal tutorial
 3343: 
 3344: You cannot @code{POSTPONE} numbers:
 3345: 
 3346: @example
 3347: : [FOO] POSTPONE 500 ; immediate
 3348: @end example
 3349: 
 3350: Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
 3351: 
 3352: @example
 3353: : [FOO] ( compilation: --; run-time: -- n )
 3354:   500 POSTPONE literal ; immediate
 3355: 
 3356: : flip [FOO] ;
 3357: flip .
 3358: see flip
 3359: @end example
 3360: 
 3361: @code{LITERAL} consumes a number at compile-time (when it's compilation
 3362: semantics are executed) and pushes it at run-time (when the code it
 3363: compiled is executed).  A frequent use of @code{LITERAL} is to compile a
 3364: number computed at compile time into the current word:
 3365: 
 3366: @example
 3367: : bar ( -- n )
 3368:   [ 2 2 + ] literal ;
 3369: see bar
 3370: @end example
 3371: 
 3372: @assignment
 3373: Write @code{]L} which allows writing the example above as @code{: bar (
 3374: -- n ) [ 2 2 + ]L ;}
 3375: @endassignment
 3376: 
 3377: @c !! @xref{Macros} for reference
 3378: 
 3379: 
 3380: @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
 3381: @section Advanced macros
 3382: @cindex macros, advanced tutorial
 3383: @cindex run-time code generation, tutorial
 3384: 
 3385: Reconsider @code{map-array} from @ref{Execution Tokens Tutorial,,
 3386: Execution Tokens}.  It frequently performs @code{execute}, a relatively
 3387: expensive operation in some Forth implementations.  You can use
 3388: @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
 3389: and produce a word that contains the word to be performed directly:
 3390: 
 3391: @c use ]] ... [[
 3392: @example
 3393: : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
 3394: \ at run-time, execute xt ( ... x -- ... ) for each element of the
 3395: \ array beginning at addr and containing u elements
 3396:   @{ xt @}
 3397:   POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
 3398:     POSTPONE i POSTPONE @@ xt compile,
 3399:   1 cells POSTPONE literal POSTPONE +loop ;
 3400: 
 3401: : sum-array ( addr u -- n )
 3402:  0 rot rot [ ' + compile-map-array ] ;
 3403: see sum-array
 3404: a 5 sum-array .
 3405: @end example
 3406: 
 3407: You can use the full power of Forth for generating the code; here's an
 3408: example where the code is generated in a loop:
 3409: 
 3410: @example
 3411: : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
 3412: \ n2=n1+(addr1)*n, addr2=addr1+cell
 3413:   POSTPONE tuck POSTPONE @@
 3414:   POSTPONE literal POSTPONE * POSTPONE +
 3415:   POSTPONE swap POSTPONE cell+ ;
 3416: 
 3417: : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
 3418: \ n=v1*v2 (inner product), where the v_i are represented as addr_i u
 3419:   0 postpone literal postpone swap
 3420:   [ ' compile-vmul-step compile-map-array ]
 3421:   postpone drop ;
 3422: see compile-vmul
 3423: 
 3424: : a-vmul ( addr -- n )
 3425: \ n=a*v, where v is a vector that's as long as a and starts at addr
 3426:  [ a 5 compile-vmul ] ;
 3427: see a-vmul
 3428: a a-vmul .
 3429: @end example
 3430: 
 3431: This example uses @code{compile-map-array} to show off, but you could
 3432: also use @code{map-array} instead (try it now!).
 3433: 
 3434: You can use this technique for efficient multiplication of large
 3435: matrices.  In matrix multiplication, you multiply every line of one
 3436: matrix with every column of the other matrix.  You can generate the code
 3437: for one line once, and use it for every column.  The only downside of
 3438: this technique is that it is cumbersome to recover the memory consumed
 3439: by the generated code when you are done (and in more complicated cases
 3440: it is not possible portably).
 3441: 
 3442: @c !! @xref{Macros} for reference
 3443: 
 3444: 
 3445: @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
 3446: @section Compilation Tokens
 3447: @cindex compilation tokens, tutorial
 3448: @cindex CT, tutorial
 3449: 
 3450: This section is Gforth-specific.  You can skip it.
 3451: 
 3452: @code{' word compile,} compiles the interpretation semantics.  For words
 3453: with default compilation semantics this is the same as performing the
 3454: compilation semantics.  To represent the compilation semantics of other
 3455: words (e.g., words like @code{if} that have no interpretation
 3456: semantics), Gforth has the concept of a compilation token (CT,
 3457: consisting of two cells), and words @code{comp'} and @code{[comp']}.
 3458: You can perform the compilation semantics represented by a CT with
 3459: @code{execute}:
 3460: 
 3461: @example
 3462: : foo2 ( n1 n2 -- n )
 3463:    [ comp' + execute ] ;
 3464: see foo
 3465: @end example
 3466: 
 3467: You can compile the compilation semantics represented by a CT with
 3468: @code{postpone,}:
 3469: 
 3470: @example
 3471: : foo3 ( -- )
 3472:   [ comp' + postpone, ] ;
 3473: see foo3
 3474: @end example
 3475: 
 3476: @code{[ comp' word postpone, ]} is equivalent to @code{POSTPONE word}.
 3477: @code{comp'} is particularly useful for words that have no
 3478: interpretation semantics:
 3479: 
 3480: @example
 3481: ' if
 3482: comp' if .s 2drop
 3483: @end example
 3484: 
 3485: Reference: @ref{Tokens for Words}.
 3486: 
 3487: 
 3488: @node Wordlists and Search Order Tutorial,  , Compilation Tokens Tutorial, Tutorial
 3489: @section Wordlists and Search Order
 3490: @cindex wordlists tutorial
 3491: @cindex search order, tutorial
 3492: 
 3493: The dictionary is not just a memory area that allows you to allocate
 3494: memory with @code{allot}, it also contains the Forth words, arranged in
 3495: several wordlists.  When searching for a word in a wordlist,
 3496: conceptually you start searching at the youngest and proceed towards
 3497: older words (in reality most systems nowadays use hash-tables); i.e., if
 3498: you define a word with the same name as an older word, the new word
 3499: shadows the older word.
 3500: 
 3501: Which wordlists are searched in which order is determined by the search
 3502: order.  You can display the search order with @code{order}.  It displays
 3503: first the search order, starting with the wordlist searched first, then
 3504: it displays the wordlist that will contain newly defined words.
 3505: 
 3506: You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
 3507: 
 3508: @example
 3509: wordlist constant mywords
 3510: @end example
 3511: 
 3512: @code{Set-current ( wid -- )} sets the wordlist that will contain newly
 3513: defined words (the @emph{current} wordlist):
 3514: 
 3515: @example
 3516: mywords set-current
 3517: order
 3518: @end example
 3519: 
 3520: Gforth does not display a name for the wordlist in @code{mywords}
 3521: because this wordlist was created anonymously with @code{wordlist}.
 3522: 
 3523: You can get the current wordlist with @code{get-current ( -- wid)}.  If
 3524: you want to put something into a specific wordlist without overall
 3525: effect on the current wordlist, this typically looks like this:
 3526: 
 3527: @example
 3528: get-current mywords set-current ( wid )
 3529: create someword
 3530: ( wid ) set-current
 3531: @end example
 3532: 
 3533: You can write the search order with @code{set-order ( wid1 .. widn n --
 3534: )} and read it with @code{get-order ( -- wid1 .. widn n )}.  The first
 3535: searched wordlist is topmost.
 3536: 
 3537: @example
 3538: get-order mywords swap 1+ set-order
 3539: order
 3540: @end example
 3541: 
 3542: Yes, the order of wordlists in the output of @code{order} is reversed
 3543: from stack comments and the output of @code{.s} and thus unintuitive.
 3544: 
 3545: @assignment
 3546: Define @code{>order ( wid -- )} with adds @code{wid} as first searched
 3547: wordlist to the search order.  Define @code{previous ( -- )}, which
 3548: removes the first searched wordlist from the search order.  Experiment
 3549: with boundary conditions (you will see some crashes or situations that
 3550: are hard or impossible to leave).
 3551: @endassignment
 3552: 
 3553: The search order is a powerful foundation for providing features similar
 3554: to Modula-2 modules and C++ namespaces.  However, trying to modularize
 3555: programs in this way has disadvantages for debugging and reuse/factoring
 3556: that overcome the advantages in my experience (I don't do huge projects,
 3557: though).  These disadvantages are not so clear in other
 3558: languages/programming environments, because these languages are not so
 3559: strong in debugging and reuse.
 3560: 
 3561: @c !! example
 3562: 
 3563: Reference: @ref{Word Lists}.
 3564: 
 3565: @c ******************************************************************
 3566: @node Introduction, Words, Tutorial, Top
 3567: @comment node-name,     next,           previous, up
 3568: @chapter An Introduction to ANS Forth
 3569: @cindex Forth - an introduction
 3570: 
 3571: The difference of this chapter from the Tutorial (@pxref{Tutorial}) is
 3572: that it is slower-paced in its examples, but uses them to dive deep into
 3573: explaining Forth internals (not covered by the Tutorial).  Apart from
 3574: that, this chapter covers far less material.  It is suitable for reading
 3575: without using a computer.
 3576: 
 3577: The primary purpose of this manual is to document Gforth. However, since
 3578: Forth is not a widely-known language and there is a lack of up-to-date
 3579: teaching material, it seems worthwhile to provide some introductory
 3580: material.  For other sources of Forth-related
 3581: information, see @ref{Forth-related information}.
 3582: 
 3583: The examples in this section should work on any ANS Forth; the
 3584: output shown was produced using Gforth. Each example attempts to
 3585: reproduce the exact output that Gforth produces. If you try out the
 3586: examples (and you should), what you should type is shown @kbd{like this}
 3587: and Gforth's response is shown @code{like this}. The single exception is
 3588: that, where the example shows @key{RET} it means that you should
 3589: press the ``carriage return'' key. Unfortunately, some output formats for
 3590: this manual cannot show the difference between @kbd{this} and
 3591: @code{this} which will make trying out the examples harder (but not
 3592: impossible).
 3593: 
 3594: Forth is an unusual language. It provides an interactive development
 3595: environment which includes both an interpreter and compiler. Forth
 3596: programming style encourages you to break a problem down into many
 3597: @cindex factoring
 3598: small fragments (@dfn{factoring}), and then to develop and test each
 3599: fragment interactively. Forth advocates assert that breaking the
 3600: edit-compile-test cycle used by conventional programming languages can
 3601: lead to great productivity improvements.
 3602: 
 3603: @menu
 3604: * Introducing the Text Interpreter::  
 3605: * Stacks and Postfix notation::  
 3606: * Your first definition::       
 3607: * How does that work?::         
 3608: * Forth is written in Forth::   
 3609: * Review - elements of a Forth system::  
 3610: * Where to go next::            
 3611: * Exercises::                   
 3612: @end menu
 3613: 
 3614: @comment ----------------------------------------------
 3615: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
 3616: @section Introducing the Text Interpreter
 3617: @cindex text interpreter
 3618: @cindex outer interpreter
 3619: 
 3620: @c IMO this is too detailed and the pace is too slow for
 3621: @c an introduction.  If you know German, take a look at
 3622: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html 
 3623: @c to see how I do it - anton 
 3624: 
 3625: @c nac-> Where I have accepted your comments 100% and modified the text
 3626: @c accordingly, I have deleted your comments. Elsewhere I have added a
 3627: @c response like this to attempt to rationalise what I have done. Of
 3628: @c course, this is a very clumsy mechanism for something that would be
 3629: @c done far more efficiently over a beer. Please delete any dialogue
 3630: @c you consider closed.
 3631: 
 3632: When you invoke the Forth image, you will see a startup banner printed
 3633: and nothing else (if you have Gforth installed on your system, try
 3634: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
 3635: its command line interpreter, which is called the @dfn{Text Interpreter}
 3636: (also known as the @dfn{Outer Interpreter}).  (You will learn a lot
 3637: about the text interpreter as you read through this chapter, for more
 3638: detail @pxref{The Text Interpreter}).
 3639: 
 3640: Although it's not obvious, Forth is actually waiting for your
 3641: input. Type a number and press the @key{RET} key:
 3642: 
 3643: @example
 3644: @kbd{45@key{RET}}  ok
 3645: @end example
 3646: 
 3647: Rather than give you a prompt to invite you to input something, the text
 3648: interpreter prints a status message @i{after} it has processed a line
 3649: of input. The status message in this case (``@code{ ok}'' followed by
 3650: carriage-return) indicates that the text interpreter was able to process
 3651: all of your input successfully. Now type something illegal:
 3652: 
 3653: @example
 3654: @kbd{qwer341@key{RET}}
 3655: :1: Undefined word
 3656: qwer341
 3657: ^^^^^^^
 3658: $400D2BA8 Bounce
 3659: $400DBDA8 no.extensions
 3660: @end example
 3661: 
 3662: The exact text, other than the ``Undefined word'' may differ slightly on
 3663: your system, but the effect is the same; when the text interpreter
 3664: detects an error, it discards any remaining text on a line, resets
 3665: certain internal state and prints an error message. For a detailed description of error messages see @ref{Error
 3666: messages}.
 3667: 
 3668: The text interpreter waits for you to press carriage-return, and then
 3669: processes your input line. Starting at the beginning of the line, it
 3670: breaks the line into groups of characters separated by spaces. For each
 3671: group of characters in turn, it makes two attempts to do something:
 3672: 
 3673: @itemize @bullet
 3674: @item
 3675: @cindex name dictionary
 3676: It tries to treat it as a command. It does this by searching a @dfn{name
 3677: dictionary}. If the group of characters matches an entry in the name
 3678: dictionary, the name dictionary provides the text interpreter with
 3679: information that allows the text interpreter perform some actions. In
 3680: Forth jargon, we say that the group
 3681: @cindex word
 3682: @cindex definition
 3683: @cindex execution token
 3684: @cindex xt
 3685: of characters names a @dfn{word}, that the dictionary search returns an
 3686: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
 3687: word, and that the text interpreter executes the xt. Often, the terms
 3688: @dfn{word} and @dfn{definition} are used interchangeably.
 3689: @item
 3690: If the text interpreter fails to find a match in the name dictionary, it
 3691: tries to treat the group of characters as a number in the current number
 3692: base (when you start up Forth, the current number base is base 10). If
 3693: the group of characters legitimately represents a number, the text
 3694: interpreter pushes the number onto a stack (we'll learn more about that
 3695: in the next section).
 3696: @end itemize
 3697: 
 3698: If the text interpreter is unable to do either of these things with any
 3699: group of characters, it discards the group of characters and the rest of
 3700: the line, then prints an error message. If the text interpreter reaches
 3701: the end of the line without error, it prints the status message ``@code{ ok}''
 3702: followed by carriage-return.
 3703: 
 3704: This is the simplest command we can give to the text interpreter:
 3705: 
 3706: @example
 3707: @key{RET}  ok
 3708: @end example
 3709: 
 3710: The text interpreter did everything we asked it to do (nothing) without
 3711: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
 3712: command:
 3713: 
 3714: @example
 3715: @kbd{12 dup fred dup@key{RET}}
 3716: :1: Undefined word
 3717: 12 dup fred dup
 3718:        ^^^^
 3719: $400D2BA8 Bounce
 3720: $400DBDA8 no.extensions
 3721: @end example
 3722: 
 3723: When you press the carriage-return key, the text interpreter starts to
 3724: work its way along the line:
 3725: 
 3726: @itemize @bullet
 3727: @item
 3728: When it gets to the space after the @code{2}, it takes the group of
 3729: characters @code{12} and looks them up in the name
 3730: dictionary@footnote{We can't tell if it found them or not, but assume
 3731: for now that it did not}. There is no match for this group of characters
 3732: in the name dictionary, so it tries to treat them as a number. It is
 3733: able to do this successfully, so it puts the number, 12, ``on the stack''
 3734: (whatever that means).
 3735: @item
 3736: The text interpreter resumes scanning the line and gets the next group
 3737: of characters, @code{dup}. It looks it up in the name dictionary and
 3738: (you'll have to take my word for this) finds it, and executes the word
 3739: @code{dup} (whatever that means).
 3740: @item
 3741: Once again, the text interpreter resumes scanning the line and gets the
 3742: group of characters @code{fred}. It looks them up in the name
 3743: dictionary, but can't find them. It tries to treat them as a number, but
 3744: they don't represent any legal number.
 3745: @end itemize
 3746: 
 3747: At this point, the text interpreter gives up and prints an error
 3748: message. The error message shows exactly how far the text interpreter
 3749: got in processing the line. In particular, it shows that the text
 3750: interpreter made no attempt to do anything with the final character
 3751: group, @code{dup}, even though we have good reason to believe that the
 3752: text interpreter would have no problem looking that word up and
 3753: executing it a second time.
 3754: 
 3755: 
 3756: @comment ----------------------------------------------
 3757: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
 3758: @section Stacks, postfix notation and parameter passing
 3759: @cindex text interpreter
 3760: @cindex outer interpreter
 3761: 
 3762: In procedural programming languages (like C and Pascal), the
 3763: building-block of programs is the @dfn{function} or @dfn{procedure}. These
 3764: functions or procedures are called with @dfn{explicit parameters}. For
 3765: example, in C we might write:
 3766: 
 3767: @example
 3768: total = total + new_volume(length,height,depth);
 3769: @end example
 3770: 
 3771: @noindent
 3772: where new_volume is a function-call to another piece of code, and total,
 3773: length, height and depth are all variables. length, height and depth are
 3774: parameters to the function-call.
 3775: 
 3776: In Forth, the equivalent of the function or procedure is the
 3777: @dfn{definition} and parameters are implicitly passed between
 3778: definitions using a shared stack that is visible to the
 3779: programmer. Although Forth does support variables, the existence of the
 3780: stack means that they are used far less often than in most other
 3781: programming languages. When the text interpreter encounters a number, it
 3782: will place (@dfn{push}) it on the stack. There are several stacks (the
 3783: actual number is implementation-dependent ...) and the particular stack
 3784: used for any operation is implied unambiguously by the operation being
 3785: performed. The stack used for all integer operations is called the @dfn{data
 3786: stack} and, since this is the stack used most commonly, references to
 3787: ``the data stack'' are often abbreviated to ``the stack''.
 3788: 
 3789: The stacks have a last-in, first-out (LIFO) organisation. If you type:
 3790: 
 3791: @example
 3792: @kbd{1 2 3@key{RET}}  ok
 3793: @end example
 3794: 
 3795: Then this instructs the text interpreter to placed three numbers on the
 3796: (data) stack. An analogy for the behaviour of the stack is to take a
 3797: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
 3798: the table. The 3 was the last card onto the pile (``last-in'') and if
 3799: you take a card off the pile then, unless you're prepared to fiddle a
 3800: bit, the card that you take off will be the 3 (``first-out''). The
 3801: number that will be first-out of the stack is called the @dfn{top of
 3802: stack}, which
 3803: @cindex TOS definition
 3804: is often abbreviated to @dfn{TOS}.
 3805: 
 3806: To understand how parameters are passed in Forth, consider the
 3807: behaviour of the definition @code{+} (pronounced ``plus''). You will not
 3808: be surprised to learn that this definition performs addition. More
 3809: precisely, it adds two number together and produces a result. Where does
 3810: it get the two numbers from? It takes the top two numbers off the
 3811: stack. Where does it place the result? On the stack. You can act-out the
 3812: behaviour of @code{+} with your playing cards like this:
 3813: 
 3814: @itemize @bullet
 3815: @item
 3816: Pick up two cards from the stack on the table
 3817: @item
 3818: Stare at them intently and ask yourself ``what @i{is} the sum of these two
 3819: numbers''
 3820: @item
 3821: Decide that the answer is 5
 3822: @item
 3823: Shuffle the two cards back into the pack and find a 5
 3824: @item
 3825: Put a 5 on the remaining ace that's on the table.
 3826: @end itemize
 3827: 
 3828: If you don't have a pack of cards handy but you do have Forth running,
 3829: you can use the definition @code{.s} to show the current state of the stack,
 3830: without affecting the stack. Type:
 3831: 
 3832: @example
 3833: @kbd{clearstack 1 2 3@key{RET}} ok
 3834: @kbd{.s@key{RET}} <3> 1 2 3  ok
 3835: @end example
 3836: 
 3837: The text interpreter looks up the word @code{clearstack} and executes
 3838: it; it tidies up the stack and removes any entries that may have been
 3839: left on it by earlier examples. The text interpreter pushes each of the
 3840: three numbers in turn onto the stack. Finally, the text interpreter
 3841: looks up the word @code{.s} and executes it. The effect of executing
 3842: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
 3843: followed by a list of all the items on the stack; the item on the far
 3844: right-hand side is the TOS.
 3845: 
 3846: You can now type:
 3847: 
 3848: @example
 3849: @kbd{+ .s@key{RET}} <2> 1 5  ok
 3850: @end example
 3851: 
 3852: @noindent
 3853: which is correct; there are now 2 items on the stack and the result of
 3854: the addition is 5.
 3855: 
 3856: If you're playing with cards, try doing a second addition: pick up the
 3857: two cards, work out that their sum is 6, shuffle them into the pack,
 3858: look for a 6 and place that on the table. You now have just one item on
 3859: the stack. What happens if you try to do a third addition? Pick up the
 3860: first card, pick up the second card -- ah! There is no second card. This
 3861: is called a @dfn{stack underflow} and consitutes an error. If you try to
 3862: do the same thing with Forth it often reports an error (probably a Stack
 3863: Underflow or an Invalid Memory Address error).
 3864: 
 3865: The opposite situation to a stack underflow is a @dfn{stack overflow},
 3866: which simply accepts that there is a finite amount of storage space
 3867: reserved for the stack. To stretch the playing card analogy, if you had
 3868: enough packs of cards and you piled the cards up on the table, you would
 3869: eventually be unable to add another card; you'd hit the ceiling. Gforth
 3870: allows you to set the maximum size of the stacks. In general, the only
 3871: time that you will get a stack overflow is because a definition has a
 3872: bug in it and is generating data on the stack uncontrollably.
 3873: 
 3874: There's one final use for the playing card analogy. If you model your
 3875: stack using a pack of playing cards, the maximum number of items on
 3876: your stack will be 52 (I assume you didn't use the Joker). The maximum
 3877: @i{value} of any item on the stack is 13 (the King). In fact, the only
 3878: possible numbers are positive integer numbers 1 through 13; you can't
 3879: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
 3880: think about some of the cards, you can accommodate different
 3881: numbers. For example, you could think of the Jack as representing 0,
 3882: the Queen as representing -1 and the King as representing -2. Your
 3883: @i{range} remains unchanged (you can still only represent a total of 13
 3884: numbers) but the numbers that you can represent are -2 through 10.
 3885: 
 3886: In that analogy, the limit was the amount of information that a single
 3887: stack entry could hold, and Forth has a similar limit. In Forth, the
 3888: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
 3889: implementation dependent and affects the maximum value that a stack
 3890: entry can hold. A Standard Forth provides a cell size of at least
 3891: 16-bits, and most desktop systems use a cell size of 32-bits.
 3892: 
 3893: Forth does not do any type checking for you, so you are free to
 3894: manipulate and combine stack items in any way you wish. A convenient way
 3895: of treating stack items is as 2's complement signed integers, and that
 3896: is what Standard words like @code{+} do. Therefore you can type:
 3897: 
 3898: @example
 3899: @kbd{-5 12 + .s@key{RET}} <1> 7  ok
 3900: @end example
 3901: 
 3902: If you use numbers and definitions like @code{+} in order to turn Forth
 3903: into a great big pocket calculator, you will realise that it's rather
 3904: different from a normal calculator. Rather than typing 2 + 3 = you had
 3905: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
 3906: result). The terminology used to describe this difference is to say that
 3907: your calculator uses @dfn{Infix Notation} (parameters and operators are
 3908: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
 3909: operators are separate), also called @dfn{Reverse Polish Notation}.
 3910: 
 3911: Whilst postfix notation might look confusing to begin with, it has
 3912: several important advantages:
 3913: 
 3914: @itemize @bullet
 3915: @item
 3916: it is unambiguous
 3917: @item
 3918: it is more concise
 3919: @item
 3920: it fits naturally with a stack-based system
 3921: @end itemize
 3922: 
 3923: To examine these claims in more detail, consider these sums:
 3924: 
 3925: @example
 3926: 6 + 5 * 4 =
 3927: 4 * 5 + 6 =
 3928: @end example
 3929: 
 3930: If you're just learning maths or your maths is very rusty, you will
 3931: probably come up with the answer 44 for the first and 26 for the
 3932: second. If you are a bit of a whizz at maths you will remember the
 3933: @i{convention} that multiplication takes precendence over addition, and
 3934: you'd come up with the answer 26 both times. To explain the answer 26
 3935: to someone who got the answer 44, you'd probably rewrite the first sum
 3936: like this:
 3937: 
 3938: @example
 3939: 6 + (5 * 4) =
 3940: @end example
 3941: 
 3942: If what you really wanted was to perform the addition before the
 3943: multiplication, you would have to use parentheses to force it.
 3944: 
 3945: If you did the first two sums on a pocket calculator you would probably
 3946: get the right answers, unless you were very cautious and entered them using
 3947: these keystroke sequences:
 3948: 
 3949: 6 + 5 = * 4 =
 3950: 4 * 5 = + 6 =
 3951: 
 3952: Postfix notation is unambiguous because the order that the operators
 3953: are applied is always explicit; that also means that parentheses are
 3954: never required. The operators are @i{active} (the act of quoting the
 3955: operator makes the operation occur) which removes the need for ``=''.
 3956: 
 3957: The sum 6 + 5 * 4 can be written (in postfix notation) in two
 3958: equivalent ways:
 3959: 
 3960: @example
 3961: 6 5 4 * +      or:
 3962: 5 4 * 6 +
 3963: @end example
 3964: 
 3965: An important thing that you should notice about this notation is that
 3966: the @i{order} of the numbers does not change; if you want to subtract
 3967: 2 from 10 you type @code{10 2 -}.
 3968: 
 3969: The reason that Forth uses postfix notation is very simple to explain: it
 3970: makes the implementation extremely simple, and it follows naturally from
 3971: using the stack as a mechanism for passing parameters. Another way of
 3972: thinking about this is to realise that all Forth definitions are
 3973: @i{active}; they execute as they are encountered by the text
 3974: interpreter. The result of this is that the syntax of Forth is trivially
 3975: simple.
 3976: 
 3977: 
 3978: 
 3979: @comment ----------------------------------------------
 3980: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
 3981: @section Your first Forth definition
 3982: @cindex first definition
 3983: 
 3984: Until now, the examples we've seen have been trivial; we've just been
 3985: using Forth as a bigger-than-pocket calculator. Also, each calculation
 3986: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
 3987: again@footnote{That's not quite true. If you press the up-arrow key on
 3988: your keyboard you should be able to scroll back to any earlier command,
 3989: edit it and re-enter it.} In this section we'll see how to add new
 3990: words to Forth's vocabulary.
 3991: 
 3992: The easiest way to create a new word is to use a @dfn{colon
 3993: definition}. We'll define a few and try them out before worrying too
 3994: much about how they work. Try typing in these examples; be careful to
 3995: copy the spaces accurately:
 3996: 
 3997: @example
 3998: : add-two 2 + . ;
 3999: : greet ." Hello and welcome" ;
 4000: : demo 5 add-two ;
 4001: @end example
 4002: 
 4003: @noindent
 4004: Now try them out:
 4005: 
 4006: @example
 4007: @kbd{greet@key{RET}} Hello and welcome  ok
 4008: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome  ok
 4009: @kbd{4 add-two@key{RET}} 6  ok
 4010: @kbd{demo@key{RET}} 7  ok
 4011: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11  ok
 4012: @end example
 4013: 
 4014: The first new thing that we've introduced here is the pair of words
 4015: @code{:} and @code{;}. These are used to start and terminate a new
 4016: definition, respectively. The first word after the @code{:} is the name
 4017: for the new definition.
 4018: 
 4019: As you can see from the examples, a definition is built up of words that
 4020: have already been defined; Forth makes no distinction between
 4021: definitions that existed when you started the system up, and those that
 4022: you define yourself.
 4023: 
 4024: The examples also introduce the words @code{.} (dot), @code{."}
 4025: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
 4026: the stack and displays it. It's like @code{.s} except that it only
 4027: displays the top item of the stack and it is destructive; after it has
 4028: executed, the number is no longer on the stack. There is always one
 4029: space printed after the number, and no spaces before it. Dot-quote
 4030: defines a string (a sequence of characters) that will be printed when
 4031: the word is executed. The string can contain any printable characters
 4032: except @code{"}. A @code{"} has a special function; it is not a Forth
 4033: word but it acts as a delimiter (the way that delimiters work is
 4034: described in the next section). Finally, @code{dup} duplicates the value
 4035: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
 4036: 
 4037: We already know that the text interpreter searches through the
 4038: dictionary to locate names. If you've followed the examples earlier, you
 4039: will already have a definition called @code{add-two}. Lets try modifying
 4040: it by typing in a new definition:
 4041: 
 4042: @example
 4043: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two  ok
 4044: @end example
 4045: 
 4046: Forth recognised that we were defining a word that already exists, and
 4047: printed a message to warn us of that fact. Let's try out the new
 4048: definition:
 4049: 
 4050: @example
 4051: @kbd{9 add-two@key{RET}} 9 + 2 =11  ok
 4052: @end example
 4053: 
 4054: @noindent
 4055: All that we've actually done here, though, is to create a new
 4056: definition, with a particular name. The fact that there was already a
 4057: definition with the same name did not make any difference to the way
 4058: that the new definition was created (except that Forth printed a warning
 4059: message). The old definition of add-two still exists (try @code{demo}
 4060: again to see that this is true). Any new definition will use the new
 4061: definition of @code{add-two}, but old definitions continue to use the
 4062: version that already existed at the time that they were @code{compiled}.
 4063: 
 4064: Before you go on to the next section, try defining and redefining some
 4065: words of your own.
 4066: 
 4067: @comment ----------------------------------------------
 4068: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
 4069: @section How does that work?
 4070: @cindex parsing words
 4071: 
 4072: @c That's pretty deep (IMO way too deep) for an introduction. - anton
 4073: 
 4074: @c Is it a good idea to talk about the interpretation semantics of a
 4075: @c number? We don't have an xt to go along with it. - anton
 4076: 
 4077: @c Now that I have eliminated execution semantics, I wonder if it would not
 4078: @c be better to keep them (or add run-time semantics), to make it easier to
 4079: @c explain what compilation semantics usually does. - anton
 4080: 
 4081: @c nac-> I removed the term ``default compilation sematics'' from the
 4082: @c introductory chapter. Removing ``execution semantics'' was making
 4083: @c everything simpler to explain, then I think the use of this term made
 4084: @c everything more complex again. I replaced it with ``default
 4085: @c semantics'' (which is used elsewhere in the manual) by which I mean
 4086: @c ``a definition that has neither the immediate nor the compile-only
 4087: @c flag set''.
 4088: 
 4089: @c anton: I have eliminated default semantics (except in one place where it
 4090: @c means "default interpretation and compilation semantics"), because it
 4091: @c makes no sense in the presence of combined words.  I reverted to
 4092: @c "execution semantics" where necessary.
 4093: 
 4094: @c nac-> I reworded big chunks of the ``how does that work''
 4095: @c section (and, unusually for me, I think I even made it shorter!).  See
 4096: @c what you think -- I know I have not addressed your primary concern
 4097: @c that it is too heavy-going for an introduction. From what I understood
 4098: @c of your course notes it looks as though they might be a good framework. 
 4099: @c Things that I've tried to capture here are some things that came as a
 4100: @c great revelation here when I first understood them. Also, I like the
 4101: @c fact that a very simple code example shows up almost all of the issues
 4102: @c that you need to understand to see how Forth works. That's unique and
 4103: @c worthwhile to emphasise.
 4104: 
 4105: @c anton: I think it's a good idea to present the details, especially those
 4106: @c that you found to be a revelation, and probably the tutorial tries to be
 4107: @c too superficial and does not get some of the things across that make
 4108: @c Forth special.  I do believe that most of the time these things should
 4109: @c be discussed at the end of a section or in separate sections instead of
 4110: @c in the middle of a section (e.g., the stuff you added in "User-defined
 4111: @c defining words" leads in a completely different direction from the rest
 4112: @c of the section).
 4113: 
 4114: Now we're going to take another look at the definition of @code{add-two}
 4115: from the previous section. From our knowledge of the way that the text
 4116: interpreter works, we would have expected this result when we tried to
 4117: define @code{add-two}:
 4118: 
 4119: @example
 4120: @kbd{: add-two 2 + . ;@key{RET}}
 4121:   ^^^^^^^
 4122: Error: Undefined word
 4123: @end example
 4124: 
 4125: The reason that this didn't happen is bound up in the way that @code{:}
 4126: works. The word @code{:} does two special things. The first special
 4127: thing that it does prevents the text interpreter from ever seeing the
 4128: characters @code{add-two}. The text interpreter uses a variable called
 4129: @cindex modifying >IN
 4130: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
 4131: input line. When it encounters the word @code{:} it behaves in exactly
 4132: the same way as it does for any other word; it looks it up in the name
 4133: dictionary, finds its xt and executes it. When @code{:} executes, it
 4134: looks at the input buffer, finds the word @code{add-two} and advances the
 4135: value of @code{>IN} to point past it. It then does some other stuff
 4136: associated with creating the new definition (including creating an entry
 4137: for @code{add-two} in the name dictionary). When the execution of @code{:}
 4138: completes, control returns to the text interpreter, which is oblivious
 4139: to the fact that it has been tricked into ignoring part of the input
 4140: line.
 4141: 
 4142: @cindex parsing words
 4143: Words like @code{:} -- words that advance the value of @code{>IN} and so
 4144: prevent the text interpreter from acting on the whole of the input line
 4145: -- are called @dfn{parsing words}.
 4146: 
 4147: @cindex @code{state} - effect on the text interpreter
 4148: @cindex text interpreter - effect of state
 4149: The second special thing that @code{:} does is change the value of a
 4150: variable called @code{state}, which affects the way that the text
 4151: interpreter behaves. When Gforth starts up, @code{state} has the value
 4152: 0, and the text interpreter is said to be @dfn{interpreting}. During a
 4153: colon definition (started with @code{:}), @code{state} is set to -1 and
 4154: the text interpreter is said to be @dfn{compiling}.
 4155: 
 4156: In this example, the text interpreter is compiling when it processes the
 4157: string ``@code{2 + . ;}''. It still breaks the string down into
 4158: character sequences in the same way. However, instead of pushing the
 4159: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
 4160: into the definition of @code{add-two} that will make the number @code{2} get
 4161: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
 4162: the behaviours of @code{+} and @code{.} are also compiled into the
 4163: definition.
 4164: 
 4165: One category of words don't get compiled. These so-called @dfn{immediate
 4166: words} get executed (performed @i{now}) regardless of whether the text
 4167: interpreter is interpreting or compiling. The word @code{;} is an
 4168: immediate word. Rather than being compiled into the definition, it
 4169: executes. Its effect is to terminate the current definition, which
 4170: includes changing the value of @code{state} back to 0.
 4171: 
 4172: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
 4173: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
 4174: definition.
 4175: 
 4176: In Forth, every word or number can be described in terms of two
 4177: properties:
 4178: 
 4179: @itemize @bullet
 4180: @item
 4181: @cindex interpretation semantics
 4182: Its @dfn{interpretation semantics} describe how it will behave when the
 4183: text interpreter encounters it in @dfn{interpret} state. The
 4184: interpretation semantics of a word are represented by an @dfn{execution
 4185: token}.
 4186: @item
 4187: @cindex compilation semantics
 4188: Its @dfn{compilation semantics} describe how it will behave when the
 4189: text interpreter encounters it in @dfn{compile} state. The compilation
 4190: semantics of a word are represented in an implementation-dependent way;
 4191: Gforth uses a @dfn{compilation token}.
 4192: @end itemize
 4193: 
 4194: @noindent
 4195: Numbers are always treated in a fixed way:
 4196: 
 4197: @itemize @bullet
 4198: @item
 4199: When the number is @dfn{interpreted}, its behaviour is to push the
 4200: number onto the stack.
 4201: @item
 4202: When the number is @dfn{compiled}, a piece of code is appended to the
 4203: current definition that pushes the number when it runs. (In other words,
 4204: the compilation semantics of a number are to postpone its interpretation
 4205: semantics until the run-time of the definition that it is being compiled
 4206: into.)
 4207: @end itemize
 4208: 
 4209: Words don't behave in such a regular way, but most have @i{default
 4210: semantics} which means that they behave like this:
 4211: 
 4212: @itemize @bullet
 4213: @item
 4214: The @dfn{interpretation semantics} of the word are to do something useful.
 4215: @item
 4216: The @dfn{compilation semantics} of the word are to append its
 4217: @dfn{interpretation semantics} to the current definition (so that its
 4218: run-time behaviour is to do something useful).
 4219: @end itemize
 4220: 
 4221: @cindex immediate words
 4222: The actual behaviour of any particular word can be controlled by using
 4223: the words @code{immediate} and @code{compile-only} when the word is
 4224: defined. These words set flags in the name dictionary entry of the most
 4225: recently defined word, and these flags are retrieved by the text
 4226: interpreter when it finds the word in the name dictionary.
 4227: 
 4228: A word that is marked as @dfn{immediate} has compilation semantics that
 4229: are identical to its interpretation semantics. In other words, it
 4230: behaves like this:
 4231: 
 4232: @itemize @bullet
 4233: @item
 4234: The @dfn{interpretation semantics} of the word are to do something useful.
 4235: @item
 4236: The @dfn{compilation semantics} of the word are to do something useful
 4237: (and actually the same thing); i.e., it is executed during compilation.
 4238: @end itemize
 4239: 
 4240: Marking a word as @dfn{compile-only} prohibits the text interpreter from
 4241: performing the interpretation semantics of the word directly; an attempt
 4242: to do so will generate an error. It is never necessary to use
 4243: @code{compile-only} (and it is not even part of ANS Forth, though it is
 4244: provided by many implementations) but it is good etiquette to apply it
 4245: to a word that will not behave correctly (and might have unexpected
 4246: side-effects) in interpret state. For example, it is only legal to use
 4247: the conditional word @code{IF} within a definition. If you forget this
 4248: and try to use it elsewhere, the fact that (in Gforth) it is marked as
 4249: @code{compile-only} allows the text interpreter to generate a helpful
 4250: error message rather than subjecting you to the consequences of your
 4251: folly.
 4252: 
 4253: This example shows the difference between an immediate and a
 4254: non-immediate word:
 4255: 
 4256: @example
 4257: : show-state state @@ . ;
 4258: : show-state-now show-state ; immediate
 4259: : word1 show-state ;
 4260: : word2 show-state-now ;
 4261: @end example
 4262: 
 4263: The word @code{immediate} after the definition of @code{show-state-now}
 4264: makes that word an immediate word. These definitions introduce a new
 4265: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
 4266: variable, and leaves it on the stack. Therefore, the behaviour of
 4267: @code{show-state} is to print a number that represents the current value
 4268: of @code{state}.
 4269: 
 4270: When you execute @code{word1}, it prints the number 0, indicating that
 4271: the system is interpreting. When the text interpreter compiled the
 4272: definition of @code{word1}, it encountered @code{show-state} whose
 4273: compilation semantics are to append its interpretation semantics to the
 4274: current definition. When you execute @code{word1}, it performs the
 4275: interpretation semantics of @code{show-state}.  At the time that @code{word1}
 4276: (and therefore @code{show-state}) are executed, the system is
 4277: interpreting.
 4278: 
 4279: When you pressed @key{RET} after entering the definition of @code{word2},
 4280: you should have seen the number -1 printed, followed by ``@code{
 4281: ok}''. When the text interpreter compiled the definition of
 4282: @code{word2}, it encountered @code{show-state-now}, an immediate word,
 4283: whose compilation semantics are therefore to perform its interpretation
 4284: semantics. It is executed straight away (even before the text
 4285: interpreter has moved on to process another group of characters; the
 4286: @code{;} in this example). The effect of executing it are to display the
 4287: value of @code{state} @i{at the time that the definition of}
 4288: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
 4289: system is compiling at this time. If you execute @code{word2} it does
 4290: nothing at all.
 4291: 
 4292: @cindex @code{."}, how it works
 4293: Before leaving the subject of immediate words, consider the behaviour of
 4294: @code{."} in the definition of @code{greet}, in the previous
 4295: section. This word is both a parsing word and an immediate word. Notice
 4296: that there is a space between @code{."} and the start of the text
 4297: @code{Hello and welcome}, but that there is no space between the last
 4298: letter of @code{welcome} and the @code{"} character. The reason for this
 4299: is that @code{."} is a Forth word; it must have a space after it so that
 4300: the text interpreter can identify it. The @code{"} is not a Forth word;
 4301: it is a @dfn{delimiter}. The examples earlier show that, when the string
 4302: is displayed, there is neither a space before the @code{H} nor after the
 4303: @code{e}. Since @code{."} is an immediate word, it executes at the time
 4304: that @code{greet} is defined. When it executes, its behaviour is to
 4305: search forward in the input line looking for the delimiter. When it
 4306: finds the delimiter, it updates @code{>IN} to point past the
 4307: delimiter. It also compiles some magic code into the definition of
 4308: @code{greet}; the xt of a run-time routine that prints a text string. It
 4309: compiles the string @code{Hello and welcome} into memory so that it is
 4310: available to be printed later. When the text interpreter gains control,
 4311: the next word it finds in the input stream is @code{;} and so it
 4312: terminates the definition of @code{greet}.
 4313: 
 4314: 
 4315: @comment ----------------------------------------------
 4316: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
 4317: @section Forth is written in Forth
 4318: @cindex structure of Forth programs
 4319: 
 4320: When you start up a Forth compiler, a large number of definitions
 4321: already exist. In Forth, you develop a new application using bottom-up
 4322: programming techniques to create new definitions that are defined in
 4323: terms of existing definitions. As you create each definition you can
 4324: test and debug it interactively.
 4325: 
 4326: If you have tried out the examples in this section, you will probably
 4327: have typed them in by hand; when you leave Gforth, your definitions will
 4328: be lost. You can avoid this by using a text editor to enter Forth source
 4329: code into a file, and then loading code from the file using
 4330: @code{include} (@pxref{Forth source files}). A Forth source file is
 4331: processed by the text interpreter, just as though you had typed it in by
 4332: hand@footnote{Actually, there are some subtle differences -- see
 4333: @ref{The Text Interpreter}.}.
 4334: 
 4335: Gforth also supports the traditional Forth alternative to using text
 4336: files for program entry (@pxref{Blocks}).
 4337: 
 4338: In common with many, if not most, Forth compilers, most of Gforth is
 4339: actually written in Forth. All of the @file{.fs} files in the
 4340: installation directory@footnote{For example,
 4341: @file{/usr/local/share/gforth...}} are Forth source files, which you can
 4342: study to see examples of Forth programming.
 4343: 
 4344: Gforth maintains a history file that records every line that you type to
 4345: the text interpreter. This file is preserved between sessions, and is
 4346: used to provide a command-line recall facility. If you enter long
 4347: definitions by hand, you can use a text editor to paste them out of the
 4348: history file into a Forth source file for reuse at a later time
 4349: (for more information @pxref{Command-line editing}).
 4350: 
 4351: 
 4352: @comment ----------------------------------------------
 4353: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
 4354: @section Review - elements of a Forth system
 4355: @cindex elements of a Forth system
 4356: 
 4357: To summarise this chapter:
 4358: 
 4359: @itemize @bullet
 4360: @item
 4361: Forth programs use @dfn{factoring} to break a problem down into small
 4362: fragments called @dfn{words} or @dfn{definitions}.
 4363: @item
 4364: Forth program development is an interactive process.
 4365: @item
 4366: The main command loop that accepts input, and controls both
 4367: interpretation and compilation, is called the @dfn{text interpreter}
 4368: (also known as the @dfn{outer interpreter}).
 4369: @item
 4370: Forth has a very simple syntax, consisting of words and numbers
 4371: separated by spaces or carriage-return characters. Any additional syntax
 4372: is imposed by @dfn{parsing words}.
 4373: @item
 4374: Forth uses a stack to pass parameters between words. As a result, it
 4375: uses postfix notation.
 4376: @item
 4377: To use a word that has previously been defined, the text interpreter
 4378: searches for the word in the @dfn{name dictionary}.
 4379: @item
 4380: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
 4381: @item
 4382: The text interpreter uses the value of @code{state} to select between
 4383: the use of the @dfn{interpretation semantics} and the  @dfn{compilation
 4384: semantics} of a word that it encounters.
 4385: @item
 4386: The relationship between the @dfn{interpretation semantics} and
 4387: @dfn{compilation semantics} for a word
 4388: depend upon the way in which the word was defined (for example, whether
 4389: it is an @dfn{immediate} word).
 4390: @item
 4391: Forth definitions can be implemented in Forth (called @dfn{high-level
 4392: definitions}) or in some other way (usually a lower-level language and
 4393: as a result often called @dfn{low-level definitions}, @dfn{code
 4394: definitions} or @dfn{primitives}).
 4395: @item
 4396: Many Forth systems are implemented mainly in Forth.
 4397: @end itemize
 4398: 
 4399: 
 4400: @comment ----------------------------------------------
 4401: @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
 4402: @section Where To Go Next
 4403: @cindex where to go next
 4404: 
 4405: Amazing as it may seem, if you have read (and understood) this far, you
 4406: know almost all the fundamentals about the inner workings of a Forth
 4407: system. You certainly know enough to be able to read and understand the
 4408: rest of this manual and the ANS Forth document, to learn more about the
 4409: facilities that Forth in general and Gforth in particular provide. Even
 4410: scarier, you know almost enough to implement your own Forth system.
 4411: However, that's not a good idea just yet... better to try writing some
 4412: programs in Gforth.
 4413: 
 4414: Forth has such a rich vocabulary that it can be hard to know where to
 4415: start in learning it. This section suggests a few sets of words that are
 4416: enough to write small but useful programs. Use the word index in this
 4417: document to learn more about each word, then try it out and try to write
 4418: small definitions using it. Start by experimenting with these words:
 4419: 
 4420: @itemize @bullet
 4421: @item
 4422: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
 4423: @item
 4424: Comparison: @code{MIN MAX =}
 4425: @item
 4426: Logic: @code{AND OR XOR NOT}
 4427: @item
 4428: Stack manipulation: @code{DUP DROP SWAP OVER}
 4429: @item
 4430: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
 4431: @item
 4432: Input/Output: @code{. ." EMIT CR KEY}
 4433: @item
 4434: Defining words: @code{: ; CREATE}
 4435: @item
 4436: Memory allocation words: @code{ALLOT ,}
 4437: @item
 4438: Tools: @code{SEE WORDS .S MARKER}
 4439: @end itemize
 4440: 
 4441: When you have mastered those, go on to:
 4442: 
 4443: @itemize @bullet
 4444: @item
 4445: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
 4446: @item
 4447: Memory access: @code{@@ !}
 4448: @end itemize
 4449: 
 4450: When you have mastered these, there's nothing for it but to read through
 4451: the whole of this manual and find out what you've missed.
 4452: 
 4453: @comment ----------------------------------------------
 4454: @node Exercises,  , Where to go next, Introduction
 4455: @section Exercises
 4456: @cindex exercises
 4457: 
 4458: TODO: provide a set of programming excercises linked into the stuff done
 4459: already and into other sections of the manual. Provide solutions to all
 4460: the exercises in a .fs file in the distribution.
 4461: 
 4462: @c Get some inspiration from Starting Forth and Kelly&Spies.
 4463: 
 4464: @c excercises:
 4465: @c 1. take inches and convert to feet and inches.
 4466: @c 2. take temperature and convert from fahrenheight to celcius;
 4467: @c    may need to care about symmetric vs floored??
 4468: @c 3. take input line and do character substitution
 4469: @c    to encipher or decipher
 4470: @c 4. as above but work on a file for in and out
 4471: @c 5. take input line and convert to pig-latin 
 4472: @c
 4473: @c thing of sets of things to exercise then come up with
 4474: @c problems that need those things.
 4475: 
 4476: 
 4477: @c ******************************************************************
 4478: @node Words, Error messages, Introduction, Top
 4479: @chapter Forth Words
 4480: @cindex words
 4481: 
 4482: @menu
 4483: * Notation::                    
 4484: * Case insensitivity::          
 4485: * Comments::                    
 4486: * Boolean Flags::               
 4487: * Arithmetic::                  
 4488: * Stack Manipulation::          
 4489: * Memory::                      
 4490: * Control Structures::          
 4491: * Defining Words::              
 4492: * Interpretation and Compilation Semantics::  
 4493: * Tokens for Words::            
 4494: * Compiling words::             
 4495: * The Text Interpreter::        
 4496: * Word Lists::                  
 4497: * Environmental Queries::       
 4498: * Files::                       
 4499: * Blocks::                      
 4500: * Other I/O::                   
 4501: * Locals::                      
 4502: * Structures::                  
 4503: * Object-oriented Forth::       
 4504: * Programming Tools::           
 4505: * Assembler and Code Words::    
 4506: * Threading Words::             
 4507: * Passing Commands to the OS::  
 4508: * Keeping track of Time::       
 4509: * Miscellaneous Words::         
 4510: @end menu
 4511: 
 4512: @node Notation, Case insensitivity, Words, Words
 4513: @section Notation
 4514: @cindex notation of glossary entries
 4515: @cindex format of glossary entries
 4516: @cindex glossary notation format
 4517: @cindex word glossary entry format
 4518: 
 4519: The Forth words are described in this section in the glossary notation
 4520: that has become a de-facto standard for Forth texts:
 4521: 
 4522: @format
 4523: @i{word}     @i{Stack effect}   @i{wordset}   @i{pronunciation}
 4524: @end format
 4525: @i{Description}
 4526: 
 4527: @table @var
 4528: @item word
 4529: The name of the word.
 4530: 
 4531: @item Stack effect
 4532: @cindex stack effect
 4533: The stack effect is written in the notation @code{@i{before} --
 4534: @i{after}}, where @i{before} and @i{after} describe the top of
 4535: stack entries before and after the execution of the word. The rest of
 4536: the stack is not touched by the word. The top of stack is rightmost,
 4537: i.e., a stack sequence is written as it is typed in. Note that Gforth
 4538: uses a separate floating point stack, but a unified stack
 4539: notation. Also, return stack effects are not shown in @i{stack
 4540: effect}, but in @i{Description}. The name of a stack item describes
 4541: the type and/or the function of the item. See below for a discussion of
 4542: the types.
 4543: 
 4544: All words have two stack effects: A compile-time stack effect and a
 4545: run-time stack effect. The compile-time stack-effect of most words is
 4546: @i{ -- }. If the compile-time stack-effect of a word deviates from
 4547: this standard behaviour, or the word does other unusual things at
 4548: compile time, both stack effects are shown; otherwise only the run-time
 4549: stack effect is shown.
 4550: 
 4551: @cindex pronounciation of words
 4552: @item pronunciation
 4553: How the word is pronounced.
 4554: 
 4555: @cindex wordset
 4556: @cindex environment wordset
 4557: @item wordset
 4558: The ANS Forth standard is divided into several word sets. A standard
 4559: system need not support all of them. Therefore, in theory, the fewer
 4560: word sets your program uses the more portable it will be. However, we
 4561: suspect that most ANS Forth systems on personal machines will feature
 4562: all word sets. Words that are not defined in ANS Forth have
 4563: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
 4564: describes words that will work in future releases of Gforth;
 4565: @code{gforth-internal} words are more volatile. Environmental query
 4566: strings are also displayed like words; you can recognize them by the
 4567: @code{environment} in the word set field.
 4568: 
 4569: @item Description
 4570: A description of the behaviour of the word.
 4571: @end table
 4572: 
 4573: @cindex types of stack items
 4574: @cindex stack item types
 4575: The type of a stack item is specified by the character(s) the name
 4576: starts with:
 4577: 
 4578: @table @code
 4579: @item f
 4580: @cindex @code{f}, stack item type
 4581: Boolean flags, i.e. @code{false} or @code{true}.
 4582: @item c
 4583: @cindex @code{c}, stack item type
 4584: Char
 4585: @item w
 4586: @cindex @code{w}, stack item type
 4587: Cell, can contain an integer or an address
 4588: @item n
 4589: @cindex @code{n}, stack item type
 4590: signed integer
 4591: @item u
 4592: @cindex @code{u}, stack item type
 4593: unsigned integer
 4594: @item d
 4595: @cindex @code{d}, stack item type
 4596: double sized signed integer
 4597: @item ud
 4598: @cindex @code{ud}, stack item type
 4599: double sized unsigned integer
 4600: @item r
 4601: @cindex @code{r}, stack item type
 4602: Float (on the FP stack)
 4603: @item a-
 4604: @cindex @code{a_}, stack item type
 4605: Cell-aligned address
 4606: @item c-
 4607: @cindex @code{c_}, stack item type
 4608: Char-aligned address (note that a Char may have two bytes in Windows NT)
 4609: @item f-
 4610: @cindex @code{f_}, stack item type
 4611: Float-aligned address
 4612: @item df-
 4613: @cindex @code{df_}, stack item type
 4614: Address aligned for IEEE double precision float
 4615: @item sf-
 4616: @cindex @code{sf_}, stack item type
 4617: Address aligned for IEEE single precision float
 4618: @item xt
 4619: @cindex @code{xt}, stack item type
 4620: Execution token, same size as Cell
 4621: @item wid
 4622: @cindex @code{wid}, stack item type
 4623: Word list ID, same size as Cell
 4624: @item ior, wior
 4625: @cindex ior type description
 4626: @cindex wior type description
 4627: I/O result code, cell-sized.  In Gforth, you can @code{throw} iors.
 4628: @item f83name
 4629: @cindex @code{f83name}, stack item type
 4630: Pointer to a name structure
 4631: @item "
 4632: @cindex @code{"}, stack item type
 4633: string in the input stream (not on the stack). The terminating character
 4634: is a blank by default. If it is not a blank, it is shown in @code{<>}
 4635: quotes.
 4636: @end table
 4637: 
 4638: @comment ----------------------------------------------
 4639: @node Case insensitivity, Comments, Notation, Words
 4640: @section Case insensitivity
 4641: @cindex case sensitivity
 4642: @cindex upper and lower case
 4643: 
 4644: Gforth is case-insensitive; you can enter definitions and invoke
 4645: Standard words using upper, lower or mixed case (however,
 4646: @pxref{core-idef, Implementation-defined options, Implementation-defined
 4647: options}).
 4648: 
 4649: ANS Forth only @i{requires} implementations to recognise Standard words
 4650: when they are typed entirely in upper case. Therefore, a Standard
 4651: program must use upper case for all Standard words. You can use whatever
 4652: case you like for words that you define, but in a Standard program you
 4653: have to use the words in the same case that you defined them.
 4654: 
 4655: Gforth supports case sensitivity through @code{table}s (case-sensitive
 4656: wordlists, @pxref{Word Lists}).
 4657: 
 4658: Two people have asked how to convert Gforth to be case-sensitive; while
 4659: we think this is a bad idea, you can change all wordlists into tables
 4660: like this:
 4661: 
 4662: @example
 4663: ' table-find forth-wordlist wordlist-map @ !
 4664: @end example
 4665: 
 4666: Note that you now have to type the predefined words in the same case
 4667: that we defined them, which are varying.  You may want to convert them
 4668: to your favourite case before doing this operation (I won't explain how,
 4669: because if you are even contemplating doing this, you'd better have
 4670: enough knowledge of Forth systems to know this already).
 4671: 
 4672: @node Comments, Boolean Flags, Case insensitivity, Words
 4673: @section Comments
 4674: @cindex comments
 4675: 
 4676: Forth supports two styles of comment; the traditional @i{in-line} comment,
 4677: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
 4678: 
 4679: 
 4680: doc-(
 4681: doc-\
 4682: doc-\G
 4683: 
 4684: 
 4685: @node Boolean Flags, Arithmetic, Comments, Words
 4686: @section Boolean Flags
 4687: @cindex Boolean flags
 4688: 
 4689: A Boolean flag is cell-sized. A cell with all bits clear represents the
 4690: flag @code{false} and a flag with all bits set represents the flag
 4691: @code{true}. Words that check a flag (for example, @code{IF}) will treat
 4692: a cell that has @i{any} bit set as @code{true}.
 4693: @c on and off to Memory? 
 4694: @c true and false to "Bitwise operations" or "Numeric comparison"?
 4695: 
 4696: doc-true
 4697: doc-false
 4698: doc-on
 4699: doc-off
 4700: 
 4701: 
 4702: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
 4703: @section Arithmetic
 4704: @cindex arithmetic words
 4705: 
 4706: @cindex division with potentially negative operands
 4707: Forth arithmetic is not checked, i.e., you will not hear about integer
 4708: overflow on addition or multiplication, you may hear about division by
 4709: zero if you are lucky. The operator is written after the operands, but
 4710: the operands are still in the original order. I.e., the infix @code{2-1}
 4711: corresponds to @code{2 1 -}. Forth offers a variety of division
 4712: operators. If you perform division with potentially negative operands,
 4713: you do not want to use @code{/} or @code{/mod} with its undefined
 4714: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
 4715: former, @pxref{Mixed precision}).
 4716: @comment TODO discuss the different division forms and the std approach
 4717: 
 4718: @menu
 4719: * Single precision::            
 4720: * Double precision::            Double-cell integer arithmetic
 4721: * Bitwise operations::          
 4722: * Numeric comparison::          
 4723: * Mixed precision::             Operations with single and double-cell integers
 4724: * Floating Point::              
 4725: @end menu
 4726: 
 4727: @node Single precision, Double precision, Arithmetic, Arithmetic
 4728: @subsection Single precision
 4729: @cindex single precision arithmetic words
 4730: 
 4731: @c !! cell undefined
 4732: 
 4733: By default, numbers in Forth are single-precision integers that are one
 4734: cell in size. They can be signed or unsigned, depending upon how you
 4735: treat them. For the rules used by the text interpreter for recognising
 4736: single-precision integers see @ref{Number Conversion}.
 4737: 
 4738: These words are all defined for signed operands, but some of them also
 4739: work for unsigned numbers: @code{+}, @code{1+}, @code{-}, @code{1-},
 4740: @code{*}.
 4741: 
 4742: doc-+
 4743: doc-1+
 4744: doc--
 4745: doc-1-
 4746: doc-*
 4747: doc-/
 4748: doc-mod
 4749: doc-/mod
 4750: doc-negate
 4751: doc-abs
 4752: doc-min
 4753: doc-max
 4754: doc-floored
 4755: 
 4756: 
 4757: @node Double precision, Bitwise operations, Single precision, Arithmetic
 4758: @subsection Double precision
 4759: @cindex double precision arithmetic words
 4760: 
 4761: For the rules used by the text interpreter for
 4762: recognising double-precision integers, see @ref{Number Conversion}.
 4763: 
 4764: A double precision number is represented by a cell pair, with the most
 4765: significant cell at the TOS. It is trivial to convert an unsigned single
 4766: to a double: simply push a @code{0} onto the TOS. Since numbers are
 4767: represented by Gforth using 2's complement arithmetic, converting a
 4768: signed single to a (signed) double requires sign-extension across the
 4769: most significant cell. This can be achieved using @code{s>d}. The moral
 4770: of the story is that you cannot convert a number without knowing whether
 4771: it represents an unsigned or a signed number.
 4772: 
 4773: These words are all defined for signed operands, but some of them also
 4774: work for unsigned numbers: @code{d+}, @code{d-}.
 4775: 
 4776: doc-s>d
 4777: doc-d>s
 4778: doc-d+
 4779: doc-d-
 4780: doc-dnegate
 4781: doc-dabs
 4782: doc-dmin
 4783: doc-dmax
 4784: 
 4785: 
 4786: @node Bitwise operations, Numeric comparison, Double precision, Arithmetic
 4787: @subsection Bitwise operations
 4788: @cindex bitwise operation words
 4789: 
 4790: 
 4791: doc-and
 4792: doc-or
 4793: doc-xor
 4794: doc-invert
 4795: doc-lshift
 4796: doc-rshift
 4797: doc-2*
 4798: doc-d2*
 4799: doc-2/
 4800: doc-d2/
 4801: 
 4802: 
 4803: @node Numeric comparison, Mixed precision, Bitwise operations, Arithmetic
 4804: @subsection Numeric comparison
 4805: @cindex numeric comparison words
 4806: 
 4807: Note that the words that compare for equality (@code{= <> 0= 0<> d= d<>
 4808: d0= d0<>}) work for for both signed and unsigned numbers.
 4809: 
 4810: doc-<
 4811: doc-<=
 4812: doc-<>
 4813: doc-=
 4814: doc->
 4815: doc->=
 4816: 
 4817: doc-0<
 4818: doc-0<=
 4819: doc-0<>
 4820: doc-0=
 4821: doc-0>
 4822: doc-0>=
 4823: 
 4824: doc-u<
 4825: doc-u<=
 4826: @c u<> and u= exist but are the same as <> and =
 4827: @c doc-u<>
 4828: @c doc-u=
 4829: doc-u>
 4830: doc-u>=
 4831: 
 4832: doc-within
 4833: 
 4834: doc-d<
 4835: doc-d<=
 4836: doc-d<>
 4837: doc-d=
 4838: doc-d>
 4839: doc-d>=
 4840: 
 4841: doc-d0<
 4842: doc-d0<=
 4843: doc-d0<>
 4844: doc-d0=
 4845: doc-d0>
 4846: doc-d0>=
 4847: 
 4848: doc-du<
 4849: doc-du<=
 4850: @c du<> and du= exist but are the same as d<> and d=
 4851: @c doc-du<>
 4852: @c doc-du=
 4853: doc-du>
 4854: doc-du>=
 4855: 
 4856: 
 4857: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
 4858: @subsection Mixed precision
 4859: @cindex mixed precision arithmetic words
 4860: 
 4861: 
 4862: doc-m+
 4863: doc-*/
 4864: doc-*/mod
 4865: doc-m*
 4866: doc-um*
 4867: doc-m*/
 4868: doc-um/mod
 4869: doc-fm/mod
 4870: doc-sm/rem
 4871: 
 4872: 
 4873: @node Floating Point,  , Mixed precision, Arithmetic
 4874: @subsection Floating Point
 4875: @cindex floating point arithmetic words
 4876: 
 4877: For the rules used by the text interpreter for
 4878: recognising floating-point numbers see @ref{Number Conversion}.
 4879: 
 4880: Gforth has a separate floating point stack, but the documentation uses
 4881: the unified notation.@footnote{It's easy to generate the separate
 4882: notation from that by just separating the floating-point numbers out:
 4883: e.g. @code{( n r1 u r2 -- r3 )} becomes @code{( n u -- ) ( F: r1 r2 --
 4884: r3 )}.}
 4885: 
 4886: @cindex floating-point arithmetic, pitfalls
 4887: Floating point numbers have a number of unpleasant surprises for the
 4888: unwary (e.g., floating point addition is not associative) and even a few
 4889: for the wary. You should not use them unless you know what you are doing
 4890: or you don't care that the results you get are totally bogus. If you
 4891: want to learn about the problems of floating point numbers (and how to
 4892: avoid them), you might start with @cite{David Goldberg,
 4893: @uref{http://www.validgh.com/goldberg/paper.ps,What Every Computer
 4894: Scientist Should Know About Floating-Point Arithmetic}, ACM Computing
 4895: Surveys 23(1):5@minus{}48, March 1991}.
 4896: 
 4897: 
 4898: doc-d>f
 4899: doc-f>d
 4900: doc-f+
 4901: doc-f-
 4902: doc-f*
 4903: doc-f/
 4904: doc-fnegate
 4905: doc-fabs
 4906: doc-fmax
 4907: doc-fmin
 4908: doc-floor
 4909: doc-fround
 4910: doc-f**
 4911: doc-fsqrt
 4912: doc-fexp
 4913: doc-fexpm1
 4914: doc-fln
 4915: doc-flnp1
 4916: doc-flog
 4917: doc-falog
 4918: doc-f2*
 4919: doc-f2/
 4920: doc-1/f
 4921: doc-precision
 4922: doc-set-precision
 4923: 
 4924: @cindex angles in trigonometric operations
 4925: @cindex trigonometric operations
 4926: Angles in floating point operations are given in radians (a full circle
 4927: has 2 pi radians).
 4928: 
 4929: doc-fsin
 4930: doc-fcos
 4931: doc-fsincos
 4932: doc-ftan
 4933: doc-fasin
 4934: doc-facos
 4935: doc-fatan
 4936: doc-fatan2
 4937: doc-fsinh
 4938: doc-fcosh
 4939: doc-ftanh
 4940: doc-fasinh
 4941: doc-facosh
 4942: doc-fatanh
 4943: doc-pi
 4944: 
 4945: @cindex equality of floats
 4946: @cindex floating-point comparisons
 4947: One particular problem with floating-point arithmetic is that comparison
 4948: for equality often fails when you would expect it to succeed.  For this
 4949: reason approximate equality is often preferred (but you still have to
 4950: know what you are doing).  Also note that IEEE NaNs may compare
 4951: differently from what you might expect.  The comparison words are:
 4952: 
 4953: doc-f~rel
 4954: doc-f~abs
 4955: doc-f~
 4956: doc-f=
 4957: doc-f<>
 4958: 
 4959: doc-f<
 4960: doc-f<=
 4961: doc-f>
 4962: doc-f>=
 4963: 
 4964: doc-f0<
 4965: doc-f0<=
 4966: doc-f0<>
 4967: doc-f0=
 4968: doc-f0>
 4969: doc-f0>=
 4970: 
 4971: 
 4972: @node Stack Manipulation, Memory, Arithmetic, Words
 4973: @section Stack Manipulation
 4974: @cindex stack manipulation words
 4975: 
 4976: @cindex floating-point stack in the standard
 4977: Gforth maintains a number of separate stacks:
 4978: 
 4979: @cindex data stack
 4980: @cindex parameter stack
 4981: @itemize @bullet
 4982: @item
 4983: A data stack (also known as the @dfn{parameter stack}) -- for
 4984: characters, cells, addresses, and double cells.
 4985: 
 4986: @cindex floating-point stack
 4987: @item
 4988: A floating point stack -- for holding floating point (FP) numbers.
 4989: 
 4990: @cindex return stack
 4991: @item
 4992: A return stack -- for holding the return addresses of colon
 4993: definitions and other (non-FP) data.
 4994: 
 4995: @cindex locals stack
 4996: @item
 4997: A locals stack -- for holding local variables.
 4998: @end itemize
 4999: 
 5000: @menu
 5001: * Data stack::                  
 5002: * Floating point stack::        
 5003: * Return stack::                
 5004: * Locals stack::                
 5005: * Stack pointer manipulation::  
 5006: @end menu
 5007: 
 5008: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
 5009: @subsection Data stack
 5010: @cindex data stack manipulation words
 5011: @cindex stack manipulations words, data stack
 5012: 
 5013: 
 5014: doc-drop
 5015: doc-nip
 5016: doc-dup
 5017: doc-over
 5018: doc-tuck
 5019: doc-swap
 5020: doc-pick
 5021: doc-rot
 5022: doc--rot
 5023: doc-?dup
 5024: doc-roll
 5025: doc-2drop
 5026: doc-2nip
 5027: doc-2dup
 5028: doc-2over
 5029: doc-2tuck
 5030: doc-2swap
 5031: doc-2rot
 5032: 
 5033: 
 5034: @node Floating point stack, Return stack, Data stack, Stack Manipulation
 5035: @subsection Floating point stack
 5036: @cindex floating-point stack manipulation words
 5037: @cindex stack manipulation words, floating-point stack
 5038: 
 5039: Whilst every sane Forth has a separate floating-point stack, it is not
 5040: strictly required; an ANS Forth system could theoretically keep
 5041: floating-point numbers on the data stack. As an additional difficulty,
 5042: you don't know how many cells a floating-point number takes. It is
 5043: reportedly possible to write words in a way that they work also for a
 5044: unified stack model, but we do not recommend trying it. Instead, just
 5045: say that your program has an environmental dependency on a separate
 5046: floating-point stack.
 5047: 
 5048: doc-floating-stack
 5049: 
 5050: doc-fdrop
 5051: doc-fnip
 5052: doc-fdup
 5053: doc-fover
 5054: doc-ftuck
 5055: doc-fswap
 5056: doc-fpick
 5057: doc-frot
 5058: 
 5059: 
 5060: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
 5061: @subsection Return stack
 5062: @cindex return stack manipulation words
 5063: @cindex stack manipulation words, return stack
 5064: 
 5065: @cindex return stack and locals
 5066: @cindex locals and return stack
 5067: A Forth system is allowed to keep local variables on the
 5068: return stack. This is reasonable, as local variables usually eliminate
 5069: the need to use the return stack explicitly. So, if you want to produce
 5070: a standard compliant program and you are using local variables in a
 5071: word, forget about return stack manipulations in that word (refer to the
 5072: standard document for the exact rules).
 5073: 
 5074: doc->r
 5075: doc-r>
 5076: doc-r@
 5077: doc-rdrop
 5078: doc-2>r
 5079: doc-2r>
 5080: doc-2r@
 5081: doc-2rdrop
 5082: 
 5083: 
 5084: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
 5085: @subsection Locals stack
 5086: 
 5087: Gforth uses an extra locals stack.  It is described, along with the
 5088: reasons for its existence, in @ref{Locals implementation}.
 5089: 
 5090: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
 5091: @subsection Stack pointer manipulation
 5092: @cindex stack pointer manipulation words
 5093: 
 5094: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
 5095: doc-sp0
 5096: doc-sp@
 5097: doc-sp!
 5098: doc-fp0
 5099: doc-fp@
 5100: doc-fp!
 5101: doc-rp0
 5102: doc-rp@
 5103: doc-rp!
 5104: doc-lp0
 5105: doc-lp@
 5106: doc-lp!
 5107: 
 5108: 
 5109: @node Memory, Control Structures, Stack Manipulation, Words
 5110: @section Memory
 5111: @cindex memory words
 5112: 
 5113: @menu
 5114: * Memory model::                
 5115: * Dictionary allocation::       
 5116: * Heap Allocation::             
 5117: * Memory Access::               
 5118: * Address arithmetic::          
 5119: * Memory Blocks::               
 5120: @end menu
 5121: 
 5122: In addition to the standard Forth memory allocation words, there is also
 5123: a @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
 5124: garbage collector}.
 5125: 
 5126: @node Memory model, Dictionary allocation, Memory, Memory
 5127: @subsection ANS Forth and Gforth memory models
 5128: 
 5129: @c The ANS Forth description is a mess (e.g., is the heap part of
 5130: @c the dictionary?), so let's not stick to closely with it.
 5131: 
 5132: ANS Forth considers a Forth system as consisting of several address
 5133: spaces, of which only @dfn{data space} is managed and accessible with
 5134: the memory words.  Memory not necessarily in data space includes the
 5135: stacks, the code (called code space) and the headers (called name
 5136: space). In Gforth everything is in data space, but the code for the
 5137: primitives is usually read-only.
 5138: 
 5139: Data space is divided into a number of areas: The (data space portion of
 5140: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
 5141: refer to the search data structure embodied in word lists and headers,
 5142: because it is used for looking up names, just as you would in a
 5143: conventional dictionary.}, the heap, and a number of system-allocated
 5144: buffers.
 5145: 
 5146: @cindex address arithmetic restrictions, ANS vs. Gforth
 5147: @cindex contiguous regions, ANS vs. Gforth
 5148: In ANS Forth data space is also divided into contiguous regions.  You
 5149: can only use address arithmetic within a contiguous region, not between
 5150: them.  Usually each allocation gives you one contiguous region, but the
 5151: dictionary allocation words have additional rules (@pxref{Dictionary
 5152: allocation}).
 5153: 
 5154: Gforth provides one big address space, and address arithmetic can be
 5155: performed between any addresses. However, in the dictionary headers or
 5156: code are interleaved with data, so almost the only contiguous data space
 5157: regions there are those described by ANS Forth as contiguous; but you
 5158: can be sure that the dictionary is allocated towards increasing
 5159: addresses even between contiguous regions.  The memory order of
 5160: allocations in the heap is platform-dependent (and possibly different
 5161: from one run to the next).
 5162: 
 5163: 
 5164: @node Dictionary allocation, Heap Allocation, Memory model, Memory
 5165: @subsection Dictionary allocation
 5166: @cindex reserving data space
 5167: @cindex data space - reserving some
 5168: 
 5169: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
 5170: you want to deallocate X, you also deallocate everything
 5171: allocated after X.
 5172: 
 5173: @cindex contiguous regions in dictionary allocation
 5174: The allocations using the words below are contiguous and grow the region
 5175: towards increasing addresses.  Other words that allocate dictionary
 5176: memory of any kind (i.e., defining words including @code{:noname}) end
 5177: the contiguous region and start a new one.
 5178: 
 5179: In ANS Forth only @code{create}d words are guaranteed to produce an
 5180: address that is the start of the following contiguous region.  In
 5181: particular, the cell allocated by @code{variable} is not guaranteed to
 5182: be contiguous with following @code{allot}ed memory.
 5183: 
 5184: You can deallocate memory by using @code{allot} with a negative argument
 5185: (with some restrictions, see @code{allot}). For larger deallocations use
 5186: @code{marker}.
 5187: 
 5188: 
 5189: doc-here
 5190: doc-unused
 5191: doc-allot
 5192: doc-c,
 5193: doc-f,
 5194: doc-,
 5195: doc-2,
 5196: 
 5197: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
 5198: course you should allocate memory in an aligned way, too. I.e., before
 5199: allocating allocating a cell, @code{here} must be cell-aligned, etc.
 5200: The words below align @code{here} if it is not already.  Basically it is
 5201: only already aligned for a type, if the last allocation was a multiple
 5202: of the size of this type and if @code{here} was aligned for this type
 5203: before.
 5204: 
 5205: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
 5206: ANS Forth (@code{maxalign}ed in Gforth).
 5207: 
 5208: doc-align
 5209: doc-falign
 5210: doc-sfalign
 5211: doc-dfalign
 5212: doc-maxalign
 5213: doc-cfalign
 5214: 
 5215: 
 5216: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
 5217: @subsection Heap allocation
 5218: @cindex heap allocation
 5219: @cindex dynamic allocation of memory
 5220: @cindex memory-allocation word set
 5221: 
 5222: @cindex contiguous regions and heap allocation
 5223: Heap allocation supports deallocation of allocated memory in any
 5224: order. Dictionary allocation is not affected by it (i.e., it does not
 5225: end a contiguous region). In Gforth, these words are implemented using
 5226: the standard C library calls malloc(), free() and resize().
 5227: 
 5228: The memory region produced by one invocation of @code{allocate} or
 5229: @code{resize} is internally contiguous.  There is no contiguity between
 5230: such a region and any other region (including others allocated from the
 5231: heap).
 5232: 
 5233: doc-allocate
 5234: doc-free
 5235: doc-resize
 5236: 
 5237: 
 5238: @node Memory Access, Address arithmetic, Heap Allocation, Memory
 5239: @subsection Memory Access
 5240: @cindex memory access words
 5241: 
 5242: doc-@
 5243: doc-!
 5244: doc-+!
 5245: doc-c@
 5246: doc-c!
 5247: doc-2@
 5248: doc-2!
 5249: doc-f@
 5250: doc-f!
 5251: doc-sf@
 5252: doc-sf!
 5253: doc-df@
 5254: doc-df!
 5255: 
 5256: 
 5257: @node Address arithmetic, Memory Blocks, Memory Access, Memory
 5258: @subsection Address arithmetic
 5259: @cindex address arithmetic words
 5260: 
 5261: Address arithmetic is the foundation on which you can build data
 5262: structures like arrays, records (@pxref{Structures}) and objects
 5263: (@pxref{Object-oriented Forth}).
 5264: 
 5265: @cindex address unit
 5266: @cindex au (address unit)
 5267: ANS Forth does not specify the sizes of the data types. Instead, it
 5268: offers a number of words for computing sizes and doing address
 5269: arithmetic. Address arithmetic is performed in terms of address units
 5270: (aus); on most systems the address unit is one byte. Note that a
 5271: character may have more than one au, so @code{chars} is no noop (on
 5272: platforms where it is a noop, it compiles to nothing).
 5273: 
 5274: The basic address arithmetic words are @code{+} and @code{-}.  E.g., if
 5275: you have the address of a cell, perform @code{1 cells +}, and you will
 5276: have the address of the next cell.
 5277: 
 5278: @cindex contiguous regions and address arithmetic
 5279: In ANS Forth you can perform address arithmetic only within a contiguous
 5280: region, i.e., if you have an address into one region, you can only add
 5281: and subtract such that the result is still within the region; you can
 5282: only subtract or compare addresses from within the same contiguous
 5283: region.  Reasons: several contiguous regions can be arranged in memory
 5284: in any way; on segmented systems addresses may have unusual
 5285: representations, such that address arithmetic only works within a
 5286: region.  Gforth provides a few more guarantees (linear address space,
 5287: dictionary grows upwards), but in general I have found it easy to stay
 5288: within contiguous regions (exception: computing and comparing to the
 5289: address just beyond the end of an array).
 5290: 
 5291: @cindex alignment of addresses for types
 5292: ANS Forth also defines words for aligning addresses for specific
 5293: types. Many computers require that accesses to specific data types
 5294: must only occur at specific addresses; e.g., that cells may only be
 5295: accessed at addresses divisible by 4. Even if a machine allows unaligned
 5296: accesses, it can usually perform aligned accesses faster. 
 5297: 
 5298: For the performance-conscious: alignment operations are usually only
 5299: necessary during the definition of a data structure, not during the
 5300: (more frequent) accesses to it.
 5301: 
 5302: ANS Forth defines no words for character-aligning addresses. This is not
 5303: an oversight, but reflects the fact that addresses that are not
 5304: char-aligned have no use in the standard and therefore will not be
 5305: created.
 5306: 
 5307: @cindex @code{CREATE} and alignment
 5308: ANS Forth guarantees that addresses returned by @code{CREATE}d words
 5309: are cell-aligned; in addition, Gforth guarantees that these addresses
 5310: are aligned for all purposes.
 5311: 
 5312: Note that the ANS Forth word @code{char} has nothing to do with address
 5313: arithmetic.
 5314: 
 5315: 
 5316: doc-chars
 5317: doc-char+
 5318: doc-cells
 5319: doc-cell+
 5320: doc-cell
 5321: doc-aligned
 5322: doc-floats
 5323: doc-float+
 5324: doc-float
 5325: doc-faligned
 5326: doc-sfloats
 5327: doc-sfloat+
 5328: doc-sfaligned
 5329: doc-dfloats
 5330: doc-dfloat+
 5331: doc-dfaligned
 5332: doc-maxaligned
 5333: doc-cfaligned
 5334: doc-address-unit-bits
 5335: 
 5336: 
 5337: @node Memory Blocks,  , Address arithmetic, Memory
 5338: @subsection Memory Blocks
 5339: @cindex memory block words
 5340: @cindex character strings - moving and copying
 5341: 
 5342: Memory blocks often represent character strings; For ways of storing
 5343: character strings in memory see @ref{String Formats}.  For other
 5344: string-processing words see @ref{Displaying characters and strings}.
 5345: 
 5346: A few of these words work on address unit blocks.  In that case, you
 5347: usually have to insert @code{CHARS} before the word when working on
 5348: character strings.  Most words work on character blocks, and expect a
 5349: char-aligned address.
 5350: 
 5351: When copying characters between overlapping memory regions, use
 5352: @code{chars move} or choose carefully between @code{cmove} and
 5353: @code{cmove>}.
 5354: 
 5355: doc-move
 5356: doc-erase
 5357: doc-cmove
 5358: doc-cmove>
 5359: doc-fill
 5360: doc-blank
 5361: doc-compare
 5362: doc-search
 5363: doc--trailing
 5364: doc-/string
 5365: doc-bounds
 5366: 
 5367: @comment TODO examples
 5368: 
 5369: 
 5370: @node Control Structures, Defining Words, Memory, Words
 5371: @section Control Structures
 5372: @cindex control structures
 5373: 
 5374: Control structures in Forth cannot be used interpretively, only in a
 5375: colon definition@footnote{To be precise, they have no interpretation
 5376: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
 5377: not like this limitation, but have not seen a satisfying way around it
 5378: yet, although many schemes have been proposed.
 5379: 
 5380: @menu
 5381: * Selection::                   IF ... ELSE ... ENDIF
 5382: * Simple Loops::                BEGIN ...
 5383: * Counted Loops::               DO
 5384: * Arbitrary control structures::  
 5385: * Calls and returns::           
 5386: * Exception Handling::          
 5387: @end menu
 5388: 
 5389: @node Selection, Simple Loops, Control Structures, Control Structures
 5390: @subsection Selection
 5391: @cindex selection control structures
 5392: @cindex control structures for selection
 5393: 
 5394: @cindex @code{IF} control structure
 5395: @example
 5396: @i{flag}
 5397: IF
 5398:   @i{code}
 5399: ENDIF
 5400: @end example
 5401: @noindent
 5402: 
 5403: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
 5404: with any bit set represents truth) @i{code} is executed.
 5405: 
 5406: @example
 5407: @i{flag}
 5408: IF
 5409:   @i{code1}
 5410: ELSE
 5411:   @i{code2}
 5412: ENDIF
 5413: @end example
 5414: 
 5415: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
 5416: executed.
 5417: 
 5418: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
 5419: standard, and @code{ENDIF} is not, although it is quite popular. We
 5420: recommend using @code{ENDIF}, because it is less confusing for people
 5421: who also know other languages (and is not prone to reinforcing negative
 5422: prejudices against Forth in these people). Adding @code{ENDIF} to a
 5423: system that only supplies @code{THEN} is simple:
 5424: @example
 5425: : ENDIF   POSTPONE then ; immediate
 5426: @end example
 5427: 
 5428: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
 5429: (adv.)}  has the following meanings:
 5430: @quotation
 5431: ... 2b: following next after in order ... 3d: as a necessary consequence
 5432: (if you were there, then you saw them).
 5433: @end quotation
 5434: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
 5435: and many other programming languages has the meaning 3d.]
 5436: 
 5437: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
 5438: you can avoid using @code{?dup}. Using these alternatives is also more
 5439: efficient than using @code{?dup}. Definitions in ANS Forth
 5440: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
 5441: @file{compat/control.fs}.
 5442: 
 5443: @cindex @code{CASE} control structure
 5444: @example
 5445: @i{n}
 5446: CASE
 5447:   @i{n1} OF @i{code1} ENDOF
 5448:   @i{n2} OF @i{code2} ENDOF
 5449:   @dots{}
 5450:   ( n ) @i{default-code} ( n )
 5451: ENDCASE
 5452: @end example
 5453: 
 5454: Executes the first @i{codei}, where the @i{ni} is equal to @i{n}.  If no
 5455: @i{ni} matches, the optional @i{default-code} is executed. The optional
 5456: default case can be added by simply writing the code after the last
 5457: @code{ENDOF}. It may use @i{n}, which is on top of the stack, but must
 5458: not consume it.
 5459: 
 5460: @progstyle
 5461: To keep the code understandable, you should ensure that on all paths
 5462: through a selection construct the stack is changed in the same way
 5463: (wrt. number and types of stack items consumed and pushed).
 5464: 
 5465: @node Simple Loops, Counted Loops, Selection, Control Structures
 5466: @subsection Simple Loops
 5467: @cindex simple loops
 5468: @cindex loops without count 
 5469: 
 5470: @cindex @code{WHILE} loop
 5471: @example
 5472: BEGIN
 5473:   @i{code1}
 5474:   @i{flag}
 5475: WHILE
 5476:   @i{code2}
 5477: REPEAT
 5478: @end example
 5479: 
 5480: @i{code1} is executed and @i{flag} is computed. If it is true,
 5481: @i{code2} is executed and the loop is restarted; If @i{flag} is
 5482: false, execution continues after the @code{REPEAT}.
 5483: 
 5484: @cindex @code{UNTIL} loop
 5485: @example
 5486: BEGIN
 5487:   @i{code}
 5488:   @i{flag}
 5489: UNTIL
 5490: @end example
 5491: 
 5492: @i{code} is executed. The loop is restarted if @code{flag} is false.
 5493: 
 5494: @progstyle
 5495: To keep the code understandable, a complete iteration of the loop should
 5496: not change the number and types of the items on the stacks.
 5497: 
 5498: @cindex endless loop
 5499: @cindex loops, endless
 5500: @example
 5501: BEGIN
 5502:   @i{code}
 5503: AGAIN
 5504: @end example
 5505: 
 5506: This is an endless loop.
 5507: 
 5508: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
 5509: @subsection Counted Loops
 5510: @cindex counted loops
 5511: @cindex loops, counted
 5512: @cindex @code{DO} loops
 5513: 
 5514: The basic counted loop is:
 5515: @example
 5516: @i{limit} @i{start}
 5517: ?DO
 5518:   @i{body}
 5519: LOOP
 5520: @end example
 5521: 
 5522: This performs one iteration for every integer, starting from @i{start}
 5523: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
 5524: accessed with @code{i}. For example, the loop:
 5525: @example
 5526: 10 0 ?DO
 5527:   i .
 5528: LOOP
 5529: @end example
 5530: @noindent
 5531: prints @code{0 1 2 3 4 5 6 7 8 9}
 5532: 
 5533: The index of the innermost loop can be accessed with @code{i}, the index
 5534: of the next loop with @code{j}, and the index of the third loop with
 5535: @code{k}.
 5536: 
 5537: 
 5538: doc-i
 5539: doc-j
 5540: doc-k
 5541: 
 5542: 
 5543: The loop control data are kept on the return stack, so there are some
 5544: restrictions on mixing return stack accesses and counted loop words. In
 5545: particuler, if you put values on the return stack outside the loop, you
 5546: cannot read them inside the loop@footnote{well, not in a way that is
 5547: portable.}. If you put values on the return stack within a loop, you
 5548: have to remove them before the end of the loop and before accessing the
 5549: index of the loop.
 5550: 
 5551: There are several variations on the counted loop:
 5552: 
 5553: @itemize @bullet
 5554: @item
 5555: @code{LEAVE} leaves the innermost counted loop immediately; execution
 5556: continues after the associated @code{LOOP} or @code{NEXT}. For example:
 5557: 
 5558: @example
 5559: 10 0 ?DO  i DUP . 3 = IF LEAVE THEN LOOP
 5560: @end example
 5561: prints @code{0 1 2 3}
 5562: 
 5563: 
 5564: @item
 5565: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
 5566: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
 5567: return stack so @code{EXIT} can get to its return address. For example:
 5568: 
 5569: @example
 5570: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
 5571: @end example
 5572: prints @code{0 1 2 3}
 5573: 
 5574: 
 5575: @item
 5576: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
 5577: (and @code{LOOP} iterates until they become equal by wrap-around
 5578: arithmetic). This behaviour is usually not what you want. Therefore,
 5579: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
 5580: @code{?DO}), which do not enter the loop if @i{start} is greater than
 5581: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
 5582: unsigned loop parameters.
 5583: 
 5584: @item
 5585: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
 5586: the loop, independent of the loop parameters. Do not use @code{DO}, even
 5587: if you know that the loop is entered in any case. Such knowledge tends
 5588: to become invalid during maintenance of a program, and then the
 5589: @code{DO} will make trouble.
 5590: 
 5591: @item
 5592: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
 5593: index by @i{n} instead of by 1. The loop is terminated when the border
 5594: between @i{limit-1} and @i{limit} is crossed. E.g.:
 5595: 
 5596: @example
 5597: 4 0 +DO  i .  2 +LOOP
 5598: @end example
 5599: @noindent
 5600: prints @code{0 2}
 5601: 
 5602: @example
 5603: 4 1 +DO  i .  2 +LOOP
 5604: @end example
 5605: @noindent
 5606: prints @code{1 3}
 5607: 
 5608: @item
 5609: @cindex negative increment for counted loops
 5610: @cindex counted loops with negative increment
 5611: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
 5612: 
 5613: @example
 5614: -1 0 ?DO  i .  -1 +LOOP
 5615: @end example
 5616: @noindent
 5617: prints @code{0 -1}
 5618: 
 5619: @example
 5620: 0 0 ?DO  i .  -1 +LOOP
 5621: @end example
 5622: prints nothing.
 5623: 
 5624: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
 5625: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
 5626: index by @i{u} each iteration. The loop is terminated when the border
 5627: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
 5628: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
 5629: 
 5630: @example
 5631: -2 0 -DO  i .  1 -LOOP
 5632: @end example
 5633: @noindent
 5634: prints @code{0 -1}
 5635: 
 5636: @example
 5637: -1 0 -DO  i .  1 -LOOP
 5638: @end example
 5639: @noindent
 5640: prints @code{0}
 5641: 
 5642: @example
 5643: 0 0 -DO  i .  1 -LOOP
 5644: @end example
 5645: @noindent
 5646: prints nothing.
 5647: 
 5648: @end itemize
 5649: 
 5650: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
 5651: @code{-LOOP} are not defined in ANS Forth. However, an implementation
 5652: for these words that uses only standard words is provided in
 5653: @file{compat/loops.fs}.
 5654: 
 5655: 
 5656: @cindex @code{FOR} loops
 5657: Another counted loop is:
 5658: @example
 5659: @i{n}
 5660: FOR
 5661:   @i{body}
 5662: NEXT
 5663: @end example
 5664: This is the preferred loop of native code compiler writers who are too
 5665: lazy to optimize @code{?DO} loops properly. This loop structure is not
 5666: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
 5667: @code{i} produces values starting with @i{n} and ending with 0. Other
 5668: Forth systems may behave differently, even if they support @code{FOR}
 5669: loops. To avoid problems, don't use @code{FOR} loops.
 5670: 
 5671: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
 5672: @subsection Arbitrary control structures
 5673: @cindex control structures, user-defined
 5674: 
 5675: @cindex control-flow stack
 5676: ANS Forth permits and supports using control structures in a non-nested
 5677: way. Information about incomplete control structures is stored on the
 5678: control-flow stack. This stack may be implemented on the Forth data
 5679: stack, and this is what we have done in Gforth.
 5680: 
 5681: @cindex @code{orig}, control-flow stack item
 5682: @cindex @code{dest}, control-flow stack item
 5683: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
 5684: entry represents a backward branch target. A few words are the basis for
 5685: building any control structure possible (except control structures that
 5686: need storage, like calls, coroutines, and backtracking).
 5687: 
 5688: 
 5689: doc-if
 5690: doc-ahead
 5691: doc-then
 5692: doc-begin
 5693: doc-until
 5694: doc-again
 5695: doc-cs-pick
 5696: doc-cs-roll
 5697: 
 5698: 
 5699: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
 5700: manipulate the control-flow stack in a portable way. Without them, you
 5701: would need to know how many stack items are occupied by a control-flow
 5702: entry (many systems use one cell. In Gforth they currently take three,
 5703: but this may change in the future).
 5704: 
 5705: Some standard control structure words are built from these words:
 5706: 
 5707: 
 5708: doc-else
 5709: doc-while
 5710: doc-repeat
 5711: 
 5712: 
 5713: @noindent
 5714: Gforth adds some more control-structure words:
 5715: 
 5716: 
 5717: doc-endif
 5718: doc-?dup-if
 5719: doc-?dup-0=-if
 5720: 
 5721: 
 5722: @noindent
 5723: Counted loop words constitute a separate group of words:
 5724: 
 5725: 
 5726: doc-?do
 5727: doc-+do
 5728: doc-u+do
 5729: doc--do
 5730: doc-u-do
 5731: doc-do
 5732: doc-for
 5733: doc-loop
 5734: doc-+loop
 5735: doc--loop
 5736: doc-next
 5737: doc-leave
 5738: doc-?leave
 5739: doc-unloop
 5740: doc-done
 5741: 
 5742: 
 5743: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
 5744: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
 5745: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
 5746: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
 5747: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
 5748: resolved (by using one of the loop-ending words or @code{DONE}).
 5749: 
 5750: @noindent
 5751: Another group of control structure words are:
 5752: 
 5753: 
 5754: doc-case
 5755: doc-endcase
 5756: doc-of
 5757: doc-endof
 5758: 
 5759: 
 5760: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
 5761: @code{CS-ROLL}.
 5762: 
 5763: @subsubsection Programming Style
 5764: @cindex control structures programming style
 5765: @cindex programming style, arbitrary control structures
 5766: 
 5767: In order to ensure readability we recommend that you do not create
 5768: arbitrary control structures directly, but define new control structure
 5769: words for the control structure you want and use these words in your
 5770: program. For example, instead of writing:
 5771: 
 5772: @example
 5773: BEGIN
 5774:   ...
 5775: IF [ 1 CS-ROLL ]
 5776:   ...
 5777: AGAIN THEN
 5778: @end example
 5779: 
 5780: @noindent
 5781: we recommend defining control structure words, e.g.,
 5782: 
 5783: @example
 5784: : WHILE ( DEST -- ORIG DEST )
 5785:  POSTPONE IF
 5786:  1 CS-ROLL ; immediate
 5787: 
 5788: : REPEAT ( orig dest -- )
 5789:  POSTPONE AGAIN
 5790:  POSTPONE THEN ; immediate
 5791: @end example
 5792: 
 5793: @noindent
 5794: and then using these to create the control structure:
 5795: 
 5796: @example
 5797: BEGIN
 5798:   ...
 5799: WHILE
 5800:   ...
 5801: REPEAT
 5802: @end example
 5803: 
 5804: That's much easier to read, isn't it? Of course, @code{REPEAT} and
 5805: @code{WHILE} are predefined, so in this example it would not be
 5806: necessary to define them.
 5807: 
 5808: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
 5809: @subsection Calls and returns
 5810: @cindex calling a definition
 5811: @cindex returning from a definition
 5812: 
 5813: @cindex recursive definitions
 5814: A definition can be called simply be writing the name of the definition
 5815: to be called. Normally a definition is invisible during its own
 5816: definition. If you want to write a directly recursive definition, you
 5817: can use @code{recursive} to make the current definition visible, or
 5818: @code{recurse} to call the current definition directly.
 5819: 
 5820: 
 5821: doc-recursive
 5822: doc-recurse
 5823: 
 5824: 
 5825: @comment TODO add example of the two recursion methods
 5826: @quotation
 5827: @progstyle
 5828: I prefer using @code{recursive} to @code{recurse}, because calling the
 5829: definition by name is more descriptive (if the name is well-chosen) than
 5830: the somewhat cryptic @code{recurse}.  E.g., in a quicksort
 5831: implementation, it is much better to read (and think) ``now sort the
 5832: partitions'' than to read ``now do a recursive call''.
 5833: @end quotation
 5834: 
 5835: For mutual recursion, use @code{Defer}red words, like this:
 5836: 
 5837: @example
 5838: Defer foo
 5839: 
 5840: : bar ( ... -- ... )
 5841:  ... foo ... ;
 5842: 
 5843: :noname ( ... -- ... )
 5844:  ... bar ... ;
 5845: IS foo
 5846: @end example
 5847: 
 5848: Deferred words are discussed in more detail in @ref{Deferred words}.
 5849: 
 5850: The current definition returns control to the calling definition when
 5851: the end of the definition is reached or @code{EXIT} is encountered.
 5852: 
 5853: doc-exit
 5854: doc-;s
 5855: 
 5856: 
 5857: @node Exception Handling,  , Calls and returns, Control Structures
 5858: @subsection Exception Handling
 5859: @cindex exceptions
 5860: 
 5861: @c quit is a very bad idea for error handling, 
 5862: @c because it does not translate into a THROW
 5863: @c it also does not belong into this chapter
 5864: 
 5865: If a word detects an error condition that it cannot handle, it can
 5866: @code{throw} an exception.  In the simplest case, this will terminate
 5867: your program, and report an appropriate error.
 5868: 
 5869: doc-throw
 5870: 
 5871: @code{Throw} consumes a cell-sized error number on the stack. There are
 5872: some predefined error numbers in ANS Forth (see @file{errors.fs}).  In
 5873: Gforth (and most other systems) you can use the iors produced by various
 5874: words as error numbers (e.g., a typical use of @code{allocate} is
 5875: @code{allocate throw}).  Gforth also provides the word @code{exception}
 5876: to define your own error numbers (with decent error reporting); an ANS
 5877: Forth version of this word (but without the error messages) is available
 5878: in @code{compat/except.fs}.  And finally, you can use your own error
 5879: numbers (anything outside the range -4095..0), but won't get nice error
 5880: messages, only numbers.  For example, try:
 5881: 
 5882: @example
 5883: -10 throw                    \ ANS defined
 5884: -267 throw                   \ system defined
 5885: s" my error" exception throw \ user defined
 5886: 7 throw                      \ arbitrary number
 5887: @end example
 5888: 
 5889: doc---exception-exception
 5890: 
 5891: A common idiom to @code{THROW} a specific error if a flag is true is
 5892: this:
 5893: 
 5894: @example
 5895: @code{( flag ) 0<> @i{errno} and throw}
 5896: @end example
 5897: 
 5898: Your program can provide exception handlers to catch exceptions.  An
 5899: exception handler can be used to correct the problem, or to clean up
 5900: some data structures and just throw the exception to the next exception
 5901: handler.  Note that @code{throw} jumps to the dynamically innermost
 5902: exception handler.  The system's exception handler is outermost, and just
 5903: prints an error and restarts command-line interpretation (or, in batch
 5904: mode (i.e., while processing the shell command line), leaves Gforth).
 5905: 
 5906: The ANS Forth way to catch exceptions is @code{catch}:
 5907: 
 5908: doc-catch
 5909: 
 5910: The most common use of exception handlers is to clean up the state when
 5911: an error happens.  E.g.,
 5912: 
 5913: @example
 5914: base @ >r hex \ actually the hex should be inside foo, or we h
 5915: ['] foo catch ( nerror|0 )
 5916: r> base !
 5917: ( nerror|0 ) throw \ pass it on
 5918: @end example
 5919: 
 5920: A use of @code{catch} for handling the error @code{myerror} might look
 5921: like this:
 5922: 
 5923: @example
 5924: ['] foo catch
 5925: CASE
 5926:   myerror OF ... ( do something about it ) ENDOF
 5927:   dup throw \ default: pass other errors on, do nothing on non-errors
 5928: ENDCASE
 5929: @end example
 5930: 
 5931: Having to wrap the code into a separate word is often cumbersome,
 5932: therefore Gforth provides an alternative syntax:
 5933: 
 5934: @example
 5935: TRY
 5936:   @i{code1}
 5937: RECOVER     \ optional
 5938:   @i{code2} \ optional
 5939: ENDTRY
 5940: @end example
 5941: 
 5942: This performs @i{Code1}.  If @i{code1} completes normally, execution
 5943: continues after the @code{endtry}.  If @i{Code1} throws, the stacks are
 5944: reset to the state during @code{try}, the throw value is pushed on the
 5945: data stack, and execution constinues at @i{code2}, and finally falls
 5946: through the @code{endtry} into the following code.
 5947: 
 5948: doc-try
 5949: doc-recover
 5950: doc-endtry
 5951: 
 5952: The cleanup example from above in this syntax:
 5953: 
 5954: @example
 5955: base @ >r TRY
 5956:   hex foo \ now the hex is placed correctly
 5957:   0       \ value for throw
 5958: RECOVER ENDTRY
 5959: r> base ! throw
 5960: @end example
 5961: 
 5962: And here's the error handling example:
 5963: 
 5964: @example
 5965: TRY
 5966:   foo
 5967: RECOVER
 5968:   CASE
 5969:     myerror OF ... ( do something about it ) ENDOF
 5970:     throw \ pass other errors on
 5971:   ENDCASE
 5972: ENDTRY
 5973: @end example
 5974: 
 5975: @progstyle
 5976: As usual, you should ensure that the stack depth is statically known at
 5977: the end: either after the @code{throw} for passing on errors, or after
 5978: the @code{ENDTRY} (or, if you use @code{catch}, after the end of the
 5979: selection construct for handling the error).
 5980: 
 5981: There are two alternatives to @code{throw}: @code{Abort"} is conditional
 5982: and you can provide an error message.  @code{Abort} just produces an
 5983: ``Aborted'' error.
 5984: 
 5985: The problem with these words is that exception handlers cannot
 5986: differentiate between different @code{abort"}s; they just look like
 5987: @code{-2 throw} to them (the error message cannot be accessed by
 5988: standard programs).  Similar @code{abort} looks like @code{-1 throw} to
 5989: exception handlers.
 5990: 
 5991: doc-abort"
 5992: doc-abort
 5993: 
 5994: 
 5995: 
 5996: @c -------------------------------------------------------------
 5997: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
 5998: @section Defining Words
 5999: @cindex defining words
 6000: 
 6001: Defining words are used to extend Forth by creating new entries in the dictionary.
 6002: 
 6003: @menu
 6004: * CREATE::                      
 6005: * Variables::                   Variables and user variables
 6006: * Constants::                   
 6007: * Values::                      Initialised variables
 6008: * Colon Definitions::           
 6009: * Anonymous Definitions::       Definitions without names
 6010: * Supplying names::             Passing definition names as strings
 6011: * User-defined Defining Words::  
 6012: * Deferred words::              Allow forward references
 6013: * Aliases::                     
 6014: @end menu
 6015: 
 6016: @node CREATE, Variables, Defining Words, Defining Words
 6017: @subsection @code{CREATE}
 6018: @cindex simple defining words
 6019: @cindex defining words, simple
 6020: 
 6021: Defining words are used to create new entries in the dictionary. The
 6022: simplest defining word is @code{CREATE}. @code{CREATE} is used like
 6023: this:
 6024: 
 6025: @example
 6026: CREATE new-word1
 6027: @end example
 6028: 
 6029: @code{CREATE} is a parsing word, i.e., it takes an argument from the
 6030: input stream (@code{new-word1} in our example).  It generates a
 6031: dictionary entry for @code{new-word1}. When @code{new-word1} is
 6032: executed, all that it does is leave an address on the stack. The address
 6033: represents the value of the data space pointer (@code{HERE}) at the time
 6034: that @code{new-word1} was defined. Therefore, @code{CREATE} is a way of
 6035: associating a name with the address of a region of memory.
 6036: 
 6037: doc-create
 6038: 
 6039: Note that in ANS Forth guarantees only for @code{create} that its body
 6040: is in dictionary data space (i.e., where @code{here}, @code{allot}
 6041: etc. work, @pxref{Dictionary allocation}).  Also, in ANS Forth only
 6042: @code{create}d words can be modified with @code{does>}
 6043: (@pxref{User-defined Defining Words}).  And in ANS Forth @code{>body}
 6044: can only be applied to @code{create}d words.
 6045: 
 6046: By extending this example to reserve some memory in data space, we end
 6047: up with something like a @i{variable}. Here are two different ways to do
 6048: it:
 6049: 
 6050: @example
 6051: CREATE new-word2 1 cells allot  \ reserve 1 cell - initial value undefined
 6052: CREATE new-word3 4 ,            \ reserve 1 cell and initialise it (to 4)
 6053: @end example
 6054: 
 6055: The variable can be examined and modified using @code{@@} (``fetch'') and
 6056: @code{!} (``store'') like this:
 6057: 
 6058: @example
 6059: new-word2 @@ .      \ get address, fetch from it and display
 6060: 1234 new-word2 !   \ new value, get address, store to it
 6061: @end example
 6062: 
 6063: @cindex arrays
 6064: A similar mechanism can be used to create arrays. For example, an
 6065: 80-character text input buffer:
 6066: 
 6067: @example
 6068: CREATE text-buf 80 chars allot
 6069: 
 6070: text-buf 0 chars c@@ \ the 1st character (offset 0)
 6071: text-buf 3 chars c@@ \ the 4th character (offset 3)
 6072: @end example
 6073: 
 6074: You can build arbitrarily complex data structures by allocating
 6075: appropriate areas of memory. For further discussions of this, and to
 6076: learn about some Gforth tools that make it easier,
 6077: @xref{Structures}.
 6078: 
 6079: 
 6080: @node Variables, Constants, CREATE, Defining Words
 6081: @subsection Variables
 6082: @cindex variables
 6083: 
 6084: The previous section showed how a sequence of commands could be used to
 6085: generate a variable.  As a final refinement, the whole code sequence can
 6086: be wrapped up in a defining word (pre-empting the subject of the next
 6087: section), making it easier to create new variables:
 6088: 
 6089: @example
 6090: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
 6091: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
 6092: 
 6093: myvariableX foo \ variable foo starts off with an unknown value
 6094: myvariable0 joe \ whilst joe is initialised to 0
 6095: 
 6096: 45 3 * foo !   \ set foo to 135
 6097: 1234 joe !     \ set joe to 1234
 6098: 3 joe +!       \ increment joe by 3.. to 1237
 6099: @end example
 6100: 
 6101: Not surprisingly, there is no need to define @code{myvariable}, since
 6102: Forth already has a definition @code{Variable}. ANS Forth does not
 6103: guarantee that a @code{Variable} is initialised when it is created
 6104: (i.e., it may behave like @code{myvariableX}). In contrast, Gforth's
 6105: @code{Variable} initialises the variable to 0 (i.e., it behaves exactly
 6106: like @code{myvariable0}). Forth also provides @code{2Variable} and
 6107: @code{fvariable} for double and floating-point variables, respectively
 6108: -- they are initialised to 0. and 0e in Gforth. If you use a @code{Variable} to
 6109: store a boolean, you can use @code{on} and @code{off} to toggle its
 6110: state.
 6111: 
 6112: doc-variable
 6113: doc-2variable
 6114: doc-fvariable
 6115: 
 6116: @cindex user variables
 6117: @cindex user space
 6118: The defining word @code{User} behaves in the same way as @code{Variable}.
 6119: The difference is that it reserves space in @i{user (data) space} rather
 6120: than normal data space. In a Forth system that has a multi-tasker, each
 6121: task has its own set of user variables.
 6122: 
 6123: doc-user
 6124: @c doc-udp
 6125: @c doc-uallot
 6126: 
 6127: @comment TODO is that stuff about user variables strictly correct? Is it
 6128: @comment just terminal tasks that have user variables?
 6129: @comment should document tasker.fs (with some examples) elsewhere
 6130: @comment in this manual, then expand on user space and user variables.
 6131: 
 6132: @node Constants, Values, Variables, Defining Words
 6133: @subsection Constants
 6134: @cindex constants
 6135: 
 6136: @code{Constant} allows you to declare a fixed value and refer to it by
 6137: name. For example:
 6138: 
 6139: @example
 6140: 12 Constant INCHES-PER-FOOT
 6141: 3E+08 fconstant SPEED-O-LIGHT
 6142: @end example
 6143: 
 6144: A @code{Variable} can be both read and written, so its run-time
 6145: behaviour is to supply an address through which its current value can be
 6146: manipulated. In contrast, the value of a @code{Constant} cannot be
 6147: changed once it has been declared@footnote{Well, often it can be -- but
 6148: not in a Standard, portable way. It's safer to use a @code{Value} (read
 6149: on).} so it's not necessary to supply the address -- it is more
 6150: efficient to return the value of the constant directly. That's exactly
 6151: what happens; the run-time effect of a constant is to put its value on
 6152: the top of the stack (You can find one
 6153: way of implementing @code{Constant} in @ref{User-defined Defining Words}).
 6154: 
 6155: Forth also provides @code{2Constant} and @code{fconstant} for defining
 6156: double and floating-point constants, respectively.
 6157: 
 6158: doc-constant
 6159: doc-2constant
 6160: doc-fconstant
 6161: 
 6162: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
 6163: @c nac-> How could that not be true in an ANS Forth? You can't define a
 6164: @c constant, use it and then delete the definition of the constant..
 6165: 
 6166: @c anton->An ANS Forth system can compile a constant to a literal; On
 6167: @c decompilation you would see only the number, just as if it had been used
 6168: @c in the first place.  The word will stay, of course, but it will only be
 6169: @c used by the text interpreter (no run-time duties, except when it is 
 6170: @c POSTPONEd or somesuch).
 6171: 
 6172: @c nac:
 6173: @c I agree that it's rather deep, but IMO it is an important difference
 6174: @c relative to other programming languages.. often it's annoying: it
 6175: @c certainly changes my programming style relative to C.
 6176: 
 6177: @c anton: In what way?
 6178: 
 6179: Constants in Forth behave differently from their equivalents in other
 6180: programming languages. In other languages, a constant (such as an EQU in
 6181: assembler or a #define in C) only exists at compile-time; in the
 6182: executable program the constant has been translated into an absolute
 6183: number and, unless you are using a symbolic debugger, it's impossible to
 6184: know what abstract thing that number represents. In Forth a constant has
 6185: an entry in the header space and remains there after the code that uses
 6186: it has been defined. In fact, it must remain in the dictionary since it
 6187: has run-time duties to perform. For example:
 6188: 
 6189: @example
 6190: 12 Constant INCHES-PER-FOOT
 6191: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
 6192: @end example
 6193: 
 6194: @cindex in-lining of constants
 6195: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
 6196: associated with the constant @code{INCHES-PER-FOOT}. If you use
 6197: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
 6198: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
 6199: attempt to optimise constants by in-lining them where they are used. You
 6200: can force Gforth to in-line a constant like this:
 6201: 
 6202: @example
 6203: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
 6204: @end example
 6205: 
 6206: If you use @code{see} to decompile @i{this} version of
 6207: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
 6208: longer present. To understand how this works, read
 6209: @ref{Interpret/Compile states}, and @ref{Literals}.
 6210: 
 6211: In-lining constants in this way might improve execution time
 6212: fractionally, and can ensure that a constant is now only referenced at
 6213: compile-time. However, the definition of the constant still remains in
 6214: the dictionary. Some Forth compilers provide a mechanism for controlling
 6215: a second dictionary for holding transient words such that this second
 6216: dictionary can be deleted later in order to recover memory
 6217: space. However, there is no standard way of doing this.
 6218: 
 6219: 
 6220: @node Values, Colon Definitions, Constants, Defining Words
 6221: @subsection Values
 6222: @cindex values
 6223: 
 6224: A @code{Value} behaves like a @code{Constant}, but it can be changed.
 6225: @code{TO} is a parsing word that changes a @code{Values}.  In Gforth
 6226: (not in ANS Forth) you can access (and change) a @code{value} also with
 6227: @code{>body}.
 6228: 
 6229: Here are some
 6230: examples:
 6231: 
 6232: @example
 6233: 12 Value APPLES     \ Define APPLES with an initial value of 12
 6234: 34 TO APPLES        \ Change the value of APPLES. TO is a parsing word
 6235: 1 ' APPLES >body +! \ Increment APPLES.  Non-standard usage.
 6236: APPLES              \ puts 35 on the top of the stack.
 6237: @end example
 6238: 
 6239: doc-value
 6240: doc-to
 6241: 
 6242: 
 6243: 
 6244: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
 6245: @subsection Colon Definitions
 6246: @cindex colon definitions
 6247: 
 6248: @example
 6249: : name ( ... -- ... )
 6250:     word1 word2 word3 ;
 6251: @end example
 6252: 
 6253: @noindent
 6254: Creates a word called @code{name} that, upon execution, executes
 6255: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
 6256: 
 6257: The explanation above is somewhat superficial. For simple examples of
 6258: colon definitions see @ref{Your first definition}.  For an in-depth
 6259: discussion of some of the issues involved, @xref{Interpretation and
 6260: Compilation Semantics}.
 6261: 
 6262: doc-:
 6263: doc-;
 6264: 
 6265: 
 6266: @node Anonymous Definitions, Supplying names, Colon Definitions, Defining Words
 6267: @subsection Anonymous Definitions
 6268: @cindex colon definitions
 6269: @cindex defining words without name
 6270: 
 6271: Sometimes you want to define an @dfn{anonymous word}; a word without a
 6272: name. You can do this with:
 6273: 
 6274: doc-:noname
 6275: 
 6276: This leaves the execution token for the word on the stack after the
 6277: closing @code{;}. Here's an example in which a deferred word is
 6278: initialised with an @code{xt} from an anonymous colon definition:
 6279: 
 6280: @example
 6281: Defer deferred
 6282: :noname ( ... -- ... )
 6283:   ... ;
 6284: IS deferred
 6285: @end example
 6286: 
 6287: @noindent
 6288: Gforth provides an alternative way of doing this, using two separate
 6289: words:
 6290: 
 6291: doc-noname
 6292: @cindex execution token of last defined word
 6293: doc-lastxt
 6294: 
 6295: @noindent
 6296: The previous example can be rewritten using @code{noname} and
 6297: @code{lastxt}:
 6298: 
 6299: @example
 6300: Defer deferred
 6301: noname : ( ... -- ... )
 6302:   ... ;
 6303: lastxt IS deferred
 6304: @end example
 6305: 
 6306: @noindent
 6307: @code{noname} works with any defining word, not just @code{:}.
 6308: 
 6309: @code{lastxt} also works when the last word was not defined as
 6310: @code{noname}.  It does not work for combined words, though.  It also has
 6311: the useful property that is is valid as soon as the header for a
 6312: definition has been built. Thus:
 6313: 
 6314: @example
 6315: lastxt . : foo [ lastxt . ] ; ' foo .
 6316: @end example
 6317: 
 6318: @noindent
 6319: prints 3 numbers; the last two are the same.
 6320: 
 6321: @node Supplying names, User-defined Defining Words, Anonymous Definitions, Defining Words
 6322: @subsection Supplying the name of a defined word
 6323: @cindex names for defined words
 6324: @cindex defining words, name given in a string
 6325: 
 6326: By default, a defining word takes the name for the defined word from the
 6327: input stream. Sometimes you want to supply the name from a string. You
 6328: can do this with:
 6329: 
 6330: doc-nextname
 6331: 
 6332: For example:
 6333: 
 6334: @example
 6335: s" foo" nextname create
 6336: @end example
 6337: 
 6338: @noindent
 6339: is equivalent to:
 6340: 
 6341: @example
 6342: create foo
 6343: @end example
 6344: 
 6345: @noindent
 6346: @code{nextname} works with any defining word.
 6347: 
 6348: 
 6349: @node User-defined Defining Words, Deferred words, Supplying names, Defining Words
 6350: @subsection User-defined Defining Words
 6351: @cindex user-defined defining words
 6352: @cindex defining words, user-defined
 6353: 
 6354: You can create a new defining word by wrapping defining-time code around
 6355: an existing defining word and putting the sequence in a colon
 6356: definition. 
 6357: 
 6358: @c anton: This example is very complex and leads in a quite different
 6359: @c direction from the CREATE-DOES> stuff that follows.  It should probably
 6360: @c be done elsewhere, or as a subsubsection of this subsection (or as a
 6361: @c subsection of Defining Words)
 6362: 
 6363: For example, suppose that you have a word @code{stats} that
 6364: gathers statistics about colon definitions given the @i{xt} of the
 6365: definition, and you want every colon definition in your application to
 6366: make a call to @code{stats}. You can define and use a new version of
 6367: @code{:} like this:
 6368: 
 6369: @example
 6370: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
 6371:   ... ;  \ other code
 6372: 
 6373: : my: : lastxt postpone literal ['] stats compile, ;
 6374: 
 6375: my: foo + - ;
 6376: @end example
 6377: 
 6378: When @code{foo} is defined using @code{my:} these steps occur:
 6379: 
 6380: @itemize @bullet
 6381: @item
 6382: @code{my:} is executed.
 6383: @item
 6384: The @code{:} within the definition (the one between @code{my:} and
 6385: @code{lastxt}) is executed, and does just what it always does; it parses
 6386: the input stream for a name, builds a dictionary header for the name
 6387: @code{foo} and switches @code{state} from interpret to compile.
 6388: @item
 6389: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
 6390: being defined -- @code{foo} -- onto the stack.
 6391: @item
 6392: The code that was produced by @code{postpone literal} is executed; this
 6393: causes the value on the stack to be compiled as a literal in the code
 6394: area of @code{foo}.
 6395: @item
 6396: The code @code{['] stats} compiles a literal into the definition of
 6397: @code{my:}. When @code{compile,} is executed, that literal -- the
 6398: execution token for @code{stats} -- is layed down in the code area of
 6399: @code{foo} , following the literal@footnote{Strictly speaking, the
 6400: mechanism that @code{compile,} uses to convert an @i{xt} into something
 6401: in the code area is implementation-dependent. A threaded implementation
 6402: might spit out the execution token directly whilst another
 6403: implementation might spit out a native code sequence.}.
 6404: @item
 6405: At this point, the execution of @code{my:} is complete, and control
 6406: returns to the text interpreter. The text interpreter is in compile
 6407: state, so subsequent text @code{+ -} is compiled into the definition of
 6408: @code{foo} and the @code{;} terminates the definition as always.
 6409: @end itemize
 6410: 
 6411: You can use @code{see} to decompile a word that was defined using
 6412: @code{my:} and see how it is different from a normal @code{:}
 6413: definition. For example:
 6414: 
 6415: @example
 6416: : bar + - ;  \ like foo but using : rather than my:
 6417: see bar
 6418: : bar
 6419:   + - ;
 6420: see foo
 6421: : foo
 6422:   107645672 stats + - ;
 6423: 
 6424: \ use ' stats . to show that 107645672 is the xt for stats
 6425: @end example
 6426: 
 6427: You can use techniques like this to make new defining words in terms of
 6428: @i{any} existing defining word.
 6429: 
 6430: 
 6431: @cindex defining defining words
 6432: @cindex @code{CREATE} ... @code{DOES>}
 6433: If you want the words defined with your defining words to behave
 6434: differently from words defined with standard defining words, you can
 6435: write your defining word like this:
 6436: 
 6437: @example
 6438: : def-word ( "name" -- )
 6439:     CREATE @i{code1}
 6440: DOES> ( ... -- ... )
 6441:     @i{code2} ;
 6442: 
 6443: def-word name
 6444: @end example
 6445: 
 6446: @cindex child words
 6447: This fragment defines a @dfn{defining word} @code{def-word} and then
 6448: executes it.  When @code{def-word} executes, it @code{CREATE}s a new
 6449: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
 6450: is not executed at this time. The word @code{name} is sometimes called a
 6451: @dfn{child} of @code{def-word}.
 6452: 
 6453: When you execute @code{name}, the address of the body of @code{name} is
 6454: put on the data stack and @i{code2} is executed (the address of the body
 6455: of @code{name} is the address @code{HERE} returns immediately after the
 6456: @code{CREATE}, i.e., the address a @code{create}d word returns by
 6457: default).
 6458: 
 6459: @c anton:
 6460: @c www.dictionary.com says:
 6461: @c at·a·vism: 1.The reappearance of a characteristic in an organism after
 6462: @c several generations of absence, usually caused by the chance
 6463: @c recombination of genes.  2.An individual or a part that exhibits
 6464: @c atavism. Also called throwback.  3.The return of a trait or recurrence
 6465: @c of previous behavior after a period of absence.
 6466: @c
 6467: @c Doesn't seem to fit.
 6468: 
 6469: @c @cindex atavism in child words
 6470: You can use @code{def-word} to define a set of child words that behave
 6471: similarly; they all have a common run-time behaviour determined by
 6472: @i{code2}. Typically, the @i{code1} sequence builds a data area in the
 6473: body of the child word. The structure of the data is common to all
 6474: children of @code{def-word}, but the data values are specific -- and
 6475: private -- to each child word. When a child word is executed, the
 6476: address of its private data area is passed as a parameter on TOS to be
 6477: used and manipulated@footnote{It is legitimate both to read and write to
 6478: this data area.} by @i{code2}.
 6479: 
 6480: The two fragments of code that make up the defining words act (are
 6481: executed) at two completely separate times:
 6482: 
 6483: @itemize @bullet
 6484: @item
 6485: At @i{define time}, the defining word executes @i{code1} to generate a
 6486: child word
 6487: @item
 6488: At @i{child execution time}, when a child word is invoked, @i{code2}
 6489: is executed, using parameters (data) that are private and specific to
 6490: the child word.
 6491: @end itemize
 6492: 
 6493: Another way of understanding the behaviour of @code{def-word} and
 6494: @code{name} is to say that, if you make the following definitions:
 6495: @example
 6496: : def-word1 ( "name" -- )
 6497:     CREATE @i{code1} ;
 6498: 
 6499: : action1 ( ... -- ... )
 6500:     @i{code2} ;
 6501: 
 6502: def-word1 name1
 6503: @end example
 6504: 
 6505: @noindent
 6506: Then using @code{name1 action1} is equivalent to using @code{name}.
 6507: 
 6508: The classic example is that you can define @code{CONSTANT} in this way:
 6509: 
 6510: @example
 6511: : CONSTANT ( w "name" -- )
 6512:     CREATE ,
 6513: DOES> ( -- w )
 6514:     @@ ;
 6515: @end example
 6516: 
 6517: @comment There is a beautiful description of how this works and what
 6518: @comment it does in the Forthwrite 100th edition.. as well as an elegant
 6519: @comment commentary on the Counting Fruits problem.
 6520: 
 6521: When you create a constant with @code{5 CONSTANT five}, a set of
 6522: define-time actions take place; first a new word @code{five} is created,
 6523: then the value 5 is laid down in the body of @code{five} with
 6524: @code{,}. When @code{five} is executed, the address of the body is put on
 6525: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
 6526: no code of its own; it simply contains a data field and a pointer to the
 6527: code that follows @code{DOES>} in its defining word. That makes words
 6528: created in this way very compact.
 6529: 
 6530: The final example in this section is intended to remind you that space
 6531: reserved in @code{CREATE}d words is @i{data} space and therefore can be
 6532: both read and written by a Standard program@footnote{Exercise: use this
 6533: example as a starting point for your own implementation of @code{Value}
 6534: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
 6535: @code{[']}.}:
 6536: 
 6537: @example
 6538: : foo ( "name" -- )
 6539:     CREATE -1 ,
 6540: DOES> ( -- )
 6541:     @@ . ;
 6542: 
 6543: foo first-word
 6544: foo second-word
 6545: 
 6546: 123 ' first-word >BODY !
 6547: @end example
 6548: 
 6549: If @code{first-word} had been a @code{CREATE}d word, we could simply
 6550: have executed it to get the address of its data field. However, since it
 6551: was defined to have @code{DOES>} actions, its execution semantics are to
 6552: perform those @code{DOES>} actions. To get the address of its data field
 6553: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
 6554: translate the xt into the address of the data field.  When you execute
 6555: @code{first-word}, it will display @code{123}. When you execute
 6556: @code{second-word} it will display @code{-1}.
 6557: 
 6558: @cindex stack effect of @code{DOES>}-parts
 6559: @cindex @code{DOES>}-parts, stack effect
 6560: In the examples above the stack comment after the @code{DOES>} specifies
 6561: the stack effect of the defined words, not the stack effect of the
 6562: following code (the following code expects the address of the body on
 6563: the top of stack, which is not reflected in the stack comment). This is
 6564: the convention that I use and recommend (it clashes a bit with using
 6565: locals declarations for stack effect specification, though).
 6566: 
 6567: @menu
 6568: * CREATE..DOES> applications::  
 6569: * CREATE..DOES> details::       
 6570: * Advanced does> usage example::  
 6571: * @code{Const-does>}::          
 6572: @end menu
 6573: 
 6574: @node CREATE..DOES> applications, CREATE..DOES> details, User-defined Defining Words, User-defined Defining Words
 6575: @subsubsection Applications of @code{CREATE..DOES>}
 6576: @cindex @code{CREATE} ... @code{DOES>}, applications
 6577: 
 6578: You may wonder how to use this feature. Here are some usage patterns:
 6579: 
 6580: @cindex factoring similar colon definitions
 6581: When you see a sequence of code occurring several times, and you can
 6582: identify a meaning, you will factor it out as a colon definition. When
 6583: you see similar colon definitions, you can factor them using
 6584: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
 6585: that look very similar:
 6586: @example
 6587: : ori, ( reg-target reg-source n -- )
 6588:     0 asm-reg-reg-imm ;
 6589: : andi, ( reg-target reg-source n -- )
 6590:     1 asm-reg-reg-imm ;
 6591: @end example
 6592: 
 6593: @noindent
 6594: This could be factored with:
 6595: @example
 6596: : reg-reg-imm ( op-code -- )
 6597:     CREATE ,
 6598: DOES> ( reg-target reg-source n -- )
 6599:     @@ asm-reg-reg-imm ;
 6600: 
 6601: 0 reg-reg-imm ori,
 6602: 1 reg-reg-imm andi,
 6603: @end example
 6604: 
 6605: @cindex currying
 6606: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
 6607: supply a part of the parameters for a word (known as @dfn{currying} in
 6608: the functional language community). E.g., @code{+} needs two
 6609: parameters. Creating versions of @code{+} with one parameter fixed can
 6610: be done like this:
 6611: 
 6612: @example
 6613: : curry+ ( n1 "name" -- )
 6614:     CREATE ,
 6615: DOES> ( n2 -- n1+n2 )
 6616:     @@ + ;
 6617: 
 6618:  3 curry+ 3+
 6619: -2 curry+ 2-
 6620: @end example
 6621: 
 6622: 
 6623: @node CREATE..DOES> details, Advanced does> usage example, CREATE..DOES> applications, User-defined Defining Words
 6624: @subsubsection The gory details of @code{CREATE..DOES>}
 6625: @cindex @code{CREATE} ... @code{DOES>}, details
 6626: 
 6627: doc-does>
 6628: 
 6629: @cindex @code{DOES>} in a separate definition
 6630: This means that you need not use @code{CREATE} and @code{DOES>} in the
 6631: same definition; you can put the @code{DOES>}-part in a separate
 6632: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
 6633: @example
 6634: : does1 
 6635: DOES> ( ... -- ... )
 6636:     ... ;
 6637: 
 6638: : does2
 6639: DOES> ( ... -- ... )
 6640:     ... ;
 6641: 
 6642: : def-word ( ... -- ... )
 6643:     create ...
 6644:     IF
 6645:        does1
 6646:     ELSE
 6647:        does2
 6648:     ENDIF ;
 6649: @end example
 6650: 
 6651: In this example, the selection of whether to use @code{does1} or
 6652: @code{does2} is made at definition-time; at the time that the child word is
 6653: @code{CREATE}d.
 6654: 
 6655: @cindex @code{DOES>} in interpretation state
 6656: In a standard program you can apply a @code{DOES>}-part only if the last
 6657: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
 6658: will override the behaviour of the last word defined in any case. In a
 6659: standard program, you can use @code{DOES>} only in a colon
 6660: definition. In Gforth, you can also use it in interpretation state, in a
 6661: kind of one-shot mode; for example:
 6662: @example
 6663: CREATE name ( ... -- ... )
 6664:   @i{initialization}
 6665: DOES>
 6666:   @i{code} ;
 6667: @end example
 6668: 
 6669: @noindent
 6670: is equivalent to the standard:
 6671: @example
 6672: :noname
 6673: DOES>
 6674:     @i{code} ;
 6675: CREATE name EXECUTE ( ... -- ... )
 6676:     @i{initialization}
 6677: @end example
 6678: 
 6679: doc->body
 6680: 
 6681: @node Advanced does> usage example, @code{Const-does>}, CREATE..DOES> details, User-defined Defining Words
 6682: @subsubsection Advanced does> usage example
 6683: 
 6684: The MIPS disassembler (@file{arch/mips/disasm.fs}) contains many words
 6685: for disassembling instructions, that follow a very repetetive scheme:
 6686: 
 6687: @example
 6688: :noname @var{disasm-operands} s" @var{inst-name}" type ;
 6689: @var{entry-num} cells @var{table} + !
 6690: @end example
 6691: 
 6692: Of course, this inspires the idea to factor out the commonalities to
 6693: allow a definition like
 6694: 
 6695: @example
 6696: @var{disasm-operands} @var{entry-num} @var{table} define-inst @var{inst-name}
 6697: @end example
 6698: 
 6699: The parameters @var{disasm-operands} and @var{table} are usually
 6700: correlated.  Moreover, before I wrote the disassembler, there already
 6701: existed code that defines instructions like this:
 6702: 
 6703: @example
 6704: @var{entry-num} @var{inst-format} @var{inst-name}
 6705: @end example
 6706: 
 6707: This code comes from the assembler and resides in
 6708: @file{arch/mips/insts.fs}.
 6709: 
 6710: So I had to define the @var{inst-format} words that performed the scheme
 6711: above when executed.  At first I chose to use run-time code-generation:
 6712: 
 6713: @example
 6714: : @var{inst-format} ( entry-num "name" -- ; compiled code: addr w -- )
 6715:   :noname Postpone @var{disasm-operands}
 6716:   name Postpone sliteral Postpone type Postpone ;
 6717:   swap cells @var{table} + ! ;
 6718: @end example
 6719: 
 6720: Note that this supplies the other two parameters of the scheme above.
 6721: 
 6722: An alternative would have been to write this using
 6723: @code{create}/@code{does>}:
 6724: 
 6725: @example
 6726: : @var{inst-format} ( entry-num "name" -- )
 6727:   here name string, ( entry-num c-addr ) \ parse and save "name"
 6728:   noname create , ( entry-num )
 6729:   lastxt swap cells @var{table} + !
 6730: does> ( addr w -- )
 6731:   \ disassemble instruction w at addr
 6732:   @@ >r 
 6733:   @var{disasm-operands}
 6734:   r> count type ;
 6735: @end example
 6736: 
 6737: Somehow the first solution is simpler, mainly because it's simpler to
 6738: shift a string from definition-time to use-time with @code{sliteral}
 6739: than with @code{string,} and friends.
 6740: 
 6741: I wrote a lot of words following this scheme and soon thought about
 6742: factoring out the commonalities among them.  Note that this uses a
 6743: two-level defining word, i.e., a word that defines ordinary defining
 6744: words.
 6745: 
 6746: This time a solution involving @code{postpone} and friends seemed more
 6747: difficult (try it as an exercise), so I decided to use a
 6748: @code{create}/@code{does>} word; since I was already at it, I also used
 6749: @code{create}/@code{does>} for the lower level (try using
 6750: @code{postpone} etc. as an exercise), resulting in the following
 6751: definition:
 6752: 
 6753: @example
 6754: : define-format ( disasm-xt table-xt -- )
 6755:     \ define an instruction format that uses disasm-xt for
 6756:     \ disassembling and enters the defined instructions into table
 6757:     \ table-xt
 6758:     create 2,
 6759: does> ( u "inst" -- )
 6760:     \ defines an anonymous word for disassembling instruction inst,
 6761:     \ and enters it as u-th entry into table-xt
 6762:     2@@ swap here name string, ( u table-xt disasm-xt c-addr ) \ remember string
 6763:     noname create 2,      \ define anonymous word
 6764:     execute lastxt swap ! \ enter xt of defined word into table-xt
 6765: does> ( addr w -- )
 6766:     \ disassemble instruction w at addr
 6767:     2@@ >r ( addr w disasm-xt R: c-addr )
 6768:     execute ( R: c-addr ) \ disassemble operands
 6769:     r> count type ; \ print name 
 6770: @end example
 6771: 
 6772: Note that the tables here (in contrast to above) do the @code{cells +}
 6773: by themselves (that's why you have to pass an xt).  This word is used in
 6774: the following way:
 6775: 
 6776: @example
 6777: ' @var{disasm-operands} ' @var{table} define-format @var{inst-format}
 6778: @end example
 6779: 
 6780: As shown above, the defined instruction format is then used like this:
 6781: 
 6782: @example
 6783: @var{entry-num} @var{inst-format} @var{inst-name}
 6784: @end example
 6785: 
 6786: In terms of currying, this kind of two-level defining word provides the
 6787: parameters in three stages: first @var{disasm-operands} and @var{table},
 6788: then @var{entry-num} and @var{inst-name}, finally @code{addr w}, i.e.,
 6789: the instruction to be disassembled.  
 6790: 
 6791: Of course this did not quite fit all the instruction format names used
 6792: in @file{insts.fs}, so I had to define a few wrappers that conditioned
 6793: the parameters into the right form.
 6794: 
 6795: If you have trouble following this section, don't worry.  First, this is
 6796: involved and takes time (and probably some playing around) to
 6797: understand; second, this is the first two-level
 6798: @code{create}/@code{does>} word I have written in seventeen years of
 6799: Forth; and if I did not have @file{insts.fs} to start with, I may well
 6800: have elected to use just a one-level defining word (with some repeating
 6801: of parameters when using the defining word). So it is not necessary to
 6802: understand this, but it may improve your understanding of Forth.
 6803: 
 6804: 
 6805: @node @code{Const-does>},  , Advanced does> usage example, User-defined Defining Words
 6806: @subsubsection @code{Const-does>}
 6807: 
 6808: A frequent use of @code{create}...@code{does>} is for transferring some
 6809: values from definition-time to run-time.  Gforth supports this use with
 6810: 
 6811: doc-const-does>
 6812: 
 6813: A typical use of this word is:
 6814: 
 6815: @example
 6816: : curry+ ( n1 "name" -- )
 6817: 1 0 CONST-DOES> ( n2 -- n1+n2 )
 6818:     + ;
 6819: 
 6820: 3 curry+ 3+
 6821: @end example
 6822: 
 6823: Here the @code{1 0} means that 1 cell and 0 floats are transferred from
 6824: definition to run-time.
 6825: 
 6826: The advantages of using @code{const-does>} are:
 6827: 
 6828: @itemize
 6829: 
 6830: @item
 6831: You don't have to deal with storing and retrieving the values, i.e.,
 6832: your program becomes more writable and readable.
 6833: 
 6834: @item
 6835: When using @code{does>}, you have to introduce a @code{@@} that cannot
 6836: be optimized away (because you could change the data using
 6837: @code{>body}...@code{!}); @code{const-does>} avoids this problem.
 6838: 
 6839: @end itemize
 6840: 
 6841: An ANS Forth implementation of @code{const-does>} is available in
 6842: @file{compat/const-does.fs}.
 6843: 
 6844: 
 6845: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
 6846: @subsection Deferred words
 6847: @cindex deferred words
 6848: 
 6849: The defining word @code{Defer} allows you to define a word by name
 6850: without defining its behaviour; the definition of its behaviour is
 6851: deferred. Here are two situation where this can be useful:
 6852: 
 6853: @itemize @bullet
 6854: @item
 6855: Where you want to allow the behaviour of a word to be altered later, and
 6856: for all precompiled references to the word to change when its behaviour
 6857: is changed.
 6858: @item
 6859: For mutual recursion; @xref{Calls and returns}.
 6860: @end itemize
 6861: 
 6862: In the following example, @code{foo} always invokes the version of
 6863: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
 6864: always invokes the version that prints ``@code{Hello}''. There is no way
 6865: of getting @code{foo} to use the later version without re-ordering the
 6866: source code and recompiling it.
 6867: 
 6868: @example
 6869: : greet ." Good morning" ;
 6870: : foo ... greet ... ;
 6871: : greet ." Hello" ;
 6872: : bar ... greet ... ;
 6873: @end example
 6874: 
 6875: This problem can be solved by defining @code{greet} as a @code{Defer}red
 6876: word. The behaviour of a @code{Defer}red word can be defined and
 6877: redefined at any time by using @code{IS} to associate the xt of a
 6878: previously-defined word with it. The previous example becomes:
 6879: 
 6880: @example
 6881: Defer greet ( -- )
 6882: : foo ... greet ... ;
 6883: : bar ... greet ... ;
 6884: : greet1 ( -- ) ." Good morning" ;
 6885: : greet2 ( -- ) ." Hello" ;
 6886: ' greet2 <IS> greet  \ make greet behave like greet2
 6887: @end example
 6888: 
 6889: @progstyle
 6890: You should write a stack comment for every deferred word, and put only
 6891: XTs into deferred words that conform to this stack effect.  Otherwise
 6892: it's too difficult to use the deferred word.
 6893: 
 6894: A deferred word can be used to improve the statistics-gathering example
 6895: from @ref{User-defined Defining Words}; rather than edit the
 6896: application's source code to change every @code{:} to a @code{my:}, do
 6897: this:
 6898: 
 6899: @example
 6900: : real: : ;     \ retain access to the original
 6901: defer :         \ redefine as a deferred word
 6902: ' my: <IS> :      \ use special version of :
 6903: \
 6904: \ load application here
 6905: \
 6906: ' real: <IS> :    \ go back to the original
 6907: @end example
 6908: 
 6909: 
 6910: One thing to note is that @code{<IS>} consumes its name when it is
 6911: executed.  If you want to specify the name at compile time, use
 6912: @code{[IS]}:
 6913: 
 6914: @example
 6915: : set-greet ( xt -- )
 6916:   [IS] greet ;
 6917: 
 6918: ' greet1 set-greet
 6919: @end example
 6920: 
 6921: A deferred word can only inherit execution semantics from the xt
 6922: (because that is all that an xt can represent -- for more discussion of
 6923: this @pxref{Tokens for Words}); by default it will have default
 6924: interpretation and compilation semantics deriving from this execution
 6925: semantics.  However, you can change the interpretation and compilation
 6926: semantics of the deferred word in the usual ways:
 6927: 
 6928: @example
 6929: : bar .... ; compile-only
 6930: Defer fred immediate
 6931: Defer jim
 6932: 
 6933: ' bar <IS> jim  \ jim has default semantics
 6934: ' bar <IS> fred \ fred is immediate
 6935: @end example
 6936: 
 6937: doc-defer
 6938: doc-<is>
 6939: doc-[is]
 6940: doc-is
 6941: @comment TODO document these: what's defers [is]
 6942: doc-what's
 6943: doc-defers
 6944: 
 6945: @c Use @code{words-deferred} to see a list of deferred words.
 6946: 
 6947: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
 6948: are provided in @file{compat/defer.fs}.
 6949: 
 6950: 
 6951: @node Aliases,  , Deferred words, Defining Words
 6952: @subsection Aliases
 6953: @cindex aliases
 6954: 
 6955: The defining word @code{Alias} allows you to define a word by name that
 6956: has the same behaviour as some other word. Here are two situation where
 6957: this can be useful:
 6958: 
 6959: @itemize @bullet
 6960: @item
 6961: When you want access to a word's definition from a different word list
 6962: (for an example of this, see the definition of the @code{Root} word list
 6963: in the Gforth source).
 6964: @item
 6965: When you want to create a synonym; a definition that can be known by
 6966: either of two names (for example, @code{THEN} and @code{ENDIF} are
 6967: aliases).
 6968: @end itemize
 6969: 
 6970: Like deferred words, an alias has default compilation and interpretation
 6971: semantics at the beginning (not the modifications of the other word),
 6972: but you can change them in the usual ways (@code{immediate},
 6973: @code{compile-only}). For example:
 6974: 
 6975: @example
 6976: : foo ... ; immediate
 6977: 
 6978: ' foo Alias bar \ bar is not an immediate word
 6979: ' foo Alias fooby immediate \ fooby is an immediate word
 6980: @end example
 6981: 
 6982: Words that are aliases have the same xt, different headers in the
 6983: dictionary, and consequently different name tokens (@pxref{Tokens for
 6984: Words}) and possibly different immediate flags.  An alias can only have
 6985: default or immediate compilation semantics; you can define aliases for
 6986: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
 6987: 
 6988: doc-alias
 6989: 
 6990: 
 6991: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
 6992: @section Interpretation and Compilation Semantics
 6993: @cindex semantics, interpretation and compilation
 6994: 
 6995: @c !! state and ' are used without explanation
 6996: @c example for immediate/compile-only? or is the tutorial enough
 6997: 
 6998: @cindex interpretation semantics
 6999: The @dfn{interpretation semantics} of a (named) word are what the text
 7000: interpreter does when it encounters the word in interpret state. It also
 7001: appears in some other contexts, e.g., the execution token returned by
 7002: @code{' @i{word}} identifies the interpretation semantics of @i{word}
 7003: (in other words, @code{' @i{word} execute} is equivalent to
 7004: interpret-state text interpretation of @code{@i{word}}).
 7005: 
 7006: @cindex compilation semantics
 7007: The @dfn{compilation semantics} of a (named) word are what the text
 7008: interpreter does when it encounters the word in compile state. It also
 7009: appears in other contexts, e.g, @code{POSTPONE @i{word}}
 7010: compiles@footnote{In standard terminology, ``appends to the current
 7011: definition''.} the compilation semantics of @i{word}.
 7012: 
 7013: @cindex execution semantics
 7014: The standard also talks about @dfn{execution semantics}. They are used
 7015: only for defining the interpretation and compilation semantics of many
 7016: words. By default, the interpretation semantics of a word are to
 7017: @code{execute} its execution semantics, and the compilation semantics of
 7018: a word are to @code{compile,} its execution semantics.@footnote{In
 7019: standard terminology: The default interpretation semantics are its
 7020: execution semantics; the default compilation semantics are to append its
 7021: execution semantics to the execution semantics of the current
 7022: definition.}
 7023: 
 7024: Unnamed words (@pxref{Anonymous Definitions}) cannot be encountered by
 7025: the text interpreter, ticked, or @code{postpone}d, so they have no
 7026: interpretation or compilation semantics.  Their behaviour is represented
 7027: by their XT (@pxref{Tokens for Words}), and we call it execution
 7028: semantics, too.
 7029: 
 7030: @comment TODO expand, make it co-operate with new sections on text interpreter.
 7031: 
 7032: @cindex immediate words
 7033: @cindex compile-only words
 7034: You can change the semantics of the most-recently defined word:
 7035: 
 7036: 
 7037: doc-immediate
 7038: doc-compile-only
 7039: doc-restrict
 7040: 
 7041: By convention, words with non-default compilation semantics (e.g.,
 7042: immediate words) often have names surrounded with brackets (e.g.,
 7043: @code{[']}, @pxref{Execution token}).
 7044: 
 7045: Note that ticking (@code{'}) a compile-only word gives an error
 7046: (``Interpreting a compile-only word'').
 7047: 
 7048: @menu
 7049: * Combined words::              
 7050: @end menu
 7051: 
 7052: 
 7053: @node Combined words,  , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
 7054: @subsection Combined Words
 7055: @cindex combined words
 7056: 
 7057: Gforth allows you to define @dfn{combined words} -- words that have an
 7058: arbitrary combination of interpretation and compilation semantics.
 7059: 
 7060: doc-interpret/compile:
 7061: 
 7062: This feature was introduced for implementing @code{TO} and @code{S"}. I
 7063: recommend that you do not define such words, as cute as they may be:
 7064: they make it hard to get at both parts of the word in some contexts.
 7065: E.g., assume you want to get an execution token for the compilation
 7066: part. Instead, define two words, one that embodies the interpretation
 7067: part, and one that embodies the compilation part.  Once you have done
 7068: that, you can define a combined word with @code{interpret/compile:} for
 7069: the convenience of your users.
 7070: 
 7071: You might try to use this feature to provide an optimizing
 7072: implementation of the default compilation semantics of a word. For
 7073: example, by defining:
 7074: @example
 7075: :noname
 7076:    foo bar ;
 7077: :noname
 7078:    POSTPONE foo POSTPONE bar ;
 7079: interpret/compile: opti-foobar
 7080: @end example
 7081: 
 7082: @noindent
 7083: as an optimizing version of:
 7084: 
 7085: @example
 7086: : foobar
 7087:     foo bar ;
 7088: @end example
 7089: 
 7090: Unfortunately, this does not work correctly with @code{[compile]},
 7091: because @code{[compile]} assumes that the compilation semantics of all
 7092: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
 7093: opti-foobar} would compile compilation semantics, whereas
 7094: @code{[compile] foobar} would compile interpretation semantics.
 7095: 
 7096: @cindex state-smart words (are a bad idea)
 7097: @anchor{state-smartness}
 7098: Some people try to use @dfn{state-smart} words to emulate the feature provided
 7099: by @code{interpret/compile:} (words are state-smart if they check
 7100: @code{STATE} during execution). E.g., they would try to code
 7101: @code{foobar} like this:
 7102: 
 7103: @example
 7104: : foobar
 7105:   STATE @@
 7106:   IF ( compilation state )
 7107:     POSTPONE foo POSTPONE bar
 7108:   ELSE
 7109:     foo bar
 7110:   ENDIF ; immediate
 7111: @end example
 7112: 
 7113: Although this works if @code{foobar} is only processed by the text
 7114: interpreter, it does not work in other contexts (like @code{'} or
 7115: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
 7116: for a state-smart word, not for the interpretation semantics of the
 7117: original @code{foobar}; when you execute this execution token (directly
 7118: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
 7119: state, the result will not be what you expected (i.e., it will not
 7120: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
 7121: write them@footnote{For a more detailed discussion of this topic, see
 7122: M. Anton Ertl,
 7123: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,@code{State}-smartness---Why
 7124: it is Evil and How to Exorcise it}}, EuroForth '98.}!
 7125: 
 7126: @cindex defining words with arbitrary semantics combinations
 7127: It is also possible to write defining words that define words with
 7128: arbitrary combinations of interpretation and compilation semantics. In
 7129: general, they look like this:
 7130: 
 7131: @example
 7132: : def-word
 7133:     create-interpret/compile
 7134:     @i{code1}
 7135: interpretation>
 7136:     @i{code2}
 7137: <interpretation
 7138: compilation>
 7139:     @i{code3}
 7140: <compilation ;
 7141: @end example
 7142: 
 7143: For a @i{word} defined with @code{def-word}, the interpretation
 7144: semantics are to push the address of the body of @i{word} and perform
 7145: @i{code2}, and the compilation semantics are to push the address of
 7146: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
 7147: can also be defined like this (except that the defined constants don't
 7148: behave correctly when @code{[compile]}d):
 7149: 
 7150: @example
 7151: : constant ( n "name" -- )
 7152:     create-interpret/compile
 7153:     ,
 7154: interpretation> ( -- n )
 7155:     @@
 7156: <interpretation
 7157: compilation> ( compilation. -- ; run-time. -- n )
 7158:     @@ postpone literal
 7159: <compilation ;
 7160: @end example
 7161: 
 7162: 
 7163: doc-create-interpret/compile
 7164: doc-interpretation>
 7165: doc-<interpretation
 7166: doc-compilation>
 7167: doc-<compilation
 7168: 
 7169: 
 7170: Words defined with @code{interpret/compile:} and
 7171: @code{create-interpret/compile} have an extended header structure that
 7172: differs from other words; however, unless you try to access them with
 7173: plain address arithmetic, you should not notice this. Words for
 7174: accessing the header structure usually know how to deal with this; e.g.,
 7175: @code{'} @i{word} @code{>body} also gives you the body of a word created
 7176: with @code{create-interpret/compile}.
 7177: 
 7178: 
 7179: @c -------------------------------------------------------------
 7180: @node Tokens for Words, Compiling words, Interpretation and Compilation Semantics, Words
 7181: @section Tokens for Words
 7182: @cindex tokens for words
 7183: 
 7184: This section describes the creation and use of tokens that represent
 7185: words.
 7186: 
 7187: @menu
 7188: * Execution token::             represents execution/interpretation semantics
 7189: * Compilation token::           represents compilation semantics
 7190: * Name token::                  represents named words
 7191: @end menu
 7192: 
 7193: @node Execution token, Compilation token, Tokens for Words, Tokens for Words
 7194: @subsection Execution token
 7195: 
 7196: @cindex xt
 7197: @cindex execution token
 7198: An @dfn{execution token} (@i{XT}) represents some behaviour of a word.
 7199: You can use @code{execute} to invoke this behaviour.
 7200: 
 7201: @cindex tick (')
 7202: You can use @code{'} to get an execution token that represents the
 7203: interpretation semantics of a named word:
 7204: 
 7205: @example
 7206: 5 ' .   ( n xt ) 
 7207: execute ( )      \ execute the xt (i.e., ".")
 7208: @end example
 7209: 
 7210: doc-'
 7211: 
 7212: @code{'} parses at run-time; there is also a word @code{[']} that parses
 7213: when it is compiled, and compiles the resulting XT:
 7214: 
 7215: @example
 7216: : foo ['] . execute ;
 7217: 5 foo
 7218: : bar ' execute ; \ by contrast,
 7219: 5 bar .           \ ' parses "." when bar executes
 7220: @end example
 7221: 
 7222: doc-[']
 7223: 
 7224: If you want the execution token of @i{word}, write @code{['] @i{word}}
 7225: in compiled code and @code{' @i{word}} in interpreted code.  Gforth's
 7226: @code{'} and @code{[']} behave somewhat unusually by complaining about
 7227: compile-only words (because these words have no interpretation
 7228: semantics).  You might get what you want by using @code{COMP' @i{word}
 7229: DROP} or @code{[COMP'] @i{word} DROP} (for details @pxref{Compilation
 7230: token}).
 7231: 
 7232: Another way to get an XT is @code{:noname} or @code{lastxt}
 7233: (@pxref{Anonymous Definitions}).  For anonymous words this gives an xt
 7234: for the only behaviour the word has (the execution semantics).  For
 7235: named words, @code{lastxt} produces an XT for the same behaviour it
 7236: would produce if the word was defined anonymously.
 7237: 
 7238: @example
 7239: :noname ." hello" ;
 7240: execute
 7241: @end example
 7242: 
 7243: An XT occupies one cell and can be manipulated like any other cell.
 7244: 
 7245: @cindex code field address
 7246: @cindex CFA
 7247: In ANS Forth the XT is just an abstract data type (i.e., defined by the
 7248: operations that produce or consume it).  For old hands: In Gforth, the
 7249: XT is implemented as a code field address (CFA).
 7250: 
 7251: doc-execute
 7252: doc-perform
 7253: 
 7254: @node Compilation token, Name token, Execution token, Tokens for Words
 7255: @subsection Compilation token
 7256: 
 7257: @cindex compilation token
 7258: @cindex CT (compilation token)
 7259: Gforth represents the compilation semantics of a named word by a
 7260: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
 7261: @i{xt} is an execution token. The compilation semantics represented by
 7262: the compilation token can be performed with @code{execute}, which
 7263: consumes the whole compilation token, with an additional stack effect
 7264: determined by the represented compilation semantics.
 7265: 
 7266: At present, the @i{w} part of a compilation token is an execution token,
 7267: and the @i{xt} part represents either @code{execute} or
 7268: @code{compile,}@footnote{Depending upon the compilation semantics of the
 7269: word. If the word has default compilation semantics, the @i{xt} will
 7270: represent @code{compile,}. Otherwise (e.g., for immediate words), the
 7271: @i{xt} will represent @code{execute}.}. However, don't rely on that
 7272: knowledge, unless necessary; future versions of Gforth may introduce
 7273: unusual compilation tokens (e.g., a compilation token that represents
 7274: the compilation semantics of a literal).
 7275: 
 7276: You can perform the compilation semantics represented by the compilation
 7277: token with @code{execute}.  You can compile the compilation semantics
 7278: with @code{postpone,}. I.e., @code{COMP' @i{word} postpone,} is
 7279: equivalent to @code{postpone @i{word}}.
 7280: 
 7281: doc-[comp']
 7282: doc-comp'
 7283: doc-postpone,
 7284: 
 7285: @node Name token,  , Compilation token, Tokens for Words
 7286: @subsection Name token
 7287: 
 7288: @cindex name token
 7289: @cindex name field address
 7290: @cindex NFA
 7291: Gforth represents named words by the @dfn{name token}, (@i{nt}). In
 7292: Gforth, the abstract data type @emph{name token} is implemented as a
 7293: name field address (NFA).
 7294: 
 7295: doc-find-name
 7296: doc-name>int
 7297: doc-name?int
 7298: doc-name>comp
 7299: doc-name>string
 7300: 
 7301: @c ----------------------------------------------------------
 7302: @node Compiling words, The Text Interpreter, Tokens for Words, Words
 7303: @section Compiling words
 7304: @cindex compiling words
 7305: @cindex macros
 7306: 
 7307: In contrast to most other languages, Forth has no strict boundary
 7308: between compilation and run-time.  E.g., you can run arbitrary code
 7309: between defining words (or for computing data used by defining words
 7310: like @code{constant}). Moreover, @code{Immediate} (@pxref{Interpretation
 7311: and Compilation Semantics} and @code{[}...@code{]} (see below) allow
 7312: running arbitrary code while compiling a colon definition (exception:
 7313: you must not allot dictionary space).
 7314: 
 7315: @menu
 7316: * Literals::                    Compiling data values
 7317: * Macros::                      Compiling words
 7318: @end menu
 7319: 
 7320: @node Literals, Macros, Compiling words, Compiling words
 7321: @subsection Literals
 7322: @cindex Literals
 7323: 
 7324: The simplest and most frequent example is to compute a literal during
 7325: compilation.  E.g., the following definition prints an array of strings,
 7326: one string per line:
 7327: 
 7328: @example
 7329: : .strings ( addr u -- ) \ gforth
 7330:     2* cells bounds U+DO
 7331: 	cr i 2@@ type
 7332:     2 cells +LOOP ;  
 7333: @end example
 7334: 
 7335: With a simple-minded compiler like Gforth's, this computes @code{2
 7336: cells} on every loop iteration.  You can compute this value once and for
 7337: all at compile time and compile it into the definition like this:
 7338: 
 7339: @example
 7340: : .strings ( addr u -- ) \ gforth
 7341:     2* cells bounds U+DO
 7342: 	cr i 2@@ type
 7343:     [ 2 cells ] literal +LOOP ;  
 7344: @end example
 7345: 
 7346: @code{[} switches the text interpreter to interpret state (you will get
 7347: an @code{ok} prompt if you type this example interactively and insert a
 7348: newline between @code{[} and @code{]}), so it performs the
 7349: interpretation semantics of @code{2 cells}; this computes a number.
 7350: @code{]} switches the text interpreter back into compile state.  It then
 7351: performs @code{Literal}'s compilation semantics, which are to compile
 7352: this number into the current word.  You can decompile the word with
 7353: @code{see .strings} to see the effect on the compiled code.
 7354: 
 7355: You can also optimize the @code{2* cells} into @code{[ 2 cells ] literal
 7356: *} in this way.
 7357: 
 7358: doc-[
 7359: doc-]
 7360: doc-literal
 7361: doc-]L
 7362: 
 7363: There are also words for compiling other data types than single cells as
 7364: literals:
 7365: 
 7366: doc-2literal
 7367: doc-fliteral
 7368: doc-sliteral
 7369: 
 7370: @cindex colon-sys, passing data across @code{:}
 7371: @cindex @code{:}, passing data across
 7372: You might be tempted to pass data from outside a colon definition to the
 7373: inside on the data stack.  This does not work, because @code{:} puhes a
 7374: colon-sys, making stuff below unaccessible.  E.g., this does not work:
 7375: 
 7376: @example
 7377: 5 : foo literal ; \ error: "unstructured"
 7378: @end example
 7379: 
 7380: Instead, you have to pass the value in some other way, e.g., through a
 7381: variable:
 7382: 
 7383: @example
 7384: variable temp
 7385: 5 temp !
 7386: : foo [ temp @@ ] literal ;
 7387: @end example
 7388: 
 7389: 
 7390: @node Macros,  , Literals, Compiling words
 7391: @subsection Macros
 7392: @cindex Macros
 7393: @cindex compiling compilation semantics
 7394: 
 7395: @code{Literal} and friends compile data values into the current
 7396: definition.  You can also write words that compile other words into the
 7397: current definition.  E.g.,
 7398: 
 7399: @example
 7400: : compile-+ ( -- ) \ compiled code: ( n1 n2 -- n )
 7401:   POSTPONE + ;
 7402: 
 7403: : foo ( n1 n2 -- n )
 7404:   [ compile-+ ] ;
 7405: 1 2 foo .
 7406: @end example
 7407: 
 7408: This is equivalent to @code{: foo + ;} (@code{see foo} to check this).
 7409: What happens in this example?  @code{Postpone} compiles the compilation
 7410: semantics of @code{+} into @code{compile-+}; later the text interpreter
 7411: executes @code{compile-+} and thus the compilation semantics of +, which
 7412: compile (the execution semantics of) @code{+} into
 7413: @code{foo}.@footnote{A recent RFI answer requires that compiling words
 7414: should only be executed in compile state, so this example is not
 7415: guaranteed to work on all standard systems, but on any decent system it
 7416: will work.}
 7417: 
 7418: doc-postpone
 7419: doc-[compile]
 7420: 
 7421: Compiling words like @code{compile-+} are usually immediate (or similar)
 7422: so you do not have to switch to interpret state to execute them;
 7423: mopifying the last example accordingly produces:
 7424: 
 7425: @example
 7426: : [compile-+] ( compilation: --; interpretation: -- )
 7427:   \ compiled code: ( n1 n2 -- n )
 7428:   POSTPONE + ; immediate
 7429: 
 7430: : foo ( n1 n2 -- n )
 7431:   [compile-+] ;
 7432: 1 2 foo .
 7433: @end example
 7434: 
 7435: Immediate compiling words are similar to macros in other languages (in
 7436: particular, Lisp).  The important differences to macros in, e.g., C are:
 7437: 
 7438: @itemize @bullet
 7439: 
 7440: @item
 7441: You use the same language for defining and processing macros, not a
 7442: separate preprocessing language and processor.
 7443: 
 7444: @item
 7445: Consequently, the full power of Forth is available in macro definitions.
 7446: E.g., you can perform arbitrarily complex computations, or generate
 7447: different code conditionally or in a loop (e.g., @pxref{Advanced macros
 7448: Tutorial}).  This power is very useful when writing a parser generators
 7449: or other code-generating software.
 7450: 
 7451: @item
 7452: Macros defined using @code{postpone} etc. deal with the language at a
 7453: higher level than strings; name binding happens at macro definition
 7454: time, so you can avoid the pitfalls of name collisions that can happen
 7455: in C macros.  Of course, Forth is a liberal language and also allows to
 7456: shoot yourself in the foot with text-interpreted macros like
 7457: 
 7458: @example
 7459: : [compile-+] s" +" evaluate ; immediate
 7460: @end example
 7461: 
 7462: Apart from binding the name at macro use time, using @code{evaluate}
 7463: also makes your definition @code{state}-smart (@pxref{state-smartness}).
 7464: @end itemize
 7465: 
 7466: You may want the macro to compile a number into a word.  The word to do
 7467: it is @code{literal}, but you have to @code{postpone} it, so its
 7468: compilation semantics take effect when the macro is executed, not when
 7469: it is compiled:
 7470: 
 7471: @example
 7472: : [compile-5] ( -- ) \ compiled code: ( -- n )
 7473:   5 POSTPONE literal ; immediate
 7474: 
 7475: : foo [compile-5] ;
 7476: foo .
 7477: @end example
 7478: 
 7479: You may want to pass parameters to a macro, that the macro should
 7480: compile into the current definition.  If the parameter is a number, then
 7481: you can use @code{postpone literal} (similar for other values).
 7482: 
 7483: If you want to pass a word that is to be compiled, the usual way is to
 7484: pass an execution token and @code{compile,} it:
 7485: 
 7486: @example
 7487: : twice1 ( xt -- ) \ compiled code: ... -- ...
 7488:   dup compile, compile, ;
 7489: 
 7490: : 2+ ( n1 -- n2 )
 7491:   [ ' 1+ twice1 ] ;
 7492: @end example
 7493: 
 7494: doc-compile,
 7495: 
 7496: An alternative available in Gforth, that allows you to pass compile-only
 7497: words as parameters is to use the compilation token (@pxref{Compilation
 7498: token}).  The same example in this technique:
 7499: 
 7500: @example
 7501: : twice ( ... ct -- ... ) \ compiled code: ... -- ...
 7502:   2dup 2>r execute 2r> execute ;
 7503: 
 7504: : 2+ ( n1 -- n2 )
 7505:   [ comp' 1+ twice ] ;
 7506: @end example
 7507: 
 7508: In the example above @code{2>r} and @code{2r>} ensure that @code{twice}
 7509: works even if the executed compilation semantics has an effect on the
 7510: data stack.
 7511: 
 7512: You can also define complete definitions with these words; this provides
 7513: an alternative to using @code{does>} (@pxref{User-defined Defining
 7514: Words}).  E.g., instead of
 7515: 
 7516: @example
 7517: : curry+ ( n1 "name" -- )
 7518:     CREATE ,
 7519: DOES> ( n2 -- n1+n2 )
 7520:     @@ + ;
 7521: @end example
 7522: 
 7523: you could define
 7524: 
 7525: @example
 7526: : curry+ ( n1 "name" -- )
 7527:   \ name execution: ( n2 -- n1+n2 )
 7528:   >r : r> POSTPONE literal POSTPONE + POSTPONE ; ;
 7529: 
 7530: -3 curry+ 3-
 7531: see 3-
 7532: @end example
 7533: 
 7534: The sequence @code{>r : r>} is necessary, because @code{:} puts a
 7535: colon-sys on the data stack that makes everything below it unaccessible.
 7536: 
 7537: This way of writing defining words is sometimes more, sometimes less
 7538: convenient than using @code{does>} (@pxref{Advanced does> usage
 7539: example}).  One advantage of this method is that it can be optimized
 7540: better, because the compiler knows that the value compiled with
 7541: @code{literal} is fixed, whereas the data associated with a
 7542: @code{create}d word can be changed.
 7543: 
 7544: @c ----------------------------------------------------------
 7545: @node The Text Interpreter, Word Lists, Compiling words, Words
 7546: @section  The Text Interpreter
 7547: @cindex interpreter - outer
 7548: @cindex text interpreter
 7549: @cindex outer interpreter
 7550: 
 7551: @c Should we really describe all these ugly details?  IMO the text
 7552: @c interpreter should be much cleaner, but that may not be possible within
 7553: @c ANS Forth. - anton
 7554: @c nac-> I wanted to explain how it works to show how you can exploit
 7555: @c it in your own programs. When I was writing a cross-compiler, figuring out
 7556: @c some of these gory details was very helpful to me. None of the textbooks
 7557: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
 7558: @c seems to positively avoid going into too much detail for some of
 7559: @c the internals.
 7560: 
 7561: @c anton: ok.  I wonder, though, if this is the right place; for some stuff
 7562: @c it is; for the ugly details, I would prefer another place.  I wonder
 7563: @c whether we should have a chapter before "Words" that describes some
 7564: @c basic concepts referred to in words, and a chapter after "Words" that
 7565: @c describes implementation details.
 7566: 
 7567: The text interpreter@footnote{This is an expanded version of the
 7568: material in @ref{Introducing the Text Interpreter}.} is an endless loop
 7569: that processes input from the current input device. It is also called
 7570: the outer interpreter, in contrast to the inner interpreter
 7571: (@pxref{Engine}) which executes the compiled Forth code on interpretive
 7572: implementations.
 7573: 
 7574: @cindex interpret state
 7575: @cindex compile state
 7576: The text interpreter operates in one of two states: @dfn{interpret
 7577: state} and @dfn{compile state}. The current state is defined by the
 7578: aptly-named variable @code{state}.
 7579: 
 7580: This section starts by describing how the text interpreter behaves when
 7581: it is in interpret state, processing input from the user input device --
 7582: the keyboard. This is the mode that a Forth system is in after it starts
 7583: up.
 7584: 
 7585: @cindex input buffer
 7586: @cindex terminal input buffer
 7587: The text interpreter works from an area of memory called the @dfn{input
 7588: buffer}@footnote{When the text interpreter is processing input from the
 7589: keyboard, this area of memory is called the @dfn{terminal input buffer}
 7590: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
 7591: @code{#TIB}.}, which stores your keyboard input when you press the
 7592: @key{RET} key. Starting at the beginning of the input buffer, it skips
 7593: leading spaces (called @dfn{delimiters}) then parses a string (a
 7594: sequence of non-space characters) until it reaches either a space
 7595: character or the end of the buffer. Having parsed a string, it makes two
 7596: attempts to process it:
 7597: 
 7598: @cindex dictionary
 7599: @itemize @bullet
 7600: @item
 7601: It looks for the string in a @dfn{dictionary} of definitions. If the
 7602: string is found, the string names a @dfn{definition} (also known as a
 7603: @dfn{word}) and the dictionary search returns information that allows
 7604: the text interpreter to perform the word's @dfn{interpretation
 7605: semantics}. In most cases, this simply means that the word will be
 7606: executed.
 7607: @item
 7608: If the string is not found in the dictionary, the text interpreter
 7609: attempts to treat it as a number, using the rules described in
 7610: @ref{Number Conversion}. If the string represents a legal number in the
 7611: current radix, the number is pushed onto a parameter stack (the data
 7612: stack for integers, the floating-point stack for floating-point
 7613: numbers).
 7614: @end itemize
 7615: 
 7616: If both attempts fail, or if the word is found in the dictionary but has
 7617: no interpretation semantics@footnote{This happens if the word was
 7618: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
 7619: remainder of the input buffer, issues an error message and waits for
 7620: more input. If one of the attempts succeeds, the text interpreter
 7621: repeats the parsing process until the whole of the input buffer has been
 7622: processed, at which point it prints the status message ``@code{ ok}''
 7623: and waits for more input.
 7624: 
 7625: @c anton: this should be in the input stream subsection (or below it)
 7626: 
 7627: @cindex parse area
 7628: The text interpreter keeps track of its position in the input buffer by
 7629: updating a variable called @code{>IN} (pronounced ``to-in''). The value
 7630: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
 7631: of the input buffer. The region from offset @code{>IN @@} to the end of
 7632: the input buffer is called the @dfn{parse area}@footnote{In other words,
 7633: the text interpreter processes the contents of the input buffer by
 7634: parsing strings from the parse area until the parse area is empty.}.
 7635: This example shows how @code{>IN} changes as the text interpreter parses
 7636: the input buffer:
 7637: 
 7638: @example
 7639: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
 7640:   CR ." ->" TYPE ." <-" ; IMMEDIATE 
 7641: 
 7642: 1 2 3 remaining + remaining . 
 7643: 
 7644: : foo 1 2 3 remaining SWAP remaining ;
 7645: @end example
 7646: 
 7647: @noindent
 7648: The result is:
 7649: 
 7650: @example
 7651: ->+ remaining .<-
 7652: ->.<-5  ok
 7653: 
 7654: ->SWAP remaining ;-<
 7655: ->;<-  ok
 7656: @end example
 7657: 
 7658: @cindex parsing words
 7659: The value of @code{>IN} can also be modified by a word in the input
 7660: buffer that is executed by the text interpreter.  This means that a word
 7661: can ``trick'' the text interpreter into either skipping a section of the
 7662: input buffer@footnote{This is how parsing words work.} or into parsing a
 7663: section twice. For example:
 7664: 
 7665: @example
 7666: : lat ." <<foo>>" ;
 7667: : flat ." <<bar>>" >IN DUP @@ 3 - SWAP ! ;
 7668: @end example
 7669: 
 7670: @noindent
 7671: When @code{flat} is executed, this output is produced@footnote{Exercise
 7672: for the reader: what would happen if the @code{3} were replaced with
 7673: @code{4}?}:
 7674: 
 7675: @example
 7676: <<bar>><<foo>>
 7677: @end example
 7678: 
 7679: This technique can be used to work around some of the interoperability
 7680: problems of parsing words.  Of course, it's better to avoid parsing
 7681: words where possible.
 7682: 
 7683: @noindent
 7684: Two important notes about the behaviour of the text interpreter:
 7685: 
 7686: @itemize @bullet
 7687: @item
 7688: It processes each input string to completion before parsing additional
 7689: characters from the input buffer.
 7690: @item
 7691: It treats the input buffer as a read-only region (and so must your code).
 7692: @end itemize
 7693: 
 7694: @noindent
 7695: When the text interpreter is in compile state, its behaviour changes in
 7696: these ways:
 7697: 
 7698: @itemize @bullet
 7699: @item
 7700: If a parsed string is found in the dictionary, the text interpreter will
 7701: perform the word's @dfn{compilation semantics}. In most cases, this
 7702: simply means that the execution semantics of the word will be appended
 7703: to the current definition.
 7704: @item
 7705: When a number is encountered, it is compiled into the current definition
 7706: (as a literal) rather than being pushed onto a parameter stack.
 7707: @item
 7708: If an error occurs, @code{state} is modified to put the text interpreter
 7709: back into interpret state.
 7710: @item
 7711: Each time a line is entered from the keyboard, Gforth prints
 7712: ``@code{ compiled}'' rather than `` @code{ok}''.
 7713: @end itemize
 7714: 
 7715: @cindex text interpreter - input sources
 7716: When the text interpreter is using an input device other than the
 7717: keyboard, its behaviour changes in these ways:
 7718: 
 7719: @itemize @bullet
 7720: @item
 7721: When the parse area is empty, the text interpreter attempts to refill
 7722: the input buffer from the input source. When the input source is
 7723: exhausted, the input source is set back to the previous input source.
 7724: @item
 7725: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
 7726: time the parse area is emptied.
 7727: @item
 7728: If an error occurs, the input source is set back to the user input
 7729: device.
 7730: @end itemize
 7731: 
 7732: You can read about this in more detail in @ref{Input Sources}.
 7733: 
 7734: doc->in
 7735: doc-source
 7736: 
 7737: doc-tib
 7738: doc-#tib
 7739: 
 7740: 
 7741: @menu
 7742: * Input Sources::               
 7743: * Number Conversion::           
 7744: * Interpret/Compile states::    
 7745: * Interpreter Directives::      
 7746: @end menu
 7747: 
 7748: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
 7749: @subsection Input Sources
 7750: @cindex input sources
 7751: @cindex text interpreter - input sources
 7752: 
 7753: By default, the text interpreter processes input from the user input
 7754: device (the keyboard) when Forth starts up. The text interpreter can
 7755: process input from any of these sources:
 7756: 
 7757: @itemize @bullet
 7758: @item
 7759: The user input device -- the keyboard.
 7760: @item
 7761: A file, using the words described in @ref{Forth source files}.
 7762: @item
 7763: A block, using the words described in @ref{Blocks}.
 7764: @item
 7765: A text string, using @code{evaluate}.
 7766: @end itemize
 7767: 
 7768: A program can identify the current input device from the values of
 7769: @code{source-id} and @code{blk}.
 7770: 
 7771: 
 7772: doc-source-id
 7773: doc-blk
 7774: 
 7775: doc-save-input
 7776: doc-restore-input
 7777: 
 7778: doc-evaluate
 7779: 
 7780: 
 7781: 
 7782: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
 7783: @subsection Number Conversion
 7784: @cindex number conversion
 7785: @cindex double-cell numbers, input format
 7786: @cindex input format for double-cell numbers
 7787: @cindex single-cell numbers, input format
 7788: @cindex input format for single-cell numbers
 7789: @cindex floating-point numbers, input format
 7790: @cindex input format for floating-point numbers
 7791: 
 7792: This section describes the rules that the text interpreter uses when it
 7793: tries to convert a string into a number.
 7794: 
 7795: Let <digit> represent any character that is a legal digit in the current
 7796: number base@footnote{For example, 0-9 when the number base is decimal or
 7797: 0-9, A-F when the number base is hexadecimal.}.
 7798: 
 7799: Let <decimal digit> represent any character in the range 0-9.
 7800: 
 7801: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
 7802: in the braces (@i{a} or @i{b} or neither).
 7803: 
 7804: Let * represent any number of instances of the previous character
 7805: (including none).
 7806: 
 7807: Let any other character represent itself.
 7808: 
 7809: @noindent
 7810: Now, the conversion rules are:
 7811: 
 7812: @itemize @bullet
 7813: @item
 7814: A string of the form <digit><digit>* is treated as a single-precision
 7815: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
 7816: @item
 7817: A string of the form -<digit><digit>* is treated as a single-precision
 7818: (cell-sized) negative integer, and is represented using 2's-complement
 7819: arithmetic. Examples are -45 -5681 -0
 7820: @item
 7821: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
 7822: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
 7823: (all three of these represent the same number).
 7824: @item
 7825: A string of the form -<digit><digit>*.<digit>* is treated as a
 7826: double-precision (double-cell-sized) negative integer, and is
 7827: represented using 2's-complement arithmetic. Examples are -3465. -3.465
 7828: -34.65 (all three of these represent the same number).
 7829: @item
 7830: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
 7831: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
 7832: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
 7833: number) +12.E-4
 7834: @end itemize
 7835: 
 7836: By default, the number base used for integer number conversion is given
 7837: by the contents of the variable @code{base}.  Note that a lot of
 7838: confusion can result from unexpected values of @code{base}.  If you
 7839: change @code{base} anywhere, make sure to save the old value and restore
 7840: it afterwards.  In general I recommend keeping @code{base} decimal, and
 7841: using the prefixes described below for the popular non-decimal bases.
 7842: 
 7843: doc-dpl
 7844: doc-base
 7845: doc-hex
 7846: doc-decimal
 7847: 
 7848: 
 7849: @cindex '-prefix for character strings
 7850: @cindex &-prefix for decimal numbers
 7851: @cindex %-prefix for binary numbers
 7852: @cindex $-prefix for hexadecimal numbers
 7853: Gforth allows you to override the value of @code{base} by using a
 7854: prefix@footnote{Some Forth implementations provide a similar scheme by
 7855: implementing @code{$} etc. as parsing words that process the subsequent
 7856: number in the input stream and push it onto the stack. For example, see
 7857: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
 7858: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
 7859: is required between the prefix and the number.} before the first digit
 7860: of an (integer) number. Four prefixes are supported:
 7861: 
 7862: @itemize @bullet
 7863: @item
 7864: @code{&} -- decimal
 7865: @item
 7866: @code{%} -- binary
 7867: @item
 7868: @code{$} -- hexadecimal
 7869: @item
 7870: @code{'} -- base @code{max-char+1}
 7871: @end itemize
 7872: 
 7873: Here are some examples, with the equivalent decimal number shown after
 7874: in braces:
 7875: 
 7876: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
 7877: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
 7878: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
 7879: &905 (905), $abc (2478), $ABC (2478).
 7880: 
 7881: @cindex number conversion - traps for the unwary
 7882: @noindent
 7883: Number conversion has a number of traps for the unwary:
 7884: 
 7885: @itemize @bullet
 7886: @item
 7887: You cannot determine the current number base using the code sequence
 7888: @code{base @@ .} -- the number base is always 10 in the current number
 7889: base. Instead, use something like @code{base @@ dec.}
 7890: @item
 7891: If the number base is set to a value greater than 14 (for example,
 7892: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
 7893: it to be intepreted as either a single-precision integer or a
 7894: floating-point number (Gforth treats it as an integer). The ambiguity
 7895: can be resolved by explicitly stating the sign of the mantissa and/or
 7896: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
 7897: ambiguity arises; either representation will be treated as a
 7898: floating-point number.
 7899: @item
 7900: There is a word @code{bin} but it does @i{not} set the number base!
 7901: It is used to specify file types.
 7902: @item
 7903: ANS Forth requires the @code{.} of a double-precision number to be the
 7904: final character in the string.  Gforth allows the @code{.} to be
 7905: anywhere after the first digit.
 7906: @item
 7907: The number conversion process does not check for overflow.
 7908: @item
 7909: In an ANS Forth program @code{base} is required to be decimal when
 7910: converting floating-point numbers.  In Gforth, number conversion to
 7911: floating-point numbers always uses base &10, irrespective of the value
 7912: of @code{base}.
 7913: @end itemize
 7914: 
 7915: You can read numbers into your programs with the words described in
 7916: @ref{Input}.
 7917: 
 7918: @node Interpret/Compile states, Interpreter Directives, Number Conversion, The Text Interpreter
 7919: @subsection Interpret/Compile states
 7920: @cindex Interpret/Compile states
 7921: 
 7922: A standard program is not permitted to change @code{state}
 7923: explicitly. However, it can change @code{state} implicitly, using the
 7924: words @code{[} and @code{]}. When @code{[} is executed it switches
 7925: @code{state} to interpret state, and therefore the text interpreter
 7926: starts interpreting. When @code{]} is executed it switches @code{state}
 7927: to compile state and therefore the text interpreter starts
 7928: compiling. The most common usage for these words is for switching into
 7929: interpret state and back from within a colon definition; this technique
 7930: can be used to compile a literal (for an example, @pxref{Literals}) or
 7931: for conditional compilation (for an example, @pxref{Interpreter
 7932: Directives}).
 7933: 
 7934: 
 7935: @c This is a bad example: It's non-standard, and it's not necessary.
 7936: @c However, I can't think of a good example for switching into compile
 7937: @c state when there is no current word (@code{state}-smart words are not a
 7938: @c good reason).  So maybe we should use an example for switching into
 7939: @c interpret @code{state} in a colon def. - anton
 7940: @c nac-> I agree. I started out by putting in the example, then realised
 7941: @c that it was non-ANS, so wrote more words around it. I hope this
 7942: @c re-written version is acceptable to you. I do want to keep the example
 7943: @c as it is helpful for showing what is and what is not portable, particularly
 7944: @c where it outlaws a style in common use.
 7945: 
 7946: @c anton: it's more important to show what's portable.  After we have done
 7947: @c that, we can also show what's not.  In any case, I have written a
 7948: @c section Compiling Words which also deals with [ ].
 7949: 
 7950: @c  !! The following example does not work in Gforth 0.5.9 or later.
 7951: 
 7952: @c  @code{[} and @code{]} also give you the ability to switch into compile
 7953: @c  state and back, but we cannot think of any useful Standard application
 7954: @c  for this ability. Pre-ANS Forth textbooks have examples like this:
 7955: 
 7956: @c  @example
 7957: @c  : AA ." this is A" ;
 7958: @c  : BB ." this is B" ;
 7959: @c  : CC ." this is C" ;
 7960: 
 7961: @c  create table ] aa bb cc [
 7962: 
 7963: @c  : go ( n -- ) \ n is offset into table.. 0 for 1st entry
 7964: @c    cells table + @@ execute ;
 7965: @c  @end example
 7966: 
 7967: @c  This example builds a jump table; @code{0 go} will display ``@code{this
 7968: @c  is A}''. Using @code{[} and @code{]} in this example is equivalent to
 7969: @c  defining @code{table} like this:
 7970: 
 7971: @c  @example
 7972: @c  create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
 7973: @c  @end example
 7974: 
 7975: @c  The problem with this code is that the definition of @code{table} is not
 7976: @c  portable -- it @i{compile}s execution tokens into code space. Whilst it
 7977: @c  @i{may} work on systems where code space and data space co-incide, the
 7978: @c  Standard only allows data space to be assigned for a @code{CREATE}d
 7979: @c  word. In addition, the Standard only allows @code{@@} to access data
 7980: @c  space, whilst this example is using it to access code space. The only
 7981: @c  portable, Standard way to build this table is to build it in data space,
 7982: @c  like this:
 7983: 
 7984: @c  @example
 7985: @c  create table ' aa , ' bb , ' cc ,
 7986: @c  @end example
 7987: 
 7988: @c  doc-state
 7989: 
 7990: 
 7991: @node Interpreter Directives,  , Interpret/Compile states, The Text Interpreter
 7992: @subsection Interpreter Directives
 7993: @cindex interpreter directives
 7994: @cindex conditional compilation
 7995: 
 7996: These words are usually used in interpret state; typically to control
 7997: which parts of a source file are processed by the text
 7998: interpreter. There are only a few ANS Forth Standard words, but Gforth
 7999: supplements these with a rich set of immediate control structure words
 8000: to compensate for the fact that the non-immediate versions can only be
 8001: used in compile state (@pxref{Control Structures}). Typical usages:
 8002: 
 8003: @example
 8004: FALSE Constant HAVE-ASSEMBLER
 8005: .
 8006: .
 8007: HAVE-ASSEMBLER [IF]
 8008: : ASSEMBLER-FEATURE
 8009:   ...
 8010: ;
 8011: [ENDIF]
 8012: .
 8013: .
 8014: : SEE
 8015:   ... \ general-purpose SEE code
 8016:   [ HAVE-ASSEMBLER [IF] ]
 8017:   ... \ assembler-specific SEE code
 8018:   [ [ENDIF] ]
 8019: ;
 8020: @end example
 8021: 
 8022: 
 8023: doc-[IF]
 8024: doc-[ELSE]
 8025: doc-[THEN]
 8026: doc-[ENDIF]
 8027: 
 8028: doc-[IFDEF]
 8029: doc-[IFUNDEF]
 8030: 
 8031: doc-[?DO]
 8032: doc-[DO]
 8033: doc-[FOR]
 8034: doc-[LOOP]
 8035: doc-[+LOOP]
 8036: doc-[NEXT]
 8037: 
 8038: doc-[BEGIN]
 8039: doc-[UNTIL]
 8040: doc-[AGAIN]
 8041: doc-[WHILE]
 8042: doc-[REPEAT]
 8043: 
 8044: 
 8045: @c -------------------------------------------------------------
 8046: @node Word Lists, Environmental Queries, The Text Interpreter, Words
 8047: @section Word Lists
 8048: @cindex word lists
 8049: @cindex header space
 8050: 
 8051: A wordlist is a list of named words; you can add new words and look up
 8052: words by name (and you can remove words in a restricted way with
 8053: markers).  Every named (and @code{reveal}ed) word is in one wordlist.
 8054: 
 8055: @cindex search order stack
 8056: The text interpreter searches the wordlists present in the search order
 8057: (a stack of wordlists), from the top to the bottom.  Within each
 8058: wordlist, the search starts conceptually at the newest word; i.e., if
 8059: two words in a wordlist have the same name, the newer word is found.
 8060: 
 8061: @cindex compilation word list
 8062: New words are added to the @dfn{compilation wordlist} (aka current
 8063: wordlist).
 8064: 
 8065: @cindex wid
 8066: A word list is identified by a cell-sized word list identifier (@i{wid})
 8067: in much the same way as a file is identified by a file handle. The
 8068: numerical value of the wid has no (portable) meaning, and might change
 8069: from session to session.
 8070: 
 8071: The ANS Forth ``Search order'' word set is intended to provide a set of
 8072: low-level tools that allow various different schemes to be
 8073: implemented. Gforth also provides @code{vocabulary}, a traditional Forth
 8074: word.  @file{compat/vocabulary.fs} provides an implementation in ANS
 8075: Forth.
 8076: 
 8077: @comment TODO: locals section refers to here, saying that every word list (aka
 8078: @comment vocabulary) has its own methods for searching etc. Need to document that.
 8079: @c anton: but better in a separate subsection on wordlist internals
 8080: 
 8081: @comment TODO: document markers, reveal, tables, mappedwordlist
 8082: 
 8083: @comment the gforthman- prefix is used to pick out the true definition of a
 8084: @comment word from the source files, rather than some alias.
 8085: 
 8086: doc-forth-wordlist
 8087: doc-definitions
 8088: doc-get-current
 8089: doc-set-current
 8090: doc-get-order
 8091: doc---gforthman-set-order
 8092: doc-wordlist
 8093: doc-table
 8094: doc->order
 8095: doc-previous
 8096: doc-also
 8097: doc---gforthman-forth
 8098: doc-only
 8099: doc---gforthman-order
 8100: 
 8101: doc-find
 8102: doc-search-wordlist
 8103: 
 8104: doc-words
 8105: doc-vlist
 8106: @c doc-words-deferred
 8107: 
 8108: @c doc-mappedwordlist @c map-structure undefined, implemantation-specific
 8109: doc-root
 8110: doc-vocabulary
 8111: doc-seal
 8112: doc-vocs
 8113: doc-current
 8114: doc-context
 8115: 
 8116: 
 8117: @menu
 8118: * Vocabularies::                
 8119: * Why use word lists?::         
 8120: * Word list example::           
 8121: @end menu
 8122: 
 8123: @node Vocabularies, Why use word lists?, Word Lists, Word Lists
 8124: @subsection Vocabularies
 8125: @cindex Vocabularies, detailed explanation
 8126: 
 8127: Here is an example of creating and using a new wordlist using ANS
 8128: Forth words:
 8129: 
 8130: @example
 8131: wordlist constant my-new-words-wordlist
 8132: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
 8133: 
 8134: \ add it to the search order
 8135: also my-new-words
 8136: 
 8137: \ alternatively, add it to the search order and make it
 8138: \ the compilation word list
 8139: also my-new-words definitions
 8140: \ type "order" to see the problem
 8141: @end example
 8142: 
 8143: The problem with this example is that @code{order} has no way to
 8144: associate the name @code{my-new-words} with the wid of the word list (in
 8145: Gforth, @code{order} and @code{vocs} will display @code{???}  for a wid
 8146: that has no associated name). There is no Standard way of associating a
 8147: name with a wid.
 8148: 
 8149: In Gforth, this example can be re-coded using @code{vocabulary}, which
 8150: associates a name with a wid:
 8151: 
 8152: @example
 8153: vocabulary my-new-words
 8154: 
 8155: \ add it to the search order
 8156: also my-new-words
 8157: 
 8158: \ alternatively, add it to the search order and make it
 8159: \ the compilation word list
 8160: my-new-words definitions
 8161: \ type "order" to see that the problem is solved
 8162: @end example
 8163: 
 8164: 
 8165: @node Why use word lists?, Word list example, Vocabularies, Word Lists
 8166: @subsection Why use word lists?
 8167: @cindex word lists - why use them?
 8168: 
 8169: Here are some reasons why people use wordlists:
 8170: 
 8171: @itemize @bullet
 8172: 
 8173: @c anton: Gforth's hashing implementation makes the search speed
 8174: @c independent from the number of words.  But it is linear with the number
 8175: @c of wordlists that have to be searched, so in effect using more wordlists
 8176: @c actually slows down compilation.
 8177: 
 8178: @c @item
 8179: @c To improve compilation speed by reducing the number of header space
 8180: @c entries that must be searched. This is achieved by creating a new
 8181: @c word list that contains all of the definitions that are used in the
 8182: @c definition of a Forth system but which would not usually be used by
 8183: @c programs running on that system. That word list would be on the search
 8184: @c list when the Forth system was compiled but would be removed from the
 8185: @c search list for normal operation. This can be a useful technique for
 8186: @c low-performance systems (for example, 8-bit processors in embedded
 8187: @c systems) but is unlikely to be necessary in high-performance desktop
 8188: @c systems.
 8189: 
 8190: @item
 8191: To prevent a set of words from being used outside the context in which
 8192: they are valid. Two classic examples of this are an integrated editor
 8193: (all of the edit commands are defined in a separate word list; the
 8194: search order is set to the editor word list when the editor is invoked;
 8195: the old search order is restored when the editor is terminated) and an
 8196: integrated assembler (the op-codes for the machine are defined in a
 8197: separate word list which is used when a @code{CODE} word is defined).
 8198: 
 8199: @item
 8200: To organize the words of an application or library into a user-visible
 8201: set (in @code{forth-wordlist} or some other common wordlist) and a set
 8202: of helper words used just for the implementation (hidden in a separate
 8203: wordlist).  This keeps @code{words}' output smaller, separates
 8204: implementation and interface, and reduces the chance of name conflicts
 8205: within the common wordlist.
 8206: 
 8207: @item
 8208: To prevent a name-space clash between multiple definitions with the same
 8209: name. For example, when building a cross-compiler you might have a word
 8210: @code{IF} that generates conditional code for your target system. By
 8211: placing this definition in a different word list you can control whether
 8212: the host system's @code{IF} or the target system's @code{IF} get used in
 8213: any particular context by controlling the order of the word lists on the
 8214: search order stack.
 8215: 
 8216: @end itemize
 8217: 
 8218: The downsides of using wordlists are:
 8219: 
 8220: @itemize
 8221: 
 8222: @item
 8223: Debugging becomes more cumbersome.
 8224: 
 8225: @item
 8226: Name conflicts worked around with wordlists are still there, and you
 8227: have to arrange the search order carefully to get the desired results;
 8228: if you forget to do that, you get hard-to-find errors (as in any case
 8229: where you read the code differently from the compiler; @code{see} can
 8230: help seeing which of several possible words the name resolves to in such
 8231: cases).  @code{See} displays just the name of the words, not what
 8232: wordlist they belong to, so it might be misleading.  Using unique names
 8233: is a better approach to avoid name conflicts.
 8234: 
 8235: @item
 8236: You have to explicitly undo any changes to the search order.  In many
 8237: cases it would be more convenient if this happened implicitly.  Gforth
 8238: currently does not provide such a feature, but it may do so in the
 8239: future.
 8240: @end itemize
 8241: 
 8242: 
 8243: @node Word list example,  , Why use word lists?, Word Lists
 8244: @subsection Word list example
 8245: @cindex word lists - example
 8246: 
 8247: The following example is from the
 8248: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
 8249: garbage collector} and uses wordlists to separate public words from
 8250: helper words:
 8251: 
 8252: @example
 8253: get-current ( wid )
 8254: vocabulary garbage-collector also garbage-collector definitions
 8255: ... \ define helper words
 8256: ( wid ) set-current \ restore original (i.e., public) compilation wordlist
 8257: ... \ define the public (i.e., API) words
 8258:     \ they can refer to the helper words
 8259: previous \ restore original search order (helper words become invisible)
 8260: @end example
 8261: 
 8262: @c -------------------------------------------------------------
 8263: @node Environmental Queries, Files, Word Lists, Words
 8264: @section Environmental Queries
 8265: @cindex environmental queries
 8266: 
 8267: ANS Forth introduced the idea of ``environmental queries'' as a way
 8268: for a program running on a system to determine certain characteristics of the system.
 8269: The Standard specifies a number of strings that might be recognised by a system.
 8270: 
 8271: The Standard requires that the header space used for environmental queries
 8272: be distinct from the header space used for definitions.
 8273: 
 8274: Typically, environmental queries are supported by creating a set of
 8275: definitions in a word list that is @i{only} used during environmental
 8276: queries; that is what Gforth does. There is no Standard way of adding
 8277: definitions to the set of recognised environmental queries, but any
 8278: implementation that supports the loading of optional word sets must have
 8279: some mechanism for doing this (after loading the word set, the
 8280: associated environmental query string must return @code{true}). In
 8281: Gforth, the word list used to honour environmental queries can be
 8282: manipulated just like any other word list.
 8283: 
 8284: 
 8285: doc-environment?
 8286: doc-environment-wordlist
 8287: 
 8288: doc-gforth
 8289: doc-os-class
 8290: 
 8291: 
 8292: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
 8293: returning two items on the stack, querying it using @code{environment?}
 8294: will return an additional item; the @code{true} flag that shows that the
 8295: string was recognised.
 8296: 
 8297: @comment TODO Document the standard strings or note where they are documented herein
 8298: 
 8299: Here are some examples of using environmental queries:
 8300: 
 8301: @example
 8302: s" address-unit-bits" environment? 0=
 8303: [IF]
 8304:      cr .( environmental attribute address-units-bits unknown... ) cr
 8305: [ELSE]
 8306:      drop \ ensure balanced stack effect
 8307: [THEN]
 8308: 
 8309: \ this might occur in the prelude of a standard program that uses THROW
 8310: s" exception" environment? [IF]
 8311:    0= [IF]
 8312:       : throw abort" exception thrown" ;
 8313:    [THEN]
 8314: [ELSE] \ we don't know, so make sure
 8315:    : throw abort" exception thrown" ;
 8316: [THEN]
 8317: 
 8318: s" gforth" environment? [IF] .( Gforth version ) TYPE
 8319:                         [ELSE] .( Not Gforth..) [THEN]
 8320: 
 8321: \ a program using v*
 8322: s" gforth" environment? [IF]
 8323:   s" 0.5.0" compare 0< [IF] \ v* is a primitive since 0.5.0
 8324:    : v* ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r )
 8325:      >r swap 2swap swap 0e r> 0 ?DO
 8326:        dup f@ over + 2swap dup f@ f* f+ over + 2swap
 8327:      LOOP
 8328:      2drop 2drop ; 
 8329:   [THEN]
 8330: [ELSE] \ 
 8331:   : v* ( f_addr1 nstride1 f_addr2 nstride2 ucount -- r )
 8332:   ...
 8333: [THEN]
 8334: @end example
 8335: 
 8336: Here is an example of adding a definition to the environment word list:
 8337: 
 8338: @example
 8339: get-current environment-wordlist set-current
 8340: true constant block
 8341: true constant block-ext
 8342: set-current
 8343: @end example
 8344: 
 8345: You can see what definitions are in the environment word list like this:
 8346: 
 8347: @example
 8348: environment-wordlist >order words previous
 8349: @end example
 8350: 
 8351: 
 8352: @c -------------------------------------------------------------
 8353: @node Files, Blocks, Environmental Queries, Words
 8354: @section Files
 8355: @cindex files
 8356: @cindex I/O - file-handling
 8357: 
 8358: Gforth provides facilities for accessing files that are stored in the
 8359: host operating system's file-system. Files that are processed by Gforth
 8360: can be divided into two categories:
 8361: 
 8362: @itemize @bullet
 8363: @item
 8364: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
 8365: @item
 8366: Files that are processed by some other program (@dfn{general files}).
 8367: @end itemize
 8368: 
 8369: @menu
 8370: * Forth source files::          
 8371: * General files::               
 8372: * Search Paths::                
 8373: @end menu
 8374: 
 8375: @c -------------------------------------------------------------
 8376: @node Forth source files, General files, Files, Files
 8377: @subsection Forth source files
 8378: @cindex including files
 8379: @cindex Forth source files
 8380: 
 8381: The simplest way to interpret the contents of a file is to use one of
 8382: these two formats:
 8383: 
 8384: @example
 8385: include mysource.fs
 8386: s" mysource.fs" included
 8387: @end example
 8388: 
 8389: You usually want to include a file only if it is not included already
 8390: (by, say, another source file). In that case, you can use one of these
 8391: three formats:
 8392: 
 8393: @example
 8394: require mysource.fs
 8395: needs mysource.fs
 8396: s" mysource.fs" required
 8397: @end example
 8398: 
 8399: @cindex stack effect of included files
 8400: @cindex including files, stack effect
 8401: It is good practice to write your source files such that interpreting them
 8402: does not change the stack. Source files designed in this way can be used with
 8403: @code{required} and friends without complications. For example:
 8404: 
 8405: @example
 8406: 1024 require foo.fs drop
 8407: @end example
 8408: 
 8409: Here you want to pass the argument 1024 (e.g., a buffer size) to
 8410: @file{foo.fs}.  Interpreting @file{foo.fs} has the stack effect ( n -- n
 8411: ), which allows its use with @code{require}.  Of course with such
 8412: parameters to required files, you have to ensure that the first
 8413: @code{require} fits for all uses (i.e., @code{require} it early in the
 8414: master load file).
 8415: 
 8416: doc-include-file
 8417: doc-included
 8418: doc-included?
 8419: doc-include
 8420: doc-required
 8421: doc-require
 8422: doc-needs
 8423: @c doc-init-included-files @c internal
 8424: doc-sourcefilename
 8425: doc-sourceline#
 8426: 
 8427: A definition in ANS Forth for @code{required} is provided in
 8428: @file{compat/required.fs}.
 8429: 
 8430: @c -------------------------------------------------------------
 8431: @node General files, Search Paths, Forth source files, Files
 8432: @subsection General files
 8433: @cindex general files
 8434: @cindex file-handling
 8435: 
 8436: Files are opened/created by name and type. The following file access
 8437: methods (FAMs) are recognised:
 8438: 
 8439: @cindex fam (file access method)
 8440: doc-r/o
 8441: doc-r/w
 8442: doc-w/o
 8443: doc-bin
 8444: 
 8445: 
 8446: When a file is opened/created, it returns a file identifier,
 8447: @i{wfileid} that is used for all other file commands. All file
 8448: commands also return a status value, @i{wior}, that is 0 for a
 8449: successful operation and an implementation-defined non-zero value in the
 8450: case of an error.
 8451: 
 8452: 
 8453: doc-open-file
 8454: doc-create-file
 8455: 
 8456: doc-close-file
 8457: doc-delete-file
 8458: doc-rename-file
 8459: doc-read-file
 8460: doc-read-line
 8461: doc-write-file
 8462: doc-write-line
 8463: doc-emit-file
 8464: doc-flush-file
 8465: 
 8466: doc-file-status
 8467: doc-file-position
 8468: doc-reposition-file
 8469: doc-file-size
 8470: doc-resize-file
 8471: 
 8472: doc-slurp-file
 8473: doc-slurp-fid
 8474: 
 8475: @c ---------------------------------------------------------
 8476: @node Search Paths,  , General files, Files
 8477: @subsection Search Paths
 8478: @cindex path for @code{included}
 8479: @cindex file search path
 8480: @cindex @code{include} search path
 8481: @cindex search path for files
 8482: 
 8483: If you specify an absolute filename (i.e., a filename starting with
 8484: @file{/} or @file{~}, or with @file{:} in the second position (as in
 8485: @samp{C:...})) for @code{included} and friends, that file is included
 8486: just as you would expect.
 8487: 
 8488: If the filename starts with @file{./}, this refers to the directory that
 8489: the present file was @code{included} from.  This allows files to include
 8490: other files relative to their own position (irrespective of the current
 8491: working directory or the absolute position).  This feature is essential
 8492: for libraries consisting of several files, where a file may include
 8493: other files from the library.  It corresponds to @code{#include "..."}
 8494: in C. If the current input source is not a file, @file{.} refers to the
 8495: directory of the innermost file being included, or, if there is no file
 8496: being included, to the current working directory.
 8497: 
 8498: For relative filenames (not starting with @file{./}), Gforth uses a
 8499: search path similar to Forth's search order (@pxref{Word Lists}). It
 8500: tries to find the given filename in the directories present in the path,
 8501: and includes the first one it finds. There are separate search paths for
 8502: Forth source files and general files.  If the search path contains the
 8503: directory @file{.}, this refers to the directory of the current file, or
 8504: the working directory, as if the file had been specified with @file{./}.
 8505: 
 8506: Use @file{~+} to refer to the current working directory (as in the
 8507: @code{bash}).
 8508: 
 8509: @c anton: fold the following subsubsections into this subsection?
 8510: 
 8511: @menu
 8512: * Source Search Paths::         
 8513: * General Search Paths::        
 8514: @end menu
 8515: 
 8516: @c ---------------------------------------------------------
 8517: @node Source Search Paths, General Search Paths, Search Paths, Search Paths
 8518: @subsubsection Source Search Paths
 8519: @cindex search path control, source files
 8520: 
 8521: The search path is initialized when you start Gforth (@pxref{Invoking
 8522: Gforth}). You can display it and change it using @code{fpath} in
 8523: combination with the general path handling words.
 8524: 
 8525: doc-fpath
 8526: @c the functionality of the following words is easily available through
 8527: @c   fpath and the general path words.  The may go away.
 8528: @c doc-.fpath
 8529: @c doc-fpath+
 8530: @c doc-fpath=
 8531: @c doc-open-fpath-file
 8532: 
 8533: @noindent
 8534: Here is an example of using @code{fpath} and @code{require}:
 8535: 
 8536: @example
 8537: fpath path= /usr/lib/forth/|./
 8538: require timer.fs
 8539: @end example
 8540: 
 8541: 
 8542: @c ---------------------------------------------------------
 8543: @node General Search Paths,  , Source Search Paths, Search Paths
 8544: @subsubsection General Search Paths
 8545: @cindex search path control, source files
 8546: 
 8547: Your application may need to search files in several directories, like
 8548: @code{included} does. To facilitate this, Gforth allows you to define
 8549: and use your own search paths, by providing generic equivalents of the
 8550: Forth search path words:
 8551: 
 8552: doc-open-path-file
 8553: doc-path-allot
 8554: doc-clear-path
 8555: doc-also-path
 8556: doc-.path
 8557: doc-path+
 8558: doc-path=
 8559: 
 8560: @c anton: better define a word for it, say "path-allot ( ucount -- path-addr )
 8561: 
 8562: Here's an example of creating an empty search path:
 8563: @c
 8564: @example
 8565: create mypath 500 path-allot \ maximum length 500 chars (is checked)
 8566: @end example
 8567: 
 8568: @c -------------------------------------------------------------
 8569: @node Blocks, Other I/O, Files, Words
 8570: @section Blocks
 8571: @cindex I/O - blocks
 8572: @cindex blocks
 8573: 
 8574: When you run Gforth on a modern desk-top computer, it runs under the
 8575: control of an operating system which provides certain services.  One of
 8576: these services is @var{file services}, which allows Forth source code
 8577: and data to be stored in files and read into Gforth (@pxref{Files}).
 8578: 
 8579: Traditionally, Forth has been an important programming language on
 8580: systems where it has interfaced directly to the underlying hardware with
 8581: no intervening operating system. Forth provides a mechanism, called
 8582: @dfn{blocks}, for accessing mass storage on such systems.
 8583: 
 8584: A block is a 1024-byte data area, which can be used to hold data or
 8585: Forth source code. No structure is imposed on the contents of the
 8586: block. A block is identified by its number; blocks are numbered
 8587: contiguously from 1 to an implementation-defined maximum.
 8588: 
 8589: A typical system that used blocks but no operating system might use a
 8590: single floppy-disk drive for mass storage, with the disks formatted to
 8591: provide 256-byte sectors. Blocks would be implemented by assigning the
 8592: first four sectors of the disk to block 1, the second four sectors to
 8593: block 2 and so on, up to the limit of the capacity of the disk. The disk
 8594: would not contain any file system information, just the set of blocks.
 8595: 
 8596: @cindex blocks file
 8597: On systems that do provide file services, blocks are typically
 8598: implemented by storing a sequence of blocks within a single @dfn{blocks
 8599: file}.  The size of the blocks file will be an exact multiple of 1024
 8600: bytes, corresponding to the number of blocks it contains. This is the
 8601: mechanism that Gforth uses.
 8602: 
 8603: @cindex @file{blocks.fb}
 8604: Only one blocks file can be open at a time. If you use block words without
 8605: having specified a blocks file, Gforth defaults to the blocks file
 8606: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
 8607: locate a blocks file (@pxref{Source Search Paths}).
 8608: 
 8609: @cindex block buffers
 8610: When you read and write blocks under program control, Gforth uses a
 8611: number of @dfn{block buffers} as intermediate storage. These buffers are
 8612: not used when you use @code{load} to interpret the contents of a block.
 8613: 
 8614: The behaviour of the block buffers is analagous to that of a cache.
 8615: Each block buffer has three states:
 8616: 
 8617: @itemize @bullet
 8618: @item
 8619: Unassigned
 8620: @item
 8621: Assigned-clean
 8622: @item
 8623: Assigned-dirty
 8624: @end itemize
 8625: 
 8626: Initially, all block buffers are @i{unassigned}. In order to access a
 8627: block, the block (specified by its block number) must be assigned to a
 8628: block buffer.
 8629: 
 8630: The assignment of a block to a block buffer is performed by @code{block}
 8631: or @code{buffer}. Use @code{block} when you wish to modify the existing
 8632: contents of a block. Use @code{buffer} when you don't care about the
 8633: existing contents of the block@footnote{The ANS Forth definition of
 8634: @code{buffer} is intended not to cause disk I/O; if the data associated
 8635: with the particular block is already stored in a block buffer due to an
 8636: earlier @code{block} command, @code{buffer} will return that block
 8637: buffer and the existing contents of the block will be
 8638: available. Otherwise, @code{buffer} will simply assign a new, empty
 8639: block buffer for the block.}.
 8640: 
 8641: Once a block has been assigned to a block buffer using @code{block} or
 8642: @code{buffer}, that block buffer becomes the @i{current block
 8643: buffer}. Data may only be manipulated (read or written) within the
 8644: current block buffer.
 8645: 
 8646: When the contents of the current block buffer has been modified it is
 8647: necessary, @emph{before calling @code{block} or @code{buffer} again}, to
 8648: either abandon the changes (by doing nothing) or mark the block as
 8649: changed (assigned-dirty), using @code{update}. Using @code{update} does
 8650: not change the blocks file; it simply changes a block buffer's state to
 8651: @i{assigned-dirty}.  The block will be written implicitly when it's
 8652: buffer is needed for another block, or explicitly by @code{flush} or
 8653: @code{save-buffers}.
 8654: 
 8655: word @code{Flush} writes all @i{assigned-dirty} blocks back to the
 8656: blocks file on disk. Leaving Gforth with @code{bye} also performs a
 8657: @code{flush}.
 8658: 
 8659: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
 8660: algorithm to assign a block buffer to a block. That means that any
 8661: particular block can only be assigned to one specific block buffer,
 8662: called (for the particular operation) the @i{victim buffer}. If the
 8663: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
 8664: the new block immediately. If it is @i{assigned-dirty} its current
 8665: contents are written back to the blocks file on disk before it is
 8666: allocated to the new block.
 8667: 
 8668: Although no structure is imposed on the contents of a block, it is
 8669: traditional to display the contents as 16 lines each of 64 characters.  A
 8670: block provides a single, continuous stream of input (for example, it
 8671: acts as a single parse area) -- there are no end-of-line characters
 8672: within a block, and no end-of-file character at the end of a
 8673: block. There are two consequences of this:
 8674: 
 8675: @itemize @bullet
 8676: @item
 8677: The last character of one line wraps straight into the first character
 8678: of the following line
 8679: @item
 8680: The word @code{\} -- comment to end of line -- requires special
 8681: treatment; in the context of a block it causes all characters until the
 8682: end of the current 64-character ``line'' to be ignored.
 8683: @end itemize
 8684: 
 8685: In Gforth, when you use @code{block} with a non-existent block number,
 8686: the current blocks file will be extended to the appropriate size and the
 8687: block buffer will be initialised with spaces.
 8688: 
 8689: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
 8690: for details) but doesn't encourage the use of blocks; the mechanism is
 8691: only provided for backward compatibility -- ANS Forth requires blocks to
 8692: be available when files are.
 8693: 
 8694: Common techniques that are used when working with blocks include:
 8695: 
 8696: @itemize @bullet
 8697: @item
 8698: A screen editor that allows you to edit blocks without leaving the Forth
 8699: environment.
 8700: @item
 8701: Shadow screens; where every code block has an associated block
 8702: containing comments (for example: code in odd block numbers, comments in
 8703: even block numbers). Typically, the block editor provides a convenient
 8704: mechanism to toggle between code and comments.
 8705: @item
 8706: Load blocks; a single block (typically block 1) contains a number of
 8707: @code{thru} commands which @code{load} the whole of the application.
 8708: @end itemize
 8709: 
 8710: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
 8711: integrated into a Forth programming environment.
 8712: 
 8713: @comment TODO what about errors on open-blocks?
 8714: 
 8715: doc-open-blocks
 8716: doc-use
 8717: doc-block-offset
 8718: doc-get-block-fid
 8719: doc-block-position
 8720: 
 8721: doc-list
 8722: doc-scr
 8723: 
 8724: doc---gforthman-block
 8725: doc-buffer
 8726: 
 8727: doc-empty-buffers
 8728: doc-empty-buffer
 8729: doc-update
 8730: doc-updated?
 8731: doc-save-buffers
 8732: doc-save-buffer
 8733: doc-flush
 8734: 
 8735: doc-load
 8736: doc-thru
 8737: doc-+load
 8738: doc-+thru
 8739: doc---gforthman--->
 8740: doc-block-included
 8741: 
 8742: 
 8743: @c -------------------------------------------------------------
 8744: @node Other I/O, Locals, Blocks, Words
 8745: @section Other I/O
 8746: @cindex I/O - keyboard and display
 8747: 
 8748: @menu
 8749: * Simple numeric output::       Predefined formats
 8750: * Formatted numeric output::    Formatted (pictured) output
 8751: * String Formats::              How Forth stores strings in memory
 8752: * Displaying characters and strings::  Other stuff
 8753: * Input::                       Input
 8754: @end menu
 8755: 
 8756: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
 8757: @subsection Simple numeric output
 8758: @cindex numeric output - simple/free-format
 8759: 
 8760: The simplest output functions are those that display numbers from the
 8761: data or floating-point stacks. Floating-point output is always displayed
 8762: using base 10. Numbers displayed from the data stack use the value stored
 8763: in @code{base}.
 8764: 
 8765: 
 8766: doc-.
 8767: doc-dec.
 8768: doc-hex.
 8769: doc-u.
 8770: doc-.r
 8771: doc-u.r
 8772: doc-d.
 8773: doc-ud.
 8774: doc-d.r
 8775: doc-ud.r
 8776: doc-f.
 8777: doc-fe.
 8778: doc-fs.
 8779: 
 8780: 
 8781: Examples of printing the number 1234.5678E23 in the different floating-point output
 8782: formats are shown below:
 8783: 
 8784: @example
 8785: f. 123456779999999000000000000.
 8786: fe. 123.456779999999E24
 8787: fs. 1.23456779999999E26
 8788: @end example
 8789: 
 8790: 
 8791: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
 8792: @subsection Formatted numeric output
 8793: @cindex formatted numeric output
 8794: @cindex pictured numeric output
 8795: @cindex numeric output - formatted
 8796: 
 8797: Forth traditionally uses a technique called @dfn{pictured numeric
 8798: output} for formatted printing of integers.  In this technique, digits
 8799: are extracted from the number (using the current output radix defined by
 8800: @code{base}), converted to ASCII codes and appended to a string that is
 8801: built in a scratch-pad area of memory (@pxref{core-idef,
 8802: Implementation-defined options, Implementation-defined
 8803: options}). Arbitrary characters can be appended to the string during the
 8804: extraction process. The completed string is specified by an address
 8805: and length and can be manipulated (@code{TYPE}ed, copied, modified)
 8806: under program control.
 8807: 
 8808: All of the integer output words described in the previous section
 8809: (@pxref{Simple numeric output}) are implemented in Gforth using pictured
 8810: numeric output.
 8811: 
 8812: Three important things to remember about pictured numeric output:
 8813: 
 8814: @itemize @bullet
 8815: @item
 8816: It always operates on double-precision numbers; to display a
 8817: single-precision number, convert it first (for ways of doing this
 8818: @pxref{Double precision}).
 8819: @item
 8820: It always treats the double-precision number as though it were
 8821: unsigned. The examples below show ways of printing signed numbers.
 8822: @item
 8823: The string is built up from right to left; least significant digit first.
 8824: @end itemize
 8825: 
 8826: 
 8827: doc-<#
 8828: doc-<<#
 8829: doc-#
 8830: doc-#s
 8831: doc-hold
 8832: doc-sign
 8833: doc-#>
 8834: doc-#>>
 8835: 
 8836: doc-represent
 8837: 
 8838: 
 8839: @noindent
 8840: Here are some examples of using pictured numeric output:
 8841: 
 8842: @example
 8843: : my-u. ( u -- )
 8844:   \ Simplest use of pns.. behaves like Standard u. 
 8845:   0              \ convert to unsigned double
 8846:   <<#            \ start conversion
 8847:   #s             \ convert all digits
 8848:   #>             \ complete conversion
 8849:   TYPE SPACE     \ display, with trailing space
 8850:   #>> ;          \ release hold area
 8851: 
 8852: : cents-only ( u -- )
 8853:   0              \ convert to unsigned double
 8854:   <<#            \ start conversion
 8855:   # #            \ convert two least-significant digits
 8856:   #>             \ complete conversion, discard other digits
 8857:   TYPE SPACE     \ display, with trailing space
 8858:   #>> ;          \ release hold area
 8859: 
 8860: : dollars-and-cents ( u -- )
 8861:   0              \ convert to unsigned double
 8862:   <<#            \ start conversion
 8863:   # #            \ convert two least-significant digits
 8864:   [char] . hold  \ insert decimal point
 8865:   #s             \ convert remaining digits
 8866:   [char] $ hold  \ append currency symbol
 8867:   #>             \ complete conversion
 8868:   TYPE SPACE     \ display, with trailing space
 8869:   #>> ;          \ release hold area
 8870: 
 8871: : my-. ( n -- )
 8872:   \ handling negatives.. behaves like Standard .
 8873:   s>d            \ convert to signed double
 8874:   swap over dabs \ leave sign byte followed by unsigned double
 8875:   <<#            \ start conversion
 8876:   #s             \ convert all digits
 8877:   rot sign       \ get at sign byte, append "-" if needed
 8878:   #>             \ complete conversion
 8879:   TYPE SPACE     \ display, with trailing space
 8880:   #>> ;          \ release hold area
 8881: 
 8882: : account. ( n -- )
 8883:   \ accountants don't like minus signs, they use parentheses
 8884:   \ for negative numbers
 8885:   s>d            \ convert to signed double
 8886:   swap over dabs \ leave sign byte followed by unsigned double
 8887:   <<#            \ start conversion
 8888:   2 pick         \ get copy of sign byte
 8889:   0< IF [char] ) hold THEN \ right-most character of output
 8890:   #s             \ convert all digits
 8891:   rot            \ get at sign byte
 8892:   0< IF [char] ( hold THEN
 8893:   #>             \ complete conversion
 8894:   TYPE SPACE     \ display, with trailing space
 8895:   #>> ;          \ release hold area
 8896: 
 8897: @end example
 8898: 
 8899: Here are some examples of using these words:
 8900: 
 8901: @example
 8902: 1 my-u. 1
 8903: hex -1 my-u. decimal FFFFFFFF
 8904: 1 cents-only 01
 8905: 1234 cents-only 34
 8906: 2 dollars-and-cents $0.02
 8907: 1234 dollars-and-cents $12.34
 8908: 123 my-. 123
 8909: -123 my. -123
 8910: 123 account. 123
 8911: -456 account. (456)
 8912: @end example
 8913: 
 8914: 
 8915: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
 8916: @subsection String Formats
 8917: @cindex strings - see character strings
 8918: @cindex character strings - formats
 8919: @cindex I/O - see character strings
 8920: @cindex counted strings
 8921: 
 8922: @c anton: this does not really belong here; maybe the memory section,
 8923: @c  or the principles chapter
 8924: 
 8925: Forth commonly uses two different methods for representing character
 8926: strings:
 8927: 
 8928: @itemize @bullet
 8929: @item
 8930: @cindex address of counted string
 8931: @cindex counted string
 8932: As a @dfn{counted string}, represented by a @i{c-addr}. The char
 8933: addressed by @i{c-addr} contains a character-count, @i{n}, of the
 8934: string and the string occupies the subsequent @i{n} char addresses in
 8935: memory.
 8936: @item
 8937: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
 8938: of the string in characters, and @i{c-addr} is the address of the
 8939: first byte of the string.
 8940: @end itemize
 8941: 
 8942: ANS Forth encourages the use of the second format when representing
 8943: strings.
 8944: 
 8945: 
 8946: doc-count
 8947: 
 8948: 
 8949: For words that move, copy and search for strings see @ref{Memory
 8950: Blocks}. For words that display characters and strings see
 8951: @ref{Displaying characters and strings}.
 8952: 
 8953: @node Displaying characters and strings, Input, String Formats, Other I/O
 8954: @subsection Displaying characters and strings
 8955: @cindex characters - compiling and displaying
 8956: @cindex character strings - compiling and displaying
 8957: 
 8958: This section starts with a glossary of Forth words and ends with a set
 8959: of examples.
 8960: 
 8961: 
 8962: doc-bl
 8963: doc-space
 8964: doc-spaces
 8965: doc-emit
 8966: doc-toupper
 8967: doc-."
 8968: doc-.(
 8969: doc-.\"
 8970: doc-type
 8971: doc-typewhite
 8972: doc-cr
 8973: @cindex cursor control
 8974: doc-at-xy
 8975: doc-page
 8976: doc-s"
 8977: doc-s\"
 8978: doc-c"
 8979: doc-char
 8980: doc-[char]
 8981: 
 8982: 
 8983: @noindent
 8984: As an example, consider the following text, stored in a file @file{test.fs}:
 8985: 
 8986: @example
 8987: .( text-1)
 8988: : my-word
 8989:   ." text-2" cr
 8990:   .( text-3)
 8991: ;
 8992: 
 8993: ." text-4"
 8994: 
 8995: : my-char
 8996:   [char] ALPHABET emit
 8997:   char emit
 8998: ;
 8999: @end example
 9000: 
 9001: When you load this code into Gforth, the following output is generated:
 9002: 
 9003: @example
 9004: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
 9005: @end example
 9006: 
 9007: @itemize @bullet
 9008: @item
 9009: Messages @code{text-1} and @code{text-3} are displayed because @code{.(} 
 9010: is an immediate word; it behaves in the same way whether it is used inside
 9011: or outside a colon definition.
 9012: @item
 9013: Message @code{text-4} is displayed because of Gforth's added interpretation
 9014: semantics for @code{."}.
 9015: @item
 9016: Message @code{text-2} is @i{not} displayed, because the text interpreter
 9017: performs the compilation semantics for @code{."} within the definition of
 9018: @code{my-word}.
 9019: @end itemize
 9020: 
 9021: Here are some examples of executing @code{my-word} and @code{my-char}:
 9022: 
 9023: @example
 9024: @kbd{my-word @key{RET}} text-2
 9025:  ok
 9026: @kbd{my-char fred @key{RET}} Af ok
 9027: @kbd{my-char jim @key{RET}} Aj ok
 9028: @end example
 9029: 
 9030: @itemize @bullet
 9031: @item
 9032: Message @code{text-2} is displayed because of the run-time behaviour of
 9033: @code{."}.
 9034: @item
 9035: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
 9036: on the stack at run-time. @code{emit} always displays the character
 9037: when @code{my-char} is executed.
 9038: @item
 9039: @code{char} parses a string at run-time and the second @code{emit} displays
 9040: the first character of the string.
 9041: @item
 9042: If you type @code{see my-char} you can see that @code{[char]} discarded
 9043: the text ``LPHABET'' and only compiled the display code for ``A'' into the
 9044: definition of @code{my-char}.
 9045: @end itemize
 9046: 
 9047: 
 9048: 
 9049: @node Input,  , Displaying characters and strings, Other I/O
 9050: @subsection Input
 9051: @cindex input
 9052: @cindex I/O - see input
 9053: @cindex parsing a string
 9054: 
 9055: For ways of storing character strings in memory see @ref{String Formats}.
 9056: 
 9057: @comment TODO examples for >number >float accept key key? pad parse word refill
 9058: @comment then index them
 9059: 
 9060: 
 9061: doc-key
 9062: doc-key?
 9063: doc-ekey
 9064: doc-ekey?
 9065: doc-ekey>char
 9066: doc->number
 9067: doc->float
 9068: doc-accept
 9069: doc-pad
 9070: @c anton: these belong in the input stream section
 9071: doc-parse
 9072: doc-word
 9073: doc-name
 9074: doc-parse-word
 9075: doc-\"-parse
 9076: doc-sword
 9077: doc-refill
 9078: @comment obsolescent words..
 9079: doc-convert
 9080: doc-query
 9081: doc-expect
 9082: doc-span
 9083: 
 9084: 
 9085: @c -------------------------------------------------------------
 9086: @node Locals, Structures, Other I/O, Words
 9087: @section Locals
 9088: @cindex locals
 9089: 
 9090: Local variables can make Forth programming more enjoyable and Forth
 9091: programs easier to read. Unfortunately, the locals of ANS Forth are
 9092: laden with restrictions. Therefore, we provide not only the ANS Forth
 9093: locals wordset, but also our own, more powerful locals wordset (we
 9094: implemented the ANS Forth locals wordset through our locals wordset).
 9095: 
 9096: The ideas in this section have also been published in M. Anton Ertl,
 9097: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz,
 9098: Automatic Scoping of Local Variables}}, EuroForth '94.
 9099: 
 9100: @menu
 9101: * Gforth locals::               
 9102: * ANS Forth locals::            
 9103: @end menu
 9104: 
 9105: @node Gforth locals, ANS Forth locals, Locals, Locals
 9106: @subsection Gforth locals
 9107: @cindex Gforth locals
 9108: @cindex locals, Gforth style
 9109: 
 9110: Locals can be defined with
 9111: 
 9112: @example
 9113: @{ local1 local2 ... -- comment @}
 9114: @end example
 9115: or
 9116: @example
 9117: @{ local1 local2 ... @}
 9118: @end example
 9119: 
 9120: E.g.,
 9121: @example
 9122: : max @{ n1 n2 -- n3 @}
 9123:  n1 n2 > if
 9124:    n1
 9125:  else
 9126:    n2
 9127:  endif ;
 9128: @end example
 9129: 
 9130: The similarity of locals definitions with stack comments is intended. A
 9131: locals definition often replaces the stack comment of a word. The order
 9132: of the locals corresponds to the order in a stack comment and everything
 9133: after the @code{--} is really a comment.
 9134: 
 9135: This similarity has one disadvantage: It is too easy to confuse locals
 9136: declarations with stack comments, causing bugs and making them hard to
 9137: find. However, this problem can be avoided by appropriate coding
 9138: conventions: Do not use both notations in the same program. If you do,
 9139: they should be distinguished using additional means, e.g. by position.
 9140: 
 9141: @cindex types of locals
 9142: @cindex locals types
 9143: The name of the local may be preceded by a type specifier, e.g.,
 9144: @code{F:} for a floating point value:
 9145: 
 9146: @example
 9147: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
 9148: \ complex multiplication
 9149:  Ar Br f* Ai Bi f* f-
 9150:  Ar Bi f* Ai Br f* f+ ;
 9151: @end example
 9152: 
 9153: @cindex flavours of locals
 9154: @cindex locals flavours
 9155: @cindex value-flavoured locals
 9156: @cindex variable-flavoured locals
 9157: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
 9158: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
 9159: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
 9160: with @code{W:}, @code{D:} etc.) produces its value and can be changed
 9161: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
 9162: produces its address (which becomes invalid when the variable's scope is
 9163: left). E.g., the standard word @code{emit} can be defined in terms of
 9164: @code{type} like this:
 9165: 
 9166: @example
 9167: : emit @{ C^ char* -- @}
 9168:     char* 1 type ;
 9169: @end example
 9170: 
 9171: @cindex default type of locals
 9172: @cindex locals, default type
 9173: A local without type specifier is a @code{W:} local. Both flavours of
 9174: locals are initialized with values from the data or FP stack.
 9175: 
 9176: Currently there is no way to define locals with user-defined data
 9177: structures, but we are working on it.
 9178: 
 9179: Gforth allows defining locals everywhere in a colon definition. This
 9180: poses the following questions:
 9181: 
 9182: @menu
 9183: * Where are locals visible by name?::  
 9184: * How long do locals live?::    
 9185: * Locals programming style::    
 9186: * Locals implementation::       
 9187: @end menu
 9188: 
 9189: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
 9190: @subsubsection Where are locals visible by name?
 9191: @cindex locals visibility
 9192: @cindex visibility of locals
 9193: @cindex scope of locals
 9194: 
 9195: Basically, the answer is that locals are visible where you would expect
 9196: it in block-structured languages, and sometimes a little longer. If you
 9197: want to restrict the scope of a local, enclose its definition in
 9198: @code{SCOPE}...@code{ENDSCOPE}.
 9199: 
 9200: 
 9201: doc-scope
 9202: doc-endscope
 9203: 
 9204: 
 9205: These words behave like control structure words, so you can use them
 9206: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
 9207: arbitrary ways.
 9208: 
 9209: If you want a more exact answer to the visibility question, here's the
 9210: basic principle: A local is visible in all places that can only be
 9211: reached through the definition of the local@footnote{In compiler
 9212: construction terminology, all places dominated by the definition of the
 9213: local.}. In other words, it is not visible in places that can be reached
 9214: without going through the definition of the local. E.g., locals defined
 9215: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
 9216: defined in @code{BEGIN}...@code{UNTIL} are visible after the
 9217: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
 9218: 
 9219: The reasoning behind this solution is: We want to have the locals
 9220: visible as long as it is meaningful. The user can always make the
 9221: visibility shorter by using explicit scoping. In a place that can
 9222: only be reached through the definition of a local, the meaning of a
 9223: local name is clear. In other places it is not: How is the local
 9224: initialized at the control flow path that does not contain the
 9225: definition? Which local is meant, if the same name is defined twice in
 9226: two independent control flow paths?
 9227: 
 9228: This should be enough detail for nearly all users, so you can skip the
 9229: rest of this section. If you really must know all the gory details and
 9230: options, read on.
 9231: 
 9232: In order to implement this rule, the compiler has to know which places
 9233: are unreachable. It knows this automatically after @code{AHEAD},
 9234: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
 9235: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
 9236: compiler that the control flow never reaches that place. If
 9237: @code{UNREACHABLE} is not used where it could, the only consequence is
 9238: that the visibility of some locals is more limited than the rule above
 9239: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
 9240: lie to the compiler), buggy code will be produced.
 9241: 
 9242: 
 9243: doc-unreachable
 9244: 
 9245: 
 9246: Another problem with this rule is that at @code{BEGIN}, the compiler
 9247: does not know which locals will be visible on the incoming
 9248: back-edge. All problems discussed in the following are due to this
 9249: ignorance of the compiler (we discuss the problems using @code{BEGIN}
 9250: loops as examples; the discussion also applies to @code{?DO} and other
 9251: loops). Perhaps the most insidious example is:
 9252: @example
 9253: AHEAD
 9254: BEGIN
 9255:   x
 9256: [ 1 CS-ROLL ] THEN
 9257:   @{ x @}
 9258:   ...
 9259: UNTIL
 9260: @end example
 9261: 
 9262: This should be legal according to the visibility rule. The use of
 9263: @code{x} can only be reached through the definition; but that appears
 9264: textually below the use.
 9265: 
 9266: From this example it is clear that the visibility rules cannot be fully
 9267: implemented without major headaches. Our implementation treats common
 9268: cases as advertised and the exceptions are treated in a safe way: The
 9269: compiler makes a reasonable guess about the locals visible after a
 9270: @code{BEGIN}; if it is too pessimistic, the
 9271: user will get a spurious error about the local not being defined; if the
 9272: compiler is too optimistic, it will notice this later and issue a
 9273: warning. In the case above the compiler would complain about @code{x}
 9274: being undefined at its use. You can see from the obscure examples in
 9275: this section that it takes quite unusual control structures to get the
 9276: compiler into trouble, and even then it will often do fine.
 9277: 
 9278: If the @code{BEGIN} is reachable from above, the most optimistic guess
 9279: is that all locals visible before the @code{BEGIN} will also be
 9280: visible after the @code{BEGIN}. This guess is valid for all loops that
 9281: are entered only through the @code{BEGIN}, in particular, for normal
 9282: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
 9283: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
 9284: compiler. When the branch to the @code{BEGIN} is finally generated by
 9285: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
 9286: warns the user if it was too optimistic:
 9287: @example
 9288: IF
 9289:   @{ x @}
 9290: BEGIN
 9291:   \ x ? 
 9292: [ 1 cs-roll ] THEN
 9293:   ...
 9294: UNTIL
 9295: @end example
 9296: 
 9297: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
 9298: optimistically assumes that it lives until the @code{THEN}. It notices
 9299: this difference when it compiles the @code{UNTIL} and issues a
 9300: warning. The user can avoid the warning, and make sure that @code{x}
 9301: is not used in the wrong area by using explicit scoping:
 9302: @example
 9303: IF
 9304:   SCOPE
 9305:   @{ x @}
 9306:   ENDSCOPE
 9307: BEGIN
 9308: [ 1 cs-roll ] THEN
 9309:   ...
 9310: UNTIL
 9311: @end example
 9312: 
 9313: Since the guess is optimistic, there will be no spurious error messages
 9314: about undefined locals.
 9315: 
 9316: If the @code{BEGIN} is not reachable from above (e.g., after
 9317: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
 9318: optimistic guess, as the locals visible after the @code{BEGIN} may be
 9319: defined later. Therefore, the compiler assumes that no locals are
 9320: visible after the @code{BEGIN}. However, the user can use
 9321: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
 9322: visible at the BEGIN as at the point where the top control-flow stack
 9323: item was created.
 9324: 
 9325: 
 9326: doc-assume-live
 9327: 
 9328: 
 9329: @noindent
 9330: E.g.,
 9331: @example
 9332: @{ x @}
 9333: AHEAD
 9334: ASSUME-LIVE
 9335: BEGIN
 9336:   x
 9337: [ 1 CS-ROLL ] THEN
 9338:   ...
 9339: UNTIL
 9340: @end example
 9341: 
 9342: Other cases where the locals are defined before the @code{BEGIN} can be
 9343: handled by inserting an appropriate @code{CS-ROLL} before the
 9344: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
 9345: behind the @code{ASSUME-LIVE}).
 9346: 
 9347: Cases where locals are defined after the @code{BEGIN} (but should be
 9348: visible immediately after the @code{BEGIN}) can only be handled by
 9349: rearranging the loop. E.g., the ``most insidious'' example above can be
 9350: arranged into:
 9351: @example
 9352: BEGIN
 9353:   @{ x @}
 9354:   ... 0=
 9355: WHILE
 9356:   x
 9357: REPEAT
 9358: @end example
 9359: 
 9360: @node How long do locals live?, Locals programming style, Where are locals visible by name?, Gforth locals
 9361: @subsubsection How long do locals live?
 9362: @cindex locals lifetime
 9363: @cindex lifetime of locals
 9364: 
 9365: The right answer for the lifetime question would be: A local lives at
 9366: least as long as it can be accessed. For a value-flavoured local this
 9367: means: until the end of its visibility. However, a variable-flavoured
 9368: local could be accessed through its address far beyond its visibility
 9369: scope. Ultimately, this would mean that such locals would have to be
 9370: garbage collected. Since this entails un-Forth-like implementation
 9371: complexities, I adopted the same cowardly solution as some other
 9372: languages (e.g., C): The local lives only as long as it is visible;
 9373: afterwards its address is invalid (and programs that access it
 9374: afterwards are erroneous).
 9375: 
 9376: @node Locals programming style, Locals implementation, How long do locals live?, Gforth locals
 9377: @subsubsection Locals programming style
 9378: @cindex locals programming style
 9379: @cindex programming style, locals
 9380: 
 9381: The freedom to define locals anywhere has the potential to change
 9382: programming styles dramatically. In particular, the need to use the
 9383: return stack for intermediate storage vanishes. Moreover, all stack
 9384: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
 9385: determined arguments) can be eliminated: If the stack items are in the
 9386: wrong order, just write a locals definition for all of them; then
 9387: write the items in the order you want.
 9388: 
 9389: This seems a little far-fetched and eliminating stack manipulations is
 9390: unlikely to become a conscious programming objective. Still, the number
 9391: of stack manipulations will be reduced dramatically if local variables
 9392: are used liberally (e.g., compare @code{max} (@pxref{Gforth locals}) with
 9393: a traditional implementation of @code{max}).
 9394: 
 9395: This shows one potential benefit of locals: making Forth programs more
 9396: readable. Of course, this benefit will only be realized if the
 9397: programmers continue to honour the principle of factoring instead of
 9398: using the added latitude to make the words longer.
 9399: 
 9400: @cindex single-assignment style for locals
 9401: Using @code{TO} can and should be avoided.  Without @code{TO},
 9402: every value-flavoured local has only a single assignment and many
 9403: advantages of functional languages apply to Forth. I.e., programs are
 9404: easier to analyse, to optimize and to read: It is clear from the
 9405: definition what the local stands for, it does not turn into something
 9406: different later.
 9407: 
 9408: E.g., a definition using @code{TO} might look like this:
 9409: @example
 9410: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 9411:  u1 u2 min 0
 9412:  ?do
 9413:    addr1 c@@ addr2 c@@ -
 9414:    ?dup-if
 9415:      unloop exit
 9416:    then
 9417:    addr1 char+ TO addr1
 9418:    addr2 char+ TO addr2
 9419:  loop
 9420:  u1 u2 - ;
 9421: @end example
 9422: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
 9423: every loop iteration. @code{strcmp} is a typical example of the
 9424: readability problems of using @code{TO}. When you start reading
 9425: @code{strcmp}, you think that @code{addr1} refers to the start of the
 9426: string. Only near the end of the loop you realize that it is something
 9427: else.
 9428: 
 9429: This can be avoided by defining two locals at the start of the loop that
 9430: are initialized with the right value for the current iteration.
 9431: @example
 9432: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 9433:  addr1 addr2
 9434:  u1 u2 min 0 
 9435:  ?do @{ s1 s2 @}
 9436:    s1 c@@ s2 c@@ -
 9437:    ?dup-if
 9438:      unloop exit
 9439:    then
 9440:    s1 char+ s2 char+
 9441:  loop
 9442:  2drop
 9443:  u1 u2 - ;
 9444: @end example
 9445: Here it is clear from the start that @code{s1} has a different value
 9446: in every loop iteration.
 9447: 
 9448: @node Locals implementation,  , Locals programming style, Gforth locals
 9449: @subsubsection Locals implementation
 9450: @cindex locals implementation
 9451: @cindex implementation of locals
 9452: 
 9453: @cindex locals stack
 9454: Gforth uses an extra locals stack. The most compelling reason for
 9455: this is that the return stack is not float-aligned; using an extra stack
 9456: also eliminates the problems and restrictions of using the return stack
 9457: as locals stack. Like the other stacks, the locals stack grows toward
 9458: lower addresses. A few primitives allow an efficient implementation:
 9459: 
 9460: 
 9461: doc-@local#
 9462: doc-f@local#
 9463: doc-laddr#
 9464: doc-lp+!#
 9465: doc-lp!
 9466: doc->l
 9467: doc-f>l
 9468: 
 9469: 
 9470: In addition to these primitives, some specializations of these
 9471: primitives for commonly occurring inline arguments are provided for
 9472: efficiency reasons, e.g., @code{@@local0} as specialization of
 9473: @code{@@local#} for the inline argument 0. The following compiling words
 9474: compile the right specialized version, or the general version, as
 9475: appropriate:
 9476: 
 9477: 
 9478: doc-compile-@local
 9479: doc-compile-f@local
 9480: doc-compile-lp+!
 9481: 
 9482: 
 9483: Combinations of conditional branches and @code{lp+!#} like
 9484: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
 9485: is taken) are provided for efficiency and correctness in loops.
 9486: 
 9487: A special area in the dictionary space is reserved for keeping the
 9488: local variable names. @code{@{} switches the dictionary pointer to this
 9489: area and @code{@}} switches it back and generates the locals
 9490: initializing code. @code{W:} etc.@ are normal defining words. This
 9491: special area is cleared at the start of every colon definition.
 9492: 
 9493: @cindex word list for defining locals
 9494: A special feature of Gforth's dictionary is used to implement the
 9495: definition of locals without type specifiers: every word list (aka
 9496: vocabulary) has its own methods for searching
 9497: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
 9498: with a special search method: When it is searched for a word, it
 9499: actually creates that word using @code{W:}. @code{@{} changes the search
 9500: order to first search the word list containing @code{@}}, @code{W:} etc.,
 9501: and then the word list for defining locals without type specifiers.
 9502: 
 9503: The lifetime rules support a stack discipline within a colon
 9504: definition: The lifetime of a local is either nested with other locals
 9505: lifetimes or it does not overlap them.
 9506: 
 9507: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
 9508: pointer manipulation is generated. Between control structure words
 9509: locals definitions can push locals onto the locals stack. @code{AGAIN}
 9510: is the simplest of the other three control flow words. It has to
 9511: restore the locals stack depth of the corresponding @code{BEGIN}
 9512: before branching. The code looks like this:
 9513: @format
 9514: @code{lp+!#} current-locals-size @minus{} dest-locals-size
 9515: @code{branch} <begin>
 9516: @end format
 9517: 
 9518: @code{UNTIL} is a little more complicated: If it branches back, it
 9519: must adjust the stack just like @code{AGAIN}. But if it falls through,
 9520: the locals stack must not be changed. The compiler generates the
 9521: following code:
 9522: @format
 9523: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
 9524: @end format
 9525: The locals stack pointer is only adjusted if the branch is taken.
 9526: 
 9527: @code{THEN} can produce somewhat inefficient code:
 9528: @format
 9529: @code{lp+!#} current-locals-size @minus{} orig-locals-size
 9530: <orig target>:
 9531: @code{lp+!#} orig-locals-size @minus{} new-locals-size
 9532: @end format
 9533: The second @code{lp+!#} adjusts the locals stack pointer from the
 9534: level at the @i{orig} point to the level after the @code{THEN}. The
 9535: first @code{lp+!#} adjusts the locals stack pointer from the current
 9536: level to the level at the orig point, so the complete effect is an
 9537: adjustment from the current level to the right level after the
 9538: @code{THEN}.
 9539: 
 9540: @cindex locals information on the control-flow stack
 9541: @cindex control-flow stack items, locals information
 9542: In a conventional Forth implementation a dest control-flow stack entry
 9543: is just the target address and an orig entry is just the address to be
 9544: patched. Our locals implementation adds a word list to every orig or dest
 9545: item. It is the list of locals visible (or assumed visible) at the point
 9546: described by the entry. Our implementation also adds a tag to identify
 9547: the kind of entry, in particular to differentiate between live and dead
 9548: (reachable and unreachable) orig entries.
 9549: 
 9550: A few unusual operations have to be performed on locals word lists:
 9551: 
 9552: 
 9553: doc-common-list
 9554: doc-sub-list?
 9555: doc-list-size
 9556: 
 9557: 
 9558: Several features of our locals word list implementation make these
 9559: operations easy to implement: The locals word lists are organised as
 9560: linked lists; the tails of these lists are shared, if the lists
 9561: contain some of the same locals; and the address of a name is greater
 9562: than the address of the names behind it in the list.
 9563: 
 9564: Another important implementation detail is the variable
 9565: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
 9566: determine if they can be reached directly or only through the branch
 9567: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
 9568: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
 9569: definition, by @code{BEGIN} and usually by @code{THEN}.
 9570: 
 9571: Counted loops are similar to other loops in most respects, but
 9572: @code{LEAVE} requires special attention: It performs basically the same
 9573: service as @code{AHEAD}, but it does not create a control-flow stack
 9574: entry. Therefore the information has to be stored elsewhere;
 9575: traditionally, the information was stored in the target fields of the
 9576: branches created by the @code{LEAVE}s, by organizing these fields into a
 9577: linked list. Unfortunately, this clever trick does not provide enough
 9578: space for storing our extended control flow information. Therefore, we
 9579: introduce another stack, the leave stack. It contains the control-flow
 9580: stack entries for all unresolved @code{LEAVE}s.
 9581: 
 9582: Local names are kept until the end of the colon definition, even if
 9583: they are no longer visible in any control-flow path. In a few cases
 9584: this may lead to increased space needs for the locals name area, but
 9585: usually less than reclaiming this space would cost in code size.
 9586: 
 9587: 
 9588: @node ANS Forth locals,  , Gforth locals, Locals
 9589: @subsection ANS Forth locals
 9590: @cindex locals, ANS Forth style
 9591: 
 9592: The ANS Forth locals wordset does not define a syntax for locals, but
 9593: words that make it possible to define various syntaxes. One of the
 9594: possible syntaxes is a subset of the syntax we used in the Gforth locals
 9595: wordset, i.e.:
 9596: 
 9597: @example
 9598: @{ local1 local2 ... -- comment @}
 9599: @end example
 9600: @noindent
 9601: or
 9602: @example
 9603: @{ local1 local2 ... @}
 9604: @end example
 9605: 
 9606: The order of the locals corresponds to the order in a stack comment. The
 9607: restrictions are:
 9608: 
 9609: @itemize @bullet
 9610: @item
 9611: Locals can only be cell-sized values (no type specifiers are allowed).
 9612: @item
 9613: Locals can be defined only outside control structures.
 9614: @item
 9615: Locals can interfere with explicit usage of the return stack. For the
 9616: exact (and long) rules, see the standard. If you don't use return stack
 9617: accessing words in a definition using locals, you will be all right. The
 9618: purpose of this rule is to make locals implementation on the return
 9619: stack easier.
 9620: @item
 9621: The whole definition must be in one line.
 9622: @end itemize
 9623: 
 9624: Locals defined in ANS Forth behave like @code{VALUE}s
 9625: (@pxref{Values}). I.e., they are initialized from the stack. Using their
 9626: name produces their value. Their value can be changed using @code{TO}.
 9627: 
 9628: Since the syntax above is supported by Gforth directly, you need not do
 9629: anything to use it. If you want to port a program using this syntax to
 9630: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
 9631: syntax on the other system.
 9632: 
 9633: Note that a syntax shown in the standard, section A.13 looks
 9634: similar, but is quite different in having the order of locals
 9635: reversed. Beware!
 9636: 
 9637: The ANS Forth locals wordset itself consists of one word:
 9638: 
 9639: doc-(local)
 9640: 
 9641: The ANS Forth locals extension wordset defines a syntax using
 9642: @code{locals|}, but it is so awful that we strongly recommend not to use
 9643: it. We have implemented this syntax to make porting to Gforth easy, but
 9644: do not document it here. The problem with this syntax is that the locals
 9645: are defined in an order reversed with respect to the standard stack
 9646: comment notation, making programs harder to read, and easier to misread
 9647: and miswrite. The only merit of this syntax is that it is easy to
 9648: implement using the ANS Forth locals wordset.
 9649: 
 9650: 
 9651: @c ----------------------------------------------------------
 9652: @node Structures, Object-oriented Forth, Locals, Words
 9653: @section  Structures
 9654: @cindex structures
 9655: @cindex records
 9656: 
 9657: This section presents the structure package that comes with Gforth. A
 9658: version of the package implemented in ANS Forth is available in
 9659: @file{compat/struct.fs}. This package was inspired by a posting on
 9660: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
 9661: possibly John Hayes). A version of this section has been published in
 9662: M. Anton Ertl,
 9663: @uref{http://www.complang.tuwien.ac.at/forth/objects/structs.html, Yet
 9664: Another Forth Structures Package}, Forth Dimensions 19(3), pages
 9665: 13--16. Marcel Hendrix provided helpful comments.
 9666: 
 9667: @menu
 9668: * Why explicit structure support?::  
 9669: * Structure Usage::             
 9670: * Structure Naming Convention::  
 9671: * Structure Implementation::    
 9672: * Structure Glossary::          
 9673: @end menu
 9674: 
 9675: @node Why explicit structure support?, Structure Usage, Structures, Structures
 9676: @subsection Why explicit structure support?
 9677: 
 9678: @cindex address arithmetic for structures
 9679: @cindex structures using address arithmetic
 9680: If we want to use a structure containing several fields, we could simply
 9681: reserve memory for it, and access the fields using address arithmetic
 9682: (@pxref{Address arithmetic}). As an example, consider a structure with
 9683: the following fields
 9684: 
 9685: @table @code
 9686: @item a
 9687: is a float
 9688: @item b
 9689: is a cell
 9690: @item c
 9691: is a float
 9692: @end table
 9693: 
 9694: Given the (float-aligned) base address of the structure we get the
 9695: address of the field
 9696: 
 9697: @table @code
 9698: @item a
 9699: without doing anything further.
 9700: @item b
 9701: with @code{float+}
 9702: @item c
 9703: with @code{float+ cell+ faligned}
 9704: @end table
 9705: 
 9706: It is easy to see that this can become quite tiring. 
 9707: 
 9708: Moreover, it is not very readable, because seeing a
 9709: @code{cell+} tells us neither which kind of structure is
 9710: accessed nor what field is accessed; we have to somehow infer the kind
 9711: of structure, and then look up in the documentation, which field of
 9712: that structure corresponds to that offset.
 9713: 
 9714: Finally, this kind of address arithmetic also causes maintenance
 9715: troubles: If you add or delete a field somewhere in the middle of the
 9716: structure, you have to find and change all computations for the fields
 9717: afterwards.
 9718: 
 9719: So, instead of using @code{cell+} and friends directly, how
 9720: about storing the offsets in constants:
 9721: 
 9722: @example
 9723: 0 constant a-offset
 9724: 0 float+ constant b-offset
 9725: 0 float+ cell+ faligned c-offset
 9726: @end example
 9727: 
 9728: Now we can get the address of field @code{x} with @code{x-offset
 9729: +}. This is much better in all respects. Of course, you still
 9730: have to change all later offset definitions if you add a field. You can
 9731: fix this by declaring the offsets in the following way:
 9732: 
 9733: @example
 9734: 0 constant a-offset
 9735: a-offset float+ constant b-offset
 9736: b-offset cell+ faligned constant c-offset
 9737: @end example
 9738: 
 9739: Since we always use the offsets with @code{+}, we could use a defining
 9740: word @code{cfield} that includes the @code{+} in the action of the
 9741: defined word:
 9742: 
 9743: @example
 9744: : cfield ( n "name" -- )
 9745:     create ,
 9746: does> ( name execution: addr1 -- addr2 )
 9747:     @@ + ;
 9748: 
 9749: 0 cfield a
 9750: 0 a float+ cfield b
 9751: 0 b cell+ faligned cfield c
 9752: @end example
 9753: 
 9754: Instead of @code{x-offset +}, we now simply write @code{x}.
 9755: 
 9756: The structure field words now can be used quite nicely. However,
 9757: their definition is still a bit cumbersome: We have to repeat the
 9758: name, the information about size and alignment is distributed before
 9759: and after the field definitions etc.  The structure package presented
 9760: here addresses these problems.
 9761: 
 9762: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
 9763: @subsection Structure Usage
 9764: @cindex structure usage
 9765: 
 9766: @cindex @code{field} usage
 9767: @cindex @code{struct} usage
 9768: @cindex @code{end-struct} usage
 9769: You can define a structure for a (data-less) linked list with:
 9770: @example
 9771: struct
 9772:     cell% field list-next
 9773: end-struct list%
 9774: @end example
 9775: 
 9776: With the address of the list node on the stack, you can compute the
 9777: address of the field that contains the address of the next node with
 9778: @code{list-next}. E.g., you can determine the length of a list
 9779: with:
 9780: 
 9781: @example
 9782: : list-length ( list -- n )
 9783: \ "list" is a pointer to the first element of a linked list
 9784: \ "n" is the length of the list
 9785:     0 BEGIN ( list1 n1 )
 9786:         over
 9787:     WHILE ( list1 n1 )
 9788:         1+ swap list-next @@ swap
 9789:     REPEAT
 9790:     nip ;
 9791: @end example
 9792: 
 9793: You can reserve memory for a list node in the dictionary with
 9794: @code{list% %allot}, which leaves the address of the list node on the
 9795: stack. For the equivalent allocation on the heap you can use @code{list%
 9796: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
 9797: use @code{list% %allocate}). You can get the the size of a list
 9798: node with @code{list% %size} and its alignment with @code{list%
 9799: %alignment}.
 9800: 
 9801: Note that in ANS Forth the body of a @code{create}d word is
 9802: @code{aligned} but not necessarily @code{faligned};
 9803: therefore, if you do a:
 9804: 
 9805: @example
 9806: create @emph{name} foo% %allot drop
 9807: @end example
 9808: 
 9809: @noindent
 9810: then the memory alloted for @code{foo%} is guaranteed to start at the
 9811: body of @code{@emph{name}} only if @code{foo%} contains only character,
 9812: cell and double fields.  Therefore, if your structure contains floats,
 9813: better use
 9814: 
 9815: @example
 9816: foo% %allot constant @emph{name}
 9817: @end example
 9818: 
 9819: @cindex structures containing structures
 9820: You can include a structure @code{foo%} as a field of
 9821: another structure, like this:
 9822: @example
 9823: struct
 9824: ...
 9825:     foo% field ...
 9826: ...
 9827: end-struct ...
 9828: @end example
 9829: 
 9830: @cindex structure extension
 9831: @cindex extended records
 9832: Instead of starting with an empty structure, you can extend an
 9833: existing structure. E.g., a plain linked list without data, as defined
 9834: above, is hardly useful; You can extend it to a linked list of integers,
 9835: like this:@footnote{This feature is also known as @emph{extended
 9836: records}. It is the main innovation in the Oberon language; in other
 9837: words, adding this feature to Modula-2 led Wirth to create a new
 9838: language, write a new compiler etc.  Adding this feature to Forth just
 9839: required a few lines of code.}
 9840: 
 9841: @example
 9842: list%
 9843:     cell% field intlist-int
 9844: end-struct intlist%
 9845: @end example
 9846: 
 9847: @code{intlist%} is a structure with two fields:
 9848: @code{list-next} and @code{intlist-int}.
 9849: 
 9850: @cindex structures containing arrays
 9851: You can specify an array type containing @emph{n} elements of
 9852: type @code{foo%} like this:
 9853: 
 9854: @example
 9855: foo% @emph{n} *
 9856: @end example
 9857: 
 9858: You can use this array type in any place where you can use a normal
 9859: type, e.g., when defining a @code{field}, or with
 9860: @code{%allot}.
 9861: 
 9862: @cindex first field optimization
 9863: The first field is at the base address of a structure and the word for
 9864: this field (e.g., @code{list-next}) actually does not change the address
 9865: on the stack. You may be tempted to leave it away in the interest of
 9866: run-time and space efficiency. This is not necessary, because the
 9867: structure package optimizes this case: If you compile a first-field
 9868: words, no code is generated. So, in the interest of readability and
 9869: maintainability you should include the word for the field when accessing
 9870: the field.
 9871: 
 9872: 
 9873: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
 9874: @subsection Structure Naming Convention
 9875: @cindex structure naming convention
 9876: 
 9877: The field names that come to (my) mind are often quite generic, and,
 9878: if used, would cause frequent name clashes. E.g., many structures
 9879: probably contain a @code{counter} field. The structure names
 9880: that come to (my) mind are often also the logical choice for the names
 9881: of words that create such a structure.
 9882: 
 9883: Therefore, I have adopted the following naming conventions: 
 9884: 
 9885: @itemize @bullet
 9886: @cindex field naming convention
 9887: @item
 9888: The names of fields are of the form
 9889: @code{@emph{struct}-@emph{field}}, where
 9890: @code{@emph{struct}} is the basic name of the structure, and
 9891: @code{@emph{field}} is the basic name of the field. You can
 9892: think of field words as converting the (address of the)
 9893: structure into the (address of the) field.
 9894: 
 9895: @cindex structure naming convention
 9896: @item
 9897: The names of structures are of the form
 9898: @code{@emph{struct}%}, where
 9899: @code{@emph{struct}} is the basic name of the structure.
 9900: @end itemize
 9901: 
 9902: This naming convention does not work that well for fields of extended
 9903: structures; e.g., the integer list structure has a field
 9904: @code{intlist-int}, but has @code{list-next}, not
 9905: @code{intlist-next}.
 9906: 
 9907: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
 9908: @subsection Structure Implementation
 9909: @cindex structure implementation
 9910: @cindex implementation of structures
 9911: 
 9912: The central idea in the implementation is to pass the data about the
 9913: structure being built on the stack, not in some global
 9914: variable. Everything else falls into place naturally once this design
 9915: decision is made.
 9916: 
 9917: The type description on the stack is of the form @emph{align
 9918: size}. Keeping the size on the top-of-stack makes dealing with arrays
 9919: very simple.
 9920: 
 9921: @code{field} is a defining word that uses @code{Create}
 9922: and @code{DOES>}. The body of the field contains the offset
 9923: of the field, and the normal @code{DOES>} action is simply:
 9924: 
 9925: @example
 9926: @@ +
 9927: @end example
 9928: 
 9929: @noindent
 9930: i.e., add the offset to the address, giving the stack effect
 9931: @i{addr1 -- addr2} for a field.
 9932: 
 9933: @cindex first field optimization, implementation
 9934: This simple structure is slightly complicated by the optimization
 9935: for fields with offset 0, which requires a different
 9936: @code{DOES>}-part (because we cannot rely on there being
 9937: something on the stack if such a field is invoked during
 9938: compilation). Therefore, we put the different @code{DOES>}-parts
 9939: in separate words, and decide which one to invoke based on the
 9940: offset. For a zero offset, the field is basically a noop; it is
 9941: immediate, and therefore no code is generated when it is compiled.
 9942: 
 9943: @node Structure Glossary,  , Structure Implementation, Structures
 9944: @subsection Structure Glossary
 9945: @cindex structure glossary
 9946: 
 9947: 
 9948: doc-%align
 9949: doc-%alignment
 9950: doc-%alloc
 9951: doc-%allocate
 9952: doc-%allot
 9953: doc-cell%
 9954: doc-char%
 9955: doc-dfloat%
 9956: doc-double%
 9957: doc-end-struct
 9958: doc-field
 9959: doc-float%
 9960: doc-naligned
 9961: doc-sfloat%
 9962: doc-%size
 9963: doc-struct
 9964: 
 9965: 
 9966: @c -------------------------------------------------------------
 9967: @node Object-oriented Forth, Programming Tools, Structures, Words
 9968: @section Object-oriented Forth
 9969: 
 9970: Gforth comes with three packages for object-oriented programming:
 9971: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
 9972: is preloaded, so you have to @code{include} them before use. The most
 9973: important differences between these packages (and others) are discussed
 9974: in @ref{Comparison with other object models}. All packages are written
 9975: in ANS Forth and can be used with any other ANS Forth.
 9976: 
 9977: @menu
 9978: * Why object-oriented programming?::  
 9979: * Object-Oriented Terminology::  
 9980: * Objects::                     
 9981: * OOF::                         
 9982: * Mini-OOF::                    
 9983: * Comparison with other object models::  
 9984: @end menu
 9985: 
 9986: @c ----------------------------------------------------------------
 9987: @node Why object-oriented programming?, Object-Oriented Terminology, Object-oriented Forth, Object-oriented Forth
 9988: @subsection Why object-oriented programming?
 9989: @cindex object-oriented programming motivation
 9990: @cindex motivation for object-oriented programming
 9991: 
 9992: Often we have to deal with several data structures (@emph{objects}),
 9993: that have to be treated similarly in some respects, but differently in
 9994: others. Graphical objects are the textbook example: circles, triangles,
 9995: dinosaurs, icons, and others, and we may want to add more during program
 9996: development. We want to apply some operations to any graphical object,
 9997: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
 9998: has to do something different for every kind of object.
 9999: @comment TODO add some other operations eg perimeter, area
10000: @comment and tie in to concrete examples later..
10001: 
10002: We could implement @code{draw} as a big @code{CASE}
10003: control structure that executes the appropriate code depending on the
10004: kind of object to be drawn. This would be not be very elegant, and,
10005: moreover, we would have to change @code{draw} every time we add
10006: a new kind of graphical object (say, a spaceship).
10007: 
10008: What we would rather do is: When defining spaceships, we would tell
10009: the system: ``Here's how you @code{draw} a spaceship; you figure
10010: out the rest''.
10011: 
10012: This is the problem that all systems solve that (rightfully) call
10013: themselves object-oriented; the object-oriented packages presented here
10014: solve this problem (and not much else).
10015: @comment TODO ?list properties of oo systems.. oo vs o-based?
10016: 
10017: @c ------------------------------------------------------------------------
10018: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
10019: @subsection Object-Oriented Terminology
10020: @cindex object-oriented terminology
10021: @cindex terminology for object-oriented programming
10022: 
10023: This section is mainly for reference, so you don't have to understand
10024: all of it right away.  The terminology is mainly Smalltalk-inspired.  In
10025: short:
10026: 
10027: @table @emph
10028: @cindex class
10029: @item class
10030: a data structure definition with some extras.
10031: 
10032: @cindex object
10033: @item object
10034: an instance of the data structure described by the class definition.
10035: 
10036: @cindex instance variables
10037: @item instance variables
10038: fields of the data structure.
10039: 
10040: @cindex selector
10041: @cindex method selector
10042: @cindex virtual function
10043: @item selector
10044: (or @emph{method selector}) a word (e.g.,
10045: @code{draw}) that performs an operation on a variety of data
10046: structures (classes). A selector describes @emph{what} operation to
10047: perform. In C++ terminology: a (pure) virtual function.
10048: 
10049: @cindex method
10050: @item method
10051: the concrete definition that performs the operation
10052: described by the selector for a specific class. A method specifies
10053: @emph{how} the operation is performed for a specific class.
10054: 
10055: @cindex selector invocation
10056: @cindex message send
10057: @cindex invoking a selector
10058: @item selector invocation
10059: a call of a selector. One argument of the call (the TOS (top-of-stack))
10060: is used for determining which method is used. In Smalltalk terminology:
10061: a message (consisting of the selector and the other arguments) is sent
10062: to the object.
10063: 
10064: @cindex receiving object
10065: @item receiving object
10066: the object used for determining the method executed by a selector
10067: invocation. In the @file{objects.fs} model, it is the object that is on
10068: the TOS when the selector is invoked. (@emph{Receiving} comes from
10069: the Smalltalk @emph{message} terminology.)
10070: 
10071: @cindex child class
10072: @cindex parent class
10073: @cindex inheritance
10074: @item child class
10075: a class that has (@emph{inherits}) all properties (instance variables,
10076: selectors, methods) from a @emph{parent class}. In Smalltalk
10077: terminology: The subclass inherits from the superclass. In C++
10078: terminology: The derived class inherits from the base class.
10079: 
10080: @end table
10081: 
10082: @c If you wonder about the message sending terminology, it comes from
10083: @c a time when each object had it's own task and objects communicated via
10084: @c message passing; eventually the Smalltalk developers realized that
10085: @c they can do most things through simple (indirect) calls. They kept the
10086: @c terminology.
10087: 
10088: @c --------------------------------------------------------------
10089: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
10090: @subsection The @file{objects.fs} model
10091: @cindex objects
10092: @cindex object-oriented programming
10093: 
10094: @cindex @file{objects.fs}
10095: @cindex @file{oof.fs}
10096: 
10097: This section describes the @file{objects.fs} package. This material also
10098: has been published in M. Anton Ertl,
10099: @cite{@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html,
10100: Yet Another Forth Objects Package}}, Forth Dimensions 19(2), pages
10101: 37--43.
10102: @c McKewan's and Zsoter's packages
10103: 
10104: This section assumes that you have read @ref{Structures}.
10105: 
10106: The techniques on which this model is based have been used to implement
10107: the parser generator, Gray, and have also been used in Gforth for
10108: implementing the various flavours of word lists (hashed or not,
10109: case-sensitive or not, special-purpose word lists for locals etc.).
10110: 
10111: 
10112: @menu
10113: * Properties of the Objects model::  
10114: * Basic Objects Usage::         
10115: * The Objects base class::      
10116: * Creating objects::            
10117: * Object-Oriented Programming Style::  
10118: * Class Binding::               
10119: * Method conveniences::         
10120: * Classes and Scoping::         
10121: * Dividing classes::            
10122: * Object Interfaces::           
10123: * Objects Implementation::      
10124: * Objects Glossary::            
10125: @end menu
10126: 
10127: Marcel Hendrix provided helpful comments on this section.
10128: 
10129: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
10130: @subsubsection Properties of the @file{objects.fs} model
10131: @cindex @file{objects.fs} properties
10132: 
10133: @itemize @bullet
10134: @item
10135: It is straightforward to pass objects on the stack. Passing
10136: selectors on the stack is a little less convenient, but possible.
10137: 
10138: @item
10139: Objects are just data structures in memory, and are referenced by their
10140: address. You can create words for objects with normal defining words
10141: like @code{constant}. Likewise, there is no difference between instance
10142: variables that contain objects and those that contain other data.
10143: 
10144: @item
10145: Late binding is efficient and easy to use.
10146: 
10147: @item
10148: It avoids parsing, and thus avoids problems with state-smartness
10149: and reduced extensibility; for convenience there are a few parsing
10150: words, but they have non-parsing counterparts. There are also a few
10151: defining words that parse. This is hard to avoid, because all standard
10152: defining words parse (except @code{:noname}); however, such
10153: words are not as bad as many other parsing words, because they are not
10154: state-smart.
10155: 
10156: @item
10157: It does not try to incorporate everything. It does a few things and does
10158: them well (IMO). In particular, this model was not designed to support
10159: information hiding (although it has features that may help); you can use
10160: a separate package for achieving this.
10161: 
10162: @item
10163: It is layered; you don't have to learn and use all features to use this
10164: model. Only a few features are necessary (@pxref{Basic Objects Usage},
10165: @pxref{The Objects base class}, @pxref{Creating objects}.), the others
10166: are optional and independent of each other.
10167: 
10168: @item
10169: An implementation in ANS Forth is available.
10170: 
10171: @end itemize
10172: 
10173: 
10174: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
10175: @subsubsection Basic @file{objects.fs} Usage
10176: @cindex basic objects usage
10177: @cindex objects, basic usage
10178: 
10179: You can define a class for graphical objects like this:
10180: 
10181: @cindex @code{class} usage
10182: @cindex @code{end-class} usage
10183: @cindex @code{selector} usage
10184: @example
10185: object class \ "object" is the parent class
10186:   selector draw ( x y graphical -- )
10187: end-class graphical
10188: @end example
10189: 
10190: This code defines a class @code{graphical} with an
10191: operation @code{draw}.  We can perform the operation
10192: @code{draw} on any @code{graphical} object, e.g.:
10193: 
10194: @example
10195: 100 100 t-rex draw
10196: @end example
10197: 
10198: @noindent
10199: where @code{t-rex} is a word (say, a constant) that produces a
10200: graphical object.
10201: 
10202: @comment TODO add a 2nd operation eg perimeter.. and use for
10203: @comment a concrete example
10204: 
10205: @cindex abstract class
10206: How do we create a graphical object? With the present definitions,
10207: we cannot create a useful graphical object. The class
10208: @code{graphical} describes graphical objects in general, but not
10209: any concrete graphical object type (C++ users would call it an
10210: @emph{abstract class}); e.g., there is no method for the selector
10211: @code{draw} in the class @code{graphical}.
10212: 
10213: For concrete graphical objects, we define child classes of the
10214: class @code{graphical}, e.g.:
10215: 
10216: @cindex @code{overrides} usage
10217: @cindex @code{field} usage in class definition
10218: @example
10219: graphical class \ "graphical" is the parent class
10220:   cell% field circle-radius
10221: 
10222: :noname ( x y circle -- )
10223:   circle-radius @@ draw-circle ;
10224: overrides draw
10225: 
10226: :noname ( n-radius circle -- )
10227:   circle-radius ! ;
10228: overrides construct
10229: 
10230: end-class circle
10231: @end example
10232: 
10233: Here we define a class @code{circle} as a child of @code{graphical},
10234: with field @code{circle-radius} (which behaves just like a field
10235: (@pxref{Structures}); it defines (using @code{overrides}) new methods
10236: for the selectors @code{draw} and @code{construct} (@code{construct} is
10237: defined in @code{object}, the parent class of @code{graphical}).
10238: 
10239: Now we can create a circle on the heap (i.e.,
10240: @code{allocate}d memory) with:
10241: 
10242: @cindex @code{heap-new} usage
10243: @example
10244: 50 circle heap-new constant my-circle
10245: @end example
10246: 
10247: @noindent
10248: @code{heap-new} invokes @code{construct}, thus
10249: initializing the field @code{circle-radius} with 50. We can draw
10250: this new circle at (100,100) with:
10251: 
10252: @example
10253: 100 100 my-circle draw
10254: @end example
10255: 
10256: @cindex selector invocation, restrictions
10257: @cindex class definition, restrictions
10258: Note: You can only invoke a selector if the object on the TOS
10259: (the receiving object) belongs to the class where the selector was
10260: defined or one of its descendents; e.g., you can invoke
10261: @code{draw} only for objects belonging to @code{graphical}
10262: or its descendents (e.g., @code{circle}).  Immediately before
10263: @code{end-class}, the search order has to be the same as
10264: immediately after @code{class}.
10265: 
10266: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
10267: @subsubsection The @file{object.fs} base class
10268: @cindex @code{object} class
10269: 
10270: When you define a class, you have to specify a parent class.  So how do
10271: you start defining classes? There is one class available from the start:
10272: @code{object}. It is ancestor for all classes and so is the
10273: only class that has no parent. It has two selectors: @code{construct}
10274: and @code{print}.
10275: 
10276: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
10277: @subsubsection Creating objects
10278: @cindex creating objects
10279: @cindex object creation
10280: @cindex object allocation options
10281: 
10282: @cindex @code{heap-new} discussion
10283: @cindex @code{dict-new} discussion
10284: @cindex @code{construct} discussion
10285: You can create and initialize an object of a class on the heap with
10286: @code{heap-new} ( ... class -- object ) and in the dictionary
10287: (allocation with @code{allot}) with @code{dict-new} (
10288: ... class -- object ). Both words invoke @code{construct}, which
10289: consumes the stack items indicated by "..." above.
10290: 
10291: @cindex @code{init-object} discussion
10292: @cindex @code{class-inst-size} discussion
10293: If you want to allocate memory for an object yourself, you can get its
10294: alignment and size with @code{class-inst-size 2@@} ( class --
10295: align size ). Once you have memory for an object, you can initialize
10296: it with @code{init-object} ( ... class object -- );
10297: @code{construct} does only a part of the necessary work.
10298: 
10299: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
10300: @subsubsection Object-Oriented Programming Style
10301: @cindex object-oriented programming style
10302: @cindex programming style, object-oriented
10303: 
10304: This section is not exhaustive.
10305: 
10306: @cindex stack effects of selectors
10307: @cindex selectors and stack effects
10308: In general, it is a good idea to ensure that all methods for the
10309: same selector have the same stack effect: when you invoke a selector,
10310: you often have no idea which method will be invoked, so, unless all
10311: methods have the same stack effect, you will not know the stack effect
10312: of the selector invocation.
10313: 
10314: One exception to this rule is methods for the selector
10315: @code{construct}. We know which method is invoked, because we
10316: specify the class to be constructed at the same place. Actually, I
10317: defined @code{construct} as a selector only to give the users a
10318: convenient way to specify initialization. The way it is used, a
10319: mechanism different from selector invocation would be more natural
10320: (but probably would take more code and more space to explain).
10321: 
10322: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
10323: @subsubsection Class Binding
10324: @cindex class binding
10325: @cindex early binding
10326: 
10327: @cindex late binding
10328: Normal selector invocations determine the method at run-time depending
10329: on the class of the receiving object. This run-time selection is called
10330: @i{late binding}.
10331: 
10332: Sometimes it's preferable to invoke a different method. For example,
10333: you might want to use the simple method for @code{print}ing
10334: @code{object}s instead of the possibly long-winded @code{print} method
10335: of the receiver class. You can achieve this by replacing the invocation
10336: of @code{print} with:
10337: 
10338: @cindex @code{[bind]} usage
10339: @example
10340: [bind] object print
10341: @end example
10342: 
10343: @noindent
10344: in compiled code or:
10345: 
10346: @cindex @code{bind} usage
10347: @example
10348: bind object print
10349: @end example
10350: 
10351: @cindex class binding, alternative to
10352: @noindent
10353: in interpreted code. Alternatively, you can define the method with a
10354: name (e.g., @code{print-object}), and then invoke it through the
10355: name. Class binding is just a (often more convenient) way to achieve
10356: the same effect; it avoids name clutter and allows you to invoke
10357: methods directly without naming them first.
10358: 
10359: @cindex superclass binding
10360: @cindex parent class binding
10361: A frequent use of class binding is this: When we define a method
10362: for a selector, we often want the method to do what the selector does
10363: in the parent class, and a little more. There is a special word for
10364: this purpose: @code{[parent]}; @code{[parent]
10365: @emph{selector}} is equivalent to @code{[bind] @emph{parent
10366: selector}}, where @code{@emph{parent}} is the parent
10367: class of the current class. E.g., a method definition might look like:
10368: 
10369: @cindex @code{[parent]} usage
10370: @example
10371: :noname
10372:   dup [parent] foo \ do parent's foo on the receiving object
10373:   ... \ do some more
10374: ; overrides foo
10375: @end example
10376: 
10377: @cindex class binding as optimization
10378: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
10379: March 1997), Andrew McKewan presents class binding as an optimization
10380: technique. I recommend not using it for this purpose unless you are in
10381: an emergency. Late binding is pretty fast with this model anyway, so the
10382: benefit of using class binding is small; the cost of using class binding
10383: where it is not appropriate is reduced maintainability.
10384: 
10385: While we are at programming style questions: You should bind
10386: selectors only to ancestor classes of the receiving object. E.g., say,
10387: you know that the receiving object is of class @code{foo} or its
10388: descendents; then you should bind only to @code{foo} and its
10389: ancestors.
10390: 
10391: @node Method conveniences, Classes and Scoping, Class Binding, Objects
10392: @subsubsection Method conveniences
10393: @cindex method conveniences
10394: 
10395: In a method you usually access the receiving object pretty often.  If
10396: you define the method as a plain colon definition (e.g., with
10397: @code{:noname}), you may have to do a lot of stack
10398: gymnastics. To avoid this, you can define the method with @code{m:
10399: ... ;m}. E.g., you could define the method for
10400: @code{draw}ing a @code{circle} with
10401: 
10402: @cindex @code{this} usage
10403: @cindex @code{m:} usage
10404: @cindex @code{;m} usage
10405: @example
10406: m: ( x y circle -- )
10407:   ( x y ) this circle-radius @@ draw-circle ;m
10408: @end example
10409: 
10410: @cindex @code{exit} in @code{m: ... ;m}
10411: @cindex @code{exitm} discussion
10412: @cindex @code{catch} in @code{m: ... ;m}
10413: When this method is executed, the receiver object is removed from the
10414: stack; you can access it with @code{this} (admittedly, in this
10415: example the use of @code{m: ... ;m} offers no advantage). Note
10416: that I specify the stack effect for the whole method (i.e. including
10417: the receiver object), not just for the code between @code{m:}
10418: and @code{;m}. You cannot use @code{exit} in
10419: @code{m:...;m}; instead, use
10420: @code{exitm}.@footnote{Moreover, for any word that calls
10421: @code{catch} and was defined before loading
10422: @code{objects.fs}, you have to redefine it like I redefined
10423: @code{catch}: @code{: catch this >r catch r> to-this ;}}
10424: 
10425: @cindex @code{inst-var} usage
10426: You will frequently use sequences of the form @code{this
10427: @emph{field}} (in the example above: @code{this
10428: circle-radius}). If you use the field only in this way, you can
10429: define it with @code{inst-var} and eliminate the
10430: @code{this} before the field name. E.g., the @code{circle}
10431: class above could also be defined with:
10432: 
10433: @example
10434: graphical class
10435:   cell% inst-var radius
10436: 
10437: m: ( x y circle -- )
10438:   radius @@ draw-circle ;m
10439: overrides draw
10440: 
10441: m: ( n-radius circle -- )
10442:   radius ! ;m
10443: overrides construct
10444: 
10445: end-class circle
10446: @end example
10447: 
10448: @code{radius} can only be used in @code{circle} and its
10449: descendent classes and inside @code{m:...;m}.
10450: 
10451: @cindex @code{inst-value} usage
10452: You can also define fields with @code{inst-value}, which is
10453: to @code{inst-var} what @code{value} is to
10454: @code{variable}.  You can change the value of such a field with
10455: @code{[to-inst]}.  E.g., we could also define the class
10456: @code{circle} like this:
10457: 
10458: @example
10459: graphical class
10460:   inst-value radius
10461: 
10462: m: ( x y circle -- )
10463:   radius draw-circle ;m
10464: overrides draw
10465: 
10466: m: ( n-radius circle -- )
10467:   [to-inst] radius ;m
10468: overrides construct
10469: 
10470: end-class circle
10471: @end example
10472: 
10473: @c !! :m is easy to confuse with m:.  Another name would be better.
10474: 
10475: @c Finally, you can define named methods with @code{:m}.  One use of this
10476: @c feature is the definition of words that occur only in one class and are
10477: @c not intended to be overridden, but which still need method context
10478: @c (e.g., for accessing @code{inst-var}s).  Another use is for methods that
10479: @c would be bound frequently, if defined anonymously.
10480: 
10481: 
10482: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
10483: @subsubsection Classes and Scoping
10484: @cindex classes and scoping
10485: @cindex scoping and classes
10486: 
10487: Inheritance is frequent, unlike structure extension. This exacerbates
10488: the problem with the field name convention (@pxref{Structure Naming
10489: Convention}): One always has to remember in which class the field was
10490: originally defined; changing a part of the class structure would require
10491: changes for renaming in otherwise unaffected code.
10492: 
10493: @cindex @code{inst-var} visibility
10494: @cindex @code{inst-value} visibility
10495: To solve this problem, I added a scoping mechanism (which was not in my
10496: original charter): A field defined with @code{inst-var} (or
10497: @code{inst-value}) is visible only in the class where it is defined and in
10498: the descendent classes of this class.  Using such fields only makes
10499: sense in @code{m:}-defined methods in these classes anyway.
10500: 
10501: This scoping mechanism allows us to use the unadorned field name,
10502: because name clashes with unrelated words become much less likely.
10503: 
10504: @cindex @code{protected} discussion
10505: @cindex @code{private} discussion
10506: Once we have this mechanism, we can also use it for controlling the
10507: visibility of other words: All words defined after
10508: @code{protected} are visible only in the current class and its
10509: descendents. @code{public} restores the compilation
10510: (i.e. @code{current}) word list that was in effect before. If you
10511: have several @code{protected}s without an intervening
10512: @code{public} or @code{set-current}, @code{public}
10513: will restore the compilation word list in effect before the first of
10514: these @code{protected}s.
10515: 
10516: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
10517: @subsubsection Dividing classes
10518: @cindex Dividing classes
10519: @cindex @code{methods}...@code{end-methods}
10520: 
10521: You may want to do the definition of methods separate from the
10522: definition of the class, its selectors, fields, and instance variables,
10523: i.e., separate the implementation from the definition.  You can do this
10524: in the following way:
10525: 
10526: @example
10527: graphical class
10528:   inst-value radius
10529: end-class circle
10530: 
10531: ... \ do some other stuff
10532: 
10533: circle methods \ now we are ready
10534: 
10535: m: ( x y circle -- )
10536:   radius draw-circle ;m
10537: overrides draw
10538: 
10539: m: ( n-radius circle -- )
10540:   [to-inst] radius ;m
10541: overrides construct
10542: 
10543: end-methods
10544: @end example
10545: 
10546: You can use several @code{methods}...@code{end-methods} sections.  The
10547: only things you can do to the class in these sections are: defining
10548: methods, and overriding the class's selectors.  You must not define new
10549: selectors or fields.
10550: 
10551: Note that you often have to override a selector before using it.  In
10552: particular, you usually have to override @code{construct} with a new
10553: method before you can invoke @code{heap-new} and friends.  E.g., you
10554: must not create a circle before the @code{overrides construct} sequence
10555: in the example above.
10556: 
10557: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
10558: @subsubsection Object Interfaces
10559: @cindex object interfaces
10560: @cindex interfaces for objects
10561: 
10562: In this model you can only call selectors defined in the class of the
10563: receiving objects or in one of its ancestors. If you call a selector
10564: with a receiving object that is not in one of these classes, the
10565: result is undefined; if you are lucky, the program crashes
10566: immediately.
10567: 
10568: @cindex selectors common to hardly-related classes
10569: Now consider the case when you want to have a selector (or several)
10570: available in two classes: You would have to add the selector to a
10571: common ancestor class, in the worst case to @code{object}. You
10572: may not want to do this, e.g., because someone else is responsible for
10573: this ancestor class.
10574: 
10575: The solution for this problem is interfaces. An interface is a
10576: collection of selectors. If a class implements an interface, the
10577: selectors become available to the class and its descendents. A class
10578: can implement an unlimited number of interfaces. For the problem
10579: discussed above, we would define an interface for the selector(s), and
10580: both classes would implement the interface.
10581: 
10582: As an example, consider an interface @code{storage} for
10583: writing objects to disk and getting them back, and a class
10584: @code{foo} that implements it. The code would look like this:
10585: 
10586: @cindex @code{interface} usage
10587: @cindex @code{end-interface} usage
10588: @cindex @code{implementation} usage
10589: @example
10590: interface
10591:   selector write ( file object -- )
10592:   selector read1 ( file object -- )
10593: end-interface storage
10594: 
10595: bar class
10596:   storage implementation
10597: 
10598: ... overrides write
10599: ... overrides read1
10600: ...
10601: end-class foo
10602: @end example
10603: 
10604: @noindent
10605: (I would add a word @code{read} @i{( file -- object )} that uses
10606: @code{read1} internally, but that's beyond the point illustrated
10607: here.)
10608: 
10609: Note that you cannot use @code{protected} in an interface; and
10610: of course you cannot define fields.
10611: 
10612: In the Neon model, all selectors are available for all classes;
10613: therefore it does not need interfaces. The price you pay in this model
10614: is slower late binding, and therefore, added complexity to avoid late
10615: binding.
10616: 
10617: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
10618: @subsubsection @file{objects.fs} Implementation
10619: @cindex @file{objects.fs} implementation
10620: 
10621: @cindex @code{object-map} discussion
10622: An object is a piece of memory, like one of the data structures
10623: described with @code{struct...end-struct}. It has a field
10624: @code{object-map} that points to the method map for the object's
10625: class.
10626: 
10627: @cindex method map
10628: @cindex virtual function table
10629: The @emph{method map}@footnote{This is Self terminology; in C++
10630: terminology: virtual function table.} is an array that contains the
10631: execution tokens (@i{xt}s) of the methods for the object's class. Each
10632: selector contains an offset into a method map.
10633: 
10634: @cindex @code{selector} implementation, class
10635: @code{selector} is a defining word that uses
10636: @code{CREATE} and @code{DOES>}. The body of the
10637: selector contains the offset; the @code{DOES>} action for a
10638: class selector is, basically:
10639: 
10640: @example
10641: ( object addr ) @@ over object-map @@ + @@ execute
10642: @end example
10643: 
10644: Since @code{object-map} is the first field of the object, it
10645: does not generate any code. As you can see, calling a selector has a
10646: small, constant cost.
10647: 
10648: @cindex @code{current-interface} discussion
10649: @cindex class implementation and representation
10650: A class is basically a @code{struct} combined with a method
10651: map. During the class definition the alignment and size of the class
10652: are passed on the stack, just as with @code{struct}s, so
10653: @code{field} can also be used for defining class
10654: fields. However, passing more items on the stack would be
10655: inconvenient, so @code{class} builds a data structure in memory,
10656: which is accessed through the variable
10657: @code{current-interface}. After its definition is complete, the
10658: class is represented on the stack by a pointer (e.g., as parameter for
10659: a child class definition).
10660: 
10661: A new class starts off with the alignment and size of its parent,
10662: and a copy of the parent's method map. Defining new fields extends the
10663: size and alignment; likewise, defining new selectors extends the
10664: method map. @code{overrides} just stores a new @i{xt} in the method
10665: map at the offset given by the selector.
10666: 
10667: @cindex class binding, implementation
10668: Class binding just gets the @i{xt} at the offset given by the selector
10669: from the class's method map and @code{compile,}s (in the case of
10670: @code{[bind]}) it.
10671: 
10672: @cindex @code{this} implementation
10673: @cindex @code{catch} and @code{this}
10674: @cindex @code{this} and @code{catch}
10675: I implemented @code{this} as a @code{value}. At the
10676: start of an @code{m:...;m} method the old @code{this} is
10677: stored to the return stack and restored at the end; and the object on
10678: the TOS is stored @code{TO this}. This technique has one
10679: disadvantage: If the user does not leave the method via
10680: @code{;m}, but via @code{throw} or @code{exit},
10681: @code{this} is not restored (and @code{exit} may
10682: crash). To deal with the @code{throw} problem, I have redefined
10683: @code{catch} to save and restore @code{this}; the same
10684: should be done with any word that can catch an exception. As for
10685: @code{exit}, I simply forbid it (as a replacement, there is
10686: @code{exitm}).
10687: 
10688: @cindex @code{inst-var} implementation
10689: @code{inst-var} is just the same as @code{field}, with
10690: a different @code{DOES>} action:
10691: @example
10692: @@ this +
10693: @end example
10694: Similar for @code{inst-value}.
10695: 
10696: @cindex class scoping implementation
10697: Each class also has a word list that contains the words defined with
10698: @code{inst-var} and @code{inst-value}, and its protected
10699: words. It also has a pointer to its parent. @code{class} pushes
10700: the word lists of the class and all its ancestors onto the search order stack,
10701: and @code{end-class} drops them.
10702: 
10703: @cindex interface implementation
10704: An interface is like a class without fields, parent and protected
10705: words; i.e., it just has a method map. If a class implements an
10706: interface, its method map contains a pointer to the method map of the
10707: interface. The positive offsets in the map are reserved for class
10708: methods, therefore interface map pointers have negative
10709: offsets. Interfaces have offsets that are unique throughout the
10710: system, unlike class selectors, whose offsets are only unique for the
10711: classes where the selector is available (invokable).
10712: 
10713: This structure means that interface selectors have to perform one
10714: indirection more than class selectors to find their method. Their body
10715: contains the interface map pointer offset in the class method map, and
10716: the method offset in the interface method map. The
10717: @code{does>} action for an interface selector is, basically:
10718: 
10719: @example
10720: ( object selector-body )
10721: 2dup selector-interface @@ ( object selector-body object interface-offset )
10722: swap object-map @@ + @@ ( object selector-body map )
10723: swap selector-offset @@ + @@ execute
10724: @end example
10725: 
10726: where @code{object-map} and @code{selector-offset} are
10727: first fields and generate no code.
10728: 
10729: As a concrete example, consider the following code:
10730: 
10731: @example
10732: interface
10733:   selector if1sel1
10734:   selector if1sel2
10735: end-interface if1
10736: 
10737: object class
10738:   if1 implementation
10739:   selector cl1sel1
10740:   cell% inst-var cl1iv1
10741: 
10742: ' m1 overrides construct
10743: ' m2 overrides if1sel1
10744: ' m3 overrides if1sel2
10745: ' m4 overrides cl1sel2
10746: end-class cl1
10747: 
10748: create obj1 object dict-new drop
10749: create obj2 cl1    dict-new drop
10750: @end example
10751: 
10752: The data structure created by this code (including the data structure
10753: for @code{object}) is shown in the
10754: @uref{objects-implementation.eps,figure}, assuming a cell size of 4.
10755: @comment TODO add this diagram..
10756: 
10757: @node Objects Glossary,  , Objects Implementation, Objects
10758: @subsubsection @file{objects.fs} Glossary
10759: @cindex @file{objects.fs} Glossary
10760: 
10761: 
10762: doc---objects-bind
10763: doc---objects-<bind>
10764: doc---objects-bind'
10765: doc---objects-[bind]
10766: doc---objects-class
10767: doc---objects-class->map
10768: doc---objects-class-inst-size
10769: doc---objects-class-override!
10770: doc---objects-class-previous
10771: doc---objects-class>order
10772: doc---objects-construct
10773: doc---objects-current'
10774: doc---objects-[current]
10775: doc---objects-current-interface
10776: doc---objects-dict-new
10777: doc---objects-end-class
10778: doc---objects-end-class-noname
10779: doc---objects-end-interface
10780: doc---objects-end-interface-noname
10781: doc---objects-end-methods
10782: doc---objects-exitm
10783: doc---objects-heap-new
10784: doc---objects-implementation
10785: doc---objects-init-object
10786: doc---objects-inst-value
10787: doc---objects-inst-var
10788: doc---objects-interface
10789: doc---objects-m:
10790: doc---objects-:m
10791: doc---objects-;m
10792: doc---objects-method
10793: doc---objects-methods
10794: doc---objects-object
10795: doc---objects-overrides
10796: doc---objects-[parent]
10797: doc---objects-print
10798: doc---objects-protected
10799: doc---objects-public
10800: doc---objects-selector
10801: doc---objects-this
10802: doc---objects-<to-inst>
10803: doc---objects-[to-inst]
10804: doc---objects-to-this
10805: doc---objects-xt-new
10806: 
10807: 
10808: @c -------------------------------------------------------------
10809: @node OOF, Mini-OOF, Objects, Object-oriented Forth
10810: @subsection The @file{oof.fs} model
10811: @cindex oof
10812: @cindex object-oriented programming
10813: 
10814: @cindex @file{objects.fs}
10815: @cindex @file{oof.fs}
10816: 
10817: This section describes the @file{oof.fs} package.
10818: 
10819: The package described in this section has been used in bigFORTH since 1991, and
10820: used for two large applications: a chromatographic system used to
10821: create new medicaments, and a graphic user interface library (MINOS).
10822: 
10823: You can find a description (in German) of @file{oof.fs} in @cite{Object
10824: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
10825: 10(2), 1994.
10826: 
10827: @menu
10828: * Properties of the OOF model::  
10829: * Basic OOF Usage::             
10830: * The OOF base class::          
10831: * Class Declaration::           
10832: * Class Implementation::        
10833: @end menu
10834: 
10835: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
10836: @subsubsection Properties of the @file{oof.fs} model
10837: @cindex @file{oof.fs} properties
10838: 
10839: @itemize @bullet
10840: @item
10841: This model combines object oriented programming with information
10842: hiding. It helps you writing large application, where scoping is
10843: necessary, because it provides class-oriented scoping.
10844: 
10845: @item
10846: Named objects, object pointers, and object arrays can be created,
10847: selector invocation uses the ``object selector'' syntax. Selector invocation
10848: to objects and/or selectors on the stack is a bit less convenient, but
10849: possible.
10850: 
10851: @item
10852: Selector invocation and instance variable usage of the active object is
10853: straightforward, since both make use of the active object.
10854: 
10855: @item
10856: Late binding is efficient and easy to use.
10857: 
10858: @item
10859: State-smart objects parse selectors. However, extensibility is provided
10860: using a (parsing) selector @code{postpone} and a selector @code{'}.
10861: 
10862: @item
10863: An implementation in ANS Forth is available.
10864: 
10865: @end itemize
10866: 
10867: 
10868: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
10869: @subsubsection Basic @file{oof.fs} Usage
10870: @cindex @file{oof.fs} usage
10871: 
10872: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
10873: 
10874: You can define a class for graphical objects like this:
10875: 
10876: @cindex @code{class} usage
10877: @cindex @code{class;} usage
10878: @cindex @code{method} usage
10879: @example
10880: object class graphical \ "object" is the parent class
10881:   method draw ( x y graphical -- )
10882: class;
10883: @end example
10884: 
10885: This code defines a class @code{graphical} with an
10886: operation @code{draw}.  We can perform the operation
10887: @code{draw} on any @code{graphical} object, e.g.:
10888: 
10889: @example
10890: 100 100 t-rex draw
10891: @end example
10892: 
10893: @noindent
10894: where @code{t-rex} is an object or object pointer, created with e.g.
10895: @code{graphical : t-rex}.
10896: 
10897: @cindex abstract class
10898: How do we create a graphical object? With the present definitions,
10899: we cannot create a useful graphical object. The class
10900: @code{graphical} describes graphical objects in general, but not
10901: any concrete graphical object type (C++ users would call it an
10902: @emph{abstract class}); e.g., there is no method for the selector
10903: @code{draw} in the class @code{graphical}.
10904: 
10905: For concrete graphical objects, we define child classes of the
10906: class @code{graphical}, e.g.:
10907: 
10908: @example
10909: graphical class circle \ "graphical" is the parent class
10910:   cell var circle-radius
10911: how:
10912:   : draw ( x y -- )
10913:     circle-radius @@ draw-circle ;
10914: 
10915:   : init ( n-radius -- (
10916:     circle-radius ! ;
10917: class;
10918: @end example
10919: 
10920: Here we define a class @code{circle} as a child of @code{graphical},
10921: with a field @code{circle-radius}; it defines new methods for the
10922: selectors @code{draw} and @code{init} (@code{init} is defined in
10923: @code{object}, the parent class of @code{graphical}).
10924: 
10925: Now we can create a circle in the dictionary with:
10926: 
10927: @example
10928: 50 circle : my-circle
10929: @end example
10930: 
10931: @noindent
10932: @code{:} invokes @code{init}, thus initializing the field
10933: @code{circle-radius} with 50. We can draw this new circle at (100,100)
10934: with:
10935: 
10936: @example
10937: 100 100 my-circle draw
10938: @end example
10939: 
10940: @cindex selector invocation, restrictions
10941: @cindex class definition, restrictions
10942: Note: You can only invoke a selector if the receiving object belongs to
10943: the class where the selector was defined or one of its descendents;
10944: e.g., you can invoke @code{draw} only for objects belonging to
10945: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
10946: mechanism will check if you try to invoke a selector that is not
10947: defined in this class hierarchy, so you'll get an error at compilation
10948: time.
10949: 
10950: 
10951: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
10952: @subsubsection The @file{oof.fs} base class
10953: @cindex @file{oof.fs} base class
10954: 
10955: When you define a class, you have to specify a parent class.  So how do
10956: you start defining classes? There is one class available from the start:
10957: @code{object}. You have to use it as ancestor for all classes. It is the
10958: only class that has no parent. Classes are also objects, except that
10959: they don't have instance variables; class manipulation such as
10960: inheritance or changing definitions of a class is handled through
10961: selectors of the class @code{object}.
10962: 
10963: @code{object} provides a number of selectors:
10964: 
10965: @itemize @bullet
10966: @item
10967: @code{class} for subclassing, @code{definitions} to add definitions
10968: later on, and @code{class?} to get type informations (is the class a
10969: subclass of the class passed on the stack?).
10970: 
10971: doc---object-class
10972: doc---object-definitions
10973: doc---object-class?
10974: 
10975: 
10976: @item
10977: @code{init} and @code{dispose} as constructor and destructor of the
10978: object. @code{init} is invocated after the object's memory is allocated,
10979: while @code{dispose} also handles deallocation. Thus if you redefine
10980: @code{dispose}, you have to call the parent's dispose with @code{super
10981: dispose}, too.
10982: 
10983: doc---object-init
10984: doc---object-dispose
10985: 
10986: 
10987: @item
10988: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
10989: @code{[]} to create named and unnamed objects and object arrays or
10990: object pointers.
10991: 
10992: doc---object-new
10993: doc---object-new[]
10994: doc---object-:
10995: doc---object-ptr
10996: doc---object-asptr
10997: doc---object-[]
10998: 
10999: 
11000: @item
11001: @code{::} and @code{super} for explicit scoping. You should use explicit
11002: scoping only for super classes or classes with the same set of instance
11003: variables. Explicitly-scoped selectors use early binding.
11004: 
11005: doc---object-::
11006: doc---object-super
11007: 
11008: 
11009: @item
11010: @code{self} to get the address of the object
11011: 
11012: doc---object-self
11013: 
11014: 
11015: @item
11016: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
11017: pointers and instance defers.
11018: 
11019: doc---object-bind
11020: doc---object-bound
11021: doc---object-link
11022: doc---object-is
11023: 
11024: 
11025: @item
11026: @code{'} to obtain selector tokens, @code{send} to invocate selectors
11027: form the stack, and @code{postpone} to generate selector invocation code.
11028: 
11029: doc---object-'
11030: doc---object-postpone
11031: 
11032: 
11033: @item
11034: @code{with} and @code{endwith} to select the active object from the
11035: stack, and enable its scope. Using @code{with} and @code{endwith}
11036: also allows you to create code using selector @code{postpone} without being
11037: trapped by the state-smart objects.
11038: 
11039: doc---object-with
11040: doc---object-endwith
11041: 
11042: 
11043: @end itemize
11044: 
11045: @node Class Declaration, Class Implementation, The OOF base class, OOF
11046: @subsubsection Class Declaration
11047: @cindex class declaration
11048: 
11049: @itemize @bullet
11050: @item
11051: Instance variables
11052: 
11053: doc---oof-var
11054: 
11055: 
11056: @item
11057: Object pointers
11058: 
11059: doc---oof-ptr
11060: doc---oof-asptr
11061: 
11062: 
11063: @item
11064: Instance defers
11065: 
11066: doc---oof-defer
11067: 
11068: 
11069: @item
11070: Method selectors
11071: 
11072: doc---oof-early
11073: doc---oof-method
11074: 
11075: 
11076: @item
11077: Class-wide variables
11078: 
11079: doc---oof-static
11080: 
11081: 
11082: @item
11083: End declaration
11084: 
11085: doc---oof-how:
11086: doc---oof-class;
11087: 
11088: 
11089: @end itemize
11090: 
11091: @c -------------------------------------------------------------
11092: @node Class Implementation,  , Class Declaration, OOF
11093: @subsubsection Class Implementation
11094: @cindex class implementation
11095: 
11096: @c -------------------------------------------------------------
11097: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
11098: @subsection The @file{mini-oof.fs} model
11099: @cindex mini-oof
11100: 
11101: Gforth's third object oriented Forth package is a 12-liner. It uses a
11102: mixture of the @file{objects.fs} and the @file{oof.fs} syntax,
11103: and reduces to the bare minimum of features. This is based on a posting
11104: of Bernd Paysan in comp.lang.forth.
11105: 
11106: @menu
11107: * Basic Mini-OOF Usage::        
11108: * Mini-OOF Example::            
11109: * Mini-OOF Implementation::     
11110: @end menu
11111: 
11112: @c -------------------------------------------------------------
11113: @node Basic Mini-OOF Usage, Mini-OOF Example, Mini-OOF, Mini-OOF
11114: @subsubsection Basic @file{mini-oof.fs} Usage
11115: @cindex mini-oof usage
11116: 
11117: There is a base class (@code{class}, which allocates one cell for the
11118: object pointer) plus seven other words: to define a method, a variable,
11119: a class; to end a class, to resolve binding, to allocate an object and
11120: to compile a class method.
11121: @comment TODO better description of the last one
11122: 
11123: 
11124: doc-object
11125: doc-method
11126: doc-var
11127: doc-class
11128: doc-end-class
11129: doc-defines
11130: doc-new
11131: doc-::
11132: 
11133: 
11134: 
11135: @c -------------------------------------------------------------
11136: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
11137: @subsubsection Mini-OOF Example
11138: @cindex mini-oof example
11139: 
11140: A short example shows how to use this package. This example, in slightly
11141: extended form, is supplied as @file{moof-exm.fs}
11142: @comment TODO could flesh this out with some comments from the Forthwrite article
11143: 
11144: @example
11145: object class
11146:   method init
11147:   method draw
11148: end-class graphical
11149: @end example
11150: 
11151: This code defines a class @code{graphical} with an
11152: operation @code{draw}.  We can perform the operation
11153: @code{draw} on any @code{graphical} object, e.g.:
11154: 
11155: @example
11156: 100 100 t-rex draw
11157: @end example
11158: 
11159: where @code{t-rex} is an object or object pointer, created with e.g.
11160: @code{graphical new Constant t-rex}.
11161: 
11162: For concrete graphical objects, we define child classes of the
11163: class @code{graphical}, e.g.:
11164: 
11165: @example
11166: graphical class
11167:   cell var circle-radius
11168: end-class circle \ "graphical" is the parent class
11169: 
11170: :noname ( x y -- )
11171:   circle-radius @@ draw-circle ; circle defines draw
11172: :noname ( r -- )
11173:   circle-radius ! ; circle defines init
11174: @end example
11175: 
11176: There is no implicit init method, so we have to define one. The creation
11177: code of the object now has to call init explicitely.
11178: 
11179: @example
11180: circle new Constant my-circle
11181: 50 my-circle init
11182: @end example
11183: 
11184: It is also possible to add a function to create named objects with
11185: automatic call of @code{init}, given that all objects have @code{init}
11186: on the same place:
11187: 
11188: @example
11189: : new: ( .. o "name" -- )
11190:     new dup Constant init ;
11191: 80 circle new: large-circle
11192: @end example
11193: 
11194: We can draw this new circle at (100,100) with:
11195: 
11196: @example
11197: 100 100 my-circle draw
11198: @end example
11199: 
11200: @node Mini-OOF Implementation,  , Mini-OOF Example, Mini-OOF
11201: @subsubsection @file{mini-oof.fs} Implementation
11202: 
11203: Object-oriented systems with late binding typically use a
11204: ``vtable''-approach: the first variable in each object is a pointer to a
11205: table, which contains the methods as function pointers. The vtable
11206: may also contain other information.
11207: 
11208: So first, let's declare selectors:
11209: 
11210: @example
11211: : method ( m v "name" -- m' v ) Create  over , swap cell+ swap
11212:   DOES> ( ... o -- ... ) @@ over @@ + @@ execute ;
11213: @end example
11214: 
11215: During selector declaration, the number of selectors and instance
11216: variables is on the stack (in address units). @code{method} creates one
11217: selector and increments the selector number. To execute a selector, it
11218: takes the object, fetches the vtable pointer, adds the offset, and
11219: executes the method @i{xt} stored there. Each selector takes the object
11220: it is invoked with as top of stack parameter; it passes the parameters
11221: (including the object) unchanged to the appropriate method which should
11222: consume that object.
11223: 
11224: Now, we also have to declare instance variables
11225: 
11226: @example
11227: : var ( m v size "name" -- m v' ) Create  over , +
11228:   DOES> ( o -- addr ) @@ + ;
11229: @end example
11230: 
11231: As before, a word is created with the current offset. Instance
11232: variables can have different sizes (cells, floats, doubles, chars), so
11233: all we do is take the size and add it to the offset. If your machine
11234: has alignment restrictions, put the proper @code{aligned} or
11235: @code{faligned} before the variable, to adjust the variable
11236: offset. That's why it is on the top of stack.
11237: 
11238: We need a starting point (the base object) and some syntactic sugar:
11239: 
11240: @example
11241: Create object  1 cells , 2 cells ,
11242: : class ( class -- class selectors vars ) dup 2@@ ;
11243: @end example
11244: 
11245: For inheritance, the vtable of the parent object has to be
11246: copied when a new, derived class is declared. This gives all the
11247: methods of the parent class, which can be overridden, though.
11248: 
11249: @example
11250: : end-class  ( class selectors vars "name" -- )
11251:   Create  here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
11252:   cell+ dup cell+ r> rot @@ 2 cells /string move ;
11253: @end example
11254: 
11255: The first line creates the vtable, initialized with
11256: @code{noop}s. The second line is the inheritance mechanism, it
11257: copies the xts from the parent vtable.
11258: 
11259: We still have no way to define new methods, let's do that now:
11260: 
11261: @example
11262: : defines ( xt class "name" -- ) ' >body @@ + ! ;
11263: @end example
11264: 
11265: To allocate a new object, we need a word, too:
11266: 
11267: @example
11268: : new ( class -- o )  here over @@ allot swap over ! ;
11269: @end example
11270: 
11271: Sometimes derived classes want to access the method of the
11272: parent object. There are two ways to achieve this with Mini-OOF:
11273: first, you could use named words, and second, you could look up the
11274: vtable of the parent object.
11275: 
11276: @example
11277: : :: ( class "name" -- ) ' >body @@ + @@ compile, ;
11278: @end example
11279: 
11280: 
11281: Nothing can be more confusing than a good example, so here is
11282: one. First let's declare a text object (called
11283: @code{button}), that stores text and position:
11284: 
11285: @example
11286: object class
11287:   cell var text
11288:   cell var len
11289:   cell var x
11290:   cell var y
11291:   method init
11292:   method draw
11293: end-class button
11294: @end example
11295: 
11296: @noindent
11297: Now, implement the two methods, @code{draw} and @code{init}:
11298: 
11299: @example
11300: :noname ( o -- )
11301:  >r r@@ x @@ r@@ y @@ at-xy  r@@ text @@ r> len @@ type ;
11302:  button defines draw
11303: :noname ( addr u o -- )
11304:  >r 0 r@@ x ! 0 r@@ y ! r@@ len ! r> text ! ;
11305:  button defines init
11306: @end example
11307: 
11308: @noindent
11309: To demonstrate inheritance, we define a class @code{bold-button}, with no
11310: new data and no new selectors:
11311: 
11312: @example
11313: button class
11314: end-class bold-button
11315: 
11316: : bold   27 emit ." [1m" ;
11317: : normal 27 emit ." [0m" ;
11318: @end example
11319: 
11320: @noindent
11321: The class @code{bold-button} has a different draw method to
11322: @code{button}, but the new method is defined in terms of the draw method
11323: for @code{button}:
11324: 
11325: @example
11326: :noname bold [ button :: draw ] normal ; bold-button defines draw
11327: @end example
11328: 
11329: @noindent
11330: Finally, create two objects and apply selectors:
11331: 
11332: @example
11333: button new Constant foo
11334: s" thin foo" foo init
11335: page
11336: foo draw
11337: bold-button new Constant bar
11338: s" fat bar" bar init
11339: 1 bar y !
11340: bar draw
11341: @end example
11342: 
11343: 
11344: @node Comparison with other object models,  , Mini-OOF, Object-oriented Forth
11345: @subsection Comparison with other object models
11346: @cindex comparison of object models
11347: @cindex object models, comparison
11348: 
11349: Many object-oriented Forth extensions have been proposed (@cite{A survey
11350: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
11351: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
11352: relation of the object models described here to two well-known and two
11353: closely-related (by the use of method maps) models.  Andras Zsoter
11354: helped us with this section.
11355: 
11356: @cindex Neon model
11357: The most popular model currently seems to be the Neon model (see
11358: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
11359: 1997) by Andrew McKewan) but this model has a number of limitations
11360: @footnote{A longer version of this critique can be
11361: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
11362: Dimensions, May 1997) by Anton Ertl.}:
11363: 
11364: @itemize @bullet
11365: @item
11366: It uses a @code{@emph{selector object}} syntax, which makes it unnatural
11367: to pass objects on the stack.
11368: 
11369: @item
11370: It requires that the selector parses the input stream (at
11371: compile time); this leads to reduced extensibility and to bugs that are
11372: hard to find.
11373: 
11374: @item
11375: It allows using every selector on every object; this eliminates the
11376: need for interfaces, but makes it harder to create efficient
11377: implementations.
11378: @end itemize
11379: 
11380: @cindex Pountain's object-oriented model
11381: Another well-known publication is @cite{Object-Oriented Forth} (Academic
11382: Press, London, 1987) by Dick Pountain. However, it is not really about
11383: object-oriented programming, because it hardly deals with late
11384: binding. Instead, it focuses on features like information hiding and
11385: overloading that are characteristic of modular languages like Ada (83).
11386: 
11387: @cindex Zsoter's object-oriented model
11388: In @uref{http://www.forth.org/oopf.html, Does late binding have to be
11389: slow?} (Forth Dimensions 18(1) 1996, pages 31-35) Andras Zsoter
11390: describes a model that makes heavy use of an active object (like
11391: @code{this} in @file{objects.fs}): The active object is not only used
11392: for accessing all fields, but also specifies the receiving object of
11393: every selector invocation; you have to change the active object
11394: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
11395: changes more or less implicitly at @code{m: ... ;m}. Such a change at
11396: the method entry point is unnecessary with Zsoter's model, because the
11397: receiving object is the active object already. On the other hand, the
11398: explicit change is absolutely necessary in that model, because otherwise
11399: no one could ever change the active object. An ANS Forth implementation
11400: of this model is available through
11401: @uref{http://www.forth.org/oopf.html}.
11402: 
11403: @cindex @file{oof.fs}, differences to other models
11404: The @file{oof.fs} model combines information hiding and overloading
11405: resolution (by keeping names in various word lists) with object-oriented
11406: programming. It sets the active object implicitly on method entry, but
11407: also allows explicit changing (with @code{>o...o>} or with
11408: @code{with...endwith}). It uses parsing and state-smart objects and
11409: classes for resolving overloading and for early binding: the object or
11410: class parses the selector and determines the method from this. If the
11411: selector is not parsed by an object or class, it performs a call to the
11412: selector for the active object (late binding), like Zsoter's model.
11413: Fields are always accessed through the active object. The big
11414: disadvantage of this model is the parsing and the state-smartness, which
11415: reduces extensibility and increases the opportunities for subtle bugs;
11416: essentially, you are only safe if you never tick or @code{postpone} an
11417: object or class (Bernd disagrees, but I (Anton) am not convinced).
11418: 
11419: @cindex @file{mini-oof.fs}, differences to other models
11420: The @file{mini-oof.fs} model is quite similar to a very stripped-down
11421: version of the @file{objects.fs} model, but syntactically it is a
11422: mixture of the @file{objects.fs} and @file{oof.fs} models.
11423: 
11424: 
11425: @c -------------------------------------------------------------
11426: @node Programming Tools, Assembler and Code Words, Object-oriented Forth, Words
11427: @section Programming Tools
11428: @cindex programming tools
11429: 
11430: @c !! move this and assembler down below OO stuff.
11431: 
11432: @menu
11433: * Examining::                   
11434: * Forgetting words::            
11435: * Debugging::                   Simple and quick.
11436: * Assertions::                  Making your programs self-checking.
11437: * Singlestep Debugger::         Executing your program word by word.
11438: @end menu
11439: 
11440: @node Examining, Forgetting words, Programming Tools, Programming Tools
11441: @subsection Examining data and code
11442: @cindex examining data and code
11443: @cindex data examination
11444: @cindex code examination
11445: 
11446: The following words inspect the stack non-destructively:
11447: 
11448: doc-.s
11449: doc-f.s
11450: 
11451: There is a word @code{.r} but it does @i{not} display the return stack!
11452: It is used for formatted numeric output (@pxref{Simple numeric output}).
11453: 
11454: doc-depth
11455: doc-fdepth
11456: doc-clearstack
11457: 
11458: The following words inspect memory.
11459: 
11460: doc-?
11461: doc-dump
11462: 
11463: And finally, @code{see} allows to inspect code:
11464: 
11465: doc-see
11466: doc-xt-see
11467: 
11468: @node Forgetting words, Debugging, Examining, Programming Tools
11469: @subsection Forgetting words
11470: @cindex words, forgetting
11471: @cindex forgeting words
11472: 
11473: @c  anton: other, maybe better places for this subsection: Defining Words;
11474: @c  Dictionary allocation.  At least a reference should be there.
11475: 
11476: Forth allows you to forget words (and everything that was alloted in the
11477: dictonary after them) in a LIFO manner.
11478: 
11479: doc-marker
11480: 
11481: The most common use of this feature is during progam development: when
11482: you change a source file, forget all the words it defined and load it
11483: again (since you also forget everything defined after the source file
11484: was loaded, you have to reload that, too).  Note that effects like
11485: storing to variables and destroyed system words are not undone when you
11486: forget words.  With a system like Gforth, that is fast enough at
11487: starting up and compiling, I find it more convenient to exit and restart
11488: Gforth, as this gives me a clean slate.
11489: 
11490: Here's an example of using @code{marker} at the start of a source file
11491: that you are debugging; it ensures that you only ever have one copy of
11492: the file's definitions compiled at any time:
11493: 
11494: @example
11495: [IFDEF] my-code
11496:     my-code
11497: [ENDIF]
11498: 
11499: marker my-code
11500: init-included-files
11501: 
11502: \ .. definitions start here
11503: \ .
11504: \ .
11505: \ end
11506: @end example
11507: 
11508: 
11509: @node Debugging, Assertions, Forgetting words, Programming Tools
11510: @subsection Debugging
11511: @cindex debugging
11512: 
11513: Languages with a slow edit/compile/link/test development loop tend to
11514: require sophisticated tracing/stepping debuggers to facilate debugging.
11515: 
11516: A much better (faster) way in fast-compiling languages is to add
11517: printing code at well-selected places, let the program run, look at
11518: the output, see where things went wrong, add more printing code, etc.,
11519: until the bug is found.
11520: 
11521: The simple debugging aids provided in @file{debugs.fs}
11522: are meant to support this style of debugging.
11523: 
11524: The word @code{~~} prints debugging information (by default the source
11525: location and the stack contents). It is easy to insert. If you use Emacs
11526: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
11527: query-replace them with nothing). The deferred words
11528: @code{printdebugdata} and @code{.debugline} control the output of
11529: @code{~~}. The default source location output format works well with
11530: Emacs' compilation mode, so you can step through the program at the
11531: source level using @kbd{C-x `} (the advantage over a stepping debugger
11532: is that you can step in any direction and you know where the crash has
11533: happened or where the strange data has occurred).
11534: 
11535: doc-~~
11536: doc-printdebugdata
11537: doc-.debugline
11538: 
11539: @cindex filenames in @code{~~} output
11540: @code{~~} (and assertions) will usually print the wrong file name if a
11541: marker is executed in the same file after their occurance.  They will
11542: print @samp{*somewhere*} as file name if a marker is executed in the
11543: same file before their occurance.
11544: 
11545: 
11546: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
11547: @subsection Assertions
11548: @cindex assertions
11549: 
11550: It is a good idea to make your programs self-checking, especially if you
11551: make an assumption that may become invalid during maintenance (for
11552: example, that a certain field of a data structure is never zero). Gforth
11553: supports @dfn{assertions} for this purpose. They are used like this:
11554: 
11555: @example
11556: assert( @i{flag} )
11557: @end example
11558: 
11559: The code between @code{assert(} and @code{)} should compute a flag, that
11560: should be true if everything is alright and false otherwise. It should
11561: not change anything else on the stack. The overall stack effect of the
11562: assertion is @code{( -- )}. E.g.
11563: 
11564: @example
11565: assert( 1 1 + 2 = ) \ what we learn in school
11566: assert( dup 0<> ) \ assert that the top of stack is not zero
11567: assert( false ) \ this code should not be reached
11568: @end example
11569: 
11570: The need for assertions is different at different times. During
11571: debugging, we want more checking, in production we sometimes care more
11572: for speed. Therefore, assertions can be turned off, i.e., the assertion
11573: becomes a comment. Depending on the importance of an assertion and the
11574: time it takes to check it, you may want to turn off some assertions and
11575: keep others turned on. Gforth provides several levels of assertions for
11576: this purpose:
11577: 
11578: 
11579: doc-assert0(
11580: doc-assert1(
11581: doc-assert2(
11582: doc-assert3(
11583: doc-assert(
11584: doc-)
11585: 
11586: 
11587: The variable @code{assert-level} specifies the highest assertions that
11588: are turned on. I.e., at the default @code{assert-level} of one,
11589: @code{assert0(} and @code{assert1(} assertions perform checking, while
11590: @code{assert2(} and @code{assert3(} assertions are treated as comments.
11591: 
11592: The value of @code{assert-level} is evaluated at compile-time, not at
11593: run-time. Therefore you cannot turn assertions on or off at run-time;
11594: you have to set the @code{assert-level} appropriately before compiling a
11595: piece of code. You can compile different pieces of code at different
11596: @code{assert-level}s (e.g., a trusted library at level 1 and
11597: newly-written code at level 3).
11598: 
11599: 
11600: doc-assert-level
11601: 
11602: 
11603: If an assertion fails, a message compatible with Emacs' compilation mode
11604: is produced and the execution is aborted (currently with @code{ABORT"}.
11605: If there is interest, we will introduce a special throw code. But if you
11606: intend to @code{catch} a specific condition, using @code{throw} is
11607: probably more appropriate than an assertion).
11608: 
11609: @cindex filenames in assertion output
11610: Assertions (and @code{~~}) will usually print the wrong file name if a
11611: marker is executed in the same file after their occurance.  They will
11612: print @samp{*somewhere*} as file name if a marker is executed in the
11613: same file before their occurance.
11614: 
11615: Definitions in ANS Forth for these assertion words are provided
11616: in @file{compat/assert.fs}.
11617: 
11618: 
11619: @node Singlestep Debugger,  , Assertions, Programming Tools
11620: @subsection Singlestep Debugger
11621: @cindex singlestep Debugger
11622: @cindex debugging Singlestep
11623: 
11624: When you create a new word there's often the need to check whether it
11625: behaves correctly or not. You can do this by typing @code{dbg
11626: badword}. A debug session might look like this:
11627: 
11628: @example
11629: : badword 0 DO i . LOOP ;  ok
11630: 2 dbg badword 
11631: : badword  
11632: Scanning code...
11633: 
11634: Nesting debugger ready!
11635: 
11636: 400D4738  8049BC4 0              -> [ 2 ] 00002 00000 
11637: 400D4740  8049F68 DO             -> [ 0 ] 
11638: 400D4744  804A0C8 i              -> [ 1 ] 00000 
11639: 400D4748 400C5E60 .              -> 0 [ 0 ] 
11640: 400D474C  8049D0C LOOP           -> [ 0 ] 
11641: 400D4744  804A0C8 i              -> [ 1 ] 00001 
11642: 400D4748 400C5E60 .              -> 1 [ 0 ] 
11643: 400D474C  8049D0C LOOP           -> [ 0 ] 
11644: 400D4758  804B384 ;              ->  ok
11645: @end example
11646: 
11647: Each line displayed is one step. You always have to hit return to
11648: execute the next word that is displayed. If you don't want to execute
11649: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
11650: an overview what keys are available:
11651: 
11652: @table @i
11653: 
11654: @item @key{RET}
11655: Next; Execute the next word.
11656: 
11657: @item n
11658: Nest; Single step through next word.
11659: 
11660: @item u
11661: Unnest; Stop debugging and execute rest of word. If we got to this word
11662: with nest, continue debugging with the calling word.
11663: 
11664: @item d
11665: Done; Stop debugging and execute rest.
11666: 
11667: @item s
11668: Stop; Abort immediately.
11669: 
11670: @end table
11671: 
11672: Debugging large application with this mechanism is very difficult, because
11673: you have to nest very deeply into the program before the interesting part
11674: begins. This takes a lot of time. 
11675: 
11676: To do it more directly put a @code{BREAK:} command into your source code.
11677: When program execution reaches @code{BREAK:} the single step debugger is
11678: invoked and you have all the features described above.
11679: 
11680: If you have more than one part to debug it is useful to know where the
11681: program has stopped at the moment. You can do this by the 
11682: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
11683: string is typed out when the ``breakpoint'' is reached.
11684: 
11685: 
11686: doc-dbg
11687: doc-break:
11688: doc-break"
11689: 
11690: 
11691: 
11692: @c -------------------------------------------------------------
11693: @node Assembler and Code Words, Threading Words, Programming Tools, Words
11694: @section Assembler and Code Words
11695: @cindex assembler
11696: @cindex code words
11697: 
11698: @menu
11699: * Code and ;code::              
11700: * Common Assembler::            Assembler Syntax
11701: * Common Disassembler::         
11702: * 386 Assembler::               Deviations and special cases
11703: * Alpha Assembler::             Deviations and special cases
11704: * MIPS assembler::              Deviations and special cases
11705: * Other assemblers::            How to write them
11706: @end menu
11707: 
11708: @node Code and ;code, Common Assembler, Assembler and Code Words, Assembler and Code Words
11709: @subsection @code{Code} and @code{;code}
11710: 
11711: Gforth provides some words for defining primitives (words written in
11712: machine code), and for defining the machine-code equivalent of
11713: @code{DOES>}-based defining words. However, the machine-independent
11714: nature of Gforth poses a few problems: First of all, Gforth runs on
11715: several architectures, so it can provide no standard assembler. What's
11716: worse is that the register allocation not only depends on the processor,
11717: but also on the @code{gcc} version and options used.
11718: 
11719: The words that Gforth offers encapsulate some system dependences (e.g.,
11720: the header structure), so a system-independent assembler may be used in
11721: Gforth. If you do not have an assembler, you can compile machine code
11722: directly with @code{,} and @code{c,}@footnote{This isn't portable,
11723: because these words emit stuff in @i{data} space; it works because
11724: Gforth has unified code/data spaces. Assembler isn't likely to be
11725: portable anyway.}.
11726: 
11727: 
11728: doc-assembler
11729: doc-init-asm
11730: doc-code
11731: doc-end-code
11732: doc-;code
11733: doc-flush-icache
11734: 
11735: 
11736: If @code{flush-icache} does not work correctly, @code{code} words
11737: etc. will not work (reliably), either.
11738: 
11739: The typical usage of these @code{code} words can be shown most easily by
11740: analogy to the equivalent high-level defining words:
11741: 
11742: @example
11743: : foo                              code foo
11744:    <high-level Forth words>              <assembler>
11745: ;                                  end-code
11746:                                 
11747: : bar                              : bar
11748:    <high-level Forth words>           <high-level Forth words>
11749:    CREATE                             CREATE
11750:       <high-level Forth words>           <high-level Forth words>
11751:    DOES>                              ;code
11752:       <high-level Forth words>           <assembler>
11753: ;                                  end-code
11754: @end example
11755: 
11756: @c anton: the following stuff is also in "Common Assembler", in less detail.
11757: 
11758: @cindex registers of the inner interpreter
11759: In the assembly code you will want to refer to the inner interpreter's
11760: registers (e.g., the data stack pointer) and you may want to use other
11761: registers for temporary storage. Unfortunately, the register allocation
11762: is installation-dependent.
11763: 
11764: In particular, @code{ip} (Forth instruction pointer) and @code{rp}
11765: (return stack pointer) may be in different places in @code{gforth} and
11766: @code{gforth-fast}, or different installations.  This means that you
11767: cannot write a @code{NEXT} routine that works reliably on both versions
11768: or different installations; so for doing @code{NEXT}, I recommend
11769: jumping to @code{' noop >code-address}, which contains nothing but a
11770: @code{NEXT}.
11771: 
11772: For general accesses to the inner interpreter's registers, the easiest
11773: solution is to use explicit register declarations (@pxref{Explicit Reg
11774: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) for
11775: all of the inner interpreter's registers: You have to compile Gforth
11776: with @code{-DFORCE_REG} (configure option @code{--enable-force-reg}) and
11777: the appropriate declarations must be present in the @code{machine.h}
11778: file (see @code{mips.h} for an example; you can find a full list of all
11779: declarable register symbols with @code{grep register engine.c}). If you
11780: give explicit registers to all variables that are declared at the
11781: beginning of @code{engine()}, you should be able to use the other
11782: caller-saved registers for temporary storage. Alternatively, you can use
11783: the @code{gcc} option @code{-ffixed-REG} (@pxref{Code Gen Options, ,
11784: Options for Code Generation Conventions, gcc.info, GNU C Manual}) to
11785: reserve a register (however, this restriction on register allocation may
11786: slow Gforth significantly).
11787: 
11788: If this solution is not viable (e.g., because @code{gcc} does not allow
11789: you to explicitly declare all the registers you need), you have to find
11790: out by looking at the code where the inner interpreter's registers
11791: reside and which registers can be used for temporary storage. You can
11792: get an assembly listing of the engine's code with @code{make engine.s}.
11793: 
11794: In any case, it is good practice to abstract your assembly code from the
11795: actual register allocation. E.g., if the data stack pointer resides in
11796: register @code{$17}, create an alias for this register called @code{sp},
11797: and use that in your assembly code.
11798: 
11799: @cindex code words, portable
11800: Another option for implementing normal and defining words efficiently
11801: is to add the desired functionality to the source of Gforth. For normal
11802: words you just have to edit @file{primitives} (@pxref{Automatic
11803: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
11804: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
11805: @file{prims2x.fs}, and possibly @file{cross.fs}.
11806: 
11807: @node Common Assembler, Common Disassembler, Code and ;code, Assembler and Code Words
11808: @subsection Common Assembler
11809: 
11810: The assemblers in Gforth generally use a postfix syntax, i.e., the
11811: instruction name follows the operands.
11812: 
11813: The operands are passed in the usual order (the same that is used in the
11814: manual of the architecture).  Since they all are Forth words, they have
11815: to be separated by spaces; you can also use Forth words to compute the
11816: operands.
11817: 
11818: The instruction names usually end with a @code{,}.  This makes it easier
11819: to visually separate instructions if you put several of them on one
11820: line; it also avoids shadowing other Forth words (e.g., @code{and}).
11821: 
11822: Registers are usually specified by number; e.g., (decimal) @code{11}
11823: specifies registers R11 and F11 on the Alpha architecture (which one,
11824: depends on the instruction).  The usual names are also available, e.g.,
11825: @code{s2} for R11 on Alpha.
11826: 
11827: Control flow is specified similar to normal Forth code (@pxref{Arbitrary
11828: control structures}), with @code{if,}, @code{ahead,}, @code{then,},
11829: @code{begin,}, @code{until,}, @code{again,}, @code{cs-roll},
11830: @code{cs-pick}, @code{else,}, @code{while,}, and @code{repeat,}.  The
11831: conditions are specified in a way specific to each assembler.
11832: 
11833: Note that the register assignments of the Gforth engine can change
11834: between Gforth versions, or even between different compilations of the
11835: same Gforth version (e.g., if you use a different GCC version).  So if
11836: you want to refer to Gforth's registers (e.g., the stack pointer or
11837: TOS), I recommend defining your own words for refering to these
11838: registers, and using them later on; then you can easily adapt to a
11839: changed register assignment.  The stability of the register assignment
11840: is usually better if you build Gforth with @code{--enable-force-reg}.
11841: 
11842: The most common use of these registers is to dispatch to the next word
11843: (the @code{next} routine).  A portable way to do this is to jump to
11844: @code{' noop >code-address} (of course, this is less efficient than
11845: integrating the @code{next} code and scheduling it well).
11846: 
11847: Another difference between Gforth version is that the top of stack is
11848: kept in memory in @code{gforth} and, on most platforms, in a register in
11849: @code{gforth-fast}.
11850: 
11851: @node  Common Disassembler, 386 Assembler, Common Assembler, Assembler and Code Words
11852: @subsection Common Disassembler
11853: 
11854: You can disassemble a @code{code} word with @code{see}
11855: (@pxref{Debugging}).  You can disassemble a section of memory with
11856: 
11857: doc-disasm
11858: 
11859: The disassembler generally produces output that can be fed into the
11860: assembler (i.e., same syntax, etc.).  It also includes additional
11861: information in comments.  In particular, the address of the instruction
11862: is given in a comment before the instruction.
11863: 
11864: @code{See} may display more or less than the actual code of the word,
11865: because the recognition of the end of the code is unreliable.  You can
11866: use @code{disasm} if it did not display enough.  It may display more, if
11867: the code word is not immediately followed by a named word.  If you have
11868: something else there, you can follow the word with @code{align last @ ,}
11869: to ensure that the end is recognized.
11870: 
11871: @node 386 Assembler, Alpha Assembler, Common Disassembler, Assembler and Code Words
11872: @subsection 386 Assembler
11873: 
11874: The 386 assembler included in Gforth was written by Bernd Paysan, it's
11875: available under GPL, and originally part of bigFORTH.
11876: 
11877: The 386 disassembler included in Gforth was written by Andrew McKewan
11878: and is in the public domain.
11879: 
11880: The disassembler displays code in an Intel-like prefix syntax.
11881: 
11882: The assembler uses a postfix syntax with reversed parameters.
11883: 
11884: The assembler includes all instruction of the Athlon, i.e. 486 core
11885: instructions, Pentium and PPro extensions, floating point, MMX, 3Dnow!,
11886: but not ISSE. It's an integrated 16- and 32-bit assembler. Default is 32
11887: bit, you can switch to 16 bit with .86 and back to 32 bit with .386.
11888: 
11889: There are several prefixes to switch between different operation sizes,
11890: @code{.b} for byte accesses, @code{.w} for word accesses, @code{.d} for
11891: double-word accesses. Addressing modes can be switched with @code{.wa}
11892: for 16 bit addresses, and @code{.da} for 32 bit addresses. You don't
11893: need a prefix for byte register names (@code{AL} et al).
11894: 
11895: For floating point operations, the prefixes are @code{.fs} (IEEE
11896: single), @code{.fl} (IEEE double), @code{.fx} (extended), @code{.fw}
11897: (word), @code{.fd} (double-word), and @code{.fq} (quad-word).
11898: 
11899: The MMX opcodes don't have size prefixes, they are spelled out like in
11900: the Intel assembler. Instead of move from and to memory, there are
11901: PLDQ/PLDD and PSTQ/PSTD.
11902: 
11903: The registers lack the 'e' prefix; even in 32 bit mode, eax is called
11904: ax.  Immediate values are indicated by postfixing them with @code{#},
11905: e.g., @code{3 #}.  Here are some examples of addressing modes in various
11906: syntaxes:
11907: 
11908: @example
11909: Gforth          Intel (NASM)   AT&T (gas)      Name
11910: .w ax           ax             %ax             register (16 bit)
11911: ax              eax            %eax            register (32 bit)
11912: 3 #             offset 3       $3              immediate
11913: 1000 #)         byte ptr 1000  1000            displacement
11914: bx )            [ebx]          (%ebx)          base
11915: 100 di d)       100[edi]       100(%edi)       base+displacement
11916: 20 ax *4 i#)    20[eax*4]      20(,%eax,4)     (index*scale)+displacement
11917: di ax *4 i)     [edi][eax*4]   (%edi,%eax,4)   base+(index*scale)
11918: 4 bx cx di)     4[ebx][ecx]    4(%ebx,%ecx)    base+index+displacement
11919: 12 sp ax *2 di) 12[esp][eax*2] 12(%esp,%eax,2) base+(index*scale)+displacement
11920: @end example
11921: 
11922: You can use @code{L)} and @code{LI)} instead of @code{D)} and
11923: @code{DI)} to enforce 32-bit displacement fields (useful for
11924: later patching).
11925: 
11926: Some example of instructions are:
11927: 
11928: @example
11929: ax bx mov             \ move ebx,eax
11930: 3 # ax mov            \ mov eax,3
11931: 100 di ) ax mov       \ mov eax,100[edi]
11932: 4 bx cx di) ax mov    \ mov eax,4[ebx][ecx]
11933: .w ax bx mov          \ mov bx,ax
11934: @end example
11935: 
11936: The following forms are supported for binary instructions:
11937: 
11938: @example
11939: <reg> <reg> <inst>
11940: <n> # <reg> <inst>
11941: <mem> <reg> <inst>
11942: <reg> <mem> <inst>
11943: @end example
11944: 
11945: Immediate to memory is not supported.  The shift/rotate syntax is:
11946: 
11947: @example
11948: <reg/mem> 1 # shl \ shortens to shift without immediate
11949: <reg/mem> 4 # shl
11950: <reg/mem> cl shl
11951: @end example
11952: 
11953: Precede string instructions (@code{movs} etc.) with @code{.b} to get
11954: the byte version.
11955: 
11956: The control structure words @code{IF} @code{UNTIL} etc. must be preceded
11957: by one of these conditions: @code{vs vc u< u>= 0= 0<> u<= u> 0< 0>= ps
11958: pc < >= <= >}. (Note that most of these words shadow some Forth words
11959: when @code{assembler} is in front of @code{forth} in the search path,
11960: e.g., in @code{code} words).  Currently the control structure words use
11961: one stack item, so you have to use @code{roll} instead of @code{cs-roll}
11962: to shuffle them (you can also use @code{swap} etc.).
11963: 
11964: Here is an example of a @code{code} word (assumes that the stack pointer
11965: is in esi and the TOS is in ebx):
11966: 
11967: @example
11968: code my+ ( n1 n2 -- n )
11969:     4 si D) bx add
11970:     4 # si add
11971:     Next
11972: end-code
11973: @end example
11974: 
11975: @node Alpha Assembler, MIPS assembler, 386 Assembler, Assembler and Code Words
11976: @subsection Alpha Assembler
11977: 
11978: The Alpha assembler and disassembler were originally written by Bernd
11979: Thallner.
11980: 
11981: The register names @code{a0}--@code{a5} are not available to avoid
11982: shadowing hex numbers.
11983: 
11984: Immediate forms of arithmetic instructions are distinguished by a
11985: @code{#} just before the @code{,}, e.g., @code{and#,} (note: @code{lda,}
11986: does not count as arithmetic instruction).
11987: 
11988: You have to specify all operands to an instruction, even those that
11989: other assemblers consider optional, e.g., the destination register for
11990: @code{br,}, or the destination register and hint for @code{jmp,}.
11991: 
11992: You can specify conditions for @code{if,} by removing the first @code{b}
11993: and the trailing @code{,} from a branch with a corresponding name; e.g.,
11994: 
11995: @example
11996: 11 fgt if, \ if F11>0e
11997:   ...
11998: endif,
11999: @end example
12000: 
12001: @code{fbgt,} gives @code{fgt}.  
12002: 
12003: @node MIPS assembler, Other assemblers, Alpha Assembler, Assembler and Code Words
12004: @subsection MIPS assembler
12005: 
12006: The MIPS assembler was originally written by Christian Pirker.
12007: 
12008: Currently the assembler and disassembler only cover the MIPS-I
12009: architecture (R3000), and don't support FP instructions.
12010: 
12011: The register names @code{$a0}--@code{$a3} are not available to avoid
12012: shadowing hex numbers.
12013: 
12014: Because there is no way to distinguish registers from immediate values,
12015: you have to explicitly use the immediate forms of instructions, i.e.,
12016: @code{addiu,}, not just @code{addu,} (@command{as} does this
12017: implicitly).
12018: 
12019: If the architecture manual specifies several formats for the instruction
12020: (e.g., for @code{jalr,}), you usually have to use the one with more
12021: arguments (i.e., two for @code{jalr,}).  When in doubt, see
12022: @code{arch/mips/testasm.fs} for an example of correct use.
12023: 
12024: Branches and jumps in the MIPS architecture have a delay slot.  You have
12025: to fill it yourself (the simplest way is to use @code{nop,}), the
12026: assembler does not do it for you (unlike @command{as}).  Even
12027: @code{if,}, @code{ahead,}, @code{until,}, @code{again,}, @code{while,},
12028: @code{else,} and @code{repeat,} need a delay slot.  Since @code{begin,}
12029: and @code{then,} just specify branch targets, they are not affected.
12030: 
12031: Note that you must not put branches, jumps, or @code{li,} into the delay
12032: slot: @code{li,} may expand to several instructions, and control flow
12033: instructions may not be put into the branch delay slot in any case.
12034: 
12035: For branches the argument specifying the target is a relative address;
12036: You have to add the address of the delay slot to get the absolute
12037: address.
12038: 
12039: The MIPS architecture also has load delay slots and restrictions on
12040: using @code{mfhi,} and @code{mflo,}; you have to order the instructions
12041: yourself to satisfy these restrictions, the assembler does not do it for
12042: you.
12043: 
12044: You can specify the conditions for @code{if,} etc. by taking a
12045: conditional branch and leaving away the @code{b} at the start and the
12046: @code{,} at the end.  E.g.,
12047: 
12048: @example
12049: 4 5 eq if,
12050:   ... \ do something if $4 equals $5
12051: then,
12052: @end example
12053: 
12054: @node Other assemblers,  , MIPS assembler, Assembler and Code Words
12055: @subsection Other assemblers
12056: 
12057: If you want to contribute another assembler/disassembler, please contact
12058: us (@email{anton@@mips.complang.tuwien.ac.at}) to check if we have such
12059: an assembler already.  If you are writing them from scratch, please use
12060: a similar syntax style as the one we use (i.e., postfix, commas at the
12061: end of the instruction names, @pxref{Common Assembler}); make the output
12062: of the disassembler be valid input for the assembler, and keep the style
12063: similar to the style we used.
12064: 
12065: Hints on implementation: The most important part is to have a good test
12066: suite that contains all instructions.  Once you have that, the rest is
12067: easy.  For actual coding you can take a look at
12068: @file{arch/mips/disasm.fs} to get some ideas on how to use data for both
12069: the assembler and disassembler, avoiding redundancy and some potential
12070: bugs.  You can also look at that file (and @pxref{Advanced does> usage
12071: example}) to get ideas how to factor a disassembler.
12072: 
12073: Start with the disassembler, because it's easier to reuse data from the
12074: disassembler for the assembler than the other way round.
12075: 
12076: For the assembler, take a look at @file{arch/alpha/asm.fs}, which shows
12077: how simple it can be.
12078: 
12079: @c -------------------------------------------------------------
12080: @node Threading Words, Passing Commands to the OS, Assembler and Code Words, Words
12081: @section Threading Words
12082: @cindex threading words
12083: 
12084: @cindex code address
12085: These words provide access to code addresses and other threading stuff
12086: in Gforth (and, possibly, other interpretive Forths). It more or less
12087: abstracts away the differences between direct and indirect threading
12088: (and, for direct threading, the machine dependences). However, at
12089: present this wordset is still incomplete. It is also pretty low-level;
12090: some day it will hopefully be made unnecessary by an internals wordset
12091: that abstracts implementation details away completely.
12092: 
12093: The terminology used here stems from indirect threaded Forth systems; in
12094: such a system, the XT of a word is represented by the CFA (code field
12095: address) of a word; the CFA points to a cell that contains the code
12096: address.  The code address is the address of some machine code that
12097: performs the run-time action of invoking the word (e.g., the
12098: @code{dovar:} routine pushes the address of the body of the word (a
12099: variable) on the stack
12100: ).
12101: 
12102: @cindex code address
12103: @cindex code field address
12104: In an indirect threaded Forth, you can get the code address of @i{name}
12105: with @code{' @i{name} @@}; in Gforth you can get it with @code{' @i{name}
12106: >code-address}, independent of the threading method.
12107: 
12108: doc-threading-method
12109: doc->code-address
12110: doc-code-address!
12111: 
12112: @cindex @code{does>}-handler
12113: @cindex @code{does>}-code
12114: For a word defined with @code{DOES>}, the code address usually points to
12115: a jump instruction (the @dfn{does-handler}) that jumps to the dodoes
12116: routine (in Gforth on some platforms, it can also point to the dodoes
12117: routine itself).  What you are typically interested in, though, is
12118: whether a word is a @code{DOES>}-defined word, and what Forth code it
12119: executes; @code{>does-code} tells you that.
12120: 
12121: doc->does-code
12122: 
12123: To create a @code{DOES>}-defined word with the following basic words,
12124: you have to set up a @code{DOES>}-handler with @code{does-handler!};
12125: @code{/does-handler} aus behind you have to place your executable Forth
12126: code.  Finally you have to create a word and modify its behaviour with
12127: @code{does-handler!}.
12128: 
12129: doc-does-code!
12130: doc-does-handler!
12131: doc-/does-handler
12132: 
12133: The code addresses produced by various defining words are produced by
12134: the following words:
12135: 
12136: doc-docol:
12137: doc-docon:
12138: doc-dovar:
12139: doc-douser:
12140: doc-dodefer:
12141: doc-dofield:
12142: 
12143: @cindex definer
12144: The following two words generalize @code{>code-address},
12145: @code{>does-code}, @code{code-address!}, and @code{does-code!}:
12146: 
12147: doc->definer
12148: doc-definer!
12149: 
12150: @c -------------------------------------------------------------
12151: @node Passing Commands to the OS, Keeping track of Time, Threading Words, Words
12152: @section Passing Commands to the Operating System
12153: @cindex operating system - passing commands
12154: @cindex shell commands
12155: 
12156: Gforth allows you to pass an arbitrary string to the host operating
12157: system shell (if such a thing exists) for execution.
12158: 
12159: 
12160: doc-sh
12161: doc-system
12162: doc-$?
12163: doc-getenv
12164: 
12165: 
12166: @c -------------------------------------------------------------
12167: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
12168: @section Keeping track of Time
12169: @cindex time-related words
12170: 
12171: doc-ms
12172: doc-time&date
12173: doc-utime
12174: doc-cputime
12175: 
12176: 
12177: @c -------------------------------------------------------------
12178: @node Miscellaneous Words,  , Keeping track of Time, Words
12179: @section Miscellaneous Words
12180: @cindex miscellaneous words
12181: 
12182: @comment TODO find homes for these
12183: 
12184: These section lists the ANS Forth words that are not documented
12185: elsewhere in this manual. Ultimately, they all need proper homes.
12186: 
12187: doc-quit
12188: 
12189: The following ANS Forth words are not currently supported by Gforth 
12190: (@pxref{ANS conformance}):
12191: 
12192: @code{EDITOR} 
12193: @code{EMIT?} 
12194: @code{FORGET} 
12195: 
12196: @c ******************************************************************
12197: @node Error messages, Tools, Words, Top
12198: @chapter Error messages
12199: @cindex error messages
12200: @cindex backtrace
12201: 
12202: A typical Gforth error message looks like this:
12203: 
12204: @example
12205: in file included from \evaluated string/:-1
12206: in file included from ./yyy.fs:1
12207: ./xxx.fs:4: Invalid memory address
12208: bar
12209: ^^^
12210: Backtrace:
12211: $400E664C @@
12212: $400E6664 foo
12213: @end example
12214: 
12215: The message identifying the error is @code{Invalid memory address}.  The
12216: error happened when text-interpreting line 4 of the file
12217: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
12218: word on the line where the error happened, is pointed out (with
12219: @code{^^^}).
12220: 
12221: The file containing the error was included in line 1 of @file{./yyy.fs},
12222: and @file{yyy.fs} was included from a non-file (in this case, by giving
12223: @file{yyy.fs} as command-line parameter to Gforth).
12224: 
12225: At the end of the error message you find a return stack dump that can be
12226: interpreted as a backtrace (possibly empty). On top you find the top of
12227: the return stack when the @code{throw} happened, and at the bottom you
12228: find the return stack entry just above the return stack of the topmost
12229: text interpreter.
12230: 
12231: To the right of most return stack entries you see a guess for the word
12232: that pushed that return stack entry as its return address. This gives a
12233: backtrace. In our case we see that @code{bar} called @code{foo}, and
12234: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
12235: address} exception).
12236: 
12237: Note that the backtrace is not perfect: We don't know which return stack
12238: entries are return addresses (so we may get false positives); and in
12239: some cases (e.g., for @code{abort"}) we cannot determine from the return
12240: address the word that pushed the return address, so for some return
12241: addresses you see no names in the return stack dump.
12242: 
12243: @cindex @code{catch} and backtraces
12244: The return stack dump represents the return stack at the time when a
12245: specific @code{throw} was executed.  In programs that make use of
12246: @code{catch}, it is not necessarily clear which @code{throw} should be
12247: used for the return stack dump (e.g., consider one @code{throw} that
12248: indicates an error, which is caught, and during recovery another error
12249: happens; which @code{throw} should be used for the stack dump?).  Gforth
12250: presents the return stack dump for the first @code{throw} after the last
12251: executed (not returned-to) @code{catch}; this works well in the usual
12252: case.
12253: 
12254: @cindex @code{gforth-fast} and backtraces
12255: @cindex @code{gforth-fast}, difference from @code{gforth}
12256: @cindex backtraces with @code{gforth-fast}
12257: @cindex return stack dump with @code{gforth-fast}
12258: @code{Gforth} is able to do a return stack dump for throws generated
12259: from primitives (e.g., invalid memory address, stack empty etc.);
12260: @code{gforth-fast} is only able to do a return stack dump from a
12261: directly called @code{throw} (including @code{abort} etc.).  Given an
12262: exception caused by a primitive in @code{gforth-fast}, you will
12263: typically see no return stack dump at all; however, if the exception is
12264: caught by @code{catch} (e.g., for restoring some state), and then
12265: @code{throw}n again, the return stack dump will be for the first such
12266: @code{throw}.
12267: 
12268: @c ******************************************************************
12269: @node Tools, ANS conformance, Error messages, Top
12270: @chapter Tools
12271: 
12272: @menu
12273: * ANS Report::                  Report the words used, sorted by wordset.
12274: @end menu
12275: 
12276: See also @ref{Emacs and Gforth}.
12277: 
12278: @node ANS Report,  , Tools, Tools
12279: @section @file{ans-report.fs}: Report the words used, sorted by wordset
12280: @cindex @file{ans-report.fs}
12281: @cindex report the words used in your program
12282: @cindex words used in your program
12283: 
12284: If you want to label a Forth program as ANS Forth Program, you must
12285: document which wordsets the program uses; for extension wordsets, it is
12286: helpful to list the words the program requires from these wordsets
12287: (because Forth systems are allowed to provide only some words of them).
12288: 
12289: The @file{ans-report.fs} tool makes it easy for you to determine which
12290: words from which wordset and which non-ANS words your application
12291: uses. You simply have to include @file{ans-report.fs} before loading the
12292: program you want to check. After loading your program, you can get the
12293: report with @code{print-ans-report}. A typical use is to run this as
12294: batch job like this:
12295: @example
12296: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
12297: @end example
12298: 
12299: The output looks like this (for @file{compat/control.fs}):
12300: @example
12301: The program uses the following words
12302: from CORE :
12303: : POSTPONE THEN ; immediate ?dup IF 0= 
12304: from BLOCK-EXT :
12305: \ 
12306: from FILE :
12307: ( 
12308: @end example
12309: 
12310: @subsection Caveats
12311: 
12312: Note that @file{ans-report.fs} just checks which words are used, not whether
12313: they are used in an ANS Forth conforming way!
12314: 
12315: Some words are defined in several wordsets in the
12316: standard. @file{ans-report.fs} reports them for only one of the
12317: wordsets, and not necessarily the one you expect. It depends on usage
12318: which wordset is the right one to specify. E.g., if you only use the
12319: compilation semantics of @code{S"}, it is a Core word; if you also use
12320: its interpretation semantics, it is a File word.
12321: 
12322: @c ******************************************************************
12323: @node ANS conformance, Standard vs Extensions, Tools, Top
12324: @chapter ANS conformance
12325: @cindex ANS conformance of Gforth
12326: 
12327: To the best of our knowledge, Gforth is an
12328: 
12329: ANS Forth System
12330: @itemize @bullet
12331: @item providing the Core Extensions word set
12332: @item providing the Block word set
12333: @item providing the Block Extensions word set
12334: @item providing the Double-Number word set
12335: @item providing the Double-Number Extensions word set
12336: @item providing the Exception word set
12337: @item providing the Exception Extensions word set
12338: @item providing the Facility word set
12339: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
12340: @item providing the File Access word set
12341: @item providing the File Access Extensions word set
12342: @item providing the Floating-Point word set
12343: @item providing the Floating-Point Extensions word set
12344: @item providing the Locals word set
12345: @item providing the Locals Extensions word set
12346: @item providing the Memory-Allocation word set
12347: @item providing the Memory-Allocation Extensions word set (that one's easy)
12348: @item providing the Programming-Tools word set
12349: @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
12350: @item providing the Search-Order word set
12351: @item providing the Search-Order Extensions word set
12352: @item providing the String word set
12353: @item providing the String Extensions word set (another easy one)
12354: @end itemize
12355: 
12356: @cindex system documentation
12357: In addition, ANS Forth systems are required to document certain
12358: implementation choices. This chapter tries to meet these
12359: requirements. In many cases it gives a way to ask the system for the
12360: information instead of providing the information directly, in
12361: particular, if the information depends on the processor, the operating
12362: system or the installation options chosen, or if they are likely to
12363: change during the maintenance of Gforth.
12364: 
12365: @comment The framework for the rest has been taken from pfe.
12366: 
12367: @menu
12368: * The Core Words::              
12369: * The optional Block word set::  
12370: * The optional Double Number word set::  
12371: * The optional Exception word set::  
12372: * The optional Facility word set::  
12373: * The optional File-Access word set::  
12374: * The optional Floating-Point word set::  
12375: * The optional Locals word set::  
12376: * The optional Memory-Allocation word set::  
12377: * The optional Programming-Tools word set::  
12378: * The optional Search-Order word set::  
12379: @end menu
12380: 
12381: 
12382: @c =====================================================================
12383: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
12384: @comment  node-name,  next,  previous,  up
12385: @section The Core Words
12386: @c =====================================================================
12387: @cindex core words, system documentation
12388: @cindex system documentation, core words
12389: 
12390: @menu
12391: * core-idef::                   Implementation Defined Options                   
12392: * core-ambcond::                Ambiguous Conditions                
12393: * core-other::                  Other System Documentation                  
12394: @end menu
12395: 
12396: @c ---------------------------------------------------------------------
12397: @node core-idef, core-ambcond, The Core Words, The Core Words
12398: @subsection Implementation Defined Options
12399: @c ---------------------------------------------------------------------
12400: @cindex core words, implementation-defined options
12401: @cindex implementation-defined options, core words
12402: 
12403: 
12404: @table @i
12405: @item (Cell) aligned addresses:
12406: @cindex cell-aligned addresses
12407: @cindex aligned addresses
12408: processor-dependent. Gforth's alignment words perform natural alignment
12409: (e.g., an address aligned for a datum of size 8 is divisible by
12410: 8). Unaligned accesses usually result in a @code{-23 THROW}.
12411: 
12412: @item @code{EMIT} and non-graphic characters:
12413: @cindex @code{EMIT} and non-graphic characters
12414: @cindex non-graphic characters and @code{EMIT}
12415: The character is output using the C library function (actually, macro)
12416: @code{putc}.
12417: 
12418: @item character editing of @code{ACCEPT} and @code{EXPECT}:
12419: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
12420: @cindex editing in @code{ACCEPT} and @code{EXPECT}
12421: @cindex @code{ACCEPT}, editing
12422: @cindex @code{EXPECT}, editing
12423: This is modeled on the GNU readline library (@pxref{Readline
12424: Interaction, , Command Line Editing, readline, The GNU Readline
12425: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
12426: producing a full word completion every time you type it (instead of
12427: producing the common prefix of all completions). @xref{Command-line editing}.
12428: 
12429: @item character set:
12430: @cindex character set
12431: The character set of your computer and display device. Gforth is
12432: 8-bit-clean (but some other component in your system may make trouble).
12433: 
12434: @item Character-aligned address requirements:
12435: @cindex character-aligned address requirements
12436: installation-dependent. Currently a character is represented by a C
12437: @code{unsigned char}; in the future we might switch to @code{wchar_t}
12438: (Comments on that requested).
12439: 
12440: @item character-set extensions and matching of names:
12441: @cindex character-set extensions and matching of names
12442: @cindex case-sensitivity for name lookup
12443: @cindex name lookup, case-sensitivity
12444: @cindex locale and case-sensitivity
12445: Any character except the ASCII NUL character can be used in a
12446: name. Matching is case-insensitive (except in @code{TABLE}s). The
12447: matching is performed using the C library function @code{strncasecmp}, whose
12448: function is probably influenced by the locale. E.g., the @code{C} locale
12449: does not know about accents and umlauts, so they are matched
12450: case-sensitively in that locale. For portability reasons it is best to
12451: write programs such that they work in the @code{C} locale. Then one can
12452: use libraries written by a Polish programmer (who might use words
12453: containing ISO Latin-2 encoded characters) and by a French programmer
12454: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
12455: funny results for some of the words (which ones, depends on the font you
12456: are using)). Also, the locale you prefer may not be available in other
12457: operating systems. Hopefully, Unicode will solve these problems one day.
12458: 
12459: @item conditions under which control characters match a space delimiter:
12460: @cindex space delimiters
12461: @cindex control characters as delimiters
12462: If @code{WORD} is called with the space character as a delimiter, all
12463: white-space characters (as identified by the C macro @code{isspace()})
12464: are delimiters. @code{PARSE}, on the other hand, treats space like other
12465: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
12466: like @code{PARSE} otherwise. @code{Name}, which is used by the outer
12467: interpreter (aka text interpreter) by default, treats all white-space
12468: characters as delimiters.
12469: 
12470: @item format of the control-flow stack:
12471: @cindex control-flow stack, format
12472: The data stack is used as control-flow stack. The size of a control-flow
12473: stack item in cells is given by the constant @code{cs-item-size}. At the
12474: time of this writing, an item consists of a (pointer to a) locals list
12475: (third), an address in the code (second), and a tag for identifying the
12476: item (TOS). The following tags are used: @code{defstart},
12477: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
12478: @code{scopestart}.
12479: 
12480: @item conversion of digits > 35
12481: @cindex digits > 35
12482: The characters @code{[\]^_'} are the digits with the decimal value
12483: 36@minus{}41. There is no way to input many of the larger digits.
12484: 
12485: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
12486: @cindex @code{EXPECT}, display after end of input
12487: @cindex @code{ACCEPT}, display after end of input
12488: The cursor is moved to the end of the entered string. If the input is
12489: terminated using the @kbd{Return} key, a space is typed.
12490: 
12491: @item exception abort sequence of @code{ABORT"}:
12492: @cindex exception abort sequence of @code{ABORT"}
12493: @cindex @code{ABORT"}, exception abort sequence
12494: The error string is stored into the variable @code{"error} and a
12495: @code{-2 throw} is performed.
12496: 
12497: @item input line terminator:
12498: @cindex input line terminator
12499: @cindex line terminator on input
12500: @cindex newline character on input
12501: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
12502: lines. One of these characters is typically produced when you type the
12503: @kbd{Enter} or @kbd{Return} key.
12504: 
12505: @item maximum size of a counted string:
12506: @cindex maximum size of a counted string
12507: @cindex counted string, maximum size
12508: @code{s" /counted-string" environment? drop .}. Currently 255 characters
12509: on all platforms, but this may change.
12510: 
12511: @item maximum size of a parsed string:
12512: @cindex maximum size of a parsed string
12513: @cindex parsed string, maximum size
12514: Given by the constant @code{/line}. Currently 255 characters.
12515: 
12516: @item maximum size of a definition name, in characters:
12517: @cindex maximum size of a definition name, in characters
12518: @cindex name, maximum length
12519: 31
12520: 
12521: @item maximum string length for @code{ENVIRONMENT?}, in characters:
12522: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
12523: @cindex @code{ENVIRONMENT?} string length, maximum
12524: 31
12525: 
12526: @item method of selecting the user input device:
12527: @cindex user input device, method of selecting
12528: The user input device is the standard input. There is currently no way to
12529: change it from within Gforth. However, the input can typically be
12530: redirected in the command line that starts Gforth.
12531: 
12532: @item method of selecting the user output device:
12533: @cindex user output device, method of selecting
12534: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
12535: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
12536: output when the user output device is a terminal, otherwise the output
12537: is buffered.
12538: 
12539: @item methods of dictionary compilation:
12540: What are we expected to document here?
12541: 
12542: @item number of bits in one address unit:
12543: @cindex number of bits in one address unit
12544: @cindex address unit, size in bits
12545: @code{s" address-units-bits" environment? drop .}. 8 in all current
12546: platforms.
12547: 
12548: @item number representation and arithmetic:
12549: @cindex number representation and arithmetic
12550: Processor-dependent. Binary two's complement on all current platforms.
12551: 
12552: @item ranges for integer types:
12553: @cindex ranges for integer types
12554: @cindex integer types, ranges
12555: Installation-dependent. Make environmental queries for @code{MAX-N},
12556: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
12557: unsigned (and positive) types is 0. The lower bound for signed types on
12558: two's complement and one's complement machines machines can be computed
12559: by adding 1 to the upper bound.
12560: 
12561: @item read-only data space regions:
12562: @cindex read-only data space regions
12563: @cindex data-space, read-only regions
12564: The whole Forth data space is writable.
12565: 
12566: @item size of buffer at @code{WORD}:
12567: @cindex size of buffer at @code{WORD}
12568: @cindex @code{WORD} buffer size
12569: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12570: shared with the pictured numeric output string. If overwriting
12571: @code{PAD} is acceptable, it is as large as the remaining dictionary
12572: space, although only as much can be sensibly used as fits in a counted
12573: string.
12574: 
12575: @item size of one cell in address units:
12576: @cindex cell size
12577: @code{1 cells .}.
12578: 
12579: @item size of one character in address units:
12580: @cindex char size
12581: @code{1 chars .}. 1 on all current platforms.
12582: 
12583: @item size of the keyboard terminal buffer:
12584: @cindex size of the keyboard terminal buffer
12585: @cindex terminal buffer, size
12586: Varies. You can determine the size at a specific time using @code{lp@@
12587: tib - .}. It is shared with the locals stack and TIBs of files that
12588: include the current file. You can change the amount of space for TIBs
12589: and locals stack at Gforth startup with the command line option
12590: @code{-l}.
12591: 
12592: @item size of the pictured numeric output buffer:
12593: @cindex size of the pictured numeric output buffer
12594: @cindex pictured numeric output buffer, size
12595: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12596: shared with @code{WORD}.
12597: 
12598: @item size of the scratch area returned by @code{PAD}:
12599: @cindex size of the scratch area returned by @code{PAD}
12600: @cindex @code{PAD} size
12601: The remainder of dictionary space. @code{unused pad here - - .}.
12602: 
12603: @item system case-sensitivity characteristics:
12604: @cindex case-sensitivity characteristics
12605: Dictionary searches are case-insensitive (except in
12606: @code{TABLE}s). However, as explained above under @i{character-set
12607: extensions}, the matching for non-ASCII characters is determined by the
12608: locale you are using. In the default @code{C} locale all non-ASCII
12609: characters are matched case-sensitively.
12610: 
12611: @item system prompt:
12612: @cindex system prompt
12613: @cindex prompt
12614: @code{ ok} in interpret state, @code{ compiled} in compile state.
12615: 
12616: @item division rounding:
12617: @cindex division rounding
12618: installation dependent. @code{s" floored" environment? drop .}. We leave
12619: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
12620: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
12621: 
12622: @item values of @code{STATE} when true:
12623: @cindex @code{STATE} values
12624: -1.
12625: 
12626: @item values returned after arithmetic overflow:
12627: On two's complement machines, arithmetic is performed modulo
12628: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12629: arithmetic (with appropriate mapping for signed types). Division by zero
12630: typically results in a @code{-55 throw} (Floating-point unidentified
12631: fault) or @code{-10 throw} (divide by zero).
12632: 
12633: @item whether the current definition can be found after @t{DOES>}:
12634: @cindex @t{DOES>}, visibility of current definition
12635: No.
12636: 
12637: @end table
12638: 
12639: @c ---------------------------------------------------------------------
12640: @node core-ambcond, core-other, core-idef, The Core Words
12641: @subsection Ambiguous conditions
12642: @c ---------------------------------------------------------------------
12643: @cindex core words, ambiguous conditions
12644: @cindex ambiguous conditions, core words
12645: 
12646: @table @i
12647: 
12648: @item a name is neither a word nor a number:
12649: @cindex name not found
12650: @cindex undefined word
12651: @code{-13 throw} (Undefined word).
12652: 
12653: @item a definition name exceeds the maximum length allowed:
12654: @cindex word name too long
12655: @code{-19 throw} (Word name too long)
12656: 
12657: @item addressing a region not inside the various data spaces of the forth system:
12658: @cindex Invalid memory address
12659: The stacks, code space and header space are accessible. Machine code space is
12660: typically readable. Accessing other addresses gives results dependent on
12661: the operating system. On decent systems: @code{-9 throw} (Invalid memory
12662: address).
12663: 
12664: @item argument type incompatible with parameter:
12665: @cindex argument type mismatch
12666: This is usually not caught. Some words perform checks, e.g., the control
12667: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
12668: mismatch).
12669: 
12670: @item attempting to obtain the execution token of a word with undefined execution semantics:
12671: @cindex Interpreting a compile-only word, for @code{'} etc.
12672: @cindex execution token of words with undefined execution semantics
12673: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
12674: get an execution token for @code{compile-only-error} (which performs a
12675: @code{-14 throw} when executed).
12676: 
12677: @item dividing by zero:
12678: @cindex dividing by zero
12679: @cindex floating point unidentified fault, integer division
12680: On some platforms, this produces a @code{-10 throw} (Division by
12681: zero); on other systems, this typically results in a @code{-55 throw}
12682: (Floating-point unidentified fault).
12683: 
12684: @item insufficient data stack or return stack space:
12685: @cindex insufficient data stack or return stack space
12686: @cindex stack overflow
12687: @cindex address alignment exception, stack overflow
12688: @cindex Invalid memory address, stack overflow
12689: Depending on the operating system, the installation, and the invocation
12690: of Gforth, this is either checked by the memory management hardware, or
12691: it is not checked. If it is checked, you typically get a @code{-3 throw}
12692: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
12693: throw} (Invalid memory address) (depending on the platform and how you
12694: achieved the overflow) as soon as the overflow happens. If it is not
12695: checked, overflows typically result in mysterious illegal memory
12696: accesses, producing @code{-9 throw} (Invalid memory address) or
12697: @code{-23 throw} (Address alignment exception); they might also destroy
12698: the internal data structure of @code{ALLOCATE} and friends, resulting in
12699: various errors in these words.
12700: 
12701: @item insufficient space for loop control parameters:
12702: @cindex insufficient space for loop control parameters
12703: Like other return stack overflows.
12704: 
12705: @item insufficient space in the dictionary:
12706: @cindex insufficient space in the dictionary
12707: @cindex dictionary overflow
12708: If you try to allot (either directly with @code{allot}, or indirectly
12709: with @code{,}, @code{create} etc.) more memory than available in the
12710: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
12711: to access memory beyond the end of the dictionary, the results are
12712: similar to stack overflows.
12713: 
12714: @item interpreting a word with undefined interpretation semantics:
12715: @cindex interpreting a word with undefined interpretation semantics
12716: @cindex Interpreting a compile-only word
12717: For some words, we have defined interpretation semantics. For the
12718: others: @code{-14 throw} (Interpreting a compile-only word).
12719: 
12720: @item modifying the contents of the input buffer or a string literal:
12721: @cindex modifying the contents of the input buffer or a string literal
12722: These are located in writable memory and can be modified.
12723: 
12724: @item overflow of the pictured numeric output string:
12725: @cindex overflow of the pictured numeric output string
12726: @cindex pictured numeric output string, overflow
12727: @code{-17 throw} (Pictured numeric ouput string overflow).
12728: 
12729: @item parsed string overflow:
12730: @cindex parsed string overflow
12731: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
12732: 
12733: @item producing a result out of range:
12734: @cindex result out of range
12735: On two's complement machines, arithmetic is performed modulo
12736: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12737: arithmetic (with appropriate mapping for signed types). Division by zero
12738: typically results in a @code{-10 throw} (divide by zero) or @code{-55
12739: throw} (floating point unidentified fault). @code{convert} and
12740: @code{>number} currently overflow silently.
12741: 
12742: @item reading from an empty data or return stack:
12743: @cindex stack empty
12744: @cindex stack underflow
12745: @cindex return stack underflow
12746: The data stack is checked by the outer (aka text) interpreter after
12747: every word executed. If it has underflowed, a @code{-4 throw} (Stack
12748: underflow) is performed. Apart from that, stacks may be checked or not,
12749: depending on operating system, installation, and invocation. If they are
12750: caught by a check, they typically result in @code{-4 throw} (Stack
12751: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
12752: (Invalid memory address), depending on the platform and which stack
12753: underflows and by how much. Note that even if the system uses checking
12754: (through the MMU), your program may have to underflow by a significant
12755: number of stack items to trigger the reaction (the reason for this is
12756: that the MMU, and therefore the checking, works with a page-size
12757: granularity).  If there is no checking, the symptoms resulting from an
12758: underflow are similar to those from an overflow.  Unbalanced return
12759: stack errors can result in a variety of symptoms, including @code{-9 throw}
12760: (Invalid memory address) and Illegal Instruction (typically @code{-260
12761: throw}).
12762: 
12763: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
12764: @cindex unexpected end of the input buffer
12765: @cindex zero-length string as a name
12766: @cindex Attempt to use zero-length string as a name
12767: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
12768: use zero-length string as a name). Words like @code{'} probably will not
12769: find what they search. Note that it is possible to create zero-length
12770: names with @code{nextname} (should it not?).
12771: 
12772: @item @code{>IN} greater than input buffer:
12773: @cindex @code{>IN} greater than input buffer
12774: The next invocation of a parsing word returns a string with length 0.
12775: 
12776: @item @code{RECURSE} appears after @code{DOES>}:
12777: @cindex @code{RECURSE} appears after @code{DOES>}
12778: Compiles a recursive call to the defining word, not to the defined word.
12779: 
12780: @item argument input source different than current input source for @code{RESTORE-INPUT}:
12781: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
12782: @cindex argument type mismatch, @code{RESTORE-INPUT}
12783: @cindex @code{RESTORE-INPUT}, Argument type mismatch
12784: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
12785: the end of the file was reached), its source-id may be
12786: reused. Therefore, restoring an input source specification referencing a
12787: closed file may lead to unpredictable results instead of a @code{-12
12788: THROW}.
12789: 
12790: In the future, Gforth may be able to restore input source specifications
12791: from other than the current input source.
12792: 
12793: @item data space containing definitions gets de-allocated:
12794: @cindex data space containing definitions gets de-allocated
12795: Deallocation with @code{allot} is not checked. This typically results in
12796: memory access faults or execution of illegal instructions.
12797: 
12798: @item data space read/write with incorrect alignment:
12799: @cindex data space read/write with incorrect alignment
12800: @cindex alignment faults
12801: @cindex address alignment exception
12802: Processor-dependent. Typically results in a @code{-23 throw} (Address
12803: alignment exception). Under Linux-Intel on a 486 or later processor with
12804: alignment turned on, incorrect alignment results in a @code{-9 throw}
12805: (Invalid memory address). There are reportedly some processors with
12806: alignment restrictions that do not report violations.
12807: 
12808: @item data space pointer not properly aligned, @code{,}, @code{C,}:
12809: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
12810: Like other alignment errors.
12811: 
12812: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
12813: Like other stack underflows.
12814: 
12815: @item loop control parameters not available:
12816: @cindex loop control parameters not available
12817: Not checked. The counted loop words simply assume that the top of return
12818: stack items are loop control parameters and behave accordingly.
12819: 
12820: @item most recent definition does not have a name (@code{IMMEDIATE}):
12821: @cindex most recent definition does not have a name (@code{IMMEDIATE})
12822: @cindex last word was headerless
12823: @code{abort" last word was headerless"}.
12824: 
12825: @item name not defined by @code{VALUE} used by @code{TO}:
12826: @cindex name not defined by @code{VALUE} used by @code{TO}
12827: @cindex @code{TO} on non-@code{VALUE}s
12828: @cindex Invalid name argument, @code{TO}
12829: @code{-32 throw} (Invalid name argument) (unless name is a local or was
12830: defined by @code{CONSTANT}; in the latter case it just changes the constant).
12831: 
12832: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
12833: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
12834: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
12835: @code{-13 throw} (Undefined word)
12836: 
12837: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
12838: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
12839: Gforth behaves as if they were of the same type. I.e., you can predict
12840: the behaviour by interpreting all parameters as, e.g., signed.
12841: 
12842: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
12843: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
12844: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
12845: compilation semantics of @code{TO}.
12846: 
12847: @item String longer than a counted string returned by @code{WORD}:
12848: @cindex string longer than a counted string returned by @code{WORD}
12849: @cindex @code{WORD}, string overflow
12850: Not checked. The string will be ok, but the count will, of course,
12851: contain only the least significant bits of the length.
12852: 
12853: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
12854: @cindex @code{LSHIFT}, large shift counts
12855: @cindex @code{RSHIFT}, large shift counts
12856: Processor-dependent. Typical behaviours are returning 0 and using only
12857: the low bits of the shift count.
12858: 
12859: @item word not defined via @code{CREATE}:
12860: @cindex @code{>BODY} of non-@code{CREATE}d words
12861: @code{>BODY} produces the PFA of the word no matter how it was defined.
12862: 
12863: @cindex @code{DOES>} of non-@code{CREATE}d words
12864: @code{DOES>} changes the execution semantics of the last defined word no
12865: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
12866: @code{CREATE , DOES>}.
12867: 
12868: @item words improperly used outside @code{<#} and @code{#>}:
12869: Not checked. As usual, you can expect memory faults.
12870: 
12871: @end table
12872: 
12873: 
12874: @c ---------------------------------------------------------------------
12875: @node core-other,  , core-ambcond, The Core Words
12876: @subsection Other system documentation
12877: @c ---------------------------------------------------------------------
12878: @cindex other system documentation, core words
12879: @cindex core words, other system documentation
12880: 
12881: @table @i
12882: @item nonstandard words using @code{PAD}:
12883: @cindex @code{PAD} use by nonstandard words
12884: None.
12885: 
12886: @item operator's terminal facilities available:
12887: @cindex operator's terminal facilities available
12888: After processing the OS's command line, Gforth goes into interactive mode,
12889: and you can give commands to Gforth interactively. The actual facilities
12890: available depend on how you invoke Gforth.
12891: 
12892: @item program data space available:
12893: @cindex program data space available
12894: @cindex data space available
12895: @code{UNUSED .} gives the remaining dictionary space. The total
12896: dictionary space can be specified with the @code{-m} switch
12897: (@pxref{Invoking Gforth}) when Gforth starts up.
12898: 
12899: @item return stack space available:
12900: @cindex return stack space available
12901: You can compute the total return stack space in cells with
12902: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
12903: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
12904: 
12905: @item stack space available:
12906: @cindex stack space available
12907: You can compute the total data stack space in cells with
12908: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
12909: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
12910: 
12911: @item system dictionary space required, in address units:
12912: @cindex system dictionary space required, in address units
12913: Type @code{here forthstart - .} after startup. At the time of this
12914: writing, this gives 80080 (bytes) on a 32-bit system.
12915: @end table
12916: 
12917: 
12918: @c =====================================================================
12919: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
12920: @section The optional Block word set
12921: @c =====================================================================
12922: @cindex system documentation, block words
12923: @cindex block words, system documentation
12924: 
12925: @menu
12926: * block-idef::                  Implementation Defined Options
12927: * block-ambcond::               Ambiguous Conditions               
12928: * block-other::                 Other System Documentation                 
12929: @end menu
12930: 
12931: 
12932: @c ---------------------------------------------------------------------
12933: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
12934: @subsection Implementation Defined Options
12935: @c ---------------------------------------------------------------------
12936: @cindex implementation-defined options, block words
12937: @cindex block words, implementation-defined options
12938: 
12939: @table @i
12940: @item the format for display by @code{LIST}:
12941: @cindex @code{LIST} display format
12942: First the screen number is displayed, then 16 lines of 64 characters,
12943: each line preceded by the line number.
12944: 
12945: @item the length of a line affected by @code{\}:
12946: @cindex length of a line affected by @code{\}
12947: @cindex @code{\}, line length in blocks
12948: 64 characters.
12949: @end table
12950: 
12951: 
12952: @c ---------------------------------------------------------------------
12953: @node block-ambcond, block-other, block-idef, The optional Block word set
12954: @subsection Ambiguous conditions
12955: @c ---------------------------------------------------------------------
12956: @cindex block words, ambiguous conditions
12957: @cindex ambiguous conditions, block words
12958: 
12959: @table @i
12960: @item correct block read was not possible:
12961: @cindex block read not possible
12962: Typically results in a @code{throw} of some OS-derived value (between
12963: -512 and -2048). If the blocks file was just not long enough, blanks are
12964: supplied for the missing portion.
12965: 
12966: @item I/O exception in block transfer:
12967: @cindex I/O exception in block transfer
12968: @cindex block transfer, I/O exception
12969: Typically results in a @code{throw} of some OS-derived value (between
12970: -512 and -2048).
12971: 
12972: @item invalid block number:
12973: @cindex invalid block number
12974: @cindex block number invalid
12975: @code{-35 throw} (Invalid block number)
12976: 
12977: @item a program directly alters the contents of @code{BLK}:
12978: @cindex @code{BLK}, altering @code{BLK}
12979: The input stream is switched to that other block, at the same
12980: position. If the storing to @code{BLK} happens when interpreting
12981: non-block input, the system will get quite confused when the block ends.
12982: 
12983: @item no current block buffer for @code{UPDATE}:
12984: @cindex @code{UPDATE}, no current block buffer
12985: @code{UPDATE} has no effect.
12986: 
12987: @end table
12988: 
12989: @c ---------------------------------------------------------------------
12990: @node block-other,  , block-ambcond, The optional Block word set
12991: @subsection Other system documentation
12992: @c ---------------------------------------------------------------------
12993: @cindex other system documentation, block words
12994: @cindex block words, other system documentation
12995: 
12996: @table @i
12997: @item any restrictions a multiprogramming system places on the use of buffer addresses:
12998: No restrictions (yet).
12999: 
13000: @item the number of blocks available for source and data:
13001: depends on your disk space.
13002: 
13003: @end table
13004: 
13005: 
13006: @c =====================================================================
13007: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
13008: @section The optional Double Number word set
13009: @c =====================================================================
13010: @cindex system documentation, double words
13011: @cindex double words, system documentation
13012: 
13013: @menu
13014: * double-ambcond::              Ambiguous Conditions              
13015: @end menu
13016: 
13017: 
13018: @c ---------------------------------------------------------------------
13019: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
13020: @subsection Ambiguous conditions
13021: @c ---------------------------------------------------------------------
13022: @cindex double words, ambiguous conditions
13023: @cindex ambiguous conditions, double words
13024: 
13025: @table @i
13026: @item @i{d} outside of range of @i{n} in @code{D>S}:
13027: @cindex @code{D>S}, @i{d} out of range of @i{n} 
13028: The least significant cell of @i{d} is produced.
13029: 
13030: @end table
13031: 
13032: 
13033: @c =====================================================================
13034: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
13035: @section The optional Exception word set
13036: @c =====================================================================
13037: @cindex system documentation, exception words
13038: @cindex exception words, system documentation
13039: 
13040: @menu
13041: * exception-idef::              Implementation Defined Options              
13042: @end menu
13043: 
13044: 
13045: @c ---------------------------------------------------------------------
13046: @node exception-idef,  , The optional Exception word set, The optional Exception word set
13047: @subsection Implementation Defined Options
13048: @c ---------------------------------------------------------------------
13049: @cindex implementation-defined options, exception words
13050: @cindex exception words, implementation-defined options
13051: 
13052: @table @i
13053: @item @code{THROW}-codes used in the system:
13054: @cindex @code{THROW}-codes used in the system
13055: The codes -256@minus{}-511 are used for reporting signals. The mapping
13056: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
13057: codes -512@minus{}-2047 are used for OS errors (for file and memory
13058: allocation operations). The mapping from OS error numbers to throw codes
13059: is -512@minus{}@code{errno}. One side effect of this mapping is that
13060: undefined OS errors produce a message with a strange number; e.g.,
13061: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
13062: @end table
13063: 
13064: @c =====================================================================
13065: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
13066: @section The optional Facility word set
13067: @c =====================================================================
13068: @cindex system documentation, facility words
13069: @cindex facility words, system documentation
13070: 
13071: @menu
13072: * facility-idef::               Implementation Defined Options               
13073: * facility-ambcond::            Ambiguous Conditions            
13074: @end menu
13075: 
13076: 
13077: @c ---------------------------------------------------------------------
13078: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
13079: @subsection Implementation Defined Options
13080: @c ---------------------------------------------------------------------
13081: @cindex implementation-defined options, facility words
13082: @cindex facility words, implementation-defined options
13083: 
13084: @table @i
13085: @item encoding of keyboard events (@code{EKEY}):
13086: @cindex keyboard events, encoding in @code{EKEY}
13087: @cindex @code{EKEY}, encoding of keyboard events
13088: Keys corresponding to ASCII characters are encoded as ASCII characters.
13089: Other keys are encoded with the constants @code{k-left}, @code{k-right},
13090: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
13091: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
13092: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
13093: 
13094: 
13095: @item duration of a system clock tick:
13096: @cindex duration of a system clock tick
13097: @cindex clock tick duration
13098: System dependent. With respect to @code{MS}, the time is specified in
13099: microseconds. How well the OS and the hardware implement this, is
13100: another question.
13101: 
13102: @item repeatability to be expected from the execution of @code{MS}:
13103: @cindex repeatability to be expected from the execution of @code{MS}
13104: @cindex @code{MS}, repeatability to be expected
13105: System dependent. On Unix, a lot depends on load. If the system is
13106: lightly loaded, and the delay is short enough that Gforth does not get
13107: swapped out, the performance should be acceptable. Under MS-DOS and
13108: other single-tasking systems, it should be good.
13109: 
13110: @end table
13111: 
13112: 
13113: @c ---------------------------------------------------------------------
13114: @node facility-ambcond,  , facility-idef, The optional Facility word set
13115: @subsection Ambiguous conditions
13116: @c ---------------------------------------------------------------------
13117: @cindex facility words, ambiguous conditions
13118: @cindex ambiguous conditions, facility words
13119: 
13120: @table @i
13121: @item @code{AT-XY} can't be performed on user output device:
13122: @cindex @code{AT-XY} can't be performed on user output device
13123: Largely terminal dependent. No range checks are done on the arguments.
13124: No errors are reported. You may see some garbage appearing, you may see
13125: simply nothing happen.
13126: 
13127: @end table
13128: 
13129: 
13130: @c =====================================================================
13131: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
13132: @section The optional File-Access word set
13133: @c =====================================================================
13134: @cindex system documentation, file words
13135: @cindex file words, system documentation
13136: 
13137: @menu
13138: * file-idef::                   Implementation Defined Options
13139: * file-ambcond::                Ambiguous Conditions                
13140: @end menu
13141: 
13142: @c ---------------------------------------------------------------------
13143: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
13144: @subsection Implementation Defined Options
13145: @c ---------------------------------------------------------------------
13146: @cindex implementation-defined options, file words
13147: @cindex file words, implementation-defined options
13148: 
13149: @table @i
13150: @item file access methods used:
13151: @cindex file access methods used
13152: @code{R/O}, @code{R/W} and @code{BIN} work as you would
13153: expect. @code{W/O} translates into the C file opening mode @code{w} (or
13154: @code{wb}): The file is cleared, if it exists, and created, if it does
13155: not (with both @code{open-file} and @code{create-file}).  Under Unix
13156: @code{create-file} creates a file with 666 permissions modified by your
13157: umask.
13158: 
13159: @item file exceptions:
13160: @cindex file exceptions
13161: The file words do not raise exceptions (except, perhaps, memory access
13162: faults when you pass illegal addresses or file-ids).
13163: 
13164: @item file line terminator:
13165: @cindex file line terminator
13166: System-dependent. Gforth uses C's newline character as line
13167: terminator. What the actual character code(s) of this are is
13168: system-dependent.
13169: 
13170: @item file name format:
13171: @cindex file name format
13172: System dependent. Gforth just uses the file name format of your OS.
13173: 
13174: @item information returned by @code{FILE-STATUS}:
13175: @cindex @code{FILE-STATUS}, returned information
13176: @code{FILE-STATUS} returns the most powerful file access mode allowed
13177: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
13178: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
13179: along with the returned mode.
13180: 
13181: @item input file state after an exception when including source:
13182: @cindex exception when including source
13183: All files that are left via the exception are closed.
13184: 
13185: @item @i{ior} values and meaning:
13186: @cindex @i{ior} values and meaning
13187: @cindex @i{wior} values and meaning
13188: The @i{ior}s returned by the file and memory allocation words are
13189: intended as throw codes. They typically are in the range
13190: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
13191: @i{ior}s is -512@minus{}@i{errno}.
13192: 
13193: @item maximum depth of file input nesting:
13194: @cindex maximum depth of file input nesting
13195: @cindex file input nesting, maximum depth
13196: limited by the amount of return stack, locals/TIB stack, and the number
13197: of open files available. This should not give you troubles.
13198: 
13199: @item maximum size of input line:
13200: @cindex maximum size of input line
13201: @cindex input line size, maximum
13202: @code{/line}. Currently 255.
13203: 
13204: @item methods of mapping block ranges to files:
13205: @cindex mapping block ranges to files
13206: @cindex files containing blocks
13207: @cindex blocks in files
13208: By default, blocks are accessed in the file @file{blocks.fb} in the
13209: current working directory. The file can be switched with @code{USE}.
13210: 
13211: @item number of string buffers provided by @code{S"}:
13212: @cindex @code{S"}, number of string buffers
13213: 1
13214: 
13215: @item size of string buffer used by @code{S"}:
13216: @cindex @code{S"}, size of string buffer
13217: @code{/line}. currently 255.
13218: 
13219: @end table
13220: 
13221: @c ---------------------------------------------------------------------
13222: @node file-ambcond,  , file-idef, The optional File-Access word set
13223: @subsection Ambiguous conditions
13224: @c ---------------------------------------------------------------------
13225: @cindex file words, ambiguous conditions
13226: @cindex ambiguous conditions, file words
13227: 
13228: @table @i
13229: @item attempting to position a file outside its boundaries:
13230: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
13231: @code{REPOSITION-FILE} is performed as usual: Afterwards,
13232: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
13233: 
13234: @item attempting to read from file positions not yet written:
13235: @cindex reading from file positions not yet written
13236: End-of-file, i.e., zero characters are read and no error is reported.
13237: 
13238: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
13239: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid 
13240: An appropriate exception may be thrown, but a memory fault or other
13241: problem is more probable.
13242: 
13243: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
13244: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
13245: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
13246: The @i{ior} produced by the operation, that discovered the problem, is
13247: thrown.
13248: 
13249: @item named file cannot be opened (@code{INCLUDED}):
13250: @cindex @code{INCLUDED}, named file cannot be opened
13251: The @i{ior} produced by @code{open-file} is thrown.
13252: 
13253: @item requesting an unmapped block number:
13254: @cindex unmapped block numbers
13255: There are no unmapped legal block numbers. On some operating systems,
13256: writing a block with a large number may overflow the file system and
13257: have an error message as consequence.
13258: 
13259: @item using @code{source-id} when @code{blk} is non-zero:
13260: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
13261: @code{source-id} performs its function. Typically it will give the id of
13262: the source which loaded the block. (Better ideas?)
13263: 
13264: @end table
13265: 
13266: 
13267: @c =====================================================================
13268: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
13269: @section The optional Floating-Point word set
13270: @c =====================================================================
13271: @cindex system documentation, floating-point words
13272: @cindex floating-point words, system documentation
13273: 
13274: @menu
13275: * floating-idef::               Implementation Defined Options
13276: * floating-ambcond::            Ambiguous Conditions            
13277: @end menu
13278: 
13279: 
13280: @c ---------------------------------------------------------------------
13281: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
13282: @subsection Implementation Defined Options
13283: @c ---------------------------------------------------------------------
13284: @cindex implementation-defined options, floating-point words
13285: @cindex floating-point words, implementation-defined options
13286: 
13287: @table @i
13288: @item format and range of floating point numbers:
13289: @cindex format and range of floating point numbers
13290: @cindex floating point numbers, format and range
13291: System-dependent; the @code{double} type of C.
13292: 
13293: @item results of @code{REPRESENT} when @i{float} is out of range:
13294: @cindex  @code{REPRESENT}, results when @i{float} is out of range
13295: System dependent; @code{REPRESENT} is implemented using the C library
13296: function @code{ecvt()} and inherits its behaviour in this respect.
13297: 
13298: @item rounding or truncation of floating-point numbers:
13299: @cindex rounding of floating-point numbers
13300: @cindex truncation of floating-point numbers
13301: @cindex floating-point numbers, rounding or truncation
13302: System dependent; the rounding behaviour is inherited from the hosting C
13303: compiler. IEEE-FP-based (i.e., most) systems by default round to
13304: nearest, and break ties by rounding to even (i.e., such that the last
13305: bit of the mantissa is 0).
13306: 
13307: @item size of floating-point stack:
13308: @cindex floating-point stack size
13309: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
13310: the floating-point stack (in floats). You can specify this on startup
13311: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
13312: 
13313: @item width of floating-point stack:
13314: @cindex floating-point stack width 
13315: @code{1 floats}.
13316: 
13317: @end table
13318: 
13319: 
13320: @c ---------------------------------------------------------------------
13321: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
13322: @subsection Ambiguous conditions
13323: @c ---------------------------------------------------------------------
13324: @cindex floating-point words, ambiguous conditions
13325: @cindex ambiguous conditions, floating-point words
13326: 
13327: @table @i
13328: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
13329: @cindex @code{df@@} or @code{df!} used with an address that is not double-float  aligned
13330: System-dependent. Typically results in a @code{-23 THROW} like other
13331: alignment violations.
13332: 
13333: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
13334: @cindex @code{f@@} used with an address that is not float aligned
13335: @cindex @code{f!} used with an address that is not float aligned
13336: System-dependent. Typically results in a @code{-23 THROW} like other
13337: alignment violations.
13338: 
13339: @item floating-point result out of range:
13340: @cindex floating-point result out of range
13341: System-dependent. Can result in a @code{-43 throw} (floating point
13342: overflow), @code{-54 throw} (floating point underflow), @code{-41 throw}
13343: (floating point inexact result), @code{-55 THROW} (Floating-point
13344: unidentified fault), or can produce a special value representing, e.g.,
13345: Infinity.
13346: 
13347: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
13348: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned
13349: System-dependent. Typically results in an alignment fault like other
13350: alignment violations.
13351: 
13352: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
13353: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
13354: The floating-point number is converted into decimal nonetheless.
13355: 
13356: @item Both arguments are equal to zero (@code{FATAN2}):
13357: @cindex @code{FATAN2}, both arguments are equal to zero
13358: System-dependent. @code{FATAN2} is implemented using the C library
13359: function @code{atan2()}.
13360: 
13361: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
13362: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
13363: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
13364: because of small errors and the tan will be a very large (or very small)
13365: but finite number.
13366: 
13367: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
13368: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
13369: The result is rounded to the nearest float.
13370: 
13371: @item dividing by zero:
13372: @cindex dividing by zero, floating-point
13373: @cindex floating-point dividing by zero
13374: @cindex floating-point unidentified fault, FP divide-by-zero
13375: Platform-dependent; can produce an Infinity, NaN, @code{-42 throw}
13376: (floating point divide by zero) or @code{-55 throw} (Floating-point
13377: unidentified fault).
13378: 
13379: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
13380: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
13381: System dependent. On IEEE-FP based systems the number is converted into
13382: an infinity.
13383: 
13384: @item @i{float}<1 (@code{FACOSH}):
13385: @cindex @code{FACOSH}, @i{float}<1
13386: @cindex floating-point unidentified fault, @code{FACOSH}
13387: Platform-dependent; on IEEE-FP systems typically produces a NaN.
13388: 
13389: @item @i{float}=<-1 (@code{FLNP1}):
13390: @cindex @code{FLNP1}, @i{float}=<-1
13391: @cindex floating-point unidentified fault, @code{FLNP1}
13392: Platform-dependent; on IEEE-FP systems typically produces a NaN (or a
13393: negative infinity for @i{float}=-1).
13394: 
13395: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
13396: @cindex @code{FLN}, @i{float}=<0
13397: @cindex @code{FLOG}, @i{float}=<0
13398: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
13399: Platform-dependent; on IEEE-FP systems typically produces a NaN (or a
13400: negative infinity for @i{float}=0).
13401: 
13402: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
13403: @cindex @code{FASINH}, @i{float}<0
13404: @cindex @code{FSQRT}, @i{float}<0
13405: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
13406: Platform-dependent; for @code{fsqrt} this typically gives a NaN, for
13407: @code{fasinh} some platforms produce a NaN, others a number (bug in the
13408: C library?).
13409: 
13410: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
13411: @cindex @code{FACOS}, |@i{float}|>1
13412: @cindex @code{FASIN}, |@i{float}|>1
13413: @cindex @code{FATANH}, |@i{float}|>1
13414: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
13415: Platform-dependent; IEEE-FP systems typically produce a NaN.
13416: 
13417: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
13418: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
13419: @cindex floating-point unidentified fault, @code{F>D}
13420: Platform-dependent; typically, some double number is produced and no
13421: error is reported.
13422: 
13423: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
13424: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
13425: @code{Precision} characters of the numeric output area are used.  If
13426: @code{precision} is too high, these words will smash the data or code
13427: close to @code{here}.
13428: @end table
13429: 
13430: @c =====================================================================
13431: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
13432: @section The optional Locals word set
13433: @c =====================================================================
13434: @cindex system documentation, locals words
13435: @cindex locals words, system documentation
13436: 
13437: @menu
13438: * locals-idef::                 Implementation Defined Options                 
13439: * locals-ambcond::              Ambiguous Conditions              
13440: @end menu
13441: 
13442: 
13443: @c ---------------------------------------------------------------------
13444: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
13445: @subsection Implementation Defined Options
13446: @c ---------------------------------------------------------------------
13447: @cindex implementation-defined options, locals words
13448: @cindex locals words, implementation-defined options
13449: 
13450: @table @i
13451: @item maximum number of locals in a definition:
13452: @cindex maximum number of locals in a definition
13453: @cindex locals, maximum number in a definition
13454: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
13455: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
13456: characters. The number of locals in a definition is bounded by the size
13457: of locals-buffer, which contains the names of the locals.
13458: 
13459: @end table
13460: 
13461: 
13462: @c ---------------------------------------------------------------------
13463: @node locals-ambcond,  , locals-idef, The optional Locals word set
13464: @subsection Ambiguous conditions
13465: @c ---------------------------------------------------------------------
13466: @cindex locals words, ambiguous conditions
13467: @cindex ambiguous conditions, locals words
13468: 
13469: @table @i
13470: @item executing a named local in interpretation state:
13471: @cindex local in interpretation state
13472: @cindex Interpreting a compile-only word, for a local
13473: Locals have no interpretation semantics. If you try to perform the
13474: interpretation semantics, you will get a @code{-14 throw} somewhere
13475: (Interpreting a compile-only word). If you perform the compilation
13476: semantics, the locals access will be compiled (irrespective of state).
13477: 
13478: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
13479: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
13480: @cindex @code{TO} on non-@code{VALUE}s and non-locals
13481: @cindex Invalid name argument, @code{TO}
13482: @code{-32 throw} (Invalid name argument)
13483: 
13484: @end table
13485: 
13486: 
13487: @c =====================================================================
13488: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
13489: @section The optional Memory-Allocation word set
13490: @c =====================================================================
13491: @cindex system documentation, memory-allocation words
13492: @cindex memory-allocation words, system documentation
13493: 
13494: @menu
13495: * memory-idef::                 Implementation Defined Options                 
13496: @end menu
13497: 
13498: 
13499: @c ---------------------------------------------------------------------
13500: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
13501: @subsection Implementation Defined Options
13502: @c ---------------------------------------------------------------------
13503: @cindex implementation-defined options, memory-allocation words
13504: @cindex memory-allocation words, implementation-defined options
13505: 
13506: @table @i
13507: @item values and meaning of @i{ior}:
13508: @cindex  @i{ior} values and meaning
13509: The @i{ior}s returned by the file and memory allocation words are
13510: intended as throw codes. They typically are in the range
13511: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
13512: @i{ior}s is -512@minus{}@i{errno}.
13513: 
13514: @end table
13515: 
13516: @c =====================================================================
13517: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
13518: @section The optional Programming-Tools word set
13519: @c =====================================================================
13520: @cindex system documentation, programming-tools words
13521: @cindex programming-tools words, system documentation
13522: 
13523: @menu
13524: * programming-idef::            Implementation Defined Options            
13525: * programming-ambcond::         Ambiguous Conditions         
13526: @end menu
13527: 
13528: 
13529: @c ---------------------------------------------------------------------
13530: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
13531: @subsection Implementation Defined Options
13532: @c ---------------------------------------------------------------------
13533: @cindex implementation-defined options, programming-tools words
13534: @cindex programming-tools words, implementation-defined options
13535: 
13536: @table @i
13537: @item ending sequence for input following @code{;CODE} and @code{CODE}:
13538: @cindex @code{;CODE} ending sequence
13539: @cindex @code{CODE} ending sequence
13540: @code{END-CODE}
13541: 
13542: @item manner of processing input following @code{;CODE} and @code{CODE}:
13543: @cindex @code{;CODE}, processing input
13544: @cindex @code{CODE}, processing input
13545: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
13546: the input is processed by the text interpreter, (starting) in interpret
13547: state.
13548: 
13549: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
13550: @cindex @code{ASSEMBLER}, search order capability
13551: The ANS Forth search order word set.
13552: 
13553: @item source and format of display by @code{SEE}:
13554: @cindex @code{SEE}, source and format of output
13555: The source for @code{see} is the executable code used by the inner
13556: interpreter.  The current @code{see} tries to output Forth source code
13557: (and on some platforms, assembly code for primitives) as well as
13558: possible.
13559: 
13560: @end table
13561: 
13562: @c ---------------------------------------------------------------------
13563: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
13564: @subsection Ambiguous conditions
13565: @c ---------------------------------------------------------------------
13566: @cindex programming-tools words, ambiguous conditions
13567: @cindex ambiguous conditions, programming-tools words
13568: 
13569: @table @i
13570: 
13571: @item deleting the compilation word list (@code{FORGET}):
13572: @cindex @code{FORGET}, deleting the compilation word list
13573: Not implemented (yet).
13574: 
13575: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
13576: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
13577: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
13578: @cindex control-flow stack underflow
13579: This typically results in an @code{abort"} with a descriptive error
13580: message (may change into a @code{-22 throw} (Control structure mismatch)
13581: in the future). You may also get a memory access error. If you are
13582: unlucky, this ambiguous condition is not caught.
13583: 
13584: @item @i{name} can't be found (@code{FORGET}):
13585: @cindex @code{FORGET}, @i{name} can't be found
13586: Not implemented (yet).
13587: 
13588: @item @i{name} not defined via @code{CREATE}:
13589: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
13590: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
13591: the execution semantics of the last defined word no matter how it was
13592: defined.
13593: 
13594: @item @code{POSTPONE} applied to @code{[IF]}:
13595: @cindex @code{POSTPONE} applied to @code{[IF]}
13596: @cindex @code{[IF]} and @code{POSTPONE}
13597: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
13598: equivalent to @code{[IF]}.
13599: 
13600: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
13601: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
13602: Continue in the same state of conditional compilation in the next outer
13603: input source. Currently there is no warning to the user about this.
13604: 
13605: @item removing a needed definition (@code{FORGET}):
13606: @cindex @code{FORGET}, removing a needed definition
13607: Not implemented (yet).
13608: 
13609: @end table
13610: 
13611: 
13612: @c =====================================================================
13613: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
13614: @section The optional Search-Order word set
13615: @c =====================================================================
13616: @cindex system documentation, search-order words
13617: @cindex search-order words, system documentation
13618: 
13619: @menu
13620: * search-idef::                 Implementation Defined Options                 
13621: * search-ambcond::              Ambiguous Conditions              
13622: @end menu
13623: 
13624: 
13625: @c ---------------------------------------------------------------------
13626: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
13627: @subsection Implementation Defined Options
13628: @c ---------------------------------------------------------------------
13629: @cindex implementation-defined options, search-order words
13630: @cindex search-order words, implementation-defined options
13631: 
13632: @table @i
13633: @item maximum number of word lists in search order:
13634: @cindex maximum number of word lists in search order
13635: @cindex search order, maximum depth
13636: @code{s" wordlists" environment? drop .}. Currently 16.
13637: 
13638: @item minimum search order:
13639: @cindex minimum search order
13640: @cindex search order, minimum
13641: @code{root root}.
13642: 
13643: @end table
13644: 
13645: @c ---------------------------------------------------------------------
13646: @node search-ambcond,  , search-idef, The optional Search-Order word set
13647: @subsection Ambiguous conditions
13648: @c ---------------------------------------------------------------------
13649: @cindex search-order words, ambiguous conditions
13650: @cindex ambiguous conditions, search-order words
13651: 
13652: @table @i
13653: @item changing the compilation word list (during compilation):
13654: @cindex changing the compilation word list (during compilation)
13655: @cindex compilation word list, change before definition ends
13656: The word is entered into the word list that was the compilation word list
13657: at the start of the definition. Any changes to the name field (e.g.,
13658: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
13659: are applied to the latest defined word (as reported by @code{last} or
13660: @code{lastxt}), if possible, irrespective of the compilation word list.
13661: 
13662: @item search order empty (@code{previous}):
13663: @cindex @code{previous}, search order empty
13664: @cindex vocstack empty, @code{previous}
13665: @code{abort" Vocstack empty"}.
13666: 
13667: @item too many word lists in search order (@code{also}):
13668: @cindex @code{also}, too many word lists in search order
13669: @cindex vocstack full, @code{also}
13670: @code{abort" Vocstack full"}.
13671: 
13672: @end table
13673: 
13674: @c ***************************************************************
13675: @node Standard vs Extensions, Model, ANS conformance, Top
13676: @chapter Should I use Gforth extensions?
13677: @cindex Gforth extensions
13678: 
13679: As you read through the rest of this manual, you will see documentation
13680: for @i{Standard} words, and documentation for some appealing Gforth
13681: @i{extensions}. You might ask yourself the question: @i{``Should I
13682: restrict myself to the standard, or should I use the extensions?''}
13683: 
13684: The answer depends on the goals you have for the program you are working
13685: on:
13686: 
13687: @itemize @bullet
13688: 
13689: @item Is it just for yourself or do you want to share it with others?
13690: 
13691: @item
13692: If you want to share it, do the others all use Gforth?
13693: 
13694: @item
13695: If it is just for yourself, do you want to restrict yourself to Gforth?
13696: 
13697: @end itemize
13698: 
13699: If restricting the program to Gforth is ok, then there is no reason not
13700: to use extensions.  It is still a good idea to keep to the standard
13701: where it is easy, in case you want to reuse these parts in another
13702: program that you want to be portable.
13703: 
13704: If you want to be able to port the program to other Forth systems, there
13705: are the following points to consider:
13706: 
13707: @itemize @bullet
13708: 
13709: @item
13710: Most Forth systems that are being maintained support the ANS Forth
13711: standard.  So if your program complies with the standard, it will be
13712: portable among many systems.
13713: 
13714: @item
13715: A number of the Gforth extensions can be implemented in ANS Forth using
13716: public-domain files provided in the @file{compat/} directory. These are
13717: mentioned in the text in passing.  There is no reason not to use these
13718: extensions, your program will still be ANS Forth compliant; just include
13719: the appropriate compat files with your program.
13720: 
13721: @item
13722: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
13723: analyse your program and determine what non-Standard words it relies
13724: upon.  However, it does not check whether you use standard words in a
13725: non-standard way.
13726: 
13727: @item
13728: Some techniques are not standardized by ANS Forth, and are hard or
13729: impossible to implement in a standard way, but can be implemented in
13730: most Forth systems easily, and usually in similar ways (e.g., accessing
13731: word headers).  Forth has a rich historical precedent for programmers
13732: taking advantage of implementation-dependent features of their tools
13733: (for example, relying on a knowledge of the dictionary
13734: structure). Sometimes these techniques are necessary to extract every
13735: last bit of performance from the hardware, sometimes they are just a
13736: programming shorthand.
13737: 
13738: @item
13739: Does using a Gforth extension save more work than the porting this part
13740: to other Forth systems (if any) will cost?
13741: 
13742: @item
13743: Is the additional functionality worth the reduction in portability and
13744: the additional porting problems?
13745: 
13746: @end itemize
13747: 
13748: In order to perform these consideratios, you need to know what's
13749: standard and what's not.  This manual generally states if something is
13750: non-standard, but the authoritative source is the
13751: @uref{http://www.taygeta.com/forth/dpans.html,standard document}.
13752: Appendix A of the Standard (@var{Rationale}) provides a valuable insight
13753: into the thought processes of the technical committee.
13754: 
13755: Note also that portability between Forth systems is not the only
13756: portability issue; there is also the issue of portability between
13757: different platforms (processor/OS combinations).
13758: 
13759: @c ***************************************************************
13760: @node Model, Integrating Gforth, Standard vs Extensions, Top
13761: @chapter Model
13762: 
13763: This chapter has yet to be written. It will contain information, on
13764: which internal structures you can rely.
13765: 
13766: @c ***************************************************************
13767: @node Integrating Gforth, Emacs and Gforth, Model, Top
13768: @chapter Integrating Gforth into C programs
13769: 
13770: This is not yet implemented.
13771: 
13772: Several people like to use Forth as scripting language for applications
13773: that are otherwise written in C, C++, or some other language.
13774: 
13775: The Forth system ATLAST provides facilities for embedding it into
13776: applications; unfortunately it has several disadvantages: most
13777: importantly, it is not based on ANS Forth, and it is apparently dead
13778: (i.e., not developed further and not supported). The facilities
13779: provided by Gforth in this area are inspired by ATLAST's facilities, so
13780: making the switch should not be hard.
13781: 
13782: We also tried to design the interface such that it can easily be
13783: implemented by other Forth systems, so that we may one day arrive at a
13784: standardized interface. Such a standard interface would allow you to
13785: replace the Forth system without having to rewrite C code.
13786: 
13787: You embed the Gforth interpreter by linking with the library
13788: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
13789: global symbols in this library that belong to the interface, have the
13790: prefix @code{forth_}. (Global symbols that are used internally have the
13791: prefix @code{gforth_}).
13792: 
13793: You can include the declarations of Forth types and the functions and
13794: variables of the interface with @code{#include <forth.h>}.
13795: 
13796: Types.
13797: 
13798: Variables.
13799: 
13800: Data and FP Stack pointer. Area sizes.
13801: 
13802: functions.
13803: 
13804: forth_init(imagefile)
13805: forth_evaluate(string) exceptions?
13806: forth_goto(address) (or forth_execute(xt)?)
13807: forth_continue() (a corountining mechanism)
13808: 
13809: Adding primitives.
13810: 
13811: No checking.
13812: 
13813: Signals?
13814: 
13815: Accessing the Stacks
13816: 
13817: @c ******************************************************************
13818: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
13819: @chapter Emacs and Gforth
13820: @cindex Emacs and Gforth
13821: 
13822: @cindex @file{gforth.el}
13823: @cindex @file{forth.el}
13824: @cindex Rydqvist, Goran
13825: @cindex comment editing commands
13826: @cindex @code{\}, editing with Emacs
13827: @cindex debug tracer editing commands
13828: @cindex @code{~~}, removal with Emacs
13829: @cindex Forth mode in Emacs
13830: Gforth comes with @file{gforth.el}, an improved version of
13831: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
13832: improvements are:
13833: 
13834: @itemize @bullet
13835: @item
13836: A better (but still not perfect) handling of indentation.
13837: @item
13838: Comment paragraph filling (@kbd{M-q})
13839: @item
13840: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
13841: @item
13842: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
13843: @item
13844: Support of the @code{info-lookup} feature for looking up the
13845: documentation of a word.
13846: @end itemize
13847: 
13848: I left the stuff I do not use alone, even though some of it only makes
13849: sense for TILE. To get a description of these features, enter Forth mode
13850: and type @kbd{C-h m}.
13851: 
13852: @cindex source location of error or debugging output in Emacs
13853: @cindex error output, finding the source location in Emacs
13854: @cindex debugging output, finding the source location in Emacs
13855: In addition, Gforth supports Emacs quite well: The source code locations
13856: given in error messages, debugging output (from @code{~~}) and failed
13857: assertion messages are in the right format for Emacs' compilation mode
13858: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
13859: Manual}) so the source location corresponding to an error or other
13860: message is only a few keystrokes away (@kbd{C-x `} for the next error,
13861: @kbd{C-c C-c} for the error under the cursor).
13862: 
13863: @cindex @file{TAGS} file
13864: @cindex @file{etags.fs}
13865: @cindex viewing the source of a word in Emacs
13866: @cindex @code{require}, placement in files
13867: @cindex @code{include}, placement in files
13868: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
13869: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
13870: contains the definitions of all words defined afterwards. You can then
13871: find the source for a word using @kbd{M-.}. Note that emacs can use
13872: several tags files at the same time (e.g., one for the Gforth sources
13873: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
13874: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
13875: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
13876: @file{/usr/local/share/gforth/0.2.0/TAGS}).  To get the best behaviour
13877: with @file{etags.fs}, you should avoid putting definitions both before
13878: and after @code{require} etc., otherwise you will see the same file
13879: visited several times by commands like @code{tags-search}.
13880: 
13881: @cindex viewing the documentation of a word in Emacs
13882: @cindex context-sensitive help
13883: Moreover, for words documented in this manual, you can look up the
13884: glossary entry quickly by using @kbd{C-h TAB}
13885: (@code{info-lookup-symbol}, @pxref{Documentation, ,Documentation
13886: Commands, emacs, Emacs Manual}).  This feature requires Emacs 20.3 or
13887: later and does not work for words containing @code{:}.
13888: 
13889: 
13890: @cindex @file{.emacs}
13891: To get all these benefits, add the following lines to your @file{.emacs}
13892: file:
13893: 
13894: @example
13895: (autoload 'forth-mode "gforth.el")
13896: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
13897: @end example
13898: 
13899: @c ******************************************************************
13900: @node Image Files, Engine, Emacs and Gforth, Top
13901: @chapter Image Files
13902: @cindex image file
13903: @cindex @file{.fi} files
13904: @cindex precompiled Forth code
13905: @cindex dictionary in persistent form
13906: @cindex persistent form of dictionary
13907: 
13908: An image file is a file containing an image of the Forth dictionary,
13909: i.e., compiled Forth code and data residing in the dictionary.  By
13910: convention, we use the extension @code{.fi} for image files.
13911: 
13912: @menu
13913: * Image Licensing Issues::      Distribution terms for images.
13914: * Image File Background::       Why have image files?
13915: * Non-Relocatable Image Files::  don't always work.
13916: * Data-Relocatable Image Files::  are better.
13917: * Fully Relocatable Image Files::  better yet.
13918: * Stack and Dictionary Sizes::  Setting the default sizes for an image.
13919: * Running Image Files::         @code{gforth -i @i{file}} or @i{file}.
13920: * Modifying the Startup Sequence::  and turnkey applications.
13921: @end menu
13922: 
13923: @node Image Licensing Issues, Image File Background, Image Files, Image Files
13924: @section Image Licensing Issues
13925: @cindex license for images
13926: @cindex image license
13927: 
13928: An image created with @code{gforthmi} (@pxref{gforthmi}) or
13929: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
13930: original image; i.e., according to copyright law it is a derived work of
13931: the original image.
13932: 
13933: Since Gforth is distributed under the GNU GPL, the newly created image
13934: falls under the GNU GPL, too. In particular, this means that if you
13935: distribute the image, you have to make all of the sources for the image
13936: available, including those you wrote.  For details see @ref{License, ,
13937: GNU General Public License (Section 3)}.
13938: 
13939: If you create an image with @code{cross} (@pxref{cross.fs}), the image
13940: contains only code compiled from the sources you gave it; if none of
13941: these sources is under the GPL, the terms discussed above do not apply
13942: to the image. However, if your image needs an engine (a gforth binary)
13943: that is under the GPL, you should make sure that you distribute both in
13944: a way that is at most a @emph{mere aggregation}, if you don't want the
13945: terms of the GPL to apply to the image.
13946: 
13947: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
13948: @section Image File Background
13949: @cindex image file background
13950: 
13951: Gforth consists not only of primitives (in the engine), but also of
13952: definitions written in Forth. Since the Forth compiler itself belongs to
13953: those definitions, it is not possible to start the system with the
13954: engine and the Forth source alone. Therefore we provide the Forth
13955: code as an image file in nearly executable form. When Gforth starts up,
13956: a C routine loads the image file into memory, optionally relocates the
13957: addresses, then sets up the memory (stacks etc.) according to
13958: information in the image file, and (finally) starts executing Forth
13959: code.
13960: 
13961: The image file variants represent different compromises between the
13962: goals of making it easy to generate image files and making them
13963: portable.
13964: 
13965: @cindex relocation at run-time
13966: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
13967: run-time. This avoids many of the complications discussed below (image
13968: files are data relocatable without further ado), but costs performance
13969: (one addition per memory access).
13970: 
13971: @cindex relocation at load-time
13972: By contrast, the Gforth loader performs relocation at image load time. The
13973: loader also has to replace tokens that represent primitive calls with the
13974: appropriate code-field addresses (or code addresses in the case of
13975: direct threading).
13976: 
13977: There are three kinds of image files, with different degrees of
13978: relocatability: non-relocatable, data-relocatable, and fully relocatable
13979: image files.
13980: 
13981: @cindex image file loader
13982: @cindex relocating loader
13983: @cindex loader for image files
13984: These image file variants have several restrictions in common; they are
13985: caused by the design of the image file loader:
13986: 
13987: @itemize @bullet
13988: @item
13989: There is only one segment; in particular, this means, that an image file
13990: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
13991: them). The contents of the stacks are not represented, either.
13992: 
13993: @item
13994: The only kinds of relocation supported are: adding the same offset to
13995: all cells that represent data addresses; and replacing special tokens
13996: with code addresses or with pieces of machine code.
13997: 
13998: If any complex computations involving addresses are performed, the
13999: results cannot be represented in the image file. Several applications that
14000: use such computations come to mind:
14001: @itemize @minus
14002: @item
14003: Hashing addresses (or data structures which contain addresses) for table
14004: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
14005: purpose, you will have no problem, because the hash tables are
14006: recomputed automatically when the system is started. If you use your own
14007: hash tables, you will have to do something similar.
14008: 
14009: @item
14010: There's a cute implementation of doubly-linked lists that uses
14011: @code{XOR}ed addresses. You could represent such lists as singly-linked
14012: in the image file, and restore the doubly-linked representation on
14013: startup.@footnote{In my opinion, though, you should think thrice before
14014: using a doubly-linked list (whatever implementation).}
14015: 
14016: @item
14017: The code addresses of run-time routines like @code{docol:} cannot be
14018: represented in the image file (because their tokens would be replaced by
14019: machine code in direct threaded implementations). As a workaround,
14020: compute these addresses at run-time with @code{>code-address} from the
14021: executions tokens of appropriate words (see the definitions of
14022: @code{docol:} and friends in @file{kernel/getdoers.fs}).
14023: 
14024: @item
14025: On many architectures addresses are represented in machine code in some
14026: shifted or mangled form. You cannot put @code{CODE} words that contain
14027: absolute addresses in this form in a relocatable image file. Workarounds
14028: are representing the address in some relative form (e.g., relative to
14029: the CFA, which is present in some register), or loading the address from
14030: a place where it is stored in a non-mangled form.
14031: @end itemize
14032: @end itemize
14033: 
14034: @node  Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
14035: @section Non-Relocatable Image Files
14036: @cindex non-relocatable image files
14037: @cindex image file, non-relocatable
14038: 
14039: These files are simple memory dumps of the dictionary. They are specific
14040: to the executable (i.e., @file{gforth} file) they were created
14041: with. What's worse, they are specific to the place on which the
14042: dictionary resided when the image was created. Now, there is no
14043: guarantee that the dictionary will reside at the same place the next
14044: time you start Gforth, so there's no guarantee that a non-relocatable
14045: image will work the next time (Gforth will complain instead of crashing,
14046: though).
14047: 
14048: You can create a non-relocatable image file with
14049: 
14050: 
14051: doc-savesystem
14052: 
14053: 
14054: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
14055: @section Data-Relocatable Image Files
14056: @cindex data-relocatable image files
14057: @cindex image file, data-relocatable
14058: 
14059: These files contain relocatable data addresses, but fixed code addresses
14060: (instead of tokens). They are specific to the executable (i.e.,
14061: @file{gforth} file) they were created with. For direct threading on some
14062: architectures (e.g., the i386), data-relocatable images do not work. You
14063: get a data-relocatable image, if you use @file{gforthmi} with a
14064: Gforth binary that is not doubly indirect threaded (@pxref{Fully
14065: Relocatable Image Files}).
14066: 
14067: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
14068: @section Fully Relocatable Image Files
14069: @cindex fully relocatable image files
14070: @cindex image file, fully relocatable
14071: 
14072: @cindex @file{kern*.fi}, relocatability
14073: @cindex @file{gforth.fi}, relocatability
14074: These image files have relocatable data addresses, and tokens for code
14075: addresses. They can be used with different binaries (e.g., with and
14076: without debugging) on the same machine, and even across machines with
14077: the same data formats (byte order, cell size, floating point
14078: format). However, they are usually specific to the version of Gforth
14079: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
14080: are fully relocatable.
14081: 
14082: There are two ways to create a fully relocatable image file:
14083: 
14084: @menu
14085: * gforthmi::                    The normal way
14086: * cross.fs::                    The hard way
14087: @end menu
14088: 
14089: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
14090: @subsection @file{gforthmi}
14091: @cindex @file{comp-i.fs}
14092: @cindex @file{gforthmi}
14093: 
14094: You will usually use @file{gforthmi}. If you want to create an
14095: image @i{file} that contains everything you would load by invoking
14096: Gforth with @code{gforth @i{options}}, you simply say:
14097: @example
14098: gforthmi @i{file} @i{options}
14099: @end example
14100: 
14101: E.g., if you want to create an image @file{asm.fi} that has the file
14102: @file{asm.fs} loaded in addition to the usual stuff, you could do it
14103: like this:
14104: 
14105: @example
14106: gforthmi asm.fi asm.fs
14107: @end example
14108: 
14109: @file{gforthmi} is implemented as a sh script and works like this: It
14110: produces two non-relocatable images for different addresses and then
14111: compares them. Its output reflects this: first you see the output (if
14112: any) of the two Gforth invocations that produce the non-relocatable image
14113: files, then you see the output of the comparing program: It displays the
14114: offset used for data addresses and the offset used for code addresses;
14115: moreover, for each cell that cannot be represented correctly in the
14116: image files, it displays a line like this:
14117: 
14118: @example
14119:      78DC         BFFFFA50         BFFFFA40
14120: @end example
14121: 
14122: This means that at offset $78dc from @code{forthstart}, one input image
14123: contains $bffffa50, and the other contains $bffffa40. Since these cells
14124: cannot be represented correctly in the output image, you should examine
14125: these places in the dictionary and verify that these cells are dead
14126: (i.e., not read before they are written).
14127: 
14128: @cindex --application, @code{gforthmi} option
14129: If you insert the option @code{--application} in front of the image file
14130: name, you will get an image that uses the @code{--appl-image} option
14131: instead of the @code{--image-file} option (@pxref{Invoking
14132: Gforth}). When you execute such an image on Unix (by typing the image
14133: name as command), the Gforth engine will pass all options to the image
14134: instead of trying to interpret them as engine options.
14135: 
14136: If you type @file{gforthmi} with no arguments, it prints some usage
14137: instructions.
14138: 
14139: @cindex @code{savesystem} during @file{gforthmi}
14140: @cindex @code{bye} during @file{gforthmi}
14141: @cindex doubly indirect threaded code
14142: @cindex environment variables
14143: @cindex @code{GFORTHD} -- environment variable
14144: @cindex @code{GFORTH} -- environment variable
14145: @cindex @code{gforth-ditc}
14146: There are a few wrinkles: After processing the passed @i{options}, the
14147: words @code{savesystem} and @code{bye} must be visible. A special doubly
14148: indirect threaded version of the @file{gforth} executable is used for
14149: creating the non-relocatable images; you can pass the exact filename of
14150: this executable through the environment variable @code{GFORTHD}
14151: (default: @file{gforth-ditc}); if you pass a version that is not doubly
14152: indirect threaded, you will not get a fully relocatable image, but a
14153: data-relocatable image (because there is no code address offset). The
14154: normal @file{gforth} executable is used for creating the relocatable
14155: image; you can pass the exact filename of this executable through the
14156: environment variable @code{GFORTH}.
14157: 
14158: @node cross.fs,  , gforthmi, Fully Relocatable Image Files
14159: @subsection @file{cross.fs}
14160: @cindex @file{cross.fs}
14161: @cindex cross-compiler
14162: @cindex metacompiler
14163: @cindex target compiler
14164: 
14165: You can also use @code{cross}, a batch compiler that accepts a Forth-like
14166: programming language (@pxref{Cross Compiler}).
14167: 
14168: @code{cross} allows you to create image files for machines with
14169: different data sizes and data formats than the one used for generating
14170: the image file. You can also use it to create an application image that
14171: does not contain a Forth compiler. These features are bought with
14172: restrictions and inconveniences in programming. E.g., addresses have to
14173: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
14174: order to make the code relocatable.
14175: 
14176: 
14177: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
14178: @section Stack and Dictionary Sizes
14179: @cindex image file, stack and dictionary sizes
14180: @cindex dictionary size default
14181: @cindex stack size default
14182: 
14183: If you invoke Gforth with a command line flag for the size
14184: (@pxref{Invoking Gforth}), the size you specify is stored in the
14185: dictionary. If you save the dictionary with @code{savesystem} or create
14186: an image with @file{gforthmi}, this size will become the default
14187: for the resulting image file. E.g., the following will create a
14188: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
14189: 
14190: @example
14191: gforthmi gforth.fi -m 1M
14192: @end example
14193: 
14194: In other words, if you want to set the default size for the dictionary
14195: and the stacks of an image, just invoke @file{gforthmi} with the
14196: appropriate options when creating the image.
14197: 
14198: @cindex stack size, cache-friendly
14199: Note: For cache-friendly behaviour (i.e., good performance), you should
14200: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
14201: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
14202: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
14203: 
14204: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
14205: @section Running Image Files
14206: @cindex running image files
14207: @cindex invoking image files
14208: @cindex image file invocation
14209: 
14210: @cindex -i, invoke image file
14211: @cindex --image file, invoke image file
14212: You can invoke Gforth with an image file @i{image} instead of the
14213: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
14214: @example
14215: gforth -i @i{image}
14216: @end example
14217: 
14218: @cindex executable image file
14219: @cindex image file, executable
14220: If your operating system supports starting scripts with a line of the
14221: form @code{#! ...}, you just have to type the image file name to start
14222: Gforth with this image file (note that the file extension @code{.fi} is
14223: just a convention). I.e., to run Gforth with the image file @i{image},
14224: you can just type @i{image} instead of @code{gforth -i @i{image}}.
14225: This works because every @code{.fi} file starts with a line of this
14226: format:
14227: 
14228: @example
14229: #! /usr/local/bin/gforth-0.4.0 -i
14230: @end example
14231: 
14232: The file and pathname for the Gforth engine specified on this line is
14233: the specific Gforth executable that it was built against; i.e. the value
14234: of the environment variable @code{GFORTH} at the time that
14235: @file{gforthmi} was executed.
14236: 
14237: You can make use of the same shell capability to make a Forth source
14238: file into an executable. For example, if you place this text in a file:
14239: 
14240: @example
14241: #! /usr/local/bin/gforth
14242: 
14243: ." Hello, world" CR
14244: bye
14245: @end example
14246: 
14247: @noindent
14248: and then make the file executable (chmod +x in Unix), you can run it
14249: directly from the command line. The sequence @code{#!} is used in two
14250: ways; firstly, it is recognised as a ``magic sequence'' by the operating
14251: system@footnote{The Unix kernel actually recognises two types of files:
14252: executable files and files of data, where the data is processed by an
14253: interpreter that is specified on the ``interpreter line'' -- the first
14254: line of the file, starting with the sequence #!. There may be a small
14255: limit (e.g., 32) on the number of characters that may be specified on
14256: the interpreter line.} secondly it is treated as a comment character by
14257: Gforth. Because of the second usage, a space is required between
14258: @code{#!} and the path to the executable (moreover, some Unixes
14259: require the sequence @code{#! /}).
14260: 
14261: The disadvantage of this latter technique, compared with using
14262: @file{gforthmi}, is that it is slightly slower; the Forth source code is
14263: compiled on-the-fly, each time the program is invoked.
14264: 
14265: doc-#!
14266: 
14267: 
14268: @node Modifying the Startup Sequence,  , Running Image Files, Image Files
14269: @section Modifying the Startup Sequence
14270: @cindex startup sequence for image file
14271: @cindex image file initialization sequence
14272: @cindex initialization sequence of image file
14273: 
14274: You can add your own initialization to the startup sequence through the
14275: deferred word @code{'cold}. @code{'cold} is invoked just before the
14276: image-specific command line processing (i.e., loading files and
14277: evaluating (@code{-e}) strings) starts.
14278: 
14279: A sequence for adding your initialization usually looks like this:
14280: 
14281: @example
14282: :noname
14283:     Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
14284:     ... \ your stuff
14285: ; IS 'cold
14286: @end example
14287: 
14288: @cindex turnkey image files
14289: @cindex image file, turnkey applications
14290: You can make a turnkey image by letting @code{'cold} execute a word
14291: (your turnkey application) that never returns; instead, it exits Gforth
14292: via @code{bye} or @code{throw}.
14293: 
14294: @cindex command-line arguments, access
14295: @cindex arguments on the command line, access
14296: You can access the (image-specific) command-line arguments through the
14297: variables @code{argc} and @code{argv}. @code{arg} provides convenient
14298: access to @code{argv}.
14299: 
14300: If @code{'cold} exits normally, Gforth processes the command-line
14301: arguments as files to be loaded and strings to be evaluated.  Therefore,
14302: @code{'cold} should remove the arguments it has used in this case.
14303: 
14304: 
14305: 
14306: doc-'cold
14307: doc-argc
14308: doc-argv
14309: doc-arg
14310: 
14311: 
14312: 
14313: @c ******************************************************************
14314: @node Engine, Binding to System Library, Image Files, Top
14315: @chapter Engine
14316: @cindex engine
14317: @cindex virtual machine
14318: 
14319: Reading this chapter is not necessary for programming with Gforth. It
14320: may be helpful for finding your way in the Gforth sources.
14321: 
14322: The ideas in this section have also been published in Bernd Paysan,
14323: @cite{ANS fig/GNU/??? Forth} (in German), Forth-Tagung '93 and M. Anton
14324: Ertl, @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z, A
14325: Portable Forth Engine}}, EuroForth '93.
14326: 
14327: @menu
14328: * Portability::                 
14329: * Threading::                   
14330: * Primitives::                  
14331: * Performance::                 
14332: @end menu
14333: 
14334: @node Portability, Threading, Engine, Engine
14335: @section Portability
14336: @cindex engine portability
14337: 
14338: An important goal of the Gforth Project is availability across a wide
14339: range of personal machines. fig-Forth, and, to a lesser extent, F83,
14340: achieved this goal by manually coding the engine in assembly language
14341: for several then-popular processors. This approach is very
14342: labor-intensive and the results are short-lived due to progress in
14343: computer architecture.
14344: 
14345: @cindex C, using C for the engine
14346: Others have avoided this problem by coding in C, e.g., Mitch Bradley
14347: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
14348: particularly popular for UNIX-based Forths due to the large variety of
14349: architectures of UNIX machines. Unfortunately an implementation in C
14350: does not mix well with the goals of efficiency and with using
14351: traditional techniques: Indirect or direct threading cannot be expressed
14352: in C, and switch threading, the fastest technique available in C, is
14353: significantly slower. Another problem with C is that it is very
14354: cumbersome to express double integer arithmetic.
14355: 
14356: @cindex GNU C for the engine
14357: @cindex long long
14358: Fortunately, there is a portable language that does not have these
14359: limitations: GNU C, the version of C processed by the GNU C compiler
14360: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
14361: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
14362: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
14363: threading possible, its @code{long long} type (@pxref{Long Long, ,
14364: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
14365: double numbers@footnote{Unfortunately, long longs are not implemented
14366: properly on all machines (e.g., on alpha-osf1, long longs are only 64
14367: bits, the same size as longs (and pointers), but they should be twice as
14368: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
14369: C Manual}). So, we had to implement doubles in C after all. Still, on
14370: most machines we can use long longs and achieve better performance than
14371: with the emulation package.}. GNU C is available for free on all
14372: important (and many unimportant) UNIX machines, VMS, 80386s running
14373: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
14374: on all these machines.
14375: 
14376: Writing in a portable language has the reputation of producing code that
14377: is slower than assembly. For our Forth engine we repeatedly looked at
14378: the code produced by the compiler and eliminated most compiler-induced
14379: inefficiencies by appropriate changes in the source code.
14380: 
14381: @cindex explicit register declarations
14382: @cindex --enable-force-reg, configuration flag
14383: @cindex -DFORCE_REG
14384: However, register allocation cannot be portably influenced by the
14385: programmer, leading to some inefficiencies on register-starved
14386: machines. We use explicit register declarations (@pxref{Explicit Reg
14387: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
14388: improve the speed on some machines. They are turned on by using the
14389: configuration flag @code{--enable-force-reg} (@code{gcc} switch
14390: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
14391: machine, but also on the compiler version: On some machines some
14392: compiler versions produce incorrect code when certain explicit register
14393: declarations are used. So by default @code{-DFORCE_REG} is not used.
14394: 
14395: @node Threading, Primitives, Portability, Engine
14396: @section Threading
14397: @cindex inner interpreter implementation
14398: @cindex threaded code implementation
14399: 
14400: @cindex labels as values
14401: GNU C's labels as values extension (available since @code{gcc-2.0},
14402: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
14403: makes it possible to take the address of @i{label} by writing
14404: @code{&&@i{label}}.  This address can then be used in a statement like
14405: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
14406: @code{goto x}.
14407: 
14408: @cindex @code{NEXT}, indirect threaded
14409: @cindex indirect threaded inner interpreter
14410: @cindex inner interpreter, indirect threaded
14411: With this feature an indirect threaded @code{NEXT} looks like:
14412: @example
14413: cfa = *ip++;
14414: ca = *cfa;
14415: goto *ca;
14416: @end example
14417: @cindex instruction pointer
14418: For those unfamiliar with the names: @code{ip} is the Forth instruction
14419: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
14420: execution token and points to the code field of the next word to be
14421: executed; The @code{ca} (code address) fetched from there points to some
14422: executable code, e.g., a primitive or the colon definition handler
14423: @code{docol}.
14424: 
14425: @cindex @code{NEXT}, direct threaded
14426: @cindex direct threaded inner interpreter
14427: @cindex inner interpreter, direct threaded
14428: Direct threading is even simpler:
14429: @example
14430: ca = *ip++;
14431: goto *ca;
14432: @end example
14433: 
14434: Of course we have packaged the whole thing neatly in macros called
14435: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
14436: 
14437: @menu
14438: * Scheduling::                  
14439: * Direct or Indirect Threaded?::  
14440: * DOES>::                       
14441: @end menu
14442: 
14443: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
14444: @subsection Scheduling
14445: @cindex inner interpreter optimization
14446: 
14447: There is a little complication: Pipelined and superscalar processors,
14448: i.e., RISC and some modern CISC machines can process independent
14449: instructions while waiting for the results of an instruction. The
14450: compiler usually reorders (schedules) the instructions in a way that
14451: achieves good usage of these delay slots. However, on our first tries
14452: the compiler did not do well on scheduling primitives. E.g., for
14453: @code{+} implemented as
14454: @example
14455: n=sp[0]+sp[1];
14456: sp++;
14457: sp[0]=n;
14458: NEXT;
14459: @end example
14460: the @code{NEXT} comes strictly after the other code, i.e., there is
14461: nearly no scheduling. After a little thought the problem becomes clear:
14462: The compiler cannot know that @code{sp} and @code{ip} point to different
14463: addresses (and the version of @code{gcc} we used would not know it even
14464: if it was possible), so it could not move the load of the cfa above the
14465: store to the TOS. Indeed the pointers could be the same, if code on or
14466: very near the top of stack were executed. In the interest of speed we
14467: chose to forbid this probably unused ``feature'' and helped the compiler
14468: in scheduling: @code{NEXT} is divided into several parts:
14469: @code{NEXT_P0}, @code{NEXT_P1} and @code{NEXT_P2}). @code{+} now looks
14470: like:
14471: @example
14472: NEXT_P0;
14473: n=sp[0]+sp[1];
14474: sp++;
14475: NEXT_P1;
14476: sp[0]=n;
14477: NEXT_P2;
14478: @end example
14479: 
14480: There are various schemes that distribute the different operations of
14481: NEXT between these parts in several ways; in general, different schemes
14482: perform best on different processors.  We use a scheme for most
14483: architectures that performs well for most processors of this
14484: architecture; in the furture we may switch to benchmarking and chosing
14485: the scheme on installation time.
14486: 
14487: 
14488: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
14489: @subsection Direct or Indirect Threaded?
14490: @cindex threading, direct or indirect?
14491: 
14492: @cindex -DDIRECT_THREADED
14493: Both! After packaging the nasty details in macro definitions we
14494: realized that we could switch between direct and indirect threading by
14495: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
14496: defining a few machine-specific macros for the direct-threading case.
14497: On the Forth level we also offer access words that hide the
14498: differences between the threading methods (@pxref{Threading Words}).
14499: 
14500: Indirect threading is implemented completely machine-independently.
14501: Direct threading needs routines for creating jumps to the executable
14502: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
14503: machine-dependent, but they do not amount to many source lines. Therefore,
14504: even porting direct threading to a new machine requires little effort.
14505: 
14506: @cindex --enable-indirect-threaded, configuration flag
14507: @cindex --enable-direct-threaded, configuration flag
14508: The default threading method is machine-dependent. You can enforce a
14509: specific threading method when building Gforth with the configuration
14510: flag @code{--enable-direct-threaded} or
14511: @code{--enable-indirect-threaded}. Note that direct threading is not
14512: supported on all machines.
14513: 
14514: @node DOES>,  , Direct or Indirect Threaded?, Threading
14515: @subsection DOES>
14516: @cindex @code{DOES>} implementation
14517: 
14518: @cindex @code{dodoes} routine
14519: @cindex @code{DOES>}-code
14520: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
14521: the chunk of code executed by every word defined by a
14522: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
14523: the Forth code to be executed, i.e. the code after the
14524: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
14525: 
14526: In fig-Forth the code field points directly to the @code{dodoes} and the
14527: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
14528: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
14529: the Forth-79 and all later standards, because in fig-Forth this address
14530: lies in the body (which is illegal in these standards). However, by
14531: making the code field larger for all words this solution becomes legal
14532: again. We use this approach for the indirect threaded version and for
14533: direct threading on some machines. Leaving a cell unused in most words
14534: is a bit wasteful, but on the machines we are targeting this is hardly a
14535: problem. The other reason for having a code field size of two cells is
14536: to avoid having different image files for direct and indirect threaded
14537: systems (direct threaded systems require two-cell code fields on many
14538: machines).
14539: 
14540: @cindex @code{DOES>}-handler
14541: The other approach is that the code field points or jumps to the cell
14542: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
14543: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
14544: @code{DOES>}-code address by computing the code address, i.e., the address of
14545: the jump to @code{dodoes}, and add the length of that jump field. A variant of
14546: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
14547: return address (which can be found in the return register on RISCs) is
14548: the @code{DOES>}-code address. Since the two cells available in the code field
14549: are used up by the jump to the code address in direct threading on many
14550: architectures, we use this approach for direct threading on these
14551: architectures. We did not want to add another cell to the code field.
14552: 
14553: @node Primitives, Performance, Threading, Engine
14554: @section Primitives
14555: @cindex primitives, implementation
14556: @cindex virtual machine instructions, implementation
14557: 
14558: @menu
14559: * Automatic Generation::        
14560: * TOS Optimization::            
14561: * Produced code::               
14562: @end menu
14563: 
14564: @node Automatic Generation, TOS Optimization, Primitives, Primitives
14565: @subsection Automatic Generation
14566: @cindex primitives, automatic generation
14567: 
14568: @cindex @file{prims2x.fs}
14569: Since the primitives are implemented in a portable language, there is no
14570: longer any need to minimize the number of primitives. On the contrary,
14571: having many primitives has an advantage: speed. In order to reduce the
14572: number of errors in primitives and to make programming them easier, we
14573: provide a tool, the primitive generator (@file{prims2x.fs}), that
14574: automatically generates most (and sometimes all) of the C code for a
14575: primitive from the stack effect notation.  The source for a primitive
14576: has the following form:
14577: 
14578: @cindex primitive source format
14579: @format
14580: @i{Forth-name}  ( @i{stack-effect} )        @i{category}    [@i{pronounc.}]
14581: [@code{""}@i{glossary entry}@code{""}]
14582: @i{C code}
14583: [@code{:}
14584: @i{Forth code}]
14585: @end format
14586: 
14587: The items in brackets are optional. The category and glossary fields
14588: are there for generating the documentation, the Forth code is there
14589: for manual implementations on machines without GNU C. E.g., the source
14590: for the primitive @code{+} is:
14591: @example
14592: +    ( n1 n2 -- n )   core    plus
14593: n = n1+n2;
14594: @end example
14595: 
14596: This looks like a specification, but in fact @code{n = n1+n2} is C
14597: code. Our primitive generation tool extracts a lot of information from
14598: the stack effect notations@footnote{We use a one-stack notation, even
14599: though we have separate data and floating-point stacks; The separate
14600: notation can be generated easily from the unified notation.}: The number
14601: of items popped from and pushed on the stack, their type, and by what
14602: name they are referred to in the C code. It then generates a C code
14603: prelude and postlude for each primitive. The final C code for @code{+}
14604: looks like this:
14605: 
14606: @example
14607: I_plus: /* + ( n1 n2 -- n ) */  /* label, stack effect */
14608: /*  */                          /* documentation */
14609: NAME("+")                       /* debugging output (with -DDEBUG) */
14610: @{
14611: DEF_CA                          /* definition of variable ca (indirect threading) */
14612: Cell n1;                        /* definitions of variables */
14613: Cell n2;
14614: Cell n;
14615: NEXT_P0;                        /* NEXT part 0 */
14616: n1 = (Cell) sp[1];              /* input */
14617: n2 = (Cell) TOS;
14618: sp += 1;                        /* stack adjustment */
14619: @{
14620: n = n1+n2;                      /* C code taken from the source */
14621: @}
14622: NEXT_P1;                        /* NEXT part 1 */
14623: TOS = (Cell)n;                  /* output */
14624: NEXT_P2;                        /* NEXT part 2 */
14625: @}
14626: @end example
14627: 
14628: This looks long and inefficient, but the GNU C compiler optimizes quite
14629: well and produces optimal code for @code{+} on, e.g., the R3000 and the
14630: HP RISC machines: Defining the @code{n}s does not produce any code, and
14631: using them as intermediate storage also adds no cost.
14632: 
14633: There are also other optimizations that are not illustrated by this
14634: example: assignments between simple variables are usually for free (copy
14635: propagation). If one of the stack items is not used by the primitive
14636: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
14637: (dead code elimination). On the other hand, there are some things that
14638: the compiler does not do, therefore they are performed by
14639: @file{prims2x.fs}: The compiler does not optimize code away that stores
14640: a stack item to the place where it just came from (e.g., @code{over}).
14641: 
14642: While programming a primitive is usually easy, there are a few cases
14643: where the programmer has to take the actions of the generator into
14644: account, most notably @code{?dup}, but also words that do not (always)
14645: fall through to @code{NEXT}.
14646: 
14647: @node TOS Optimization, Produced code, Automatic Generation, Primitives
14648: @subsection TOS Optimization
14649: @cindex TOS optimization for primitives
14650: @cindex primitives, keeping the TOS in a register
14651: 
14652: An important optimization for stack machine emulators, e.g., Forth
14653: engines, is keeping  one or more of the top stack items in
14654: registers.  If a word has the stack effect @i{in1}...@i{inx} @code{--}
14655: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
14656: @itemize @bullet
14657: @item
14658: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
14659: due to fewer loads from and stores to the stack.
14660: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
14661: @i{y<n}, due to additional moves between registers.
14662: @end itemize
14663: 
14664: @cindex -DUSE_TOS
14665: @cindex -DUSE_NO_TOS
14666: In particular, keeping one item in a register is never a disadvantage,
14667: if there are enough registers. Keeping two items in registers is a
14668: disadvantage for frequent words like @code{?branch}, constants,
14669: variables, literals and @code{i}. Therefore our generator only produces
14670: code that keeps zero or one items in registers. The generated C code
14671: covers both cases; the selection between these alternatives is made at
14672: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
14673: code for @code{+} is just a simple variable name in the one-item case,
14674: otherwise it is a macro that expands into @code{sp[0]}. Note that the
14675: GNU C compiler tries to keep simple variables like @code{TOS} in
14676: registers, and it usually succeeds, if there are enough registers.
14677: 
14678: @cindex -DUSE_FTOS
14679: @cindex -DUSE_NO_FTOS
14680: The primitive generator performs the TOS optimization for the
14681: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
14682: operations the benefit of this optimization is even larger:
14683: floating-point operations take quite long on most processors, but can be
14684: performed in parallel with other operations as long as their results are
14685: not used. If the FP-TOS is kept in a register, this works. If
14686: it is kept on the stack, i.e., in memory, the store into memory has to
14687: wait for the result of the floating-point operation, lengthening the
14688: execution time of the primitive considerably.
14689: 
14690: The TOS optimization makes the automatic generation of primitives a
14691: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
14692: @code{TOS} is not sufficient. There are some special cases to
14693: consider:
14694: @itemize @bullet
14695: @item In the case of @code{dup ( w -- w w )} the generator must not
14696: eliminate the store to the original location of the item on the stack,
14697: if the TOS optimization is turned on.
14698: @item Primitives with stack effects of the form @code{--}
14699: @i{out1}...@i{outy} must store the TOS to the stack at the start.
14700: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
14701: must load the TOS from the stack at the end. But for the null stack
14702: effect @code{--} no stores or loads should be generated.
14703: @end itemize
14704: 
14705: @node Produced code,  , TOS Optimization, Primitives
14706: @subsection Produced code
14707: @cindex primitives, assembly code listing
14708: 
14709: @cindex @file{engine.s}
14710: To see what assembly code is produced for the primitives on your machine
14711: with your compiler and your flag settings, type @code{make engine.s} and
14712: look at the resulting file @file{engine.s}.  Alternatively, you can also
14713: disassemble the code of primitives with @code{see} on some architectures.
14714: 
14715: @node  Performance,  , Primitives, Engine
14716: @section Performance
14717: @cindex performance of some Forth interpreters
14718: @cindex engine performance
14719: @cindex benchmarking Forth systems
14720: @cindex Gforth performance
14721: 
14722: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
14723: impossible to write a significantly faster engine.
14724: 
14725: On register-starved machines like the 386 architecture processors
14726: improvements are possible, because @code{gcc} does not utilize the
14727: registers as well as a human, even with explicit register declarations;
14728: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
14729: and hand-tuned it for the 486; this system is 1.19 times faster on the
14730: Sieve benchmark on a 486DX2/66 than Gforth compiled with
14731: @code{gcc-2.6.3} with @code{-DFORCE_REG}.  The situation has improved
14732: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
14733: registers fit in real registers (and we can even afford to use the TOS
14734: optimization), resulting in a speedup of 1.14 on the sieve over the
14735: earlier results.
14736: 
14737: @cindex Win32Forth performance
14738: @cindex NT Forth performance
14739: @cindex eforth performance
14740: @cindex ThisForth performance
14741: @cindex PFE performance
14742: @cindex TILE performance
14743: The potential advantage of assembly language implementations is not
14744: necessarily realized in complete Forth systems: We compared Gforth-0.4.9
14745: (direct threaded, compiled with @code{gcc-2.95.1} and
14746: @code{-DFORCE_REG}) with Win32Forth 1.2093 (newer versions are
14747: reportedly much faster), LMI's NT Forth (Beta, May 1994) and Eforth
14748: (with and without peephole (aka pinhole) optimization of the threaded
14749: code); all these systems were written in assembly language. We also
14750: compared Gforth with three systems written in C: PFE-0.9.14 (compiled
14751: with @code{gcc-2.6.3} with the default configuration for Linux:
14752: @code{-O2 -fomit-frame-pointer -DUSE_REGS -DUNROLL_NEXT}), ThisForth
14753: Beta (compiled with @code{gcc-2.6.3 -O3 -fomit-frame-pointer}; ThisForth
14754: employs peephole optimization of the threaded code) and TILE (compiled
14755: with @code{make opt}). We benchmarked Gforth, PFE, ThisForth and TILE on
14756: a 486DX2/66 under Linux. Kenneth O'Heskin kindly provided the results
14757: for Win32Forth and NT Forth on a 486DX2/66 with similar memory
14758: performance under Windows NT. Marcel Hendrix ported Eforth to Linux,
14759: then extended it to run the benchmarks, added the peephole optimizer,
14760: ran the benchmarks and reported the results.
14761: 
14762: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
14763: matrix multiplication come from the Stanford integer benchmarks and have
14764: been translated into Forth by Martin Fraeman; we used the versions
14765: included in the TILE Forth package, but with bigger data set sizes; and
14766: a recursive Fibonacci number computation for benchmarking calling
14767: performance. The following table shows the time taken for the benchmarks
14768: scaled by the time taken by Gforth (in other words, it shows the speedup
14769: factor that Gforth achieved over the other systems).
14770: 
14771: @example
14772: relative      Win32-    NT       eforth       This-      
14773:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
14774: sieve     1.00  1.60  1.32   1.60  0.98  1.82  3.67  9.91
14775: bubble    1.00  1.55  1.66   1.75  1.04  1.78        4.58
14776: matmul    1.00  1.71  1.57   1.69  0.86  1.83        4.74
14777: fib       1.00  1.76  1.54   1.41  1.00  2.01  3.45  4.96
14778: @end example
14779: 
14780: You may be quite surprised by the good performance of Gforth when
14781: compared with systems written in assembly language. One important reason
14782: for the disappointing performance of these other systems is probably
14783: that they are not written optimally for the 486 (e.g., they use the
14784: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
14785: but costly method for relocating the Forth image: like @code{cforth}, it
14786: computes the actual addresses at run time, resulting in two address
14787: computations per @code{NEXT} (@pxref{Image File Background}).
14788: 
14789: Only Eforth with the peephole optimizer performs comparable to
14790: Gforth. The speedups achieved with peephole optimization of threaded
14791: code are quite remarkable. Adding a peephole optimizer to Gforth should
14792: cause similar speedups.
14793: 
14794: The speedup of Gforth over PFE, ThisForth and TILE can be easily
14795: explained with the self-imposed restriction of the latter systems to
14796: standard C, which makes efficient threading impossible (however, the
14797: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
14798: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
14799: Moreover, current C compilers have a hard time optimizing other aspects
14800: of the ThisForth and the TILE source.
14801: 
14802: The performance of Gforth on 386 architecture processors varies widely
14803: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
14804: allocate any of the virtual machine registers into real machine
14805: registers by itself and would not work correctly with explicit register
14806: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
14807: the Sieve) than the one measured above.
14808: 
14809: Note that there have been several releases of Win32Forth since the
14810: release presented here, so the results presented above may have little
14811: predictive value for the performance of Win32Forth today (results for
14812: the current release on an i486DX2/66 are welcome).
14813: 
14814: @cindex @file{Benchres}
14815: In
14816: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz,
14817: Translating Forth to Efficient C}} by M. Anton Ertl and Martin
14818: Maierhofer (presented at EuroForth '95), an indirect threaded version of
14819: Gforth is compared with Win32Forth, NT Forth, PFE, ThisForth, and
14820: several native code systems; that version of Gforth is slower on a 486
14821: than the direct threaded version used here. You can find a newer version
14822: of these measurements at
14823: @uref{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
14824: find numbers for Gforth on various machines in @file{Benchres}.
14825: 
14826: @c ******************************************************************
14827: @node Binding to System Library, Cross Compiler, Engine, Top
14828: @chapter Binding to System Library
14829: 
14830: @node Cross Compiler, Bugs, Binding to System Library, Top
14831: @chapter Cross Compiler
14832: @cindex @file{cross.fs}
14833: @cindex cross-compiler
14834: @cindex metacompiler
14835: @cindex target compiler
14836: 
14837: The cross compiler is used to bootstrap a Forth kernel. Since Gforth is
14838: mostly written in Forth, including crucial parts like the outer
14839: interpreter and compiler, it needs compiled Forth code to get
14840: started. The cross compiler allows to create new images for other
14841: architectures, even running under another Forth system.
14842: 
14843: @menu
14844: * Using the Cross Compiler::    
14845: * How the Cross Compiler Works::  
14846: @end menu
14847: 
14848: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
14849: @section Using the Cross Compiler
14850: 
14851: The cross compiler uses a language that resembles Forth, but isn't. The
14852: main difference is that you can execute Forth code after definition,
14853: while you usually can't execute the code compiled by cross, because the
14854: code you are compiling is typically for a different computer than the
14855: one you are compiling on.
14856: 
14857: @c anton: This chapter is somewhat different from waht I would expect: I
14858: @c would expect an explanation of the cross language and how to create an
14859: @c application image with it.  The section explains some aspects of
14860: @c creating a Gforth kernel.
14861: 
14862: The Makefile is already set up to allow you to create kernels for new
14863: architectures with a simple make command. The generic kernels using the
14864: GCC compiled virtual machine are created in the normal build process
14865: with @code{make}. To create a embedded Gforth executable for e.g. the
14866: 8086 processor (running on a DOS machine), type
14867: 
14868: @example
14869: make kernl-8086.fi
14870: @end example
14871: 
14872: This will use the machine description from the @file{arch/8086}
14873: directory to create a new kernel. A machine file may look like that:
14874: 
14875: @example
14876: \ Parameter for target systems                         06oct92py
14877: 
14878:     4 Constant cell             \ cell size in bytes
14879:     2 Constant cell<<           \ cell shift to bytes
14880:     5 Constant cell>bit         \ cell shift to bits
14881:     8 Constant bits/char        \ bits per character
14882:     8 Constant bits/byte        \ bits per byte [default: 8]
14883:     8 Constant float            \ bytes per float
14884:     8 Constant /maxalign        \ maximum alignment in bytes
14885: false Constant bigendian        \ byte order
14886: ( true=big, false=little )
14887: 
14888: include machpc.fs               \ feature list
14889: @end example
14890: 
14891: This part is obligatory for the cross compiler itself, the feature list
14892: is used by the kernel to conditionally compile some features in and out,
14893: depending on whether the target supports these features.
14894: 
14895: There are some optional features, if you define your own primitives,
14896: have an assembler, or need special, nonstandard preparation to make the
14897: boot process work. @code{asm-include} includes an assembler,
14898: @code{prims-include} includes primitives, and @code{>boot} prepares for
14899: booting.
14900: 
14901: @example
14902: : asm-include    ." Include assembler" cr
14903:   s" arch/8086/asm.fs" included ;
14904: 
14905: : prims-include  ." Include primitives" cr
14906:   s" arch/8086/prim.fs" included ;
14907: 
14908: : >boot          ." Prepare booting" cr
14909:   s" ' boot >body into-forth 1+ !" evaluate ;
14910: @end example
14911: 
14912: These words are used as sort of macro during the cross compilation in
14913: the file @file{kernel/main.fs}. Instead of using these macros, it would
14914: be possible --- but more complicated --- to write a new kernel project
14915: file, too.
14916: 
14917: @file{kernel/main.fs} expects the machine description file name on the
14918: stack; the cross compiler itself (@file{cross.fs}) assumes that either
14919: @code{mach-file} leaves a counted string on the stack, or
14920: @code{machine-file} leaves an address, count pair of the filename on the
14921: stack.
14922: 
14923: The feature list is typically controlled using @code{SetValue}, generic
14924: files that are used by several projects can use @code{DefaultValue}
14925: instead. Both functions work like @code{Value}, when the value isn't
14926: defined, but @code{SetValue} works like @code{to} if the value is
14927: defined, and @code{DefaultValue} doesn't set anything, if the value is
14928: defined.
14929: 
14930: @example
14931: \ generic mach file for pc gforth                       03sep97jaw
14932: 
14933: true DefaultValue NIL  \ relocating
14934: 
14935: >ENVIRON
14936: 
14937: true DefaultValue file          \ controls the presence of the
14938:                                 \ file access wordset
14939: true DefaultValue OS            \ flag to indicate a operating system
14940: 
14941: true DefaultValue prims         \ true: primitives are c-code
14942: 
14943: true DefaultValue floating      \ floating point wordset is present
14944: 
14945: true DefaultValue glocals       \ gforth locals are present
14946:                                 \ will be loaded
14947: true DefaultValue dcomps        \ double number comparisons
14948: 
14949: true DefaultValue hash          \ hashing primitives are loaded/present
14950: 
14951: true DefaultValue xconds        \ used together with glocals,
14952:                                 \ special conditionals supporting gforths'
14953:                                 \ local variables
14954: true DefaultValue header        \ save a header information
14955: 
14956: true DefaultValue backtrace     \ enables backtrace code
14957: 
14958: false DefaultValue ec
14959: false DefaultValue crlf
14960: 
14961: cell 2 = [IF] &32 [ELSE] &256 [THEN] KB DefaultValue kernel-size
14962: 
14963: &16 KB          DefaultValue stack-size
14964: &15 KB &512 +   DefaultValue fstack-size
14965: &15 KB          DefaultValue rstack-size
14966: &14 KB &512 +   DefaultValue lstack-size
14967: @end example
14968: 
14969: @node How the Cross Compiler Works,  , Using the Cross Compiler, Cross Compiler
14970: @section How the Cross Compiler Works
14971: 
14972: @node Bugs, Origin, Cross Compiler, Top
14973: @appendix Bugs
14974: @cindex bug reporting
14975: 
14976: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
14977: 
14978: If you find a bug, please submit a bug report through
14979: @uref{https://savannah.gnu.org/bugs/?func=addbug&group=gforth}.
14980: 
14981: @itemize @bullet
14982: @item
14983: A program (or a sequence of keyboard commands) that reproduces the bug.
14984: @item
14985: A description of what you think constitutes the buggy behaviour.
14986: @item
14987: The Gforth version used (it is announced at the start of an
14988: interactive Gforth session).
14989: @item
14990: The machine and operating system (on Unix
14991: systems @code{uname -a} will report this information).
14992: @item
14993: The installation options (you can find the configure options at the
14994: start of @file{config.status}) and configuration (@code{configure}
14995: output or @file{config.cache}).
14996: @item
14997: A complete list of changes (if any) you (or your installer) have made to the
14998: Gforth sources.
14999: @end itemize
15000: 
15001: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
15002: to Report Bugs, gcc.info, GNU C Manual}.
15003: 
15004: 
15005: @node Origin, Forth-related information, Bugs, Top
15006: @appendix Authors and Ancestors of Gforth
15007: 
15008: @section Authors and Contributors
15009: @cindex authors of Gforth
15010: @cindex contributors to Gforth
15011: 
15012: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
15013: Ertl. The third major author was Jens Wilke.  Neal Crook contributed a
15014: lot to the manual.  Assemblers and disassemblers were contributed by
15015: Andrew McKewan, Christian Pirker, and Bernd Thallner.  Lennart Benschop
15016: (who was one of Gforth's first users, in mid-1993) and Stuart Ramsden
15017: inspired us with their continuous feedback. Lennart Benshop contributed
15018: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
15019: support for calling C libraries. Helpful comments also came from Paul
15020: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
15021: Wavrik, Barrie Stott, Marc de Groot, Jorge Acerada, Bruce Hoyt, and
15022: Robert Epprecht. Since the release of Gforth-0.2.1 there were also
15023: helpful comments from many others; thank you all, sorry for not listing
15024: you here (but digging through my mailbox to extract your names is on my
15025: to-do list).
15026: 
15027: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
15028: and autoconf, among others), and to the creators of the Internet: Gforth
15029: was developed across the Internet, and its authors did not meet
15030: physically for the first 4 years of development.
15031: 
15032: @section Pedigree
15033: @cindex pedigree of Gforth
15034: 
15035: Gforth descends from bigFORTH (1993) and fig-Forth.  Of course, a
15036: significant part of the design of Gforth was prescribed by ANS Forth.
15037: 
15038: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
15039: 32 bit native code version of VolksForth for the Atari ST, written
15040: mostly by Dietrich Weineck.
15041: 
15042: VolksForth was written by Klaus Schleisiek, Bernd Pennemann, Georg
15043: Rehfeld and Dietrich Weineck for the C64 (called UltraForth there) in
15044: the mid-80s and ported to the Atari ST in 1986.  It descends from F83.
15045: 
15046: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
15047: Forth-83 standard. !! Pedigree? When?
15048: 
15049: A team led by Bill Ragsdale implemented fig-Forth on many processors in
15050: 1979. Robert Selzer and Bill Ragsdale developed the original
15051: implementation of fig-Forth for the 6502 based on microForth.
15052: 
15053: The principal architect of microForth was Dean Sanderson. microForth was
15054: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
15055: the 1802, and subsequently implemented on the 8080, the 6800 and the
15056: Z80.
15057: 
15058: All earlier Forth systems were custom-made, usually by Charles Moore,
15059: who discovered (as he puts it) Forth during the late 60s. The first full
15060: Forth existed in 1971.
15061: 
15062: A part of the information in this section comes from
15063: @cite{@uref{http://www.forth.com/Content/History/History1.htm,The
15064: Evolution of Forth}} by Elizabeth D. Rather, Donald R. Colburn and
15065: Charles H. Moore, presented at the HOPL-II conference and preprinted in
15066: SIGPLAN Notices 28(3), 1993.  You can find more historical and
15067: genealogical information about Forth there.
15068: 
15069: @c ------------------------------------------------------------------
15070: @node Forth-related information, Word Index, Origin, Top
15071: @appendix Other Forth-related information
15072: @cindex Forth-related information
15073: 
15074: @c anton: I threw most of this stuff out, because it can be found through
15075: @c the FAQ and the FAQ is more likely to be up-to-date.
15076: 
15077: @cindex comp.lang.forth
15078: @cindex frequently asked questions
15079: There is an active news group (comp.lang.forth) discussing Forth
15080: (including Gforth) and Forth-related issues. Its
15081: @uref{http://www.complang.tuwien.ac.at/forth/faq/faq-general-2.html,FAQs}
15082: (frequently asked questions and their answers) contains a lot of
15083: information on Forth.  You should read it before posting to
15084: comp.lang.forth.
15085: 
15086: The ANS Forth standard is most usable in its
15087: @uref{http://www.taygeta.com/forth/dpans.html, HTML form}.
15088: 
15089: @c ------------------------------------------------------------------
15090: @node Word Index, Concept Index, Forth-related information, Top
15091: @unnumbered Word Index
15092: 
15093: This index is a list of Forth words that have ``glossary'' entries
15094: within this manual. Each word is listed with its stack effect and
15095: wordset.
15096: 
15097: @printindex fn
15098: 
15099: @c anton: the name index seems superfluous given the word and concept indices.
15100: 
15101: @c @node Name Index, Concept Index, Word Index, Top
15102: @c @unnumbered Name Index
15103: 
15104: @c This index is a list of Forth words that have ``glossary'' entries
15105: @c within this manual.
15106: 
15107: @c @printindex ky
15108: 
15109: @node Concept Index,  , Word Index, Top
15110: @unnumbered Concept and Word Index
15111: 
15112: Not all entries listed in this index are present verbatim in the
15113: text. This index also duplicates, in abbreviated form, all of the words
15114: listed in the Word Index (only the names are listed for the words here).
15115: 
15116: @printindex cp
15117: 
15118: @contents
15119: @bye
15120: 
15121: 
15122: 

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