File:  [gforth] / gforth / Attic / gforth.ds
Revision 1.48: download - view: text, annotated - select for diffs
Thu Apr 10 15:00:06 1997 UTC (23 years, 9 months ago) by anton
Branches: MAIN
CVS tags: HEAD
bumped the version number to 0.3.1
gforth-makeimage now makes an executable file and uses $GFORTH
documentation changes
fixed bug involving locals and recurse

    1: \input texinfo   @c -*-texinfo-*-
    2: @comment The source is gforth.ds, from which gforth.texi is generated
    3: @comment %**start of header (This is for running Texinfo on a region.)
    4: @setfilename
    5: @settitle Gforth Manual
    6: @dircategory GNU programming tools
    7: @direntry
    8: * Gforth: (gforth).             A fast interpreter for the Forth language.
    9: @end direntry
   10: @comment @setchapternewpage odd
   11: @comment %**end of header (This is for running Texinfo on a region.)
   13: @ifinfo
   14: This file documents Gforth 0.3
   16: Copyright @copyright{} 1995-1997 Free Software Foundation, Inc.
   18:      Permission is granted to make and distribute verbatim copies of
   19:      this manual provided the copyright notice and this permission notice
   20:      are preserved on all copies.
   22: @ignore
   23:      Permission is granted to process this file through TeX and print the
   24:      results, provided the printed document carries a copying permission
   25:      notice identical to this one except for the removal of this paragraph
   26:      (this paragraph not being relevant to the printed manual).
   28: @end ignore
   29:      Permission is granted to copy and distribute modified versions of this
   30:      manual under the conditions for verbatim copying, provided also that the
   31:      sections entitled "Distribution" and "General Public License" are
   32:      included exactly as in the original, and provided that the entire
   33:      resulting derived work is distributed under the terms of a permission
   34:      notice identical to this one.
   36:      Permission is granted to copy and distribute translations of this manual
   37:      into another language, under the above conditions for modified versions,
   38:      except that the sections entitled "Distribution" and "General Public
   39:      License" may be included in a translation approved by the author instead
   40:      of in the original English.
   41: @end ifinfo
   43: @finalout
   44: @titlepage
   45: @sp 10
   46: @center @titlefont{Gforth Manual}
   47: @sp 2
   48: @center for version 0.3
   49: @sp 2
   50: @center Anton Ertl
   51: @center Bernd Paysan
   52: @sp 3
   53: @center This manual is under construction
   55: @comment  The following two commands start the copyright page.
   56: @page
   57: @vskip 0pt plus 1filll
   58: Copyright @copyright{} 1995--1997 Free Software Foundation, Inc.
   60: @comment !! Published by ... or You can get a copy of this manual ...
   62:      Permission is granted to make and distribute verbatim copies of
   63:      this manual provided the copyright notice and this permission notice
   64:      are preserved on all copies.
   66:      Permission is granted to copy and distribute modified versions of this
   67:      manual under the conditions for verbatim copying, provided also that the
   68:      sections entitled "Distribution" and "General Public License" are
   69:      included exactly as in the original, and provided that the entire
   70:      resulting derived work is distributed under the terms of a permission
   71:      notice identical to this one.
   73:      Permission is granted to copy and distribute translations of this manual
   74:      into another language, under the above conditions for modified versions,
   75:      except that the sections entitled "Distribution" and "General Public
   76:      License" may be included in a translation approved by the author instead
   77:      of in the original English.
   78: @end titlepage
   81: @node Top, License, (dir), (dir)
   82: @ifinfo
   83: Gforth is a free implementation of ANS Forth available on many
   84: personal machines. This manual corresponds to version 0.3.
   85: @end ifinfo
   87: @menu
   88: * License::                     
   89: * Goals::                       About the Gforth Project
   90: * Other Books::                 Things you might want to read
   91: * Invoking Gforth::             Starting Gforth
   92: * Words::                       Forth words available in Gforth
   93: * Tools::                       Programming tools
   94: * ANS conformance::             Implementation-defined options etc.
   95: * Model::                       The abstract machine of Gforth
   96: * Integrating Gforth::          Forth as scripting language for applications
   97: * Emacs and Gforth::            The Gforth Mode
   98: * Image Files::                 @code{.fi} files contain compiled code
   99: * Engine::                      The inner interpreter and the primitives
  100: * Bugs::                        How to report them
  101: * Origin::                      Authors and ancestors of Gforth
  102: * Word Index::                  An item for each Forth word
  103: * Concept Index::               A menu covering many topics
  104: @end menu
  106: @node License, Goals, Top, Top
  108: @center Version 2, June 1991
  110: @display
  111: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
  112: 675 Mass Ave, Cambridge, MA 02139, USA
  114: Everyone is permitted to copy and distribute verbatim copies
  115: of this license document, but changing it is not allowed.
  116: @end display
  118: @unnumberedsec Preamble
  120:   The licenses for most software are designed to take away your
  121: freedom to share and change it.  By contrast, the GNU General Public
  122: License is intended to guarantee your freedom to share and change free
  123: software---to make sure the software is free for all its users.  This
  124: General Public License applies to most of the Free Software
  125: Foundation's software and to any other program whose authors commit to
  126: using it.  (Some other Free Software Foundation software is covered by
  127: the GNU Library General Public License instead.)  You can apply it to
  128: your programs, too.
  130:   When we speak of free software, we are referring to freedom, not
  131: price.  Our General Public Licenses are designed to make sure that you
  132: have the freedom to distribute copies of free software (and charge for
  133: this service if you wish), that you receive source code or can get it
  134: if you want it, that you can change the software or use pieces of it
  135: in new free programs; and that you know you can do these things.
  137:   To protect your rights, we need to make restrictions that forbid
  138: anyone to deny you these rights or to ask you to surrender the rights.
  139: These restrictions translate to certain responsibilities for you if you
  140: distribute copies of the software, or if you modify it.
  142:   For example, if you distribute copies of such a program, whether
  143: gratis or for a fee, you must give the recipients all the rights that
  144: you have.  You must make sure that they, too, receive or can get the
  145: source code.  And you must show them these terms so they know their
  146: rights.
  148:   We protect your rights with two steps: (1) copyright the software, and
  149: (2) offer you this license which gives you legal permission to copy,
  150: distribute and/or modify the software.
  152:   Also, for each author's protection and ours, we want to make certain
  153: that everyone understands that there is no warranty for this free
  154: software.  If the software is modified by someone else and passed on, we
  155: want its recipients to know that what they have is not the original, so
  156: that any problems introduced by others will not reflect on the original
  157: authors' reputations.
  159:   Finally, any free program is threatened constantly by software
  160: patents.  We wish to avoid the danger that redistributors of a free
  161: program will individually obtain patent licenses, in effect making the
  162: program proprietary.  To prevent this, we have made it clear that any
  163: patent must be licensed for everyone's free use or not licensed at all.
  165:   The precise terms and conditions for copying, distribution and
  166: modification follow.
  168: @iftex
  170: @end iftex
  171: @ifinfo
  173: @end ifinfo
  175: @enumerate 0
  176: @item
  177: This License applies to any program or other work which contains
  178: a notice placed by the copyright holder saying it may be distributed
  179: under the terms of this General Public License.  The ``Program'', below,
  180: refers to any such program or work, and a ``work based on the Program''
  181: means either the Program or any derivative work under copyright law:
  182: that is to say, a work containing the Program or a portion of it,
  183: either verbatim or with modifications and/or translated into another
  184: language.  (Hereinafter, translation is included without limitation in
  185: the term ``modification''.)  Each licensee is addressed as ``you''.
  187: Activities other than copying, distribution and modification are not
  188: covered by this License; they are outside its scope.  The act of
  189: running the Program is not restricted, and the output from the Program
  190: is covered only if its contents constitute a work based on the
  191: Program (independent of having been made by running the Program).
  192: Whether that is true depends on what the Program does.
  194: @item
  195: You may copy and distribute verbatim copies of the Program's
  196: source code as you receive it, in any medium, provided that you
  197: conspicuously and appropriately publish on each copy an appropriate
  198: copyright notice and disclaimer of warranty; keep intact all the
  199: notices that refer to this License and to the absence of any warranty;
  200: and give any other recipients of the Program a copy of this License
  201: along with the Program.
  203: You may charge a fee for the physical act of transferring a copy, and
  204: you may at your option offer warranty protection in exchange for a fee.
  206: @item
  207: You may modify your copy or copies of the Program or any portion
  208: of it, thus forming a work based on the Program, and copy and
  209: distribute such modifications or work under the terms of Section 1
  210: above, provided that you also meet all of these conditions:
  212: @enumerate a
  213: @item
  214: You must cause the modified files to carry prominent notices
  215: stating that you changed the files and the date of any change.
  217: @item
  218: You must cause any work that you distribute or publish, that in
  219: whole or in part contains or is derived from the Program or any
  220: part thereof, to be licensed as a whole at no charge to all third
  221: parties under the terms of this License.
  223: @item
  224: If the modified program normally reads commands interactively
  225: when run, you must cause it, when started running for such
  226: interactive use in the most ordinary way, to print or display an
  227: announcement including an appropriate copyright notice and a
  228: notice that there is no warranty (or else, saying that you provide
  229: a warranty) and that users may redistribute the program under
  230: these conditions, and telling the user how to view a copy of this
  231: License.  (Exception: if the Program itself is interactive but
  232: does not normally print such an announcement, your work based on
  233: the Program is not required to print an announcement.)
  234: @end enumerate
  236: These requirements apply to the modified work as a whole.  If
  237: identifiable sections of that work are not derived from the Program,
  238: and can be reasonably considered independent and separate works in
  239: themselves, then this License, and its terms, do not apply to those
  240: sections when you distribute them as separate works.  But when you
  241: distribute the same sections as part of a whole which is a work based
  242: on the Program, the distribution of the whole must be on the terms of
  243: this License, whose permissions for other licensees extend to the
  244: entire whole, and thus to each and every part regardless of who wrote it.
  246: Thus, it is not the intent of this section to claim rights or contest
  247: your rights to work written entirely by you; rather, the intent is to
  248: exercise the right to control the distribution of derivative or
  249: collective works based on the Program.
  251: In addition, mere aggregation of another work not based on the Program
  252: with the Program (or with a work based on the Program) on a volume of
  253: a storage or distribution medium does not bring the other work under
  254: the scope of this License.
  256: @item
  257: You may copy and distribute the Program (or a work based on it,
  258: under Section 2) in object code or executable form under the terms of
  259: Sections 1 and 2 above provided that you also do one of the following:
  261: @enumerate a
  262: @item
  263: Accompany it with the complete corresponding machine-readable
  264: source code, which must be distributed under the terms of Sections
  265: 1 and 2 above on a medium customarily used for software interchange; or,
  267: @item
  268: Accompany it with a written offer, valid for at least three
  269: years, to give any third party, for a charge no more than your
  270: cost of physically performing source distribution, a complete
  271: machine-readable copy of the corresponding source code, to be
  272: distributed under the terms of Sections 1 and 2 above on a medium
  273: customarily used for software interchange; or,
  275: @item
  276: Accompany it with the information you received as to the offer
  277: to distribute corresponding source code.  (This alternative is
  278: allowed only for noncommercial distribution and only if you
  279: received the program in object code or executable form with such
  280: an offer, in accord with Subsection b above.)
  281: @end enumerate
  283: The source code for a work means the preferred form of the work for
  284: making modifications to it.  For an executable work, complete source
  285: code means all the source code for all modules it contains, plus any
  286: associated interface definition files, plus the scripts used to
  287: control compilation and installation of the executable.  However, as a
  288: special exception, the source code distributed need not include
  289: anything that is normally distributed (in either source or binary
  290: form) with the major components (compiler, kernel, and so on) of the
  291: operating system on which the executable runs, unless that component
  292: itself accompanies the executable.
  294: If distribution of executable or object code is made by offering
  295: access to copy from a designated place, then offering equivalent
  296: access to copy the source code from the same place counts as
  297: distribution of the source code, even though third parties are not
  298: compelled to copy the source along with the object code.
  300: @item
  301: You may not copy, modify, sublicense, or distribute the Program
  302: except as expressly provided under this License.  Any attempt
  303: otherwise to copy, modify, sublicense or distribute the Program is
  304: void, and will automatically terminate your rights under this License.
  305: However, parties who have received copies, or rights, from you under
  306: this License will not have their licenses terminated so long as such
  307: parties remain in full compliance.
  309: @item
  310: You are not required to accept this License, since you have not
  311: signed it.  However, nothing else grants you permission to modify or
  312: distribute the Program or its derivative works.  These actions are
  313: prohibited by law if you do not accept this License.  Therefore, by
  314: modifying or distributing the Program (or any work based on the
  315: Program), you indicate your acceptance of this License to do so, and
  316: all its terms and conditions for copying, distributing or modifying
  317: the Program or works based on it.
  319: @item
  320: Each time you redistribute the Program (or any work based on the
  321: Program), the recipient automatically receives a license from the
  322: original licensor to copy, distribute or modify the Program subject to
  323: these terms and conditions.  You may not impose any further
  324: restrictions on the recipients' exercise of the rights granted herein.
  325: You are not responsible for enforcing compliance by third parties to
  326: this License.
  328: @item
  329: If, as a consequence of a court judgment or allegation of patent
  330: infringement or for any other reason (not limited to patent issues),
  331: conditions are imposed on you (whether by court order, agreement or
  332: otherwise) that contradict the conditions of this License, they do not
  333: excuse you from the conditions of this License.  If you cannot
  334: distribute so as to satisfy simultaneously your obligations under this
  335: License and any other pertinent obligations, then as a consequence you
  336: may not distribute the Program at all.  For example, if a patent
  337: license would not permit royalty-free redistribution of the Program by
  338: all those who receive copies directly or indirectly through you, then
  339: the only way you could satisfy both it and this License would be to
  340: refrain entirely from distribution of the Program.
  342: If any portion of this section is held invalid or unenforceable under
  343: any particular circumstance, the balance of the section is intended to
  344: apply and the section as a whole is intended to apply in other
  345: circumstances.
  347: It is not the purpose of this section to induce you to infringe any
  348: patents or other property right claims or to contest validity of any
  349: such claims; this section has the sole purpose of protecting the
  350: integrity of the free software distribution system, which is
  351: implemented by public license practices.  Many people have made
  352: generous contributions to the wide range of software distributed
  353: through that system in reliance on consistent application of that
  354: system; it is up to the author/donor to decide if he or she is willing
  355: to distribute software through any other system and a licensee cannot
  356: impose that choice.
  358: This section is intended to make thoroughly clear what is believed to
  359: be a consequence of the rest of this License.
  361: @item
  362: If the distribution and/or use of the Program is restricted in
  363: certain countries either by patents or by copyrighted interfaces, the
  364: original copyright holder who places the Program under this License
  365: may add an explicit geographical distribution limitation excluding
  366: those countries, so that distribution is permitted only in or among
  367: countries not thus excluded.  In such case, this License incorporates
  368: the limitation as if written in the body of this License.
  370: @item
  371: The Free Software Foundation may publish revised and/or new versions
  372: of the General Public License from time to time.  Such new versions will
  373: be similar in spirit to the present version, but may differ in detail to
  374: address new problems or concerns.
  376: Each version is given a distinguishing version number.  If the Program
  377: specifies a version number of this License which applies to it and ``any
  378: later version'', you have the option of following the terms and conditions
  379: either of that version or of any later version published by the Free
  380: Software Foundation.  If the Program does not specify a version number of
  381: this License, you may choose any version ever published by the Free Software
  382: Foundation.
  384: @item
  385: If you wish to incorporate parts of the Program into other free
  386: programs whose distribution conditions are different, write to the author
  387: to ask for permission.  For software which is copyrighted by the Free
  388: Software Foundation, write to the Free Software Foundation; we sometimes
  389: make exceptions for this.  Our decision will be guided by the two goals
  390: of preserving the free status of all derivatives of our free software and
  391: of promoting the sharing and reuse of software generally.
  393: @iftex
  394: @heading NO WARRANTY
  395: @end iftex
  396: @ifinfo
  397: @center NO WARRANTY
  398: @end ifinfo
  400: @item
  411: @item
  421: @end enumerate
  423: @iftex
  425: @end iftex
  426: @ifinfo
  428: @end ifinfo
  430: @page
  431: @unnumberedsec How to Apply These Terms to Your New Programs
  433:   If you develop a new program, and you want it to be of the greatest
  434: possible use to the public, the best way to achieve this is to make it
  435: free software which everyone can redistribute and change under these terms.
  437:   To do so, attach the following notices to the program.  It is safest
  438: to attach them to the start of each source file to most effectively
  439: convey the exclusion of warranty; and each file should have at least
  440: the ``copyright'' line and a pointer to where the full notice is found.
  442: @smallexample
  443: @var{one line to give the program's name and a brief idea of what it does.}
  444: Copyright (C) 19@var{yy}  @var{name of author}
  446: This program is free software; you can redistribute it and/or modify 
  447: it under the terms of the GNU General Public License as published by 
  448: the Free Software Foundation; either version 2 of the License, or 
  449: (at your option) any later version.
  451: This program is distributed in the hope that it will be useful,
  452: but WITHOUT ANY WARRANTY; without even the implied warranty of
  454: GNU General Public License for more details.
  456: You should have received a copy of the GNU General Public License
  457: along with this program; if not, write to the Free Software
  458: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  459: @end smallexample
  461: Also add information on how to contact you by electronic and paper mail.
  463: If the program is interactive, make it output a short notice like this
  464: when it starts in an interactive mode:
  466: @smallexample
  467: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
  468: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
  469: type `show w'.  
  470: This is free software, and you are welcome to redistribute it 
  471: under certain conditions; type `show c' for details.
  472: @end smallexample
  474: The hypothetical commands @samp{show w} and @samp{show c} should show
  475: the appropriate parts of the General Public License.  Of course, the
  476: commands you use may be called something other than @samp{show w} and
  477: @samp{show c}; they could even be mouse-clicks or menu items---whatever
  478: suits your program.
  480: You should also get your employer (if you work as a programmer) or your
  481: school, if any, to sign a ``copyright disclaimer'' for the program, if
  482: necessary.  Here is a sample; alter the names:
  484: @smallexample
  485: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
  486: `Gnomovision' (which makes passes at compilers) written by James Hacker.
  488: @var{signature of Ty Coon}, 1 April 1989
  489: Ty Coon, President of Vice
  490: @end smallexample
  492: This General Public License does not permit incorporating your program into
  493: proprietary programs.  If your program is a subroutine library, you may
  494: consider it more useful to permit linking proprietary applications with the
  495: library.  If this is what you want to do, use the GNU Library General
  496: Public License instead of this License.
  498: @iftex
  499: @unnumbered Preface
  500: @cindex Preface
  501: This manual documents Gforth. The reader is expected to know
  502: Forth. This manual is primarily a reference manual. @xref{Other Books}
  503: for introductory material.
  504: @end iftex
  506: @node    Goals, Other Books, License, Top
  507: @comment node-name,     next,           previous, up
  508: @chapter Goals of Gforth
  509: @cindex Goals
  510: The goal of the Gforth Project is to develop a standard model for
  511: ANS Forth. This can be split into several subgoals:
  513: @itemize @bullet
  514: @item
  515: Gforth should conform to the Forth standard (ANS Forth).
  516: @item
  517: It should be a model, i.e. it should define all the
  518: implementation-dependent things.
  519: @item
  520: It should become standard, i.e. widely accepted and used. This goal
  521: is the most difficult one.
  522: @end itemize
  524: To achieve these goals Gforth should be
  525: @itemize @bullet
  526: @item
  527: Similar to previous models (fig-Forth, F83)
  528: @item
  529: Powerful. It should provide for all the things that are considered
  530: necessary today and even some that are not yet considered necessary.
  531: @item
  532: Efficient. It should not get the reputation of being exceptionally
  533: slow.
  534: @item
  535: Free.
  536: @item
  537: Available on many machines/easy to port.
  538: @end itemize
  540: Have we achieved these goals? Gforth conforms to the ANS Forth
  541: standard. It may be considered a model, but we have not yet documented
  542: which parts of the model are stable and which parts we are likely to
  543: change. It certainly has not yet become a de facto standard. It has some
  544: similarities and some differences to previous models. It has some
  545: powerful features, but not yet everything that we envisioned. We
  546: certainly have achieved our execution speed goals (@pxref{Performance}).
  547: It is free and available on many machines.
  549: @node Other Books, Invoking Gforth, Goals, Top
  550: @chapter Other books on ANS Forth
  551: @cindex books on Forth
  553: As the standard is relatively new, there are not many books out yet. It
  554: is not recommended to learn Forth by using Gforth and a book that is
  555: not written for ANS Forth, as you will not know your mistakes from the
  556: deviations of the book.
  558: @cindex standard document for ANS Forth
  559: @cindex ANS Forth document
  560: There is, of course, the standard, the definite reference if you want to
  561: write ANS Forth programs. It is available in printed form from the
  562: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
  563: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
  564: can also get it from Global Engineering Documents (Tel.: USA (800)
  565: 854-7179; Fax.: (303) 843-9880) for about $300.
  567: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
  568: for publication is available electronically and for free in some MS Word
  569: format, and it has been converted to HTML. Some pointers to these
  570: versions can be found through
  571: @*@url{}.
  573: @cindex introductory book
  574: @cindex book, introductory
  575: @cindex Woehr, Jack: @cite{Forth: The New Model}
  576: @cindex @cite{Forth: The new model} (book)
  577: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
  578: introductory book based on a draft version of the standard. It does not
  579: cover the whole standard. It also contains interesting background
  580: information (Jack Woehr was in the ANS Forth Technical Committee). It is
  581: not appropriate for complete newbies, but programmers experienced in
  582: other languages should find it ok.
  584: @node Invoking Gforth, Words, Other Books, Top
  585: @chapter Invoking Gforth
  586: @cindex invoking Gforth
  587: @cindex running Gforth
  588: @cindex command-line options
  589: @cindex options on the command line
  590: @cindex flags on the command line
  592: You will usually just say @code{gforth}. In many other cases the default
  593: Gforth image will be invoked like this:
  595: @example
  596: gforth [files] [-e forth-code]
  597: @end example
  599: executing the contents of the files and the Forth code in the order they
  600: are given.
  602: In general, the command line looks like this:
  604: @example
  605: gforth [initialization options] [image-specific options]
  606: @end example
  608: The initialization options must come before the rest of the command
  609: line. They are:
  611: @table @code
  612: @cindex -i, command-line option
  613: @cindex --image-file, command-line option
  614: @item --image-file @var{file}
  615: @itemx -i @var{file}
  616: Loads the Forth image @var{file} instead of the default
  617: @file{} (@pxref{Image Files}).
  619: @cindex --path, command-line option
  620: @cindex -p, command-line option
  621: @item --path @var{path}
  622: @itemx -p @var{path}
  623: Uses @var{path} for searching the image file and Forth source code files
  624: instead of the default in the environment variable @code{GFORTHPATH} or
  625: the path specified at installation time (e.g.,
  626: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
  627: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
  629: @cindex --dictionary-size, command-line option
  630: @cindex -m, command-line option
  631: @cindex @var{size} parameters for command-line options
  632: @cindex size of the dictionary and the stacks
  633: @item --dictionary-size @var{size}
  634: @itemx -m @var{size}
  635: Allocate @var{size} space for the Forth dictionary space instead of
  636: using the default specified in the image (typically 256K). The
  637: @var{size} specification consists of an integer and a unit (e.g.,
  638: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
  639: size, in this case Cells), @code{k} (kilobytes), and @code{M}
  640: (Megabytes). If no unit is specified, @code{e} is used.
  642: @cindex --data-stack-size, command-line option
  643: @cindex -d, command-line option
  644: @item --data-stack-size @var{size}
  645: @itemx -d @var{size}
  646: Allocate @var{size} space for the data stack instead of using the
  647: default specified in the image (typically 16K).
  649: @cindex --return-stack-size, command-line option
  650: @cindex -r, command-line option
  651: @item --return-stack-size @var{size}
  652: @itemx -r @var{size}
  653: Allocate @var{size} space for the return stack instead of using the
  654: default specified in the image (typically 15K).
  656: @cindex --fp-stack-size, command-line option
  657: @cindex -f, command-line option
  658: @item --fp-stack-size @var{size}
  659: @itemx -f @var{size}
  660: Allocate @var{size} space for the floating point stack instead of
  661: using the default specified in the image (typically 15.5K). In this case
  662: the unit specifier @code{e} refers to floating point numbers.
  664: @cindex --locals-stack-size, command-line option
  665: @cindex -l, command-line option
  666: @item --locals-stack-size @var{size}
  667: @itemx -l @var{size}
  668: Allocate @var{size} space for the locals stack instead of using the
  669: default specified in the image (typically 14.5K).
  671: @cindex -h, command-line option
  672: @cindex --help, command-line option
  673: @item --help
  674: @itemx -h
  675: Print a message about the command-line options
  677: @cindex -v, command-line option
  678: @cindex --version, command-line option
  679: @item --version
  680: @itemx -v
  681: Print version and exit
  683: @cindex --debug, command-line option
  684: @item --debug
  685: Print some information useful for debugging on startup.
  687: @cindex --offset-image, command-line option
  688: @item --offset-image
  689: Start the dictionary at a slightly different position than would be used
  690: otherwise (useful for creating data-relocatable images,
  691: @pxref{Data-Relocatable Image Files}).
  693: @cindex --clear-dictionary, command-line option
  694: @item --clear-dictionary
  695: Initialize all bytes in the dictionary to 0 before loading the image
  696: (@pxref{Data-Relocatable Image Files}).
  697: @end table
  699: @cindex loading files at startup
  700: @cindex executing code on startup
  701: @cindex batch processing with Gforth
  702: As explained above, the image-specific command-line arguments for the
  703: default image @file{} consist of a sequence of filenames and
  704: @code{-e @var{forth-code}} options that are interpreted in the sequence
  705: in which they are given. The @code{-e @var{forth-code}} or
  706: @code{--evaluate @var{forth-code}} option evaluates the forth
  707: code. This option takes only one argument; if you want to evaluate more
  708: Forth words, you have to quote them or use several @code{-e}s. To exit
  709: after processing the command line (instead of entering interactive mode)
  710: append @code{-e bye} to the command line.
  712: @cindex versions, invoking other versions of Gforth
  713: If you have several versions of Gforth installed, @code{gforth} will
  714: invoke the version that was installed last. @code{gforth-@var{version}}
  715: invokes a specific version. You may want to use the option
  716: @code{--path}, if your environment contains the variable
  717: @code{GFORTHPATH}.
  719: Not yet implemented:
  720: On startup the system first executes the system initialization file
  721: (unless the option @code{--no-init-file} is given; note that the system
  722: resulting from using this option may not be ANS Forth conformant). Then
  723: the user initialization file @file{.gforth.fs} is executed, unless the
  724: option @code{--no-rc} is given; this file is first searched in @file{.},
  725: then in @file{~}, then in the normal path (see above).
  727: @node Words, Tools, Invoking Gforth, Top
  728: @chapter Forth Words
  729: @cindex Words
  731: @menu
  732: * Notation::                    
  733: * Arithmetic::                  
  734: * Stack Manipulation::          
  735: * Memory::               
  736: * Control Structures::          
  737: * Locals::                      
  738: * Defining Words::              
  739: * Tokens for Words::            
  740: * Wordlists::                   
  741: * Files::                       
  742: * Blocks::                      
  743: * Other I/O::                   
  744: * Programming Tools::           
  745: * Assembler and Code words::    
  746: * Threading Words::             
  747: @end menu
  749: @node Notation, Arithmetic, Words, Words
  750: @section Notation
  751: @cindex notation of glossary entries
  752: @cindex format of glossary entries
  753: @cindex glossary notation format
  754: @cindex word glossary entry format
  756: The Forth words are described in this section in the glossary notation
  757: that has become a de-facto standard for Forth texts, i.e.,
  759: @format
  760: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
  761: @end format
  762: @var{Description}
  764: @table @var
  765: @item word
  766: @cindex case insensitivity
  767: The name of the word. BTW, Gforth is case insensitive, so you can
  768: type the words in in lower case (However, @pxref{core-idef}).
  770: @item Stack effect
  771: @cindex stack effect
  772: The stack effect is written in the notation @code{@var{before} --
  773: @var{after}}, where @var{before} and @var{after} describe the top of
  774: stack entries before and after the execution of the word. The rest of
  775: the stack is not touched by the word. The top of stack is rightmost,
  776: i.e., a stack sequence is written as it is typed in. Note that Gforth
  777: uses a separate floating point stack, but a unified stack
  778: notation. Also, return stack effects are not shown in @var{stack
  779: effect}, but in @var{Description}. The name of a stack item describes
  780: the type and/or the function of the item. See below for a discussion of
  781: the types.
  783: All words have two stack effects: A compile-time stack effect and a
  784: run-time stack effect. The compile-time stack-effect of most words is
  785: @var{ -- }. If the compile-time stack-effect of a word deviates from
  786: this standard behaviour, or the word does other unusual things at
  787: compile time, both stack effects are shown; otherwise only the run-time
  788: stack effect is shown.
  790: @cindex pronounciation of words
  791: @item pronunciation
  792: How the word is pronounced.
  794: @cindex wordset
  795: @item wordset
  796: The ANS Forth standard is divided into several wordsets. A standard
  797: system need not support all of them. So, the fewer wordsets your program
  798: uses the more portable it will be in theory. However, we suspect that
  799: most ANS Forth systems on personal machines will feature all
  800: wordsets. Words that are not defined in the ANS standard have
  801: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
  802: describes words that will work in future releases of Gforth;
  803: @code{gforth-internal} words are more volatile. Environmental query
  804: strings are also displayed like words; you can recognize them by the
  805: @code{environment} in the wordset field.
  807: @item Description
  808: A description of the behaviour of the word.
  809: @end table
  811: @cindex types of stack items
  812: @cindex stack item types
  813: The type of a stack item is specified by the character(s) the name
  814: starts with:
  816: @table @code
  817: @item f
  818: @cindex @code{f}, stack item type
  819: Boolean flags, i.e. @code{false} or @code{true}.
  820: @item c
  821: @cindex @code{c}, stack item type
  822: Char
  823: @item w
  824: @cindex @code{w}, stack item type
  825: Cell, can contain an integer or an address
  826: @item n
  827: @cindex @code{n}, stack item type
  828: signed integer
  829: @item u
  830: @cindex @code{u}, stack item type
  831: unsigned integer
  832: @item d
  833: @cindex @code{d}, stack item type
  834: double sized signed integer
  835: @item ud
  836: @cindex @code{ud}, stack item type
  837: double sized unsigned integer
  838: @item r
  839: @cindex @code{r}, stack item type
  840: Float (on the FP stack)
  841: @item a_
  842: @cindex @code{a_}, stack item type
  843: Cell-aligned address
  844: @item c_
  845: @cindex @code{c_}, stack item type
  846: Char-aligned address (note that a Char may have two bytes in Windows NT)
  847: @item f_
  848: @cindex @code{f_}, stack item type
  849: Float-aligned address
  850: @item df_
  851: @cindex @code{df_}, stack item type
  852: Address aligned for IEEE double precision float
  853: @item sf_
  854: @cindex @code{sf_}, stack item type
  855: Address aligned for IEEE single precision float
  856: @item xt
  857: @cindex @code{xt}, stack item type
  858: Execution token, same size as Cell
  859: @item wid
  860: @cindex @code{wid}, stack item type
  861: Wordlist ID, same size as Cell
  862: @item f83name
  863: @cindex @code{f83name}, stack item type
  864: Pointer to a name structure
  865: @item "
  866: @cindex @code{"}, stack item type
  867: string in the input stream (not the stack). The terminating character is
  868: a blank by default. If it is not a blank, it is shown in @code{<>}
  869: quotes.
  870: @end table
  872: @node Arithmetic, Stack Manipulation, Notation, Words
  873: @section Arithmetic
  874: @cindex arithmetic words
  876: @cindex division with potentially negative operands
  877: Forth arithmetic is not checked, i.e., you will not hear about integer
  878: overflow on addition or multiplication, you may hear about division by
  879: zero if you are lucky. The operator is written after the operands, but
  880: the operands are still in the original order. I.e., the infix @code{2-1}
  881: corresponds to @code{2 1 -}. Forth offers a variety of division
  882: operators. If you perform division with potentially negative operands,
  883: you do not want to use @code{/} or @code{/mod} with its undefined
  884: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
  885: former, @pxref{Mixed precision}).
  887: @menu
  888: * Single precision::            
  889: * Bitwise operations::          
  890: * Mixed precision::             operations with single and double-cell integers
  891: * Double precision::            Double-cell integer arithmetic
  892: * Floating Point::              
  893: @end menu
  895: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
  896: @subsection Single precision
  897: @cindex single precision arithmetic words
  899: doc-+
  900: doc--
  901: doc-*
  902: doc-/
  903: doc-mod
  904: doc-/mod
  905: doc-negate
  906: doc-abs
  907: doc-min
  908: doc-max
  910: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
  911: @subsection Bitwise operations
  912: @cindex bitwise operation words
  914: doc-and
  915: doc-or
  916: doc-xor
  917: doc-invert
  918: doc-2*
  919: doc-2/
  921: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
  922: @subsection Mixed precision
  923: @cindex mixed precision arithmetic words
  925: doc-m+
  926: doc-*/
  927: doc-*/mod
  928: doc-m*
  929: doc-um*
  930: doc-m*/
  931: doc-um/mod
  932: doc-fm/mod
  933: doc-sm/rem
  935: @node Double precision, Floating Point, Mixed precision, Arithmetic
  936: @subsection Double precision
  937: @cindex double precision arithmetic words
  939: @cindex double-cell numbers, input format
  940: @cindex input format for double-cell numbers
  941: The outer (aka text) interpreter converts numbers containing a dot into
  942: a double precision number. Note that only numbers with the dot as last
  943: character are standard-conforming.
  945: doc-d+
  946: doc-d-
  947: doc-dnegate
  948: doc-dabs
  949: doc-dmin
  950: doc-dmax
  952: @node Floating Point,  , Double precision, Arithmetic
  953: @subsection Floating Point
  954: @cindex floating point arithmetic words
  956: @cindex floating-point numbers, input format
  957: @cindex input format for floating-point numbers
  958: The format of floating point numbers recognized by the outer (aka text)
  959: interpreter is: a signed decimal number, possibly containing a decimal
  960: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
  961: by a signed integer (the exponent). E.g., @code{1e} is the same as
  962: @code{+1.0e+0}. Note that a number without @code{e}
  963: is not interpreted as floating-point number, but as double (if the
  964: number contains a @code{.}) or single precision integer. Also,
  965: conversions between string and floating point numbers always use base
  966: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
  967: value greater then 14, the @code{E} may be interpreted as digit and the
  968: number will be interpreted as integer, unless it has a signed exponent
  969: (both @code{+} and @code{-} are allowed as signs).
  971: @cindex angles in trigonometric operations
  972: @cindex trigonometric operations
  973: Angles in floating point operations are given in radians (a full circle
  974: has 2 pi radians). Note, that Gforth has a separate floating point
  975: stack, but we use the unified notation.
  977: @cindex floating-point arithmetic, pitfalls
  978: Floating point numbers have a number of unpleasant surprises for the
  979: unwary (e.g., floating point addition is not associative) and even a few
  980: for the wary. You should not use them unless you know what you are doing
  981: or you don't care that the results you get are totally bogus. If you
  982: want to learn about the problems of floating point numbers (and how to
  983: avoid them), you might start with @cite{David Goldberg, What Every
  984: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
  985: Computing Surveys 23(1):5@minus{}48, March 1991}.
  987: doc-f+
  988: doc-f-
  989: doc-f*
  990: doc-f/
  991: doc-fnegate
  992: doc-fabs
  993: doc-fmax
  994: doc-fmin
  995: doc-floor
  996: doc-fround
  997: doc-f**
  998: doc-fsqrt
  999: doc-fexp
 1000: doc-fexpm1
 1001: doc-fln
 1002: doc-flnp1
 1003: doc-flog
 1004: doc-falog
 1005: doc-fsin
 1006: doc-fcos
 1007: doc-fsincos
 1008: doc-ftan
 1009: doc-fasin
 1010: doc-facos
 1011: doc-fatan
 1012: doc-fatan2
 1013: doc-fsinh
 1014: doc-fcosh
 1015: doc-ftanh
 1016: doc-fasinh
 1017: doc-facosh
 1018: doc-fatanh
 1020: @node Stack Manipulation, Memory, Arithmetic, Words
 1021: @section Stack Manipulation
 1022: @cindex stack manipulation words
 1024: @cindex floating-point stack in the standard
 1025: Gforth has a data stack (aka parameter stack) for characters, cells,
 1026: addresses, and double cells, a floating point stack for floating point
 1027: numbers, a return stack for storing the return addresses of colon
 1028: definitions and other data, and a locals stack for storing local
 1029: variables. Note that while every sane Forth has a separate floating
 1030: point stack, this is not strictly required; an ANS Forth system could
 1031: theoretically keep floating point numbers on the data stack. As an
 1032: additional difficulty, you don't know how many cells a floating point
 1033: number takes. It is reportedly possible to write words in a way that
 1034: they work also for a unified stack model, but we do not recommend trying
 1035: it. Instead, just say that your program has an environmental dependency
 1036: on a separate FP stack.
 1038: @cindex return stack and locals
 1039: @cindex locals and return stack
 1040: Also, a Forth system is allowed to keep the local variables on the
 1041: return stack. This is reasonable, as local variables usually eliminate
 1042: the need to use the return stack explicitly. So, if you want to produce
 1043: a standard complying program and if you are using local variables in a
 1044: word, forget about return stack manipulations in that word (see the
 1045: standard document for the exact rules).
 1047: @menu
 1048: * Data stack::                  
 1049: * Floating point stack::        
 1050: * Return stack::                
 1051: * Locals stack::                
 1052: * Stack pointer manipulation::  
 1053: @end menu
 1055: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
 1056: @subsection Data stack
 1057: @cindex data stack manipulation words
 1058: @cindex stack manipulations words, data stack
 1060: doc-drop
 1061: doc-nip
 1062: doc-dup
 1063: doc-over
 1064: doc-tuck
 1065: doc-swap
 1066: doc-rot
 1067: doc--rot
 1068: doc-?dup
 1069: doc-pick
 1070: doc-roll
 1071: doc-2drop
 1072: doc-2nip
 1073: doc-2dup
 1074: doc-2over
 1075: doc-2tuck
 1076: doc-2swap
 1077: doc-2rot
 1079: @node Floating point stack, Return stack, Data stack, Stack Manipulation
 1080: @subsection Floating point stack
 1081: @cindex floating-point stack manipulation words
 1082: @cindex stack manipulation words, floating-point stack
 1084: doc-fdrop
 1085: doc-fnip
 1086: doc-fdup
 1087: doc-fover
 1088: doc-ftuck
 1089: doc-fswap
 1090: doc-frot
 1092: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
 1093: @subsection Return stack
 1094: @cindex return stack manipulation words
 1095: @cindex stack manipulation words, return stack
 1097: doc->r
 1098: doc-r>
 1099: doc-r@
 1100: doc-rdrop
 1101: doc-2>r
 1102: doc-2r>
 1103: doc-2r@
 1104: doc-2rdrop
 1106: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
 1107: @subsection Locals stack
 1109: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
 1110: @subsection Stack pointer manipulation
 1111: @cindex stack pointer manipulation words
 1113: doc-sp@
 1114: doc-sp!
 1115: doc-fp@
 1116: doc-fp!
 1117: doc-rp@
 1118: doc-rp!
 1119: doc-lp@
 1120: doc-lp!
 1122: @node Memory, Control Structures, Stack Manipulation, Words
 1123: @section Memory
 1124: @cindex Memory words
 1126: @menu
 1127: * Memory Access::      
 1128: * Address arithmetic::          
 1129: * Memory Blocks::         
 1130: @end menu
 1132: @node Memory Access, Address arithmetic, Memory, Memory
 1133: @subsection Memory Access
 1134: @cindex memory access words
 1136: doc-@
 1137: doc-!
 1138: doc-+!
 1139: doc-c@
 1140: doc-c!
 1141: doc-2@
 1142: doc-2!
 1143: doc-f@
 1144: doc-f!
 1145: doc-sf@
 1146: doc-sf!
 1147: doc-df@
 1148: doc-df!
 1150: @node Address arithmetic, Memory Blocks, Memory Access, Memory
 1151: @subsection Address arithmetic
 1152: @cindex address arithmetic words
 1154: ANS Forth does not specify the sizes of the data types. Instead, it
 1155: offers a number of words for computing sizes and doing address
 1156: arithmetic. Basically, address arithmetic is performed in terms of
 1157: address units (aus); on most systems the address unit is one byte. Note
 1158: that a character may have more than one au, so @code{chars} is no noop
 1159: (on systems where it is a noop, it compiles to nothing).
 1161: @cindex alignment of addresses for types
 1162: ANS Forth also defines words for aligning addresses for specific
 1163: types. Many computers require that accesses to specific data types
 1164: must only occur at specific addresses; e.g., that cells may only be
 1165: accessed at addresses divisible by 4. Even if a machine allows unaligned
 1166: accesses, it can usually perform aligned accesses faster. 
 1168: For the performance-conscious: alignment operations are usually only
 1169: necessary during the definition of a data structure, not during the
 1170: (more frequent) accesses to it.
 1172: ANS Forth defines no words for character-aligning addresses. This is not
 1173: an oversight, but reflects the fact that addresses that are not
 1174: char-aligned have no use in the standard and therefore will not be
 1175: created.
 1177: @cindex @code{CREATE} and alignment
 1178: The standard guarantees that addresses returned by @code{CREATE}d words
 1179: are cell-aligned; in addition, Gforth guarantees that these addresses
 1180: are aligned for all purposes.
 1182: Note that the standard defines a word @code{char}, which has nothing to
 1183: do with address arithmetic.
 1185: doc-chars
 1186: doc-char+
 1187: doc-cells
 1188: doc-cell+
 1189: doc-cell
 1190: doc-align
 1191: doc-aligned
 1192: doc-floats
 1193: doc-float+
 1194: doc-float
 1195: doc-falign
 1196: doc-faligned
 1197: doc-sfloats
 1198: doc-sfloat+
 1199: doc-sfalign
 1200: doc-sfaligned
 1201: doc-dfloats
 1202: doc-dfloat+
 1203: doc-dfalign
 1204: doc-dfaligned
 1205: doc-maxalign
 1206: doc-maxaligned
 1207: doc-cfalign
 1208: doc-cfaligned
 1209: doc-address-unit-bits
 1211: @node Memory Blocks,  , Address arithmetic, Memory
 1212: @subsection Memory Blocks
 1213: @cindex memory block words
 1215: doc-move
 1216: doc-erase
 1218: While the previous words work on address units, the rest works on
 1219: characters.
 1221: doc-cmove
 1222: doc-cmove>
 1223: doc-fill
 1224: doc-blank
 1226: @node Control Structures, Locals, Memory, Words
 1227: @section Control Structures
 1228: @cindex control structures
 1230: Control structures in Forth cannot be used in interpret state, only in
 1231: compile state@footnote{More precisely, they have no interpretation
 1232: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
 1233: a colon definition. We do not like this limitation, but have not seen a
 1234: satisfying way around it yet, although many schemes have been proposed.
 1236: @menu
 1237: * Selection::                   
 1238: * Simple Loops::                
 1239: * Counted Loops::               
 1240: * Arbitrary control structures::  
 1241: * Calls and returns::           
 1242: * Exception Handling::          
 1243: @end menu
 1245: @node Selection, Simple Loops, Control Structures, Control Structures
 1246: @subsection Selection
 1247: @cindex selection control structures
 1248: @cindex control structures for selection
 1250: @cindex @code{IF} control structure
 1251: @example
 1252: @var{flag}
 1253: IF
 1254:   @var{code}
 1255: ENDIF
 1256: @end example
 1257: or
 1258: @example
 1259: @var{flag}
 1260: IF
 1261:   @var{code1}
 1262: ELSE
 1263:   @var{code2}
 1264: ENDIF
 1265: @end example
 1267: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
 1268: standard, and @code{ENDIF} is not, although it is quite popular. We
 1269: recommend using @code{ENDIF}, because it is less confusing for people
 1270: who also know other languages (and is not prone to reinforcing negative
 1271: prejudices against Forth in these people). Adding @code{ENDIF} to a
 1272: system that only supplies @code{THEN} is simple:
 1273: @example
 1274: : endif   POSTPONE then ; immediate
 1275: @end example
 1277: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
 1278: (adv.)}  has the following meanings:
 1279: @quotation
 1280: ... 2b: following next after in order ... 3d: as a necessary consequence
 1281: (if you were there, then you saw them).
 1282: @end quotation
 1283: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
 1284: and many other programming languages has the meaning 3d.]
 1286: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
 1287: you can avoid using @code{?dup}. Using these alternatives is also more
 1288: efficient than using @code{?dup}. Definitions in plain standard Forth
 1289: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
 1290: @file{compat/control.fs}.
 1292: @cindex @code{CASE} control structure
 1293: @example
 1294: @var{n}
 1295: CASE
 1296:   @var{n1} OF @var{code1} ENDOF
 1297:   @var{n2} OF @var{code2} ENDOF
 1298:   @dots{}
 1299: ENDCASE
 1300: @end example
 1302: Executes the first @var{codei}, where the @var{ni} is equal to
 1303: @var{n}. A default case can be added by simply writing the code after
 1304: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
 1305: but must not consume it.
 1307: @node Simple Loops, Counted Loops, Selection, Control Structures
 1308: @subsection Simple Loops
 1309: @cindex simple loops
 1310: @cindex loops without count 
 1312: @cindex @code{WHILE} loop
 1313: @example
 1314: BEGIN
 1315:   @var{code1}
 1316:   @var{flag}
 1317: WHILE
 1318:   @var{code2}
 1319: REPEAT
 1320: @end example
 1322: @var{code1} is executed and @var{flag} is computed. If it is true,
 1323: @var{code2} is executed and the loop is restarted; If @var{flag} is
 1324: false, execution continues after the @code{REPEAT}.
 1326: @cindex @code{UNTIL} loop
 1327: @example
 1328: BEGIN
 1329:   @var{code}
 1330:   @var{flag}
 1331: UNTIL
 1332: @end example
 1334: @var{code} is executed. The loop is restarted if @code{flag} is false.
 1336: @cindex endless loop
 1337: @cindex loops, endless
 1338: @example
 1339: BEGIN
 1340:   @var{code}
 1341: AGAIN
 1342: @end example
 1344: This is an endless loop.
 1346: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
 1347: @subsection Counted Loops
 1348: @cindex counted loops
 1349: @cindex loops, counted
 1350: @cindex @code{DO} loops
 1352: The basic counted loop is:
 1353: @example
 1354: @var{limit} @var{start}
 1355: ?DO
 1356:   @var{body}
 1357: LOOP
 1358: @end example
 1360: This performs one iteration for every integer, starting from @var{start}
 1361: and up to, but excluding @var{limit}. The counter, aka index, can be
 1362: accessed with @code{i}. E.g., the loop
 1363: @example
 1364: 10 0 ?DO
 1365:   i .
 1366: LOOP
 1367: @end example
 1368: prints
 1369: @example
 1370: 0 1 2 3 4 5 6 7 8 9
 1371: @end example
 1372: The index of the innermost loop can be accessed with @code{i}, the index
 1373: of the next loop with @code{j}, and the index of the third loop with
 1374: @code{k}.
 1376: doc-i
 1377: doc-j
 1378: doc-k
 1380: The loop control data are kept on the return stack, so there are some
 1381: restrictions on mixing return stack accesses and counted loop
 1382: words. E.g., if you put values on the return stack outside the loop, you
 1383: cannot read them inside the loop. If you put values on the return stack
 1384: within a loop, you have to remove them before the end of the loop and
 1385: before accessing the index of the loop.
 1387: There are several variations on the counted loop:
 1389: @code{LEAVE} leaves the innermost counted loop immediately.
 1391: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
 1392: (and @code{LOOP} iterates until they become equal by wrap-around
 1393: arithmetic). This behaviour is usually not what you want. Therefore,
 1394: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
 1395: @code{?DO}), which do not enter the loop if @var{start} is greater than
 1396: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
 1397: unsigned loop parameters.
 1399: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
 1400: index by @var{n} instead of by 1. The loop is terminated when the border
 1401: between @var{limit-1} and @var{limit} is crossed. E.g.:
 1403: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
 1405: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
 1407: @cindex negative increment for counted loops
 1408: @cindex counted loops with negative increment
 1409: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
 1411: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
 1413: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
 1415: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
 1416: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
 1417: index by @var{u} each iteration. The loop is terminated when the border
 1418: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
 1419: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
 1421: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
 1423: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
 1425: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
 1427: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
 1428: @code{-LOOP} are not in the ANS Forth standard. However, an
 1429: implementation for these words that uses only standard words is provided
 1430: in @file{compat/loops.fs}.
 1432: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
 1433: the loop, independent of the loop parameters. Do not use @code{DO}, even
 1434: if you know that the loop is entered in any case. Such knowledge tends
 1435: to become invalid during maintenance of a program, and then the
 1436: @code{DO} will make trouble.
 1438: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
 1439: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
 1440: return stack so @code{EXIT} can get to its return address.
 1442: @cindex @code{FOR} loops
 1443: Another counted loop is
 1444: @example
 1445: @var{n}
 1446: FOR
 1447:   @var{body}
 1448: NEXT
 1449: @end example
 1450: This is the preferred loop of native code compiler writers who are too
 1451: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
 1452: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
 1453: and ending with 0. Other Forth systems may behave differently, even if
 1454: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
 1455: loops.
 1457: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
 1458: @subsection Arbitrary control structures
 1459: @cindex control structures, user-defined
 1461: @cindex control-flow stack
 1462: ANS Forth permits and supports using control structures in a non-nested
 1463: way. Information about incomplete control structures is stored on the
 1464: control-flow stack. This stack may be implemented on the Forth data
 1465: stack, and this is what we have done in Gforth.
 1467: @cindex @code{orig}, control-flow stack item
 1468: @cindex @code{dest}, control-flow stack item
 1469: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
 1470: entry represents a backward branch target. A few words are the basis for
 1471: building any control structure possible (except control structures that
 1472: need storage, like calls, coroutines, and backtracking).
 1474: doc-if
 1475: doc-ahead
 1476: doc-then
 1477: doc-begin
 1478: doc-until
 1479: doc-again
 1480: doc-cs-pick
 1481: doc-cs-roll
 1483: On many systems control-flow stack items take one word, in Gforth they
 1484: currently take three (this may change in the future). Therefore it is a
 1485: really good idea to manipulate the control flow stack with
 1486: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
 1487: words.
 1489: Some standard control structure words are built from these words:
 1491: doc-else
 1492: doc-while
 1493: doc-repeat
 1495: Gforth adds some more control-structure words:
 1497: doc-endif
 1498: doc-?dup-if
 1499: doc-?dup-0=-if
 1501: Counted loop words constitute a separate group of words:
 1503: doc-?do
 1504: doc-+do
 1505: doc-u+do
 1506: doc--do
 1507: doc-u-do
 1508: doc-do
 1509: doc-for
 1510: doc-loop
 1511: doc-+loop
 1512: doc--loop
 1513: doc-next
 1514: doc-leave
 1515: doc-?leave
 1516: doc-unloop
 1517: doc-done
 1519: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
 1520: @i{do-sys}. Our system allows it, but it's your job to ensure that for
 1521: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
 1522: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
 1523: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
 1524: resolved (by using one of the loop-ending words or @code{DONE}).
 1526: Another group of control structure words are
 1528: doc-case
 1529: doc-endcase
 1530: doc-of
 1531: doc-endof
 1533: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
 1534: @code{cs-roll}.
 1536: @subsubsection Programming Style
 1538: In order to ensure readability we recommend that you do not create
 1539: arbitrary control structures directly, but define new control structure
 1540: words for the control structure you want and use these words in your
 1541: program.
 1543: E.g., instead of writing
 1545: @example
 1546: begin
 1547:   ...
 1548: if [ 1 cs-roll ]
 1549:   ...
 1550: again then
 1551: @end example
 1553: we recommend defining control structure words, e.g.,
 1555: @example
 1556: : while ( dest -- orig dest )
 1557:  POSTPONE if
 1558:  1 cs-roll ; immediate
 1560: : repeat ( orig dest -- )
 1561:  POSTPONE again
 1562:  POSTPONE then ; immediate
 1563: @end example
 1565: and then using these to create the control structure:
 1567: @example
 1568: begin
 1569:   ...
 1570: while
 1571:   ...
 1572: repeat
 1573: @end example
 1575: That's much easier to read, isn't it? Of course, @code{REPEAT} and
 1576: @code{WHILE} are predefined, so in this example it would not be
 1577: necessary to define them.
 1579: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
 1580: @subsection Calls and returns
 1581: @cindex calling a definition
 1582: @cindex returning from a definition
 1584: A definition can be called simply be writing the name of the
 1585: definition. When the end of the definition is reached, it returns. An
 1586: earlier return can be forced using
 1588: doc-exit
 1590: Don't forget to clean up the return stack and @code{UNLOOP} any
 1591: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
 1592: primitive compiled by @code{EXIT} is
 1594: doc-;s
 1596: @node Exception Handling,  , Calls and returns, Control Structures
 1597: @subsection Exception Handling
 1598: @cindex Exceptions
 1600: doc-catch
 1601: doc-throw
 1603: @node Locals, Defining Words, Control Structures, Words
 1604: @section Locals
 1605: @cindex locals
 1607: Local variables can make Forth programming more enjoyable and Forth
 1608: programs easier to read. Unfortunately, the locals of ANS Forth are
 1609: laden with restrictions. Therefore, we provide not only the ANS Forth
 1610: locals wordset, but also our own, more powerful locals wordset (we
 1611: implemented the ANS Forth locals wordset through our locals wordset).
 1613: The ideas in this section have also been published in the paper
 1614: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
 1615: at EuroForth '94; it is available at
 1616: @*@url{}.
 1618: @menu
 1619: * Gforth locals::               
 1620: * ANS Forth locals::            
 1621: @end menu
 1623: @node Gforth locals, ANS Forth locals, Locals, Locals
 1624: @subsection Gforth locals
 1625: @cindex Gforth locals
 1626: @cindex locals, Gforth style
 1628: Locals can be defined with
 1630: @example
 1631: @{ local1 local2 ... -- comment @}
 1632: @end example
 1633: or
 1634: @example
 1635: @{ local1 local2 ... @}
 1636: @end example
 1638: E.g.,
 1639: @example
 1640: : max @{ n1 n2 -- n3 @}
 1641:  n1 n2 > if
 1642:    n1
 1643:  else
 1644:    n2
 1645:  endif ;
 1646: @end example
 1648: The similarity of locals definitions with stack comments is intended. A
 1649: locals definition often replaces the stack comment of a word. The order
 1650: of the locals corresponds to the order in a stack comment and everything
 1651: after the @code{--} is really a comment.
 1653: This similarity has one disadvantage: It is too easy to confuse locals
 1654: declarations with stack comments, causing bugs and making them hard to
 1655: find. However, this problem can be avoided by appropriate coding
 1656: conventions: Do not use both notations in the same program. If you do,
 1657: they should be distinguished using additional means, e.g. by position.
 1659: @cindex types of locals
 1660: @cindex locals types
 1661: The name of the local may be preceded by a type specifier, e.g.,
 1662: @code{F:} for a floating point value:
 1664: @example
 1665: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
 1666: \ complex multiplication
 1667:  Ar Br f* Ai Bi f* f-
 1668:  Ar Bi f* Ai Br f* f+ ;
 1669: @end example
 1671: @cindex flavours of locals
 1672: @cindex locals flavours
 1673: @cindex value-flavoured locals
 1674: @cindex variable-flavoured locals
 1675: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
 1676: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
 1677: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
 1678: with @code{W:}, @code{D:} etc.) produces its value and can be changed
 1679: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
 1680: produces its address (which becomes invalid when the variable's scope is
 1681: left). E.g., the standard word @code{emit} can be defined in terms of
 1682: @code{type} like this:
 1684: @example
 1685: : emit @{ C^ char* -- @}
 1686:     char* 1 type ;
 1687: @end example
 1689: @cindex default type of locals
 1690: @cindex locals, default type
 1691: A local without type specifier is a @code{W:} local. Both flavours of
 1692: locals are initialized with values from the data or FP stack.
 1694: Currently there is no way to define locals with user-defined data
 1695: structures, but we are working on it.
 1697: Gforth allows defining locals everywhere in a colon definition. This
 1698: poses the following questions:
 1700: @menu
 1701: * Where are locals visible by name?::  
 1702: * How long do locals live?::    
 1703: * Programming Style::           
 1704: * Implementation::              
 1705: @end menu
 1707: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
 1708: @subsubsection Where are locals visible by name?
 1709: @cindex locals visibility
 1710: @cindex visibility of locals
 1711: @cindex scope of locals
 1713: Basically, the answer is that locals are visible where you would expect
 1714: it in block-structured languages, and sometimes a little longer. If you
 1715: want to restrict the scope of a local, enclose its definition in
 1716: @code{SCOPE}...@code{ENDSCOPE}.
 1718: doc-scope
 1719: doc-endscope
 1721: These words behave like control structure words, so you can use them
 1722: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
 1723: arbitrary ways.
 1725: If you want a more exact answer to the visibility question, here's the
 1726: basic principle: A local is visible in all places that can only be
 1727: reached through the definition of the local@footnote{In compiler
 1728: construction terminology, all places dominated by the definition of the
 1729: local.}. In other words, it is not visible in places that can be reached
 1730: without going through the definition of the local. E.g., locals defined
 1731: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
 1732: defined in @code{BEGIN}...@code{UNTIL} are visible after the
 1733: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
 1735: The reasoning behind this solution is: We want to have the locals
 1736: visible as long as it is meaningful. The user can always make the
 1737: visibility shorter by using explicit scoping. In a place that can
 1738: only be reached through the definition of a local, the meaning of a
 1739: local name is clear. In other places it is not: How is the local
 1740: initialized at the control flow path that does not contain the
 1741: definition? Which local is meant, if the same name is defined twice in
 1742: two independent control flow paths?
 1744: This should be enough detail for nearly all users, so you can skip the
 1745: rest of this section. If you really must know all the gory details and
 1746: options, read on.
 1748: In order to implement this rule, the compiler has to know which places
 1749: are unreachable. It knows this automatically after @code{AHEAD},
 1750: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
 1751: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
 1752: compiler that the control flow never reaches that place. If
 1753: @code{UNREACHABLE} is not used where it could, the only consequence is
 1754: that the visibility of some locals is more limited than the rule above
 1755: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
 1756: lie to the compiler), buggy code will be produced.
 1758: doc-unreachable
 1760: Another problem with this rule is that at @code{BEGIN}, the compiler
 1761: does not know which locals will be visible on the incoming
 1762: back-edge. All problems discussed in the following are due to this
 1763: ignorance of the compiler (we discuss the problems using @code{BEGIN}
 1764: loops as examples; the discussion also applies to @code{?DO} and other
 1765: loops). Perhaps the most insidious example is:
 1766: @example
 1767: AHEAD
 1768: BEGIN
 1769:   x
 1770: [ 1 CS-ROLL ] THEN
 1771:   @{ x @}
 1772:   ...
 1773: UNTIL
 1774: @end example
 1776: This should be legal according to the visibility rule. The use of
 1777: @code{x} can only be reached through the definition; but that appears
 1778: textually below the use.
 1780: From this example it is clear that the visibility rules cannot be fully
 1781: implemented without major headaches. Our implementation treats common
 1782: cases as advertised and the exceptions are treated in a safe way: The
 1783: compiler makes a reasonable guess about the locals visible after a
 1784: @code{BEGIN}; if it is too pessimistic, the
 1785: user will get a spurious error about the local not being defined; if the
 1786: compiler is too optimistic, it will notice this later and issue a
 1787: warning. In the case above the compiler would complain about @code{x}
 1788: being undefined at its use. You can see from the obscure examples in
 1789: this section that it takes quite unusual control structures to get the
 1790: compiler into trouble, and even then it will often do fine.
 1792: If the @code{BEGIN} is reachable from above, the most optimistic guess
 1793: is that all locals visible before the @code{BEGIN} will also be
 1794: visible after the @code{BEGIN}. This guess is valid for all loops that
 1795: are entered only through the @code{BEGIN}, in particular, for normal
 1796: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
 1797: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
 1798: compiler. When the branch to the @code{BEGIN} is finally generated by
 1799: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
 1800: warns the user if it was too optimistic:
 1801: @example
 1802: IF
 1803:   @{ x @}
 1804: BEGIN
 1805:   \ x ? 
 1806: [ 1 cs-roll ] THEN
 1807:   ...
 1808: UNTIL
 1809: @end example
 1811: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
 1812: optimistically assumes that it lives until the @code{THEN}. It notices
 1813: this difference when it compiles the @code{UNTIL} and issues a
 1814: warning. The user can avoid the warning, and make sure that @code{x}
 1815: is not used in the wrong area by using explicit scoping:
 1816: @example
 1817: IF
 1818:   SCOPE
 1819:   @{ x @}
 1820:   ENDSCOPE
 1821: BEGIN
 1822: [ 1 cs-roll ] THEN
 1823:   ...
 1824: UNTIL
 1825: @end example
 1827: Since the guess is optimistic, there will be no spurious error messages
 1828: about undefined locals.
 1830: If the @code{BEGIN} is not reachable from above (e.g., after
 1831: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
 1832: optimistic guess, as the locals visible after the @code{BEGIN} may be
 1833: defined later. Therefore, the compiler assumes that no locals are
 1834: visible after the @code{BEGIN}. However, the user can use
 1835: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
 1836: visible at the BEGIN as at the point where the top control-flow stack
 1837: item was created.
 1839: doc-assume-live
 1841: E.g.,
 1842: @example
 1843: @{ x @}
 1844: AHEAD
 1846: BEGIN
 1847:   x
 1848: [ 1 CS-ROLL ] THEN
 1849:   ...
 1850: UNTIL
 1851: @end example
 1853: Other cases where the locals are defined before the @code{BEGIN} can be
 1854: handled by inserting an appropriate @code{CS-ROLL} before the
 1855: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
 1856: behind the @code{ASSUME-LIVE}).
 1858: Cases where locals are defined after the @code{BEGIN} (but should be
 1859: visible immediately after the @code{BEGIN}) can only be handled by
 1860: rearranging the loop. E.g., the ``most insidious'' example above can be
 1861: arranged into:
 1862: @example
 1863: BEGIN
 1864:   @{ x @}
 1865:   ... 0=
 1866: WHILE
 1867:   x
 1868: REPEAT
 1869: @end example
 1871: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
 1872: @subsubsection How long do locals live?
 1873: @cindex locals lifetime
 1874: @cindex lifetime of locals
 1876: The right answer for the lifetime question would be: A local lives at
 1877: least as long as it can be accessed. For a value-flavoured local this
 1878: means: until the end of its visibility. However, a variable-flavoured
 1879: local could be accessed through its address far beyond its visibility
 1880: scope. Ultimately, this would mean that such locals would have to be
 1881: garbage collected. Since this entails un-Forth-like implementation
 1882: complexities, I adopted the same cowardly solution as some other
 1883: languages (e.g., C): The local lives only as long as it is visible;
 1884: afterwards its address is invalid (and programs that access it
 1885: afterwards are erroneous).
 1887: @node Programming Style, Implementation, How long do locals live?, Gforth locals
 1888: @subsubsection Programming Style
 1889: @cindex locals programming style
 1890: @cindex programming style, locals
 1892: The freedom to define locals anywhere has the potential to change
 1893: programming styles dramatically. In particular, the need to use the
 1894: return stack for intermediate storage vanishes. Moreover, all stack
 1895: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
 1896: determined arguments) can be eliminated: If the stack items are in the
 1897: wrong order, just write a locals definition for all of them; then
 1898: write the items in the order you want.
 1900: This seems a little far-fetched and eliminating stack manipulations is
 1901: unlikely to become a conscious programming objective. Still, the number
 1902: of stack manipulations will be reduced dramatically if local variables
 1903: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
 1904: a traditional implementation of @code{max}).
 1906: This shows one potential benefit of locals: making Forth programs more
 1907: readable. Of course, this benefit will only be realized if the
 1908: programmers continue to honour the principle of factoring instead of
 1909: using the added latitude to make the words longer.
 1911: @cindex single-assignment style for locals
 1912: Using @code{TO} can and should be avoided.  Without @code{TO},
 1913: every value-flavoured local has only a single assignment and many
 1914: advantages of functional languages apply to Forth. I.e., programs are
 1915: easier to analyse, to optimize and to read: It is clear from the
 1916: definition what the local stands for, it does not turn into something
 1917: different later.
 1919: E.g., a definition using @code{TO} might look like this:
 1920: @example
 1921: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1922:  u1 u2 min 0
 1923:  ?do
 1924:    addr1 c@@ addr2 c@@ -
 1925:    ?dup-if
 1926:      unloop exit
 1927:    then
 1928:    addr1 char+ TO addr1
 1929:    addr2 char+ TO addr2
 1930:  loop
 1931:  u1 u2 - ;
 1932: @end example
 1933: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
 1934: every loop iteration. @code{strcmp} is a typical example of the
 1935: readability problems of using @code{TO}. When you start reading
 1936: @code{strcmp}, you think that @code{addr1} refers to the start of the
 1937: string. Only near the end of the loop you realize that it is something
 1938: else.
 1940: This can be avoided by defining two locals at the start of the loop that
 1941: are initialized with the right value for the current iteration.
 1942: @example
 1943: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1944:  addr1 addr2
 1945:  u1 u2 min 0 
 1946:  ?do @{ s1 s2 @}
 1947:    s1 c@@ s2 c@@ -
 1948:    ?dup-if
 1949:      unloop exit
 1950:    then
 1951:    s1 char+ s2 char+
 1952:  loop
 1953:  2drop
 1954:  u1 u2 - ;
 1955: @end example
 1956: Here it is clear from the start that @code{s1} has a different value
 1957: in every loop iteration.
 1959: @node Implementation,  , Programming Style, Gforth locals
 1960: @subsubsection Implementation
 1961: @cindex locals implementation
 1962: @cindex implementation of locals
 1964: @cindex locals stack
 1965: Gforth uses an extra locals stack. The most compelling reason for
 1966: this is that the return stack is not float-aligned; using an extra stack
 1967: also eliminates the problems and restrictions of using the return stack
 1968: as locals stack. Like the other stacks, the locals stack grows toward
 1969: lower addresses. A few primitives allow an efficient implementation:
 1971: doc-@local#
 1972: doc-f@local#
 1973: doc-laddr#
 1974: doc-lp+!#
 1975: doc-lp!
 1976: doc->l
 1977: doc-f>l
 1979: In addition to these primitives, some specializations of these
 1980: primitives for commonly occurring inline arguments are provided for
 1981: efficiency reasons, e.g., @code{@@local0} as specialization of
 1982: @code{@@local#} for the inline argument 0. The following compiling words
 1983: compile the right specialized version, or the general version, as
 1984: appropriate:
 1986: doc-compile-@local
 1987: doc-compile-f@local
 1988: doc-compile-lp+!
 1990: Combinations of conditional branches and @code{lp+!#} like
 1991: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
 1992: is taken) are provided for efficiency and correctness in loops.
 1994: A special area in the dictionary space is reserved for keeping the
 1995: local variable names. @code{@{} switches the dictionary pointer to this
 1996: area and @code{@}} switches it back and generates the locals
 1997: initializing code. @code{W:} etc.@ are normal defining words. This
 1998: special area is cleared at the start of every colon definition.
 2000: @cindex wordlist for defining locals
 2001: A special feature of Gforth's dictionary is used to implement the
 2002: definition of locals without type specifiers: every wordlist (aka
 2003: vocabulary) has its own methods for searching
 2004: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
 2005: with a special search method: When it is searched for a word, it
 2006: actually creates that word using @code{W:}. @code{@{} changes the search
 2007: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
 2008: and then the wordlist for defining locals without type specifiers.
 2010: The lifetime rules support a stack discipline within a colon
 2011: definition: The lifetime of a local is either nested with other locals
 2012: lifetimes or it does not overlap them.
 2014: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
 2015: pointer manipulation is generated. Between control structure words
 2016: locals definitions can push locals onto the locals stack. @code{AGAIN}
 2017: is the simplest of the other three control flow words. It has to
 2018: restore the locals stack depth of the corresponding @code{BEGIN}
 2019: before branching. The code looks like this:
 2020: @format
 2021: @code{lp+!#} current-locals-size @minus{} dest-locals-size
 2022: @code{branch} <begin>
 2023: @end format
 2025: @code{UNTIL} is a little more complicated: If it branches back, it
 2026: must adjust the stack just like @code{AGAIN}. But if it falls through,
 2027: the locals stack must not be changed. The compiler generates the
 2028: following code:
 2029: @format
 2030: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
 2031: @end format
 2032: The locals stack pointer is only adjusted if the branch is taken.
 2034: @code{THEN} can produce somewhat inefficient code:
 2035: @format
 2036: @code{lp+!#} current-locals-size @minus{} orig-locals-size
 2037: <orig target>:
 2038: @code{lp+!#} orig-locals-size @minus{} new-locals-size
 2039: @end format
 2040: The second @code{lp+!#} adjusts the locals stack pointer from the
 2041: level at the @var{orig} point to the level after the @code{THEN}. The
 2042: first @code{lp+!#} adjusts the locals stack pointer from the current
 2043: level to the level at the orig point, so the complete effect is an
 2044: adjustment from the current level to the right level after the
 2045: @code{THEN}.
 2047: @cindex locals information on the control-flow stack
 2048: @cindex control-flow stack items, locals information
 2049: In a conventional Forth implementation a dest control-flow stack entry
 2050: is just the target address and an orig entry is just the address to be
 2051: patched. Our locals implementation adds a wordlist to every orig or dest
 2052: item. It is the list of locals visible (or assumed visible) at the point
 2053: described by the entry. Our implementation also adds a tag to identify
 2054: the kind of entry, in particular to differentiate between live and dead
 2055: (reachable and unreachable) orig entries.
 2057: A few unusual operations have to be performed on locals wordlists:
 2059: doc-common-list
 2060: doc-sub-list?
 2061: doc-list-size
 2063: Several features of our locals wordlist implementation make these
 2064: operations easy to implement: The locals wordlists are organised as
 2065: linked lists; the tails of these lists are shared, if the lists
 2066: contain some of the same locals; and the address of a name is greater
 2067: than the address of the names behind it in the list.
 2069: Another important implementation detail is the variable
 2070: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
 2071: determine if they can be reached directly or only through the branch
 2072: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
 2073: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
 2074: definition, by @code{BEGIN} and usually by @code{THEN}.
 2076: Counted loops are similar to other loops in most respects, but
 2077: @code{LEAVE} requires special attention: It performs basically the same
 2078: service as @code{AHEAD}, but it does not create a control-flow stack
 2079: entry. Therefore the information has to be stored elsewhere;
 2080: traditionally, the information was stored in the target fields of the
 2081: branches created by the @code{LEAVE}s, by organizing these fields into a
 2082: linked list. Unfortunately, this clever trick does not provide enough
 2083: space for storing our extended control flow information. Therefore, we
 2084: introduce another stack, the leave stack. It contains the control-flow
 2085: stack entries for all unresolved @code{LEAVE}s.
 2087: Local names are kept until the end of the colon definition, even if
 2088: they are no longer visible in any control-flow path. In a few cases
 2089: this may lead to increased space needs for the locals name area, but
 2090: usually less than reclaiming this space would cost in code size.
 2093: @node ANS Forth locals,  , Gforth locals, Locals
 2094: @subsection ANS Forth locals
 2095: @cindex locals, ANS Forth style
 2097: The ANS Forth locals wordset does not define a syntax for locals, but
 2098: words that make it possible to define various syntaxes. One of the
 2099: possible syntaxes is a subset of the syntax we used in the Gforth locals
 2100: wordset, i.e.:
 2102: @example
 2103: @{ local1 local2 ... -- comment @}
 2104: @end example
 2105: or
 2106: @example
 2107: @{ local1 local2 ... @}
 2108: @end example
 2110: The order of the locals corresponds to the order in a stack comment. The
 2111: restrictions are:
 2113: @itemize @bullet
 2114: @item
 2115: Locals can only be cell-sized values (no type specifiers are allowed).
 2116: @item
 2117: Locals can be defined only outside control structures.
 2118: @item
 2119: Locals can interfere with explicit usage of the return stack. For the
 2120: exact (and long) rules, see the standard. If you don't use return stack
 2121: accessing words in a definition using locals, you will be all right. The
 2122: purpose of this rule is to make locals implementation on the return
 2123: stack easier.
 2124: @item
 2125: The whole definition must be in one line.
 2126: @end itemize
 2128: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
 2129: Defining Words}). I.e., they are initialized from the stack. Using their
 2130: name produces their value. Their value can be changed using @code{TO}.
 2132: Since this syntax is supported by Gforth directly, you need not do
 2133: anything to use it. If you want to port a program using this syntax to
 2134: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
 2135: syntax on the other system.
 2137: Note that a syntax shown in the standard, section A.13 looks
 2138: similar, but is quite different in having the order of locals
 2139: reversed. Beware!
 2141: The ANS Forth locals wordset itself consists of the following word
 2143: doc-(local)
 2145: The ANS Forth locals extension wordset defines a syntax, but it is so
 2146: awful that we strongly recommend not to use it. We have implemented this
 2147: syntax to make porting to Gforth easy, but do not document it here. The
 2148: problem with this syntax is that the locals are defined in an order
 2149: reversed with respect to the standard stack comment notation, making
 2150: programs harder to read, and easier to misread and miswrite. The only
 2151: merit of this syntax is that it is easy to implement using the ANS Forth
 2152: locals wordset.
 2154: @node Defining Words, Tokens for Words, Locals, Words
 2155: @section Defining Words
 2156: @cindex defining words
 2158: @menu
 2159: * Simple Defining Words::       
 2160: * Colon Definitions::           
 2161: * User-defined Defining Words::  
 2162: * Supplying names::             
 2163: * Interpretation and Compilation Semantics::  
 2164: @end menu
 2166: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
 2167: @subsection Simple Defining Words
 2168: @cindex simple defining words
 2169: @cindex defining words, simple
 2171: doc-constant
 2172: doc-2constant
 2173: doc-fconstant
 2174: doc-variable
 2175: doc-2variable
 2176: doc-fvariable
 2177: doc-create
 2178: doc-user
 2179: doc-value
 2180: doc-to
 2181: doc-defer
 2182: doc-is
 2184: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
 2185: @subsection Colon Definitions
 2186: @cindex colon definitions
 2188: @example
 2189: : name ( ... -- ... )
 2190:     word1 word2 word3 ;
 2191: @end example
 2193: creates a word called @code{name}, that, upon execution, executes
 2194: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
 2196: The explanation above is somewhat superficial. @xref{Interpretation and
 2197: Compilation Semantics} for an in-depth discussion of some of the issues
 2198: involved.
 2200: doc-:
 2201: doc-;
 2203: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
 2204: @subsection User-defined Defining Words
 2205: @cindex user-defined defining words
 2206: @cindex defining words, user-defined
 2208: You can create new defining words simply by wrapping defining-time code
 2209: around existing defining words and putting the sequence in a colon
 2210: definition.
 2212: @cindex @code{CREATE} ... @code{DOES>}
 2213: If you want the words defined with your defining words to behave
 2214: differently from words defined with standard defining words, you can
 2215: write your defining word like this:
 2217: @example
 2218: : def-word ( "name" -- )
 2219:     Create @var{code1}
 2220: DOES> ( ... -- ... )
 2221:     @var{code2} ;
 2223: def-word name
 2224: @end example
 2226: Technically, this fragment defines a defining word @code{def-word}, and
 2227: a word @code{name}; when you execute @code{name}, the address of the
 2228: body of @code{name} is put on the data stack and @var{code2} is executed
 2229: (the address of the body of @code{name} is the address @code{HERE}
 2230: returns immediately after the @code{CREATE}).
 2232: In other words, if you make the following definitions:
 2234: @example
 2235: : def-word1 ( "name" -- )
 2236:     Create @var{code1} ;
 2238: : action1 ( ... -- ... )
 2239:     @var{code2} ;
 2241: def-word name1
 2242: @end example
 2244: Using @code{name1 action1} is equivalent to using @code{name}.
 2246: E.g., you can implement @code{Constant} in this way:
 2248: @example
 2249: : constant ( w "name" -- )
 2250:     create ,
 2251: DOES> ( -- w )
 2252:     @@ ;
 2253: @end example
 2255: When you create a constant with @code{5 constant five}, first a new word
 2256: @code{five} is created, then the value 5 is laid down in the body of
 2257: @code{five} with @code{,}. When @code{five} is invoked, the address of
 2258: the body is put on the stack, and @code{@@} retrieves the value 5.
 2260: @cindex stack effect of @code{DOES>}-parts
 2261: @cindex @code{DOES>}-parts, stack effect
 2262: In the example above the stack comment after the @code{DOES>} specifies
 2263: the stack effect of the defined words, not the stack effect of the
 2264: following code (the following code expects the address of the body on
 2265: the top of stack, which is not reflected in the stack comment). This is
 2266: the convention that I use and recommend (it clashes a bit with using
 2267: locals declarations for stack effect specification, though).
 2269: @subsubsection Applications of @code{CREATE..DOES>}
 2270: @cindex @code{CREATE} ... @code{DOES>}, applications
 2272: You may wonder how to use this feature. Here are some usage patterns:
 2274: @cindex factoring similar colon definitions
 2275: When you see a sequence of code occurring several times, and you can
 2276: identify a meaning, you will factor it out as a colon definition. When
 2277: you see similar colon definitions, you can factor them using
 2278: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
 2279: that look very similar:
 2280: @example
 2281: : ori, ( reg-target reg-source n -- )
 2282:     0 asm-reg-reg-imm ;
 2283: : andi, ( reg-target reg-source n -- )
 2284:     1 asm-reg-reg-imm ;
 2285: @end example
 2287: This could be factored with:
 2288: @example
 2289: : reg-reg-imm ( op-code -- )
 2290:     create ,
 2291: DOES> ( reg-target reg-source n -- )
 2292:     @@ asm-reg-reg-imm ;
 2294: 0 reg-reg-imm ori,
 2295: 1 reg-reg-imm andi,
 2296: @end example
 2298: @cindex currying
 2299: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
 2300: supply a part of the parameters for a word (known as @dfn{currying} in
 2301: the functional language community). E.g., @code{+} needs two
 2302: parameters. Creating versions of @code{+} with one parameter fixed can
 2303: be done like this:
 2304: @example
 2305: : curry+ ( n1 -- )
 2306:     create ,
 2307: DOES> ( n2 -- n1+n2 )
 2308:     @@ + ;
 2310:  3 curry+ 3+
 2311: -2 curry+ 2-
 2312: @end example
 2314: @subsubsection The gory details of @code{CREATE..DOES>}
 2315: @cindex @code{CREATE} ... @code{DOES>}, details
 2317: doc-does>
 2319: @cindex @code{DOES>} in a separate definition
 2320: This means that you need not use @code{CREATE} and @code{DOES>} in the
 2321: same definition; E.g., you can put the @code{DOES>}-part in a separate
 2322: definition. This allows us to, e.g., select among different DOES>-parts:
 2323: @example
 2324: : does1 
 2325: DOES> ( ... -- ... )
 2326:     ... ;
 2328: : does2
 2329: DOES> ( ... -- ... )
 2330:     ... ;
 2332: : def-word ( ... -- ... )
 2333:     create ...
 2334:     IF
 2335:        does1
 2336:     ELSE
 2337:        does2
 2338:     ENDIF ;
 2339: @end example
 2341: @cindex @code{DOES>} in interpretation state
 2342: In a standard program you can apply a @code{DOES>}-part only if the last
 2343: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
 2344: will override the behaviour of the last word defined in any case. In a
 2345: standard program, you can use @code{DOES>} only in a colon
 2346: definition. In Gforth, you can also use it in interpretation state, in a
 2347: kind of one-shot mode:
 2348: @example
 2349: CREATE name ( ... -- ... )
 2350:   @var{initialization}
 2351: DOES>
 2352:   @var{code} ;
 2353: @end example
 2354: This is equivalent to the standard
 2355: @example
 2356: :noname
 2357: DOES>
 2358:     @var{code} ;
 2359: CREATE name EXECUTE ( ... -- ... )
 2360:     @var{initialization}
 2361: @end example
 2363: You can get the address of the body of a word with
 2365: doc->body
 2367: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
 2368: @subsection Supplying names for the defined words
 2369: @cindex names for defined words
 2370: @cindex defining words, name parameter
 2372: @cindex defining words, name given in a string
 2373: By default, defining words take the names for the defined words from the
 2374: input stream. Sometimes you want to supply the name from a string. You
 2375: can do this with
 2377: doc-nextname
 2379: E.g.,
 2381: @example
 2382: s" foo" nextname create
 2383: @end example
 2384: is equivalent to
 2385: @example
 2386: create foo
 2387: @end example
 2389: @cindex defining words without name
 2390: Sometimes you want to define a word without a name. You can do this with
 2392: doc-noname
 2394: @cindex execution token of last defined word
 2395: To make any use of the newly defined word, you need its execution
 2396: token. You can get it with
 2398: doc-lastxt
 2400: E.g., you can initialize a deferred word with an anonymous colon
 2401: definition:
 2402: @example
 2403: Defer deferred
 2404: noname : ( ... -- ... )
 2405:   ... ;
 2406: lastxt IS deferred
 2407: @end example
 2409: @code{lastxt} also works when the last word was not defined as
 2410: @code{noname}. 
 2412: The standard has also recognized the need for anonymous words and
 2413: provides
 2415: doc-:noname
 2417: This leaves the execution token for the word on the stack after the
 2418: closing @code{;}. You can rewrite the last example with @code{:noname}:
 2419: @example
 2420: Defer deferred
 2421: :noname ( ... -- ... )
 2422:   ... ;
 2423: IS deferred
 2424: @end example
 2426: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
 2427: @subsection Interpretation and Compilation Semantics
 2428: @cindex semantics, interpretation and compilation
 2430: @cindex interpretation semantics
 2431: The @dfn{interpretation semantics} of a word are what the text
 2432: interpreter does when it encounters the word in interpret state. It also
 2433: appears in some other contexts, e.g., the execution token returned by
 2434: @code{' @var{word}} identifies the interpretation semantics of
 2435: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
 2436: interpret-state text interpretation of @code{@var{word}}).
 2438: @cindex compilation semantics
 2439: The @dfn{compilation semantics} of a word are what the text interpreter
 2440: does when it encounters the word in compile state. It also appears in
 2441: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
 2442: standard terminology, ``appends to the current definition''.} the
 2443: compilation semantics of @var{word}.
 2445: @cindex execution semantics
 2446: The standard also talks about @dfn{execution semantics}. They are used
 2447: only for defining the interpretation and compilation semantics of many
 2448: words. By default, the interpretation semantics of a word are to
 2449: @code{execute} its execution semantics, and the compilation semantics of
 2450: a word are to @code{compile,} its execution semantics.@footnote{In
 2451: standard terminology: The default interpretation semantics are its
 2452: execution semantics; the default compilation semantics are to append its
 2453: execution semantics to the execution semantics of the current
 2454: definition.}
 2456: @cindex immediate words
 2457: You can change the compilation semantics into @code{execute}ing the
 2458: execution semantics with
 2460: doc-immediate
 2462: @cindex compile-only words
 2463: You can remove the interpretation semantics of a word with
 2465: doc-compile-only
 2466: doc-restrict
 2468: Note that ticking (@code{'}) compile-only words gives an error
 2469: (``Interpreting a compile-only word'').
 2471: Gforth also allows you to define words with arbitrary combinations of
 2472: interpretation and compilation semantics.
 2474: doc-interpret/compile:
 2476: This feature was introduced for implementing @code{TO} and @code{S"}. I
 2477: recommend that you do not define such words, as cute as they may be:
 2478: they make it hard to get at both parts of the word in some contexts.
 2479: E.g., assume you want to get an execution token for the compilation
 2480: part. Instead, define two words, one that embodies the interpretation
 2481: part, and one that embodies the compilation part.
 2483: There is, however, a potentially useful application of this feature:
 2484: Providing differing implementations for the default semantics. While
 2485: this introduces redundancy and is therefore usually a bad idea, a
 2486: performance improvement may be worth the trouble. E.g., consider the
 2487: word @code{foobar}:
 2489: @example
 2490: : foobar
 2491:     foo bar ;
 2492: @end example
 2494: Let us assume that @code{foobar} is called so frequently that the
 2495: calling overhead would take a significant amount of the run-time. We can
 2496: optimize it with @code{interpret/compile:}:
 2498: @example
 2499: :noname
 2500:    foo bar ;
 2501: :noname
 2502:    POSTPONE foo POSTPONE bar ;
 2503: interpret/compile: foobar
 2504: @end example
 2506: This definition has the same interpretation semantics and essentially
 2507: the same compilation semantics as the simple definition of
 2508: @code{foobar}, but the implementation of the compilation semantics is
 2509: more efficient with respect to run-time.
 2511: @cindex state-smart words are a bad idea
 2512: Some people try to use state-smart words to emulate the feature provided
 2513: by @code{interpret/compile:} (words are state-smart if they check
 2514: @code{STATE} during execution). E.g., they would try to code
 2515: @code{foobar} like this:
 2517: @example
 2518: : foobar
 2519:   STATE @@
 2520:   IF ( compilation state )
 2521:     POSTPONE foo POSTPONE bar
 2522:   ELSE
 2523:     foo bar
 2524:   ENDIF ; immediate
 2525: @end example
 2527: While this works if @code{foobar} is processed only by the text
 2528: interpreter, it does not work in other contexts (like @code{'} or
 2529: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
 2530: for a state-smart word, not for the interpretation semantics of the
 2531: original @code{foobar}; when you execute this execution token (directly
 2532: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
 2533: state, the result will not be what you expected (i.e., it will not
 2534: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
 2535: write them!
 2537: @cindex defining words with arbitrary semantics combinations
 2538: It is also possible to write defining words that define words with
 2539: arbitrary combinations of interpretation and compilation semantics (or,
 2540: preferably, arbitrary combinations of implementations of the default
 2541: semantics). In general, this looks like:
 2543: @example
 2544: : def-word
 2545:     create-interpret/compile
 2546:     @var{code1}
 2547: interpretation>
 2548:     @var{code2}
 2549: <interpretation
 2550: compilation>
 2551:     @var{code3}
 2552: <compilation ;
 2553: @end example
 2555: For a @var{word} defined with @code{def-word}, the interpretation
 2556: semantics are to push the address of the body of @var{word} and perform
 2557: @var{code2}, and the compilation semantics are to push the address of
 2558: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
 2559: can also be defined like this:
 2561: @example
 2562: : constant ( n "name" -- )
 2563:     create-interpret/compile
 2564:     ,
 2565: interpretation> ( -- n )
 2566:     @@
 2567: <interpretation
 2568: compilation> ( compilation. -- ; run-time. -- n )
 2569:     @@ postpone literal
 2570: <compilation ;
 2571: @end example
 2573: doc-create-interpret/compile
 2574: doc-interpretation>
 2575: doc-<interpretation
 2576: doc-compilation>
 2577: doc-<compilation
 2579: Note that words defined with @code{interpret/compile:} and
 2580: @code{create-interpret/compile} have an extended header structure that
 2581: differs from other words; however, unless you try to access them with
 2582: plain address arithmetic, you should not notice this. Words for
 2583: accessing the header structure usually know how to deal with this; e.g.,
 2584: @code{' word >body} also gives you the body of a word created with
 2585: @code{create-interpret/compile}.
 2587: @node Tokens for Words, Wordlists, Defining Words, Words
 2588: @section Tokens for Words
 2589: @cindex tokens for words
 2591: This chapter describes the creation and use of tokens that represent
 2592: words on the stack (and in data space).
 2594: Named words have interpretation and compilation semantics. Unnamed words
 2595: just have execution semantics.
 2597: @cindex execution token
 2598: An @dfn{execution token} represents the execution semantics of an
 2599: unnamed word. An execution token occupies one cell. As explained in
 2600: section @ref{Supplying names}, the execution token of the last words
 2601: defined can be produced with
 2603: short-lastxt
 2605: You can perform the semantics represented by an execution token with
 2606: doc-execute
 2607: You can compile the word with
 2608: doc-compile,
 2610: @cindex code field address
 2611: @cindex CFA
 2612: In Gforth, the abstract data type @emph{execution token} is implemented
 2613: as CFA (code field address).
 2615: The interpretation semantics of a named word are also represented by an
 2616: execution token. You can get it with
 2618: doc-[']
 2619: doc-'
 2621: For literals, you use @code{'} in interpreted code and @code{[']} in
 2622: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
 2623: by complaining about compile-only words. To get an execution token for a
 2624: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
 2625: @var{X} drop}.
 2627: @cindex compilation token
 2628: The compilation semantics are represented by a @dfn{compilation token}
 2629: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
 2630: execution token. The compilation semantics represented by the
 2631: compilation token can be performed with @code{execute}, which consumes
 2632: the whole compilation token, with an additional stack effect determined
 2633: by the represented compilation semantics.
 2635: doc-[comp']
 2636: doc-comp'
 2638: You can compile the compilation semantics with @code{postpone,}. I.e.,
 2639: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
 2640: @var{word}}.
 2642: doc-postpone,
 2644: At present, the @var{w} part of a compilation token is an execution
 2645: token, and the @var{xt} part represents either @code{execute} or
 2646: @code{compile,}. However, don't rely on that knowledge, unless necessary;
 2647: we may introduce unusual compilation tokens in the future (e.g.,
 2648: compilation tokens representing the compilation semantics of literals).
 2650: @cindex name token
 2651: @cindex name field address
 2652: @cindex NFA
 2653: Named words are also represented by the @dfn{name token}. The abstract
 2654: data type @emph{name token} is implemented as NFA (name field address).
 2656: doc-find-name
 2657: doc-name>int
 2658: doc-name?int
 2659: doc-name>comp
 2660: doc-name>string
 2662: @node Wordlists, Files, Tokens for Words, Words
 2663: @section Wordlists
 2665: @node Files, Blocks, Wordlists, Words
 2666: @section Files
 2668: @node Blocks, Other I/O, Files, Words
 2669: @section Blocks
 2671: @node Other I/O, Programming Tools, Blocks, Words
 2672: @section Other I/O
 2674: @node Programming Tools, Assembler and Code words, Other I/O, Words
 2675: @section Programming Tools
 2676: @cindex programming tools
 2678: @menu
 2679: * Debugging::                   Simple and quick.
 2680: * Assertions::                  Making your programs self-checking.
 2681: @end menu
 2683: @node Debugging, Assertions, Programming Tools, Programming Tools
 2684: @subsection Debugging
 2685: @cindex debugging
 2687: The simple debugging aids provided in @file{debugging.fs}
 2688: are meant to support a different style of debugging than the
 2689: tracing/stepping debuggers used in languages with long turn-around
 2690: times.
 2692: A much better (faster) way in fast-compiling languages is to add
 2693: printing code at well-selected places, let the program run, look at
 2694: the output, see where things went wrong, add more printing code, etc.,
 2695: until the bug is found.
 2697: The word @code{~~} is easy to insert. It just prints debugging
 2698: information (by default the source location and the stack contents). It
 2699: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
 2700: query-replace them with nothing). The deferred words
 2701: @code{printdebugdata} and @code{printdebugline} control the output of
 2702: @code{~~}. The default source location output format works well with
 2703: Emacs' compilation mode, so you can step through the program at the
 2704: source level using @kbd{C-x `} (the advantage over a stepping debugger
 2705: is that you can step in any direction and you know where the crash has
 2706: happened or where the strange data has occurred).
 2708: Note that the default actions clobber the contents of the pictured
 2709: numeric output string, so you should not use @code{~~}, e.g., between
 2710: @code{<#} and @code{#>}.
 2712: doc-~~
 2713: doc-printdebugdata
 2714: doc-printdebugline
 2716: @node Assertions,  , Debugging, Programming Tools
 2717: @subsection Assertions
 2718: @cindex assertions
 2720: It is a good idea to make your programs self-checking, in particular, if
 2721: you use an assumption (e.g., that a certain field of a data structure is
 2722: never zero) that may become wrong during maintenance. Gforth supports
 2723: assertions for this purpose. They are used like this:
 2725: @example
 2726: assert( @var{flag} )
 2727: @end example
 2729: The code between @code{assert(} and @code{)} should compute a flag, that
 2730: should be true if everything is alright and false otherwise. It should
 2731: not change anything else on the stack. The overall stack effect of the
 2732: assertion is @code{( -- )}. E.g.
 2734: @example
 2735: assert( 1 1 + 2 = ) \ what we learn in school
 2736: assert( dup 0<> ) \ assert that the top of stack is not zero
 2737: assert( false ) \ this code should not be reached
 2738: @end example
 2740: The need for assertions is different at different times. During
 2741: debugging, we want more checking, in production we sometimes care more
 2742: for speed. Therefore, assertions can be turned off, i.e., the assertion
 2743: becomes a comment. Depending on the importance of an assertion and the
 2744: time it takes to check it, you may want to turn off some assertions and
 2745: keep others turned on. Gforth provides several levels of assertions for
 2746: this purpose:
 2748: doc-assert0(
 2749: doc-assert1(
 2750: doc-assert2(
 2751: doc-assert3(
 2752: doc-assert(
 2753: doc-)
 2755: @code{Assert(} is the same as @code{assert1(}. The variable
 2756: @code{assert-level} specifies the highest assertions that are turned
 2757: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
 2758: @code{assert1(} assertions perform checking, while @code{assert2(} and
 2759: @code{assert3(} assertions are treated as comments.
 2761: Note that the @code{assert-level} is evaluated at compile-time, not at
 2762: run-time. I.e., you cannot turn assertions on or off at run-time, you
 2763: have to set the @code{assert-level} appropriately before compiling a
 2764: piece of code. You can compile several pieces of code at several
 2765: @code{assert-level}s (e.g., a trusted library at level 1 and newly
 2766: written code at level 3).
 2768: doc-assert-level
 2770: If an assertion fails, a message compatible with Emacs' compilation mode
 2771: is produced and the execution is aborted (currently with @code{ABORT"}.
 2772: If there is interest, we will introduce a special throw code. But if you
 2773: intend to @code{catch} a specific condition, using @code{throw} is
 2774: probably more appropriate than an assertion).
 2776: @node Assembler and Code words, Threading Words, Programming Tools, Words
 2777: @section Assembler and Code words
 2778: @cindex assembler
 2779: @cindex code words
 2781: Gforth provides some words for defining primitives (words written in
 2782: machine code), and for defining the the machine-code equivalent of
 2783: @code{DOES>}-based defining words. However, the machine-independent
 2784: nature of Gforth poses a few problems: First of all, Gforth runs on
 2785: several architectures, so it can provide no standard assembler. What's
 2786: worse is that the register allocation not only depends on the processor,
 2787: but also on the @code{gcc} version and options used.
 2789: The words that Gforth offers encapsulate some system dependences (e.g., the
 2790: header structure), so a system-independent assembler may be used in
 2791: Gforth. If you do not have an assembler, you can compile machine code
 2792: directly with @code{,} and @code{c,}.
 2794: doc-assembler
 2795: doc-code
 2796: doc-end-code
 2797: doc-;code
 2798: doc-flush-icache
 2800: If @code{flush-icache} does not work correctly, @code{code} words
 2801: etc. will not work (reliably), either.
 2803: These words are rarely used. Therefore they reside in @code{code.fs},
 2804: which is usually not loaded (except @code{flush-icache}, which is always
 2805: present). You can load them with @code{require code.fs}.
 2807: @cindex registers of the inner interpreter
 2808: In the assembly code you will want to refer to the inner interpreter's
 2809: registers (e.g., the data stack pointer) and you may want to use other
 2810: registers for temporary storage. Unfortunately, the register allocation
 2811: is installation-dependent.
 2813: The easiest solution is to use explicit register declarations
 2814: (@pxref{Explicit Reg Vars, , Variables in Specified Registers,,
 2815: GNU C Manual}) for all of the inner interpreter's registers: You have to
 2816: compile Gforth with @code{-DFORCE_REG} (configure option
 2817: @code{--enable-force-reg}) and the appropriate declarations must be
 2818: present in the @code{machine.h} file (see @code{mips.h} for an example;
 2819: you can find a full list of all declarable register symbols with
 2820: @code{grep register engine.c}). If you give explicit registers to all
 2821: variables that are declared at the beginning of @code{engine()}, you
 2822: should be able to use the other caller-saved registers for temporary
 2823: storage. Alternatively, you can use the @code{gcc} option
 2824: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
 2825: Generation Conventions,, GNU C Manual}) to reserve a register
 2826: (however, this restriction on register allocation may slow Gforth
 2827: significantly).
 2829: If this solution is not viable (e.g., because @code{gcc} does not allow
 2830: you to explicitly declare all the registers you need), you have to find
 2831: out by looking at the code where the inner interpreter's registers
 2832: reside and which registers can be used for temporary storage. You can
 2833: get an assembly listing of the engine's code with @code{make engine.s}.
 2835: In any case, it is good practice to abstract your assembly code from the
 2836: actual register allocation. E.g., if the data stack pointer resides in
 2837: register @code{$17}, create an alias for this register called @code{sp},
 2838: and use that in your assembly code.
 2840: @cindex code words, portable
 2841: Another option for implementing normal and defining words efficiently
 2842: is: adding the wanted functionality to the source of Gforth. For normal
 2843: words you just have to edit @file{primitives} (@pxref{Automatic
 2844: Generation}), defining words (equivalent to @code{;CODE} words, for fast
 2845: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
 2846: @file{prims2x.fs}, and possibly @file{cross.fs}.
 2849: @node Threading Words,  , Assembler and Code words, Words
 2850: @section Threading Words
 2851: @cindex threading words
 2853: @cindex code address
 2854: These words provide access to code addresses and other threading stuff
 2855: in Gforth (and, possibly, other interpretive Forths). It more or less
 2856: abstracts away the differences between direct and indirect threading
 2857: (and, for direct threading, the machine dependences). However, at
 2858: present this wordset is still incomplete. It is also pretty low-level;
 2859: some day it will hopefully be made unnecessary by an internals wordset
 2860: that abstracts implementation details away completely.
 2862: doc->code-address
 2863: doc->does-code
 2864: doc-code-address!
 2865: doc-does-code!
 2866: doc-does-handler!
 2867: doc-/does-handler
 2869: The code addresses produced by various defining words are produced by
 2870: the following words:
 2872: doc-docol:
 2873: doc-docon:
 2874: doc-dovar:
 2875: doc-douser:
 2876: doc-dodefer:
 2877: doc-dofield:
 2879: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
 2880: with @code{>DOES-CODE}. If the word was defined in that way, the value
 2881: returned is different from 0 and identifies the @code{DOES>} used by the
 2882: defining word.
 2884: @node Tools, ANS conformance, Words, Top
 2885: @chapter Tools
 2887: @menu
 2888: * ANS Report::                  Report the words used, sorted by wordset.
 2889: @end menu
 2891: See also @ref{Emacs and Gforth}.
 2893: @node ANS Report,  , Tools, Tools
 2894: @section @file{ans-report.fs}: Report the words used, sorted by wordset
 2895: @cindex @file{ans-report.fs}
 2896: @cindex report the words used in your program
 2897: @cindex words used in your program
 2899: If you want to label a Forth program as ANS Forth Program, you must
 2900: document which wordsets the program uses; for extension wordsets, it is
 2901: helpful to list the words the program requires from these wordsets
 2902: (because Forth systems are allowed to provide only some words of them).
 2904: The @file{ans-report.fs} tool makes it easy for you to determine which
 2905: words from which wordset and which non-ANS words your application
 2906: uses. You simply have to include @file{ans-report.fs} before loading the
 2907: program you want to check. After loading your program, you can get the
 2908: report with @code{print-ans-report}. A typical use is to run this as
 2909: batch job like this:
 2910: @example
 2911: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
 2912: @end example
 2914: The output looks like this (for @file{compat/control.fs}):
 2915: @example
 2916: The program uses the following words
 2917: from CORE :
 2918: : POSTPONE THEN ; immediate ?dup IF 0= 
 2919: from BLOCK-EXT :
 2920: \ 
 2921: from FILE :
 2922: ( 
 2923: @end example
 2925: @subsection Caveats
 2927: Note that @file{ans-report.fs} just checks which words are used, not whether
 2928: they are used in an ANS Forth conforming way!
 2930: Some words are defined in several wordsets in the
 2931: standard. @file{ans-report.fs} reports them for only one of the
 2932: wordsets, and not necessarily the one you expect. It depends on usage
 2933: which wordset is the right one to specify. E.g., if you only use the
 2934: compilation semantics of @code{S"}, it is a Core word; if you also use
 2935: its interpretation semantics, it is a File word.
 2937: @c ******************************************************************
 2938: @node ANS conformance, Model, Tools, Top
 2939: @chapter ANS conformance
 2940: @cindex ANS conformance of Gforth
 2942: To the best of our knowledge, Gforth is an
 2944: ANS Forth System
 2945: @itemize @bullet
 2946: @item providing the Core Extensions word set
 2947: @item providing the Block word set
 2948: @item providing the Block Extensions word set
 2949: @item providing the Double-Number word set
 2950: @item providing the Double-Number Extensions word set
 2951: @item providing the Exception word set
 2952: @item providing the Exception Extensions word set
 2953: @item providing the Facility word set
 2954: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
 2955: @item providing the File Access word set
 2956: @item providing the File Access Extensions word set
 2957: @item providing the Floating-Point word set
 2958: @item providing the Floating-Point Extensions word set
 2959: @item providing the Locals word set
 2960: @item providing the Locals Extensions word set
 2961: @item providing the Memory-Allocation word set
 2962: @item providing the Memory-Allocation Extensions word set (that one's easy)
 2963: @item providing the Programming-Tools word set
 2964: @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
 2965: @item providing the Search-Order word set
 2966: @item providing the Search-Order Extensions word set
 2967: @item providing the String word set
 2968: @item providing the String Extensions word set (another easy one)
 2969: @end itemize
 2971: @cindex system documentation
 2972: In addition, ANS Forth systems are required to document certain
 2973: implementation choices. This chapter tries to meet these
 2974: requirements. In many cases it gives a way to ask the system for the
 2975: information instead of providing the information directly, in
 2976: particular, if the information depends on the processor, the operating
 2977: system or the installation options chosen, or if they are likely to
 2978: change during the maintenance of Gforth.
 2980: @comment The framework for the rest has been taken from pfe.
 2982: @menu
 2983: * The Core Words::              
 2984: * The optional Block word set::  
 2985: * The optional Double Number word set::  
 2986: * The optional Exception word set::  
 2987: * The optional Facility word set::  
 2988: * The optional File-Access word set::  
 2989: * The optional Floating-Point word set::  
 2990: * The optional Locals word set::  
 2991: * The optional Memory-Allocation word set::  
 2992: * The optional Programming-Tools word set::  
 2993: * The optional Search-Order word set::  
 2994: @end menu
 2997: @c =====================================================================
 2998: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
 2999: @comment  node-name,  next,  previous,  up
 3000: @section The Core Words
 3001: @c =====================================================================
 3002: @cindex core words, system documentation
 3003: @cindex system documentation, core words
 3005: @menu
 3006: * core-idef::                   Implementation Defined Options                   
 3007: * core-ambcond::                Ambiguous Conditions                
 3008: * core-other::                  Other System Documentation                  
 3009: @end menu
 3011: @c ---------------------------------------------------------------------
 3012: @node core-idef, core-ambcond, The Core Words, The Core Words
 3013: @subsection Implementation Defined Options
 3014: @c ---------------------------------------------------------------------
 3015: @cindex core words, implementation-defined options
 3016: @cindex implementation-defined options, core words
 3019: @table @i
 3020: @item (Cell) aligned addresses:
 3021: @cindex cell-aligned addresses
 3022: @cindex aligned addresses
 3023: processor-dependent. Gforth's alignment words perform natural alignment
 3024: (e.g., an address aligned for a datum of size 8 is divisible by
 3025: 8). Unaligned accesses usually result in a @code{-23 THROW}.
 3027: @item @code{EMIT} and non-graphic characters:
 3028: @cindex @code{EMIT} and non-graphic characters
 3029: @cindex non-graphic characters and @code{EMIT}
 3030: The character is output using the C library function (actually, macro)
 3031: @code{putc}.
 3033: @item character editing of @code{ACCEPT} and @code{EXPECT}:
 3034: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
 3035: @cindex editing in @code{ACCEPT} and @code{EXPECT}
 3036: @cindex @code{ACCEPT}, editing
 3037: @cindex @code{EXPECT}, editing
 3038: This is modeled on the GNU readline library (@pxref{Readline
 3039: Interaction, , Command Line Editing, readline, The GNU Readline
 3040: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
 3041: producing a full word completion every time you type it (instead of
 3042: producing the common prefix of all completions).
 3044: @item character set:
 3045: @cindex character set
 3046: The character set of your computer and display device. Gforth is
 3047: 8-bit-clean (but some other component in your system may make trouble).
 3049: @item Character-aligned address requirements:
 3050: @cindex character-aligned address requirements
 3051: installation-dependent. Currently a character is represented by a C
 3052: @code{unsigned char}; in the future we might switch to @code{wchar_t}
 3053: (Comments on that requested).
 3055: @item character-set extensions and matching of names:
 3056: @cindex character-set extensions and matching of names
 3057: @cindex case sensitivity for name lookup
 3058: @cindex name lookup, case sensitivity
 3059: @cindex locale and case sensitivity
 3060: Any character except the ASCII NUL charcter can be used in a
 3061: name. Matching is case-insensitive (except in @code{TABLE}s). The
 3062: matching is performed using the C function @code{strncasecmp}, whose
 3063: function is probably influenced by the locale. E.g., the @code{C} locale
 3064: does not know about accents and umlauts, so they are matched
 3065: case-sensitively in that locale. For portability reasons it is best to
 3066: write programs such that they work in the @code{C} locale. Then one can
 3067: use libraries written by a Polish programmer (who might use words
 3068: containing ISO Latin-2 encoded characters) and by a French programmer
 3069: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
 3070: funny results for some of the words (which ones, depends on the font you
 3071: are using)). Also, the locale you prefer may not be available in other
 3072: operating systems. Hopefully, Unicode will solve these problems one day.
 3074: @item conditions under which control characters match a space delimiter:
 3075: @cindex space delimiters
 3076: @cindex control characters as delimiters
 3077: If @code{WORD} is called with the space character as a delimiter, all
 3078: white-space characters (as identified by the C macro @code{isspace()})
 3079: are delimiters. @code{PARSE}, on the other hand, treats space like other
 3080: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
 3081: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
 3082: interpreter (aka text interpreter) by default, treats all white-space
 3083: characters as delimiters.
 3085: @item format of the control flow stack:
 3086: @cindex control flow stack, format
 3087: The data stack is used as control flow stack. The size of a control flow
 3088: stack item in cells is given by the constant @code{cs-item-size}. At the
 3089: time of this writing, an item consists of a (pointer to a) locals list
 3090: (third), an address in the code (second), and a tag for identifying the
 3091: item (TOS). The following tags are used: @code{defstart},
 3092: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
 3093: @code{scopestart}.
 3095: @item conversion of digits > 35
 3096: @cindex digits > 35
 3097: The characters @code{[\]^_'} are the digits with the decimal value
 3098: 36@minus{}41. There is no way to input many of the larger digits.
 3100: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
 3101: @cindex @code{EXPECT}, display after end of input
 3102: @cindex @code{ACCEPT}, display after end of input
 3103: The cursor is moved to the end of the entered string. If the input is
 3104: terminated using the @kbd{Return} key, a space is typed.
 3106: @item exception abort sequence of @code{ABORT"}:
 3107: @cindex exception abort sequence of @code{ABORT"}
 3108: @cindex @code{ABORT"}, exception abort sequence
 3109: The error string is stored into the variable @code{"error} and a
 3110: @code{-2 throw} is performed.
 3112: @item input line terminator:
 3113: @cindex input line terminator
 3114: @cindex line terminator on input
 3115: @cindex newline charcter on input
 3116: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
 3117: lines. One of these characters is typically produced when you type the
 3118: @kbd{Enter} or @kbd{Return} key.
 3120: @item maximum size of a counted string:
 3121: @cindex maximum size of a counted string
 3122: @cindex counted string, maximum size
 3123: @code{s" /counted-string" environment? drop .}. Currently 255 characters
 3124: on all ports, but this may change.
 3126: @item maximum size of a parsed string:
 3127: @cindex maximum size of a parsed string
 3128: @cindex parsed string, maximum size
 3129: Given by the constant @code{/line}. Currently 255 characters.
 3131: @item maximum size of a definition name, in characters:
 3132: @cindex maximum size of a definition name, in characters
 3133: @cindex name, maximum length
 3134: 31
 3136: @item maximum string length for @code{ENVIRONMENT?}, in characters:
 3137: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
 3138: @cindex @code{ENVIRONMENT?} string length, maximum
 3139: 31
 3141: @item method of selecting the user input device:
 3142: @cindex user input device, method of selecting
 3143: The user input device is the standard input. There is currently no way to
 3144: change it from within Gforth. However, the input can typically be
 3145: redirected in the command line that starts Gforth.
 3147: @item method of selecting the user output device:
 3148: @cindex user output device, method of selecting
 3149: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
 3150: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
 3151: output, so output on a terminal does not become visible before the next
 3152: newline or buffer overflow. Output on non-terminals is invisible until
 3153: the buffer overflows.
 3155: @item methods of dictionary compilation:
 3156: What are we expected to document here?
 3158: @item number of bits in one address unit:
 3159: @cindex number of bits in one address unit
 3160: @cindex address unit, size in bits
 3161: @code{s" address-units-bits" environment? drop .}. 8 in all current
 3162: ports.
 3164: @item number representation and arithmetic:
 3165: @cindex number representation and arithmetic
 3166: Processor-dependent. Binary two's complement on all current ports.
 3168: @item ranges for integer types:
 3169: @cindex ranges for integer types
 3170: @cindex integer types, ranges
 3171: Installation-dependent. Make environmental queries for @code{MAX-N},
 3172: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
 3173: unsigned (and positive) types is 0. The lower bound for signed types on
 3174: two's complement and one's complement machines machines can be computed
 3175: by adding 1 to the upper bound.
 3177: @item read-only data space regions:
 3178: @cindex read-only data space regions
 3179: @cindex data-space, read-only regions
 3180: The whole Forth data space is writable.
 3182: @item size of buffer at @code{WORD}:
 3183: @cindex size of buffer at @code{WORD}
 3184: @cindex @code{WORD} buffer size
 3185: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 3186: shared with the pictured numeric output string. If overwriting
 3187: @code{PAD} is acceptable, it is as large as the remaining dictionary
 3188: space, although only as much can be sensibly used as fits in a counted
 3189: string.
 3191: @item size of one cell in address units:
 3192: @cindex cell size
 3193: @code{1 cells .}.
 3195: @item size of one character in address units:
 3196: @cindex char size
 3197: @code{1 chars .}. 1 on all current ports.
 3199: @item size of the keyboard terminal buffer:
 3200: @cindex size of the keyboard terminal buffer
 3201: @cindex terminal buffer, size
 3202: Varies. You can determine the size at a specific time using @code{lp@@
 3203: tib - .}. It is shared with the locals stack and TIBs of files that
 3204: include the current file. You can change the amount of space for TIBs
 3205: and locals stack at Gforth startup with the command line option
 3206: @code{-l}.
 3208: @item size of the pictured numeric output buffer:
 3209: @cindex size of the pictured numeric output buffer
 3210: @cindex pictured numeric output buffer, size
 3211: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 3212: shared with @code{WORD}.
 3214: @item size of the scratch area returned by @code{PAD}:
 3215: @cindex size of the scratch area returned by @code{PAD}
 3216: @cindex @code{PAD} size
 3217: The remainder of dictionary space. @code{unused pad here - - .}.
 3219: @item system case-sensitivity characteristics:
 3220: @cindex case-sensitivity characteristics
 3221: Dictionary searches are case insensitive (except in
 3222: @code{TABLE}s). However, as explained above under @i{character-set
 3223: extensions}, the matching for non-ASCII characters is determined by the
 3224: locale you are using. In the default @code{C} locale all non-ASCII
 3225: characters are matched case-sensitively.
 3227: @item system prompt:
 3228: @cindex system prompt
 3229: @cindex prompt
 3230: @code{ ok} in interpret state, @code{ compiled} in compile state.
 3232: @item division rounding:
 3233: @cindex division rounding
 3234: installation dependent. @code{s" floored" environment? drop .}. We leave
 3235: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
 3236: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
 3238: @item values of @code{STATE} when true:
 3239: @cindex @code{STATE} values
 3240: -1.
 3242: @item values returned after arithmetic overflow:
 3243: On two's complement machines, arithmetic is performed modulo
 3244: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 3245: arithmetic (with appropriate mapping for signed types). Division by zero
 3246: typically results in a @code{-55 throw} (Floating-point unidentified
 3247: fault), although a @code{-10 throw} (divide by zero) would be more
 3248: appropriate.
 3250: @item whether the current definition can be found after @t{DOES>}:
 3251: @cindex @t{DOES>}, visibility of current definition
 3252: No.
 3254: @end table
 3256: @c ---------------------------------------------------------------------
 3257: @node core-ambcond, core-other, core-idef, The Core Words
 3258: @subsection Ambiguous conditions
 3259: @c ---------------------------------------------------------------------
 3260: @cindex core words, ambiguous conditions
 3261: @cindex ambiguous conditions, core words
 3263: @table @i
 3265: @item a name is neither a word nor a number:
 3266: @cindex name not found
 3267: @cindex Undefined word
 3268: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
 3269: preserves the data and FP stack, so you don't lose more work than
 3270: necessary.
 3272: @item a definition name exceeds the maximum length allowed:
 3273: @cindex Word name too long
 3274: @code{-19 throw} (Word name too long)
 3276: @item addressing a region not inside the various data spaces of the forth system:
 3277: @cindex Invalid memory address
 3278: The stacks, code space and name space are accessible. Machine code space is
 3279: typically readable. Accessing other addresses gives results dependent on
 3280: the operating system. On decent systems: @code{-9 throw} (Invalid memory
 3281: address).
 3283: @item argument type incompatible with parameter:
 3284: @cindex Argument type mismatch
 3285: This is usually not caught. Some words perform checks, e.g., the control
 3286: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
 3287: mismatch).
 3289: @item attempting to obtain the execution token of a word with undefined execution semantics:
 3290: @cindex Interpreting a compile-only word, for @code{'} etc.
 3291: @cindex execution token of words with undefined execution semantics
 3292: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
 3293: get an execution token for @code{compile-only-error} (which performs a
 3294: @code{-14 throw} when executed).
 3296: @item dividing by zero:
 3297: @cindex dividing by zero
 3298: @cindex floating point unidentified fault, integer division
 3299: @cindex divide by zero
 3300: typically results in a @code{-55 throw} (floating point unidentified
 3301: fault), although a @code{-10 throw} (divide by zero) would be more
 3302: appropriate.
 3304: @item insufficient data stack or return stack space:
 3305: @cindex insufficient data stack or return stack space
 3306: @cindex stack overflow
 3307: @cindex Address alignment exception, stack overflow
 3308: @cindex Invalid memory address, stack overflow
 3309: Depending on the operating system, the installation, and the invocation
 3310: of Gforth, this is either checked by the memory management hardware, or
 3311: it is not checked. If it is checked, you typically get a @code{-9 throw}
 3312: (Invalid memory address) as soon as the overflow happens. If it is not
 3313: check, overflows typically result in mysterious illegal memory accesses,
 3314: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
 3315: (Address alignment exception); they might also destroy the internal data
 3316: structure of @code{ALLOCATE} and friends, resulting in various errors in
 3317: these words.
 3319: @item insufficient space for loop control parameters:
 3320: @cindex insufficient space for loop control parameters
 3321: like other return stack overflows.
 3323: @item insufficient space in the dictionary:
 3324: @cindex insufficient space in the dictionary
 3325: @cindex dictionary overflow
 3326: Depending on the operating system, the installation, and the invocation
 3327: of Gforth, this is either checked by the memory management hardware, or
 3328: it is not checked. Similar results as stack overflows. However,
 3329: typically the error appears at a different place when one inserts or
 3330: removes code. Also, the @code{THROW} does not relieve the situation (it
 3331: does for stack overflows).
 3333: @item interpreting a word with undefined interpretation semantics:
 3334: @cindex interpreting a word with undefined interpretation semantics
 3335: @cindex Interpreting a compile-only word
 3336: For some words, we have defined interpretation semantics. For the
 3337: others: @code{-14 throw} (Interpreting a compile-only word).
 3339: @item modifying the contents of the input buffer or a string literal:
 3340: @cindex modifying the contents of the input buffer or a string literal
 3341: These are located in writable memory and can be modified.
 3343: @item overflow of the pictured numeric output string:
 3344: @cindex overflow of the pictured numeric output string
 3345: @cindex pictured numeric output string, overflow
 3346: Not checked. Runs into the dictionary and destroys it (at least,
 3347: partially).
 3349: @item parsed string overflow:
 3350: @cindex parsed string overflow
 3351: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
 3353: @item producing a result out of range:
 3354: @cindex result out of range
 3355: On two's complement machines, arithmetic is performed modulo
 3356: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 3357: arithmetic (with appropriate mapping for signed types). Division by zero
 3358: typically results in a @code{-55 throw} (floatingpoint unidentified
 3359: fault), although a @code{-10 throw} (divide by zero) would be more
 3360: appropriate. @code{convert} and @code{>number} currently overflow
 3361: silently.
 3363: @item reading from an empty data or return stack:
 3364: @cindex stack empty
 3365: @cindex stack underflow
 3366: The data stack is checked by the outer (aka text) interpreter after
 3367: every word executed. If it has underflowed, a @code{-4 throw} (Stack
 3368: underflow) is performed. Apart from that, stacks may be checked or not,
 3369: depending on operating system, installation, and invocation. The
 3370: consequences of stack underflows are similar to the consequences of
 3371: stack overflows. Note that even if the system uses checking (through the
 3372: MMU), your program may have to underflow by a significant number of
 3373: stack items to trigger the reaction (the reason for this is that the
 3374: MMU, and therefore the checking, works with a page-size granularity).
 3376: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
 3377: @cindex unexpected end of the input buffer
 3378: @cindex zero-length string as a name
 3379: @cindex Attempt to use zero-length string as a name
 3380: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
 3381: use zero-length string as a name). Words like @code{'} probably will not
 3382: find what they search. Note that it is possible to create zero-length
 3383: names with @code{nextname} (should it not?).
 3385: @item @code{>IN} greater than input buffer:
 3386: @cindex @code{>IN} greater than input buffer
 3387: The next invocation of a parsing word returns a string with length 0.
 3389: @item @code{RECURSE} appears after @code{DOES>}:
 3390: @cindex @code{RECURSE} appears after @code{DOES>}
 3391: Compiles a recursive call to the defining word, not to the defined word.
 3393: @item argument input source different than current input source for @code{RESTORE-INPUT}:
 3394: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
 3395: @cindex Argument type mismatch, @code{RESTORE-INPUT}
 3396: @cindex @code{RESTORE-INPUT}, Argument type mismatch
 3397: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
 3398: the end of the file was reached), its source-id may be
 3399: reused. Therefore, restoring an input source specification referencing a
 3400: closed file may lead to unpredictable results instead of a @code{-12
 3401: THROW}.
 3403: In the future, Gforth may be able to restore input source specifications
 3404: from other than the current input source.
 3406: @item data space containing definitions gets de-allocated:
 3407: @cindex data space containing definitions gets de-allocated
 3408: Deallocation with @code{allot} is not checked. This typically results in
 3409: memory access faults or execution of illegal instructions.
 3411: @item data space read/write with incorrect alignment:
 3412: @cindex data space read/write with incorrect alignment
 3413: @cindex alignment faults
 3414: @cindex Address alignment exception
 3415: Processor-dependent. Typically results in a @code{-23 throw} (Address
 3416: alignment exception). Under Linux on a 486 or later processor with
 3417: alignment turned on, incorrect alignment results in a @code{-9 throw}
 3418: (Invalid memory address). There are reportedly some processors with
 3419: alignment restrictions that do not report them.
 3421: @item data space pointer not properly aligned, @code{,}, @code{C,}:
 3422: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
 3423: Like other alignment errors.
 3425: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
 3426: Like other stack underflows.
 3428: @item loop control parameters not available:
 3429: @cindex loop control parameters not available
 3430: Not checked. The counted loop words simply assume that the top of return
 3431: stack items are loop control parameters and behave accordingly.
 3433: @item most recent definition does not have a name (@code{IMMEDIATE}):
 3434: @cindex most recent definition does not have a name (@code{IMMEDIATE})
 3435: @cindex last word was headerless
 3436: @code{abort" last word was headerless"}.
 3438: @item name not defined by @code{VALUE} used by @code{TO}:
 3439: @cindex name not defined by @code{VALUE} used by @code{TO}
 3440: @cindex @code{TO} on non-@code{VALUE}s
 3441: @cindex Invalid name argument, @code{TO}
 3442: @code{-32 throw} (Invalid name argument) (unless name is a local or was
 3443: defined by @code{CONSTANT}; in the latter case it just changes the constant).
 3445: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
 3446: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
 3447: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
 3448: @code{-13 throw} (Undefined word)
 3450: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
 3451: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
 3452: Gforth behaves as if they were of the same type. I.e., you can predict
 3453: the behaviour by interpreting all parameters as, e.g., signed.
 3455: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
 3456: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
 3457: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
 3458: compilation semantics of @code{TO}.
 3460: @item String longer than a counted string returned by @code{WORD}:
 3461: @cindex String longer than a counted string returned by @code{WORD}
 3462: @cindex @code{WORD}, string overflow
 3463: Not checked. The string will be ok, but the count will, of course,
 3464: contain only the least significant bits of the length.
 3466: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
 3467: @cindex @code{LSHIFT}, large shift counts
 3468: @cindex @code{RSHIFT}, large shift counts
 3469: Processor-dependent. Typical behaviours are returning 0 and using only
 3470: the low bits of the shift count.
 3472: @item word not defined via @code{CREATE}:
 3473: @cindex @code{>BODY} of non-@code{CREATE}d words
 3474: @code{>BODY} produces the PFA of the word no matter how it was defined.
 3476: @cindex @code{DOES>} of non-@code{CREATE}d words
 3477: @code{DOES>} changes the execution semantics of the last defined word no
 3478: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
 3479: @code{CREATE , DOES>}.
 3481: @item words improperly used outside @code{<#} and @code{#>}:
 3482: Not checked. As usual, you can expect memory faults.
 3484: @end table
 3487: @c ---------------------------------------------------------------------
 3488: @node core-other,  , core-ambcond, The Core Words
 3489: @subsection Other system documentation
 3490: @c ---------------------------------------------------------------------
 3491: @cindex other system documentation, core words
 3492: @cindex core words, other system documentation
 3494: @table @i
 3495: @item nonstandard words using @code{PAD}:
 3496: @cindex @code{PAD} use by nonstandard words
 3497: None.
 3499: @item operator's terminal facilities available:
 3500: @cindex operator's terminal facilities available
 3501: After processing the command line, Gforth goes into interactive mode,
 3502: and you can give commands to Gforth interactively. The actual facilities
 3503: available depend on how you invoke Gforth.
 3505: @item program data space available:
 3506: @cindex program data space available
 3507: @cindex data space available
 3508: @code{UNUSED .} gives the remaining dictionary space. The total
 3509: dictionary space can be specified with the @code{-m} switch
 3510: (@pxref{Invoking Gforth}) when Gforth starts up.
 3512: @item return stack space available:
 3513: @cindex return stack space available
 3514: You can compute the total return stack space in cells with
 3515: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
 3516: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
 3518: @item stack space available:
 3519: @cindex stack space available
 3520: You can compute the total data stack space in cells with
 3521: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
 3522: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
 3524: @item system dictionary space required, in address units:
 3525: @cindex system dictionary space required, in address units
 3526: Type @code{here forthstart - .} after startup. At the time of this
 3527: writing, this gives 80080 (bytes) on a 32-bit system.
 3528: @end table
 3531: @c =====================================================================
 3532: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
 3533: @section The optional Block word set
 3534: @c =====================================================================
 3535: @cindex system documentation, block words
 3536: @cindex block words, system documentation
 3538: @menu
 3539: * block-idef::                  Implementation Defined Options
 3540: * block-ambcond::               Ambiguous Conditions               
 3541: * block-other::                 Other System Documentation                 
 3542: @end menu
 3545: @c ---------------------------------------------------------------------
 3546: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
 3547: @subsection Implementation Defined Options
 3548: @c ---------------------------------------------------------------------
 3549: @cindex implementation-defined options, block words
 3550: @cindex block words, implementation-defined options
 3552: @table @i
 3553: @item the format for display by @code{LIST}:
 3554: @cindex @code{LIST} display format
 3555: First the screen number is displayed, then 16 lines of 64 characters,
 3556: each line preceded by the line number.
 3558: @item the length of a line affected by @code{\}:
 3559: @cindex length of a line affected by @code{\}
 3560: @cindex @code{\}, line length in blocks
 3561: 64 characters.
 3562: @end table
 3565: @c ---------------------------------------------------------------------
 3566: @node block-ambcond, block-other, block-idef, The optional Block word set
 3567: @subsection Ambiguous conditions
 3568: @c ---------------------------------------------------------------------
 3569: @cindex block words, ambiguous conditions
 3570: @cindex ambiguous conditions, block words
 3572: @table @i
 3573: @item correct block read was not possible:
 3574: @cindex block read not possible
 3575: Typically results in a @code{throw} of some OS-derived value (between
 3576: -512 and -2048). If the blocks file was just not long enough, blanks are
 3577: supplied for the missing portion.
 3579: @item I/O exception in block transfer:
 3580: @cindex I/O exception in block transfer
 3581: @cindex block transfer, I/O exception
 3582: Typically results in a @code{throw} of some OS-derived value (between
 3583: -512 and -2048).
 3585: @item invalid block number:
 3586: @cindex invalid block number
 3587: @cindex block number invalid
 3588: @code{-35 throw} (Invalid block number)
 3590: @item a program directly alters the contents of @code{BLK}:
 3591: @cindex @code{BLK}, altering @code{BLK}
 3592: The input stream is switched to that other block, at the same
 3593: position. If the storing to @code{BLK} happens when interpreting
 3594: non-block input, the system will get quite confused when the block ends.
 3596: @item no current block buffer for @code{UPDATE}:
 3597: @cindex @code{UPDATE}, no current block buffer
 3598: @code{UPDATE} has no effect.
 3600: @end table
 3602: @c ---------------------------------------------------------------------
 3603: @node block-other,  , block-ambcond, The optional Block word set
 3604: @subsection Other system documentation
 3605: @c ---------------------------------------------------------------------
 3606: @cindex other system documentation, block words
 3607: @cindex block words, other system documentation
 3609: @table @i
 3610: @item any restrictions a multiprogramming system places on the use of buffer addresses:
 3611: No restrictions (yet).
 3613: @item the number of blocks available for source and data:
 3614: depends on your disk space.
 3616: @end table
 3619: @c =====================================================================
 3620: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
 3621: @section The optional Double Number word set
 3622: @c =====================================================================
 3623: @cindex system documentation, double words
 3624: @cindex double words, system documentation
 3626: @menu
 3627: * double-ambcond::              Ambiguous Conditions              
 3628: @end menu
 3631: @c ---------------------------------------------------------------------
 3632: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
 3633: @subsection Ambiguous conditions
 3634: @c ---------------------------------------------------------------------
 3635: @cindex double words, ambiguous conditions
 3636: @cindex ambiguous conditions, double words
 3638: @table @i
 3639: @item @var{d} outside of range of @var{n} in @code{D>S}:
 3640: @cindex @code{D>S}, @var{d} out of range of @var{n} 
 3641: The least significant cell of @var{d} is produced.
 3643: @end table
 3646: @c =====================================================================
 3647: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
 3648: @section The optional Exception word set
 3649: @c =====================================================================
 3650: @cindex system documentation, exception words
 3651: @cindex exception words, system documentation
 3653: @menu
 3654: * exception-idef::              Implementation Defined Options              
 3655: @end menu
 3658: @c ---------------------------------------------------------------------
 3659: @node exception-idef,  , The optional Exception word set, The optional Exception word set
 3660: @subsection Implementation Defined Options
 3661: @c ---------------------------------------------------------------------
 3662: @cindex implementation-defined options, exception words
 3663: @cindex exception words, implementation-defined options
 3665: @table @i
 3666: @item @code{THROW}-codes used in the system:
 3667: @cindex @code{THROW}-codes used in the system
 3668: The codes -256@minus{}-511 are used for reporting signals. The mapping
 3669: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
 3670: codes -512@minus{}-2047 are used for OS errors (for file and memory
 3671: allocation operations). The mapping from OS error numbers to throw codes
 3672: is -512@minus{}@code{errno}. One side effect of this mapping is that
 3673: undefined OS errors produce a message with a strange number; e.g.,
 3674: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
 3675: @end table
 3677: @c =====================================================================
 3678: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
 3679: @section The optional Facility word set
 3680: @c =====================================================================
 3681: @cindex system documentation, facility words
 3682: @cindex facility words, system documentation
 3684: @menu
 3685: * facility-idef::               Implementation Defined Options               
 3686: * facility-ambcond::            Ambiguous Conditions            
 3687: @end menu
 3690: @c ---------------------------------------------------------------------
 3691: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
 3692: @subsection Implementation Defined Options
 3693: @c ---------------------------------------------------------------------
 3694: @cindex implementation-defined options, facility words
 3695: @cindex facility words, implementation-defined options
 3697: @table @i
 3698: @item encoding of keyboard events (@code{EKEY}):
 3699: @cindex keyboard events, encoding in @code{EKEY}
 3700: @cindex @code{EKEY}, encoding of keyboard events
 3701: Not yet implemented.
 3703: @item duration of a system clock tick:
 3704: @cindex duration of a system clock tick
 3705: @cindex clock tick duration
 3706: System dependent. With respect to @code{MS}, the time is specified in
 3707: microseconds. How well the OS and the hardware implement this, is
 3708: another question.
 3710: @item repeatability to be expected from the execution of @code{MS}:
 3711: @cindex repeatability to be expected from the execution of @code{MS}
 3712: @cindex @code{MS}, repeatability to be expected
 3713: System dependent. On Unix, a lot depends on load. If the system is
 3714: lightly loaded, and the delay is short enough that Gforth does not get
 3715: swapped out, the performance should be acceptable. Under MS-DOS and
 3716: other single-tasking systems, it should be good.
 3718: @end table
 3721: @c ---------------------------------------------------------------------
 3722: @node facility-ambcond,  , facility-idef, The optional Facility word set
 3723: @subsection Ambiguous conditions
 3724: @c ---------------------------------------------------------------------
 3725: @cindex facility words, ambiguous conditions
 3726: @cindex ambiguous conditions, facility words
 3728: @table @i
 3729: @item @code{AT-XY} can't be performed on user output device:
 3730: @cindex @code{AT-XY} can't be performed on user output device
 3731: Largely terminal dependent. No range checks are done on the arguments.
 3732: No errors are reported. You may see some garbage appearing, you may see
 3733: simply nothing happen.
 3735: @end table
 3738: @c =====================================================================
 3739: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
 3740: @section The optional File-Access word set
 3741: @c =====================================================================
 3742: @cindex system documentation, file words
 3743: @cindex file words, system documentation
 3745: @menu
 3746: * file-idef::                   Implementation Defined Options
 3747: * file-ambcond::                Ambiguous Conditions                
 3748: @end menu
 3750: @c ---------------------------------------------------------------------
 3751: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
 3752: @subsection Implementation Defined Options
 3753: @c ---------------------------------------------------------------------
 3754: @cindex implementation-defined options, file words
 3755: @cindex file words, implementation-defined options
 3757: @table @i
 3758: @item file access methods used:
 3759: @cindex file access methods used
 3760: @code{R/O}, @code{R/W} and @code{BIN} work as you would
 3761: expect. @code{W/O} translates into the C file opening mode @code{w} (or
 3762: @code{wb}): The file is cleared, if it exists, and created, if it does
 3763: not (with both @code{open-file} and @code{create-file}).  Under Unix
 3764: @code{create-file} creates a file with 666 permissions modified by your
 3765: umask.
 3767: @item file exceptions:
 3768: @cindex file exceptions
 3769: The file words do not raise exceptions (except, perhaps, memory access
 3770: faults when you pass illegal addresses or file-ids).
 3772: @item file line terminator:
 3773: @cindex file line terminator
 3774: System-dependent. Gforth uses C's newline character as line
 3775: terminator. What the actual character code(s) of this are is
 3776: system-dependent.
 3778: @item file name format:
 3779: @cindex file name format
 3780: System dependent. Gforth just uses the file name format of your OS.
 3782: @item information returned by @code{FILE-STATUS}:
 3783: @cindex @code{FILE-STATUS}, returned information
 3784: @code{FILE-STATUS} returns the most powerful file access mode allowed
 3785: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
 3786: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
 3787: along with the returned mode.
 3789: @item input file state after an exception when including source:
 3790: @cindex exception when including source
 3791: All files that are left via the exception are closed.
 3793: @item @var{ior} values and meaning:
 3794: @cindex @var{ior} values and meaning
 3795: The @var{ior}s returned by the file and memory allocation words are
 3796: intended as throw codes. They typically are in the range
 3797: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 3798: @var{ior}s is -512@minus{}@var{errno}.
 3800: @item maximum depth of file input nesting:
 3801: @cindex maximum depth of file input nesting
 3802: @cindex file input nesting, maximum depth
 3803: limited by the amount of return stack, locals/TIB stack, and the number
 3804: of open files available. This should not give you troubles.
 3806: @item maximum size of input line:
 3807: @cindex maximum size of input line
 3808: @cindex input line size, maximum
 3809: @code{/line}. Currently 255.
 3811: @item methods of mapping block ranges to files:
 3812: @cindex mapping block ranges to files
 3813: @cindex files containing blocks
 3814: @cindex blocks in files
 3815: By default, blocks are accessed in the file @file{blocks.fb} in the
 3816: current working directory. The file can be switched with @code{USE}.
 3818: @item number of string buffers provided by @code{S"}:
 3819: @cindex @code{S"}, number of string buffers
 3820: 1
 3822: @item size of string buffer used by @code{S"}:
 3823: @cindex @code{S"}, size of string buffer
 3824: @code{/line}. currently 255.
 3826: @end table
 3828: @c ---------------------------------------------------------------------
 3829: @node file-ambcond,  , file-idef, The optional File-Access word set
 3830: @subsection Ambiguous conditions
 3831: @c ---------------------------------------------------------------------
 3832: @cindex file words, ambiguous conditions
 3833: @cindex ambiguous conditions, file words
 3835: @table @i
 3836: @item attempting to position a file outside its boundaries:
 3837: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
 3838: @code{REPOSITION-FILE} is performed as usual: Afterwards,
 3839: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
 3841: @item attempting to read from file positions not yet written:
 3842: @cindex reading from file positions not yet written
 3843: End-of-file, i.e., zero characters are read and no error is reported.
 3845: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
 3846: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid 
 3847: An appropriate exception may be thrown, but a memory fault or other
 3848: problem is more probable.
 3850: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
 3851: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
 3852: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
 3853: The @var{ior} produced by the operation, that discovered the problem, is
 3854: thrown.
 3856: @item named file cannot be opened (@code{INCLUDED}):
 3857: @cindex @code{INCLUDED}, named file cannot be opened
 3858: The @var{ior} produced by @code{open-file} is thrown.
 3860: @item requesting an unmapped block number:
 3861: @cindex unmapped block numbers
 3862: There are no unmapped legal block numbers. On some operating systems,
 3863: writing a block with a large number may overflow the file system and
 3864: have an error message as consequence.
 3866: @item using @code{source-id} when @code{blk} is non-zero:
 3867: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
 3868: @code{source-id} performs its function. Typically it will give the id of
 3869: the source which loaded the block. (Better ideas?)
 3871: @end table
 3874: @c =====================================================================
 3875: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
 3876: @section The optional Floating-Point word set
 3877: @c =====================================================================
 3878: @cindex system documentation, floating-point words
 3879: @cindex floating-point words, system documentation
 3881: @menu
 3882: * floating-idef::               Implementation Defined Options
 3883: * floating-ambcond::            Ambiguous Conditions            
 3884: @end menu
 3887: @c ---------------------------------------------------------------------
 3888: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
 3889: @subsection Implementation Defined Options
 3890: @c ---------------------------------------------------------------------
 3891: @cindex implementation-defined options, floating-point words
 3892: @cindex floating-point words, implementation-defined options
 3894: @table @i
 3895: @item format and range of floating point numbers:
 3896: @cindex format and range of floating point numbers
 3897: @cindex floating point numbers, format and range
 3898: System-dependent; the @code{double} type of C.
 3900: @item results of @code{REPRESENT} when @var{float} is out of range:
 3901: @cindex  @code{REPRESENT}, results when @var{float} is out of range
 3902: System dependent; @code{REPRESENT} is implemented using the C library
 3903: function @code{ecvt()} and inherits its behaviour in this respect.
 3905: @item rounding or truncation of floating-point numbers:
 3906: @cindex rounding of floating-point numbers
 3907: @cindex truncation of floating-point numbers
 3908: @cindex floating-point numbers, rounding or truncation
 3909: System dependent; the rounding behaviour is inherited from the hosting C
 3910: compiler. IEEE-FP-based (i.e., most) systems by default round to
 3911: nearest, and break ties by rounding to even (i.e., such that the last
 3912: bit of the mantissa is 0).
 3914: @item size of floating-point stack:
 3915: @cindex floating-point stack size
 3916: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
 3917: the floating-point stack (in floats). You can specify this on startup
 3918: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
 3920: @item width of floating-point stack:
 3921: @cindex floating-point stack width 
 3922: @code{1 floats}.
 3924: @end table
 3927: @c ---------------------------------------------------------------------
 3928: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
 3929: @subsection Ambiguous conditions
 3930: @c ---------------------------------------------------------------------
 3931: @cindex floating-point words, ambiguous conditions
 3932: @cindex ambiguous conditions, floating-point words
 3934: @table @i
 3935: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
 3936: @cindex @code{df@@} or @code{df!} used with an address that is not double-float  aligned
 3937: System-dependent. Typically results in a @code{-23 THROW} like other
 3938: alignment violations.
 3940: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
 3941: @cindex @code{f@@} used with an address that is not float aligned
 3942: @cindex @code{f!} used with an address that is not float aligned
 3943: System-dependent. Typically results in a @code{-23 THROW} like other
 3944: alignment violations.
 3946: @item floating-point result out of range:
 3947: @cindex floating-point result out of range
 3948: System-dependent. Can result in a @code{-55 THROW} (Floating-point
 3949: unidentified fault), or can produce a special value representing, e.g.,
 3950: Infinity.
 3952: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
 3953: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned
 3954: System-dependent. Typically results in an alignment fault like other
 3955: alignment violations.
 3957: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
 3958: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
 3959: The floating-point number is converted into decimal nonetheless.
 3961: @item Both arguments are equal to zero (@code{FATAN2}):
 3962: @cindex @code{FATAN2}, both arguments are equal to zero
 3963: System-dependent. @code{FATAN2} is implemented using the C library
 3964: function @code{atan2()}.
 3966: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
 3967: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
 3968: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
 3969: because of small errors and the tan will be a very large (or very small)
 3970: but finite number.
 3972: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
 3973: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
 3974: The result is rounded to the nearest float.
 3976: @item dividing by zero:
 3977: @cindex dividing by zero, floating-point
 3978: @cindex floating-point dividing by zero
 3979: @cindex floating-point unidentified fault, FP divide-by-zero
 3980: @code{-55 throw} (Floating-point unidentified fault)
 3982: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
 3983: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
 3984: System dependent. On IEEE-FP based systems the number is converted into
 3985: an infinity.
 3987: @item @var{float}<1 (@code{FACOSH}):
 3988: @cindex @code{FACOSH}, @var{float}<1
 3989: @cindex floating-point unidentified fault, @code{FACOSH}
 3990: @code{-55 throw} (Floating-point unidentified fault)
 3992: @item @var{float}=<-1 (@code{FLNP1}):
 3993: @cindex @code{FLNP1}, @var{float}=<-1
 3994: @cindex floating-point unidentified fault, @code{FLNP1}
 3995: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 3996: negative infinity is typically produced for @var{float}=-1.
 3998: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
 3999: @cindex @code{FLN}, @var{float}=<0
 4000: @cindex @code{FLOG}, @var{float}=<0
 4001: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
 4002: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 4003: negative infinity is typically produced for @var{float}=0.
 4005: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
 4006: @cindex @code{FASINH}, @var{float}<0
 4007: @cindex @code{FSQRT}, @var{float}<0
 4008: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
 4009: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
 4010: produces values for these inputs on my Linux box (Bug in the C library?)
 4012: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
 4013: @cindex @code{FACOS}, |@var{float}|>1
 4014: @cindex @code{FASIN}, |@var{float}|>1
 4015: @cindex @code{FATANH}, |@var{float}|>1
 4016: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
 4017: @code{-55 throw} (Floating-point unidentified fault).
 4019: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
 4020: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
 4021: @cindex floating-point unidentified fault, @code{F>D}
 4022: @code{-55 throw} (Floating-point unidentified fault).
 4024: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
 4025: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
 4026: This does not happen.
 4027: @end table
 4029: @c =====================================================================
 4030: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
 4031: @section The optional Locals word set
 4032: @c =====================================================================
 4033: @cindex system documentation, locals words
 4034: @cindex locals words, system documentation
 4036: @menu
 4037: * locals-idef::                 Implementation Defined Options                 
 4038: * locals-ambcond::              Ambiguous Conditions              
 4039: @end menu
 4042: @c ---------------------------------------------------------------------
 4043: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
 4044: @subsection Implementation Defined Options
 4045: @c ---------------------------------------------------------------------
 4046: @cindex implementation-defined options, locals words
 4047: @cindex locals words, implementation-defined options
 4049: @table @i
 4050: @item maximum number of locals in a definition:
 4051: @cindex maximum number of locals in a definition
 4052: @cindex locals, maximum number in a definition
 4053: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
 4054: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
 4055: characters. The number of locals in a definition is bounded by the size
 4056: of locals-buffer, which contains the names of the locals.
 4058: @end table
 4061: @c ---------------------------------------------------------------------
 4062: @node locals-ambcond,  , locals-idef, The optional Locals word set
 4063: @subsection Ambiguous conditions
 4064: @c ---------------------------------------------------------------------
 4065: @cindex locals words, ambiguous conditions
 4066: @cindex ambiguous conditions, locals words
 4068: @table @i
 4069: @item executing a named local in interpretation state:
 4070: @cindex local in interpretation state
 4071: @cindex Interpreting a compile-only word, for a local
 4072: Locals have no interpretation semantics. If you try to perform the
 4073: interpretation semantics, you will get a @code{-14 throw} somewhere
 4074: (Interpreting a compile-only word). If you perform the compilation
 4075: semantics, the locals access will be compiled (irrespective of state).
 4077: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
 4078: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
 4079: @cindex @code{TO} on non-@code{VALUE}s and non-locals
 4080: @cindex Invalid name argument, @code{TO}
 4081: @code{-32 throw} (Invalid name argument)
 4083: @end table
 4086: @c =====================================================================
 4087: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
 4088: @section The optional Memory-Allocation word set
 4089: @c =====================================================================
 4090: @cindex system documentation, memory-allocation words
 4091: @cindex memory-allocation words, system documentation
 4093: @menu
 4094: * memory-idef::                 Implementation Defined Options                 
 4095: @end menu
 4098: @c ---------------------------------------------------------------------
 4099: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
 4100: @subsection Implementation Defined Options
 4101: @c ---------------------------------------------------------------------
 4102: @cindex implementation-defined options, memory-allocation words
 4103: @cindex memory-allocation words, implementation-defined options
 4105: @table @i
 4106: @item values and meaning of @var{ior}:
 4107: @cindex  @var{ior} values and meaning
 4108: The @var{ior}s returned by the file and memory allocation words are
 4109: intended as throw codes. They typically are in the range
 4110: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 4111: @var{ior}s is -512@minus{}@var{errno}.
 4113: @end table
 4115: @c =====================================================================
 4116: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
 4117: @section The optional Programming-Tools word set
 4118: @c =====================================================================
 4119: @cindex system documentation, programming-tools words
 4120: @cindex programming-tools words, system documentation
 4122: @menu
 4123: * programming-idef::            Implementation Defined Options            
 4124: * programming-ambcond::         Ambiguous Conditions         
 4125: @end menu
 4128: @c ---------------------------------------------------------------------
 4129: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
 4130: @subsection Implementation Defined Options
 4131: @c ---------------------------------------------------------------------
 4132: @cindex implementation-defined options, programming-tools words
 4133: @cindex programming-tools words, implementation-defined options
 4135: @table @i
 4136: @item ending sequence for input following @code{;CODE} and @code{CODE}:
 4137: @cindex @code{;CODE} ending sequence
 4138: @cindex @code{CODE} ending sequence
 4139: @code{END-CODE}
 4141: @item manner of processing input following @code{;CODE} and @code{CODE}:
 4142: @cindex @code{;CODE}, processing input
 4143: @cindex @code{CODE}, processing input
 4144: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
 4145: the input is processed by the text interpreter, (starting) in interpret
 4146: state.
 4148: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
 4149: @cindex @code{ASSEMBLER}, search order capability
 4150: The ANS Forth search order word set.
 4152: @item source and format of display by @code{SEE}:
 4153: @cindex @code{SEE}, source and format of output
 4154: The source for @code{see} is the intermediate code used by the inner
 4155: interpreter.  The current @code{see} tries to output Forth source code
 4156: as well as possible.
 4158: @end table
 4160: @c ---------------------------------------------------------------------
 4161: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
 4162: @subsection Ambiguous conditions
 4163: @c ---------------------------------------------------------------------
 4164: @cindex programming-tools words, ambiguous conditions
 4165: @cindex ambiguous conditions, programming-tools words
 4167: @table @i
 4169: @item deleting the compilation wordlist (@code{FORGET}):
 4170: @cindex @code{FORGET}, deleting the compilation wordlist
 4171: Not implemented (yet).
 4173: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
 4174: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
 4175: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
 4176: @cindex control-flow stack underflow
 4177: This typically results in an @code{abort"} with a descriptive error
 4178: message (may change into a @code{-22 throw} (Control structure mismatch)
 4179: in the future). You may also get a memory access error. If you are
 4180: unlucky, this ambiguous condition is not caught.
 4182: @item @var{name} can't be found (@code{FORGET}):
 4183: @cindex @code{FORGET}, @var{name} can't be found
 4184: Not implemented (yet).
 4186: @item @var{name} not defined via @code{CREATE}:
 4187: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
 4188: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
 4189: the execution semantics of the last defined word no matter how it was
 4190: defined.
 4192: @item @code{POSTPONE} applied to @code{[IF]}:
 4193: @cindex @code{POSTPONE} applied to @code{[IF]}
 4194: @cindex @code{[IF]} and @code{POSTPONE}
 4195: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
 4196: equivalent to @code{[IF]}.
 4198: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
 4199: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
 4200: Continue in the same state of conditional compilation in the next outer
 4201: input source. Currently there is no warning to the user about this.
 4203: @item removing a needed definition (@code{FORGET}):
 4204: @cindex @code{FORGET}, removing a needed definition
 4205: Not implemented (yet).
 4207: @end table
 4210: @c =====================================================================
 4211: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
 4212: @section The optional Search-Order word set
 4213: @c =====================================================================
 4214: @cindex system documentation, search-order words
 4215: @cindex search-order words, system documentation
 4217: @menu
 4218: * search-idef::                 Implementation Defined Options                 
 4219: * search-ambcond::              Ambiguous Conditions              
 4220: @end menu
 4223: @c ---------------------------------------------------------------------
 4224: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
 4225: @subsection Implementation Defined Options
 4226: @c ---------------------------------------------------------------------
 4227: @cindex implementation-defined options, search-order words
 4228: @cindex search-order words, implementation-defined options
 4230: @table @i
 4231: @item maximum number of word lists in search order:
 4232: @cindex maximum number of word lists in search order
 4233: @cindex search order, maximum depth
 4234: @code{s" wordlists" environment? drop .}. Currently 16.
 4236: @item minimum search order:
 4237: @cindex minimum search order
 4238: @cindex search order, minimum
 4239: @code{root root}.
 4241: @end table
 4243: @c ---------------------------------------------------------------------
 4244: @node search-ambcond,  , search-idef, The optional Search-Order word set
 4245: @subsection Ambiguous conditions
 4246: @c ---------------------------------------------------------------------
 4247: @cindex search-order words, ambiguous conditions
 4248: @cindex ambiguous conditions, search-order words
 4250: @table @i
 4251: @item changing the compilation wordlist (during compilation):
 4252: @cindex changing the compilation wordlist (during compilation)
 4253: @cindex compilation wordlist, change before definition ends
 4254: The word is entered into the wordlist that was the compilation wordlist
 4255: at the start of the definition. Any changes to the name field (e.g.,
 4256: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
 4257: are applied to the latest defined word (as reported by @code{last} or
 4258: @code{lastxt}), if possible, irrespective of the compilation wordlist.
 4260: @item search order empty (@code{previous}):
 4261: @cindex @code{previous}, search order empty
 4262: @cindex Vocstack empty, @code{previous}
 4263: @code{abort" Vocstack empty"}.
 4265: @item too many word lists in search order (@code{also}):
 4266: @cindex @code{also}, too many word lists in search order
 4267: @cindex Vocstack full, @code{also}
 4268: @code{abort" Vocstack full"}.
 4270: @end table
 4272: @c ***************************************************************
 4273: @node Model, Integrating Gforth, ANS conformance, Top
 4274: @chapter Model
 4276: This chapter has yet to be written. It will contain information, on
 4277: which internal structures you can rely.
 4279: @c ***************************************************************
 4280: @node Integrating Gforth, Emacs and Gforth, Model, Top
 4281: @chapter Integrating Gforth into C programs
 4283: This is not yet implemented.
 4285: Several people like to use Forth as scripting language for applications
 4286: that are otherwise written in C, C++, or some other language.
 4288: The Forth system ATLAST provides facilities for embedding it into
 4289: applications; unfortunately it has several disadvantages: most
 4290: importantly, it is not based on ANS Forth, and it is apparently dead
 4291: (i.e., not developed further and not supported). The facilities
 4292: provided by Gforth in this area are inspired by ATLASTs facilities, so
 4293: making the switch should not be hard.
 4295: We also tried to design the interface such that it can easily be
 4296: implemented by other Forth systems, so that we may one day arrive at a
 4297: standardized interface. Such a standard interface would allow you to
 4298: replace the Forth system without having to rewrite C code.
 4300: You embed the Gforth interpreter by linking with the library
 4301: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
 4302: global symbols in this library that belong to the interface, have the
 4303: prefix @code{forth_}. (Global symbols that are used internally have the
 4304: prefix @code{gforth_}).
 4306: You can include the declarations of Forth types and the functions and
 4307: variables of the interface with @code{#include <forth.h>}.
 4309: Types.
 4311: Variables.
 4313: Data and FP Stack pointer. Area sizes.
 4315: functions.
 4317: forth_init(imagefile)
 4318: forth_evaluate(string) exceptions?
 4319: forth_goto(address) (or forth_execute(xt)?)
 4320: forth_continue() (a corountining mechanism)
 4322: Adding primitives.
 4324: No checking.
 4326: Signals?
 4328: Accessing the Stacks
 4330: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
 4331: @chapter Emacs and Gforth
 4332: @cindex Emacs and Gforth
 4334: @cindex @file{gforth.el}
 4335: @cindex @file{forth.el}
 4336: @cindex Rydqvist, Goran
 4337: @cindex comment editing commands
 4338: @cindex @code{\}, editing with Emacs
 4339: @cindex debug tracer editing commands
 4340: @cindex @code{~~}, removal with Emacs
 4341: @cindex Forth mode in Emacs
 4342: Gforth comes with @file{gforth.el}, an improved version of
 4343: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
 4344: improvements are a better (but still not perfect) handling of
 4345: indentation. I have also added comment paragraph filling (@kbd{M-q}),
 4346: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
 4347: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
 4348: stuff I do not use alone, even though some of it only makes sense for
 4349: TILE. To get a description of these features, enter Forth mode and type
 4350: @kbd{C-h m}.
 4352: @cindex source location of error or debugging output in Emacs
 4353: @cindex error output, finding the source location in Emacs
 4354: @cindex debugging output, finding the source location in Emacs
 4355: In addition, Gforth supports Emacs quite well: The source code locations
 4356: given in error messages, debugging output (from @code{~~}) and failed
 4357: assertion messages are in the right format for Emacs' compilation mode
 4358: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
 4359: Manual}) so the source location corresponding to an error or other
 4360: message is only a few keystrokes away (@kbd{C-x `} for the next error,
 4361: @kbd{C-c C-c} for the error under the cursor).
 4363: @cindex @file{TAGS} file
 4364: @cindex @file{etags.fs}
 4365: @cindex viewing the source of a word in Emacs
 4366: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
 4367: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
 4368: contains the definitions of all words defined afterwards. You can then
 4369: find the source for a word using @kbd{M-.}. Note that emacs can use
 4370: several tags files at the same time (e.g., one for the Gforth sources
 4371: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
 4372: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
 4373: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
 4374: @file{/usr/local/share/gforth/0.2.0/TAGS}).
 4376: @cindex @file{.emacs}
 4377: To get all these benefits, add the following lines to your @file{.emacs}
 4378: file:
 4380: @example
 4381: (autoload 'forth-mode "gforth.el")
 4382: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
 4383: @end example
 4385: @node Image Files, Engine, Emacs and Gforth, Top
 4386: @chapter Image Files
 4387: @cindex image files
 4388: @cindex @code{.fi} files
 4389: @cindex precompiled Forth code
 4390: @cindex dictionary in persistent form
 4391: @cindex persistent form of dictionary
 4393: An image file is a file containing an image of the Forth dictionary,
 4394: i.e., compiled Forth code and data residing in the dictionary.  By
 4395: convention, we use the extension @code{.fi} for image files.
 4397: @menu
 4398: * Image File Background::          Why have image files?
 4399: * Non-Relocatable Image Files::    don't always work.
 4400: * Data-Relocatable Image Files::   are better.
 4401: * Fully Relocatable Image Files::  better yet.
 4402: * Stack and Dictionary Sizes::     Setting the default sizes for an image.
 4403: * Running Image Files::            @code{gforth -i @var{file}} or @var{file}.
 4404: * Modifying the Startup Sequence:: and turnkey applications.
 4405: @end menu
 4407: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
 4408: @section Image File Background
 4409: @cindex image file background
 4411: Our Forth system consists not only of primitives, but also of
 4412: definitions written in Forth. Since the Forth compiler itself belongs to
 4413: those definitions, it is not possible to start the system with the
 4414: primitives and the Forth source alone. Therefore we provide the Forth
 4415: code as an image file in nearly executable form. At the start of the
 4416: system a C routine loads the image file into memory, optionally
 4417: relocates the addresses, then sets up the memory (stacks etc.) according
 4418: to information in the image file, and starts executing Forth code.
 4420: The image file variants represent different compromises between the
 4421: goals of making it easy to generate image files and making them
 4422: portable.
 4424: @cindex relocation at run-time
 4425: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
 4426: run-time. This avoids many of the complications discussed below (image
 4427: files are data relocatable without further ado), but costs performance
 4428: (one addition per memory access).
 4430: @cindex relocation at load-time
 4431: By contrast, our loader performs relocation at image load time. The
 4432: loader also has to replace tokens standing for primitive calls with the
 4433: appropriate code-field addresses (or code addresses in the case of
 4434: direct threading).
 4436: There are three kinds of image files, with different degrees of
 4437: relocatability: non-relocatable, data-relocatable, and fully relocatable
 4438: image files.
 4440: @cindex image file loader
 4441: @cindex relocating loader
 4442: @cindex loader for image files
 4443: These image file variants have several restrictions in common; they are
 4444: caused by the design of the image file loader:
 4446: @itemize @bullet
 4447: @item
 4448: There is only one segment; in particular, this means, that an image file
 4449: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
 4450: them). And the contents of the stacks are not represented, either.
 4452: @item
 4453: The only kinds of relocation supported are: adding the same offset to
 4454: all cells that represent data addresses; and replacing special tokens
 4455: with code addresses or with pieces of machine code.
 4457: If any complex computations involving addresses are performed, the
 4458: results cannot be represented in the image file. Several applications that
 4459: use such computations come to mind:
 4460: @itemize @minus
 4461: @item
 4462: Hashing addresses (or data structures which contain addresses) for table
 4463: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
 4464: purpose, you will have no problem, because the hash tables are
 4465: recomputed automatically when the system is started. If you use your own
 4466: hash tables, you will have to do something similar.
 4468: @item
 4469: There's a cute implementation of doubly-linked lists that uses
 4470: @code{XOR}ed addresses. You could represent such lists as singly-linked
 4471: in the image file, and restore the doubly-linked representation on
 4472: startup.@footnote{In my opinion, though, you should think thrice before
 4473: using a doubly-linked list (whatever implementation).}
 4475: @item
 4476: The code addresses of run-time routines like @code{docol:} cannot be
 4477: represented in the image file (because their tokens would be replaced by
 4478: machine code in direct threaded implementations). As a workaround,
 4479: compute these addresses at run-time with @code{>code-address} from the
 4480: executions tokens of appropriate words (see the definitions of
 4481: @code{docol:} and friends in @file{kernel.fs}).
 4483: @item
 4484: On many architectures addresses are represented in machine code in some
 4485: shifted or mangled form. You cannot put @code{CODE} words that contain
 4486: absolute addresses in this form in a relocatable image file. Workarounds
 4487: are representing the address in some relative form (e.g., relative to
 4488: the CFA, which is present in some register), or loading the address from
 4489: a place where it is stored in a non-mangled form.
 4490: @end itemize
 4491: @end itemize
 4493: @node  Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
 4494: @section Non-Relocatable Image Files
 4495: @cindex non-relocatable image files
 4496: @cindex image files, non-relocatable
 4498: These files are simple memory dumps of the dictionary. They are specific
 4499: to the executable (i.e., @file{gforth} file) they were created
 4500: with. What's worse, they are specific to the place on which the
 4501: dictionary resided when the image was created. Now, there is no
 4502: guarantee that the dictionary will reside at the same place the next
 4503: time you start Gforth, so there's no guarantee that a non-relocatable
 4504: image will work the next time (Gforth will complain instead of crashing,
 4505: though).
 4507: You can create a non-relocatable image file with
 4509: doc-savesystem
 4511: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
 4512: @section Data-Relocatable Image Files
 4513: @cindex data-relocatable image files
 4514: @cindex image files, data-relocatable
 4516: These files contain relocatable data addresses, but fixed code addresses
 4517: (instead of tokens). They are specific to the executable (i.e.,
 4518: @file{gforth} file) they were created with. For direct threading on some
 4519: architectures (e.g., the i386), data-relocatable images do not work. You
 4520: get a data-relocatable image, if you use @file{gforth-makeimage} with a
 4521: Gforth binary that is not doubly indirect threaded (@pxref{Fully
 4522: Relocatable Image Files}).
 4524: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
 4525: @section Fully Relocatable Image Files
 4526: @cindex fully relocatable image files
 4527: @cindex image files, fully relocatable
 4529: @cindex @file{kern*.fi}, relocatability
 4530: @cindex @file{}, relocatability
 4531: These image files have relocatable data addresses, and tokens for code
 4532: addresses. They can be used with different binaries (e.g., with and
 4533: without debugging) on the same machine, and even across machines with
 4534: the same data formats (byte order, cell size, floating point
 4535: format). However, they are usually specific to the version of Gforth
 4536: they were created with. The files @file{} and @file{kernl*.fi}
 4537: are fully relocatable.
 4539: There are two ways to create a fully relocatable image file:
 4541: @menu
 4542: * gforth-makeimage::            The normal way
 4543: * cross.fs::                    The hard way
 4544: @end menu
 4546: @node gforth-makeimage, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
 4547: @subsection @file{gforth-makeimage}
 4548: @cindex @file{comp-image.fs}
 4549: @cindex @file{gforth-makeimage}
 4551: You will usually use @file{gforth-makeimage}. If you want to create an
 4552: image @var{file} that contains everything you would load by invoking
 4553: Gforth with @code{gforth @var{options}}, you simply say
 4554: @example
 4555: gforth-makeimage @var{file} @var{options}
 4556: @end example
 4558: E.g., if you want to create an image @file{} that has the file
 4559: @file{asm.fs} loaded in addition to the usual stuff, you could do it
 4560: like this:
 4562: @example
 4563: gforth-makeimage asm.fs
 4564: @end example
 4566: @file{gforth-makeimage} works like this: It produces two non-relocatable
 4567: images for different addresses and then compares them. Its output
 4568: reflects this: first you see the output (if any) of the two Gforth
 4569: invocations that produce the nonrelocatable image files, then you see
 4570: the output of the comparing program: It displays the offset used for
 4571: data addresses and the offset used for code addresses;
 4572: moreover, for each cell that cannot be represented correctly in the
 4573: image files, it displays a line like the following one:
 4575: @example
 4576:      78DC         BFFFFA50         BFFFFA40
 4577: @end example
 4579: This means that at offset $78dc from @code{forthstart}, one input image
 4580: contains $bffffa50, and the other contains $bffffa40. Since these cells
 4581: cannot be represented correctly in the output image, you should examine
 4582: these places in the dictionary and verify that these cells are dead
 4583: (i.e., not read before they are written).
 4585: @cindex @code{savesystem} during @file{gforth-makeimage}
 4586: @cindex @code{bye} during @file{gforth-makeimage}
 4587: @cindex doubly indirect threaded code
 4588: @cindex environment variable @code{GFORTHD}
 4589: @cindex @code{GFORTHD} environment variable
 4590: @cindex @code{gforth-ditc}
 4591: There are a few wrinkles: After processing the passed @var{options}, the
 4592: words @code{savesystem} and @code{bye} must be visible. A special doubly
 4593: indirect threaded version of the @file{gforth} executable is used for
 4594: creating the nonrelocatable images; you can pass the exact filename of
 4595: this executable through the environment variable @code{GFORTHD}
 4596: (default: @file{gforth-ditc}); if you pass a version that is not doubly
 4597: indirect threaded, you will not get a fully relocatable image, but a
 4598: data-relocatable image (because there is no code address offset).
 4600: @node cross.fs,  , gforth-makeimage, Fully Relocatable Image Files
 4601: @subsection @file{cross.fs}
 4602: @cindex @file{cross.fs}
 4603: @cindex cross-compiler
 4604: @cindex metacompiler
 4606: You can also use @code{cross}, a batch compiler that accepts a Forth-like
 4607: programming language. This @code{cross} language has to be documented
 4608: yet.
 4610: @cindex target compiler
 4611: @code{cross} also allows you to create image files for machines with
 4612: different data sizes and data formats than the one used for generating
 4613: the image file. You can also use it to create an application image that
 4614: does not contain a Forth compiler. These features are bought with
 4615: restrictions and inconveniences in programming. E.g., addresses have to
 4616: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
 4617: order to make the code relocatable.
 4620: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
 4621: @section Stack and Dictionary Sizes
 4622: @cindex image file, stack and dictionary sizes
 4623: @cindex dictionary size default
 4624: @cindex stack size default
 4626: If you invoke Gforth with a command line flag for the size
 4627: (@pxref{Invoking Gforth}), the size you specify is stored in the
 4628: dictionary. If you save the dictionary with @code{savesystem} or create
 4629: an image with @file{gforth-makeimage}, this size will become the default
 4630: for the resulting image file. E.g., the following will create a
 4631: fully relocatable version of with a 1MB dictionary:
 4633: @example
 4634: gforth-makeimage -m 1M
 4635: @end example
 4637: In other words, if you want to set the default size for the dictionary
 4638: and the stacks of an image, just invoke @file{gforth-makeimage} with the
 4639: appropriate options when creating the image.
 4641: @cindex stack size, cache-friendly
 4642: Note: For cache-friendly behaviour (i.e., good performance), you should
 4643: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
 4644: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
 4645: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
 4647: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
 4648: @section Running Image Files
 4649: @cindex running image files
 4650: @cindex invoking image files
 4651: @cindex image file invocation
 4653: @cindex -i, invoke image file
 4654: @cindex --image file, invoke image file
 4655: You can invoke Gforth with an image file @var{image} instead of the
 4656: default @file{} with the @code{-i} flag (@pxref{Invoking Gforth}):
 4657: @example
 4658: gforth -i @var{image}
 4659: @end example
 4661: @cindex executable image file
 4662: @cindex image files, executable
 4663: If your operating system supports starting scripts with a line of the
 4664: form @code{#! ...}, you just have to type the image file name to start
 4665: Gforth with this image file (note that the file extension @code{.fi} is
 4666: just a convention). I.e., to run Gforth with the image file @var{image},
 4667: you can just type @var{image} instead of @code{gforth -i @var{image}}.
 4669: doc-#!
 4671: @node Modifying the Startup Sequence,  , Running Image Files, Image Files
 4672: @section Modifying the Startup Sequence
 4673: @cindex startup sequence for image file
 4674: @cindex image file initialization sequence
 4675: @cindex initialization sequence of image file
 4677: You can add your own initialization to the startup sequence through the
 4678: deferred word
 4680: doc-'cold
 4682: @code{'cold} is invoked just before the image-specific command line
 4683: processing (by default, loading files and evaluating (@code{-e}) strings)
 4684: starts.
 4686: A sequence for adding your initialization usually looks like this:
 4688: @example
 4689: :noname
 4690:     Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
 4691:     ... \ your stuff
 4692: ; IS 'cold
 4693: @end example
 4695: @cindex turnkey image files
 4696: @cindex image files, turnkey applications
 4697: You can make a turnkey image by letting @code{'cold} execute a word
 4698: (your turnkey application) that never returns; instead, it exits Gforth
 4699: via @code{bye} or @code{throw}.
 4701: @cindex command-line arguments, access
 4702: @cindex arguments on the command line, access
 4703: You can access the (image-specific) command-line arguments through the
 4704: variables @code{argc} and @code{argv}. @code{arg} provides conventient
 4705: access to @code{argv}.
 4707: doc-argc
 4708: doc-argv
 4709: doc-arg
 4711: If @code{'cold} exits normally, Gforth processes the command-line
 4712: arguments as files to be loaded and strings to be evaluated.  Therefore,
 4713: @code{'cold} should remove the arguments it has used in this case.
 4715: @c ******************************************************************
 4716: @node Engine, Bugs, Image Files, Top
 4717: @chapter Engine
 4718: @cindex engine
 4719: @cindex virtual machine
 4721: Reading this section is not necessary for programming with Gforth. It
 4722: may be helpful for finding your way in the Gforth sources.
 4724: The ideas in this section have also been published in the papers
 4725: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
 4726: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
 4727: Ertl, presented at EuroForth '93; the latter is available at
 4728: @*@url{}.
 4730: @menu
 4731: * Portability::                 
 4732: * Threading::                   
 4733: * Primitives::                  
 4734: * Performance::                 
 4735: @end menu
 4737: @node Portability, Threading, Engine, Engine
 4738: @section Portability
 4739: @cindex engine portability
 4741: One of the main goals of the effort is availability across a wide range
 4742: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
 4743: this goal by manually coding the engine in assembly language for several
 4744: then-popular processors. This approach is very labor-intensive and the
 4745: results are short-lived due to progress in computer architecture.
 4747: @cindex C, using C for the engine
 4748: Others have avoided this problem by coding in C, e.g., Mitch Bradley
 4749: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
 4750: particularly popular for UNIX-based Forths due to the large variety of
 4751: architectures of UNIX machines. Unfortunately an implementation in C
 4752: does not mix well with the goals of efficiency and with using
 4753: traditional techniques: Indirect or direct threading cannot be expressed
 4754: in C, and switch threading, the fastest technique available in C, is
 4755: significantly slower. Another problem with C is that it is very
 4756: cumbersome to express double integer arithmetic.
 4758: @cindex GNU C for the engine
 4759: @cindex long long
 4760: Fortunately, there is a portable language that does not have these
 4761: limitations: GNU C, the version of C processed by the GNU C compiler
 4762: (@pxref{C Extensions, , Extensions to the C Language Family,,
 4763: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
 4764: Labels as Values,, GNU C Manual}) makes direct and indirect
 4765: threading possible, its @code{long long} type (@pxref{Long Long, ,
 4766: Double-Word Integers,, GNU C Manual}) corresponds to Forth's
 4767: double numbers@footnote{Unfortunately, long longs are not implemented
 4768: properly on all machines (e.g., on alpha-osf1, long longs are only 64
 4769: bits, the same size as longs (and pointers), but they should be twice as
 4770: long according to @ref{Long Long, , Double-Word Integers,, GNU
 4771: C Manual}). So, we had to implement doubles in C after all. Still, on
 4772: most machines we can use long longs and achieve better performance than
 4773: with the emulation package.}. GNU C is available for free on all
 4774: important (and many unimportant) UNIX machines, VMS, 80386s running
 4775: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
 4776: on all these machines.
 4778: Writing in a portable language has the reputation of producing code that
 4779: is slower than assembly. For our Forth engine we repeatedly looked at
 4780: the code produced by the compiler and eliminated most compiler-induced
 4781: inefficiencies by appropriate changes in the source code.
 4783: @cindex explicit register declarations
 4784: @cindex --enable-force-reg, configuration flag
 4785: @cindex -DFORCE_REG
 4786: However, register allocation cannot be portably influenced by the
 4787: programmer, leading to some inefficiencies on register-starved
 4788: machines. We use explicit register declarations (@pxref{Explicit Reg
 4789: Vars, , Variables in Specified Registers,, GNU C Manual}) to
 4790: improve the speed on some machines. They are turned on by using the
 4791: configuration flag @code{--enable-force-reg} (@code{gcc} switch
 4792: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
 4793: machine, but also on the compiler version: On some machines some
 4794: compiler versions produce incorrect code when certain explicit register
 4795: declarations are used. So by default @code{-DFORCE_REG} is not used.
 4797: @node Threading, Primitives, Portability, Engine
 4798: @section Threading
 4799: @cindex inner interpreter implementation
 4800: @cindex threaded code implementation
 4802: @cindex labels as values
 4803: GNU C's labels as values extension (available since @code{gcc-2.0},
 4804: @pxref{Labels as Values, , Labels as Values,, GNU C Manual})
 4805: makes it possible to take the address of @var{label} by writing
 4806: @code{&&@var{label}}.  This address can then be used in a statement like
 4807: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
 4808: @code{goto x}.
 4810: @cindex NEXT, indirect threaded
 4811: @cindex indirect threaded inner interpreter
 4812: @cindex inner interpreter, indirect threaded
 4813: With this feature an indirect threaded NEXT looks like:
 4814: @example
 4815: cfa = *ip++;
 4816: ca = *cfa;
 4817: goto *ca;
 4818: @end example
 4819: @cindex instruction pointer
 4820: For those unfamiliar with the names: @code{ip} is the Forth instruction
 4821: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
 4822: execution token and points to the code field of the next word to be
 4823: executed; The @code{ca} (code address) fetched from there points to some
 4824: executable code, e.g., a primitive or the colon definition handler
 4825: @code{docol}.
 4827: @cindex NEXT, direct threaded
 4828: @cindex direct threaded inner interpreter
 4829: @cindex inner interpreter, direct threaded
 4830: Direct threading is even simpler:
 4831: @example
 4832: ca = *ip++;
 4833: goto *ca;
 4834: @end example
 4836: Of course we have packaged the whole thing neatly in macros called
 4837: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
 4839: @menu
 4840: * Scheduling::                  
 4841: * Direct or Indirect Threaded?::  
 4842: * DOES>::                       
 4843: @end menu
 4845: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
 4846: @subsection Scheduling
 4847: @cindex inner interpreter optimization
 4849: There is a little complication: Pipelined and superscalar processors,
 4850: i.e., RISC and some modern CISC machines can process independent
 4851: instructions while waiting for the results of an instruction. The
 4852: compiler usually reorders (schedules) the instructions in a way that
 4853: achieves good usage of these delay slots. However, on our first tries
 4854: the compiler did not do well on scheduling primitives. E.g., for
 4855: @code{+} implemented as
 4856: @example
 4857: n=sp[0]+sp[1];
 4858: sp++;
 4859: sp[0]=n;
 4860: NEXT;
 4861: @end example
 4862: the NEXT comes strictly after the other code, i.e., there is nearly no
 4863: scheduling. After a little thought the problem becomes clear: The
 4864: compiler cannot know that sp and ip point to different addresses (and
 4865: the version of @code{gcc} we used would not know it even if it was
 4866: possible), so it could not move the load of the cfa above the store to
 4867: the TOS. Indeed the pointers could be the same, if code on or very near
 4868: the top of stack were executed. In the interest of speed we chose to
 4869: forbid this probably unused ``feature'' and helped the compiler in
 4870: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
 4871: the goto part (@code{NEXT_P2}). @code{+} now looks like:
 4872: @example
 4873: n=sp[0]+sp[1];
 4874: sp++;
 4875: NEXT_P1;
 4876: sp[0]=n;
 4877: NEXT_P2;
 4878: @end example
 4879: This can be scheduled optimally by the compiler.
 4881: This division can be turned off with the switch @code{-DCISC_NEXT}. This
 4882: switch is on by default on machines that do not profit from scheduling
 4883: (e.g., the 80386), in order to preserve registers.
 4885: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
 4886: @subsection Direct or Indirect Threaded?
 4887: @cindex threading, direct or indirect?
 4889: @cindex -DDIRECT_THREADED
 4890: Both! After packaging the nasty details in macro definitions we
 4891: realized that we could switch between direct and indirect threading by
 4892: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
 4893: defining a few machine-specific macros for the direct-threading case.
 4894: On the Forth level we also offer access words that hide the
 4895: differences between the threading methods (@pxref{Threading Words}).
 4897: Indirect threading is implemented completely machine-independently.
 4898: Direct threading needs routines for creating jumps to the executable
 4899: code (e.g. to docol or dodoes). These routines are inherently
 4900: machine-dependent, but they do not amount to many source lines. I.e.,
 4901: even porting direct threading to a new machine is a small effort.
 4903: @cindex --enable-indirect-threaded, configuration flag
 4904: @cindex --enable-direct-threaded, configuration flag
 4905: The default threading method is machine-dependent. You can enforce a
 4906: specific threading method when building Gforth with the configuration
 4907: flag @code{--enable-direct-threaded} or
 4908: @code{--enable-indirect-threaded}. Note that direct threading is not
 4909: supported on all machines.
 4911: @node DOES>,  , Direct or Indirect Threaded?, Threading
 4912: @subsection DOES>
 4913: @cindex @code{DOES>} implementation
 4915: @cindex dodoes routine
 4916: @cindex DOES-code
 4917: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
 4918: the chunk of code executed by every word defined by a
 4919: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
 4920: the Forth code to be executed, i.e. the code after the
 4921: @code{DOES>} (the DOES-code)? There are two solutions:
 4923: In fig-Forth the code field points directly to the dodoes and the
 4924: DOES-code address is stored in the cell after the code address (i.e. at
 4925: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
 4926: the Forth-79 and all later standards, because in fig-Forth this address
 4927: lies in the body (which is illegal in these standards). However, by
 4928: making the code field larger for all words this solution becomes legal
 4929: again. We use this approach for the indirect threaded version and for
 4930: direct threading on some machines. Leaving a cell unused in most words
 4931: is a bit wasteful, but on the machines we are targeting this is hardly a
 4932: problem. The other reason for having a code field size of two cells is
 4933: to avoid having different image files for direct and indirect threaded
 4934: systems (direct threaded systems require two-cell code fields on many
 4935: machines).
 4937: @cindex DOES-handler
 4938: The other approach is that the code field points or jumps to the cell
 4939: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
 4940: this address (the DOES-handler). @code{dodoes} can then get the
 4941: DOES-code address by computing the code address, i.e., the address of
 4942: the jump to dodoes, and add the length of that jump field. A variant of
 4943: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
 4944: return address (which can be found in the return register on RISCs) is
 4945: the DOES-code address. Since the two cells available in the code field
 4946: are used up by the jump to the code address in direct threading on many
 4947: architectures, we use this approach for direct threading on these
 4948: architectures. We did not want to add another cell to the code field.
 4950: @node Primitives, Performance, Threading, Engine
 4951: @section Primitives
 4952: @cindex primitives, implementation
 4953: @cindex virtual machine instructions, implementation
 4955: @menu
 4956: * Automatic Generation::        
 4957: * TOS Optimization::            
 4958: * Produced code::               
 4959: @end menu
 4961: @node Automatic Generation, TOS Optimization, Primitives, Primitives
 4962: @subsection Automatic Generation
 4963: @cindex primitives, automatic generation
 4965: @cindex @file{prims2x.fs}
 4966: Since the primitives are implemented in a portable language, there is no
 4967: longer any need to minimize the number of primitives. On the contrary,
 4968: having many primitives has an advantage: speed. In order to reduce the
 4969: number of errors in primitives and to make programming them easier, we
 4970: provide a tool, the primitive generator (@file{prims2x.fs}), that
 4971: automatically generates most (and sometimes all) of the C code for a
 4972: primitive from the stack effect notation.  The source for a primitive
 4973: has the following form:
 4975: @cindex primitive source format
 4976: @format
 4977: @var{Forth-name}	@var{stack-effect}	@var{category}	[@var{pronounc.}]
 4978: [@code{""}@var{glossary entry}@code{""}]
 4979: @var{C code}
 4980: [@code{:}
 4981: @var{Forth code}]
 4982: @end format
 4984: The items in brackets are optional. The category and glossary fields
 4985: are there for generating the documentation, the Forth code is there
 4986: for manual implementations on machines without GNU C. E.g., the source
 4987: for the primitive @code{+} is:
 4988: @example
 4989: +    n1 n2 -- n    core    plus
 4990: n = n1+n2;
 4991: @end example
 4993: This looks like a specification, but in fact @code{n = n1+n2} is C
 4994: code. Our primitive generation tool extracts a lot of information from
 4995: the stack effect notations@footnote{We use a one-stack notation, even
 4996: though we have separate data and floating-point stacks; The separate
 4997: notation can be generated easily from the unified notation.}: The number
 4998: of items popped from and pushed on the stack, their type, and by what
 4999: name they are referred to in the C code. It then generates a C code
 5000: prelude and postlude for each primitive. The final C code for @code{+}
 5001: looks like this:
 5003: @example
 5004: I_plus:	/* + ( n1 n2 -- n ) */  /* label, stack effect */
 5005: /*  */                          /* documentation */
 5006: @{
 5007: DEF_CA                          /* definition of variable ca (indirect threading) */
 5008: Cell n1;                        /* definitions of variables */
 5009: Cell n2;
 5010: Cell n;
 5011: n1 = (Cell) sp[1];              /* input */
 5012: n2 = (Cell) TOS;
 5013: sp += 1;                        /* stack adjustment */
 5014: NAME("+")                       /* debugging output (with -DDEBUG) */
 5015: @{
 5016: n = n1+n2;                      /* C code taken from the source */
 5017: @}
 5018: NEXT_P1;                        /* NEXT part 1 */
 5019: TOS = (Cell)n;                  /* output */
 5020: NEXT_P2;                        /* NEXT part 2 */
 5021: @}
 5022: @end example
 5024: This looks long and inefficient, but the GNU C compiler optimizes quite
 5025: well and produces optimal code for @code{+} on, e.g., the R3000 and the
 5026: HP RISC machines: Defining the @code{n}s does not produce any code, and
 5027: using them as intermediate storage also adds no cost.
 5029: There are also other optimizations, that are not illustrated by this
 5030: example: Assignments between simple variables are usually for free (copy
 5031: propagation). If one of the stack items is not used by the primitive
 5032: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
 5033: (dead code elimination). On the other hand, there are some things that
 5034: the compiler does not do, therefore they are performed by
 5035: @file{prims2x.fs}: The compiler does not optimize code away that stores
 5036: a stack item to the place where it just came from (e.g., @code{over}).
 5038: While programming a primitive is usually easy, there are a few cases
 5039: where the programmer has to take the actions of the generator into
 5040: account, most notably @code{?dup}, but also words that do not (always)
 5041: fall through to NEXT.
 5043: @node TOS Optimization, Produced code, Automatic Generation, Primitives
 5044: @subsection TOS Optimization
 5045: @cindex TOS optimization for primitives
 5046: @cindex primitives, keeping the TOS in a register
 5048: An important optimization for stack machine emulators, e.g., Forth
 5049: engines, is keeping  one or more of the top stack items in
 5050: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
 5051: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
 5052: @itemize @bullet
 5053: @item
 5054: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
 5055: due to fewer loads from and stores to the stack.
 5056: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
 5057: @var{y<n}, due to additional moves between registers.
 5058: @end itemize
 5060: @cindex -DUSE_TOS
 5061: @cindex -DUSE_NO_TOS
 5062: In particular, keeping one item in a register is never a disadvantage,
 5063: if there are enough registers. Keeping two items in registers is a
 5064: disadvantage for frequent words like @code{?branch}, constants,
 5065: variables, literals and @code{i}. Therefore our generator only produces
 5066: code that keeps zero or one items in registers. The generated C code
 5067: covers both cases; the selection between these alternatives is made at
 5068: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
 5069: code for @code{+} is just a simple variable name in the one-item case,
 5070: otherwise it is a macro that expands into @code{sp[0]}. Note that the
 5071: GNU C compiler tries to keep simple variables like @code{TOS} in
 5072: registers, and it usually succeeds, if there are enough registers.
 5074: @cindex -DUSE_FTOS
 5075: @cindex -DUSE_NO_FTOS
 5076: The primitive generator performs the TOS optimization for the
 5077: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
 5078: operations the benefit of this optimization is even larger:
 5079: floating-point operations take quite long on most processors, but can be
 5080: performed in parallel with other operations as long as their results are
 5081: not used. If the FP-TOS is kept in a register, this works. If
 5082: it is kept on the stack, i.e., in memory, the store into memory has to
 5083: wait for the result of the floating-point operation, lengthening the
 5084: execution time of the primitive considerably.
 5086: The TOS optimization makes the automatic generation of primitives a
 5087: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
 5088: @code{TOS} is not sufficient. There are some special cases to
 5089: consider:
 5090: @itemize @bullet
 5091: @item In the case of @code{dup ( w -- w w )} the generator must not
 5092: eliminate the store to the original location of the item on the stack,
 5093: if the TOS optimization is turned on.
 5094: @item Primitives with stack effects of the form @code{--}
 5095: @var{out1}...@var{outy} must store the TOS to the stack at the start.
 5096: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
 5097: must load the TOS from the stack at the end. But for the null stack
 5098: effect @code{--} no stores or loads should be generated.
 5099: @end itemize
 5101: @node Produced code,  , TOS Optimization, Primitives
 5102: @subsection Produced code
 5103: @cindex primitives, assembly code listing
 5105: @cindex @file{engine.s}
 5106: To see what assembly code is produced for the primitives on your machine
 5107: with your compiler and your flag settings, type @code{make engine.s} and
 5108: look at the resulting file @file{engine.s}.
 5110: @node  Performance,  , Primitives, Engine
 5111: @section Performance
 5112: @cindex performance of some Forth interpreters
 5113: @cindex engine performance
 5114: @cindex benchmarking Forth systems
 5115: @cindex Gforth performance
 5117: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
 5118: impossible to write a significantly faster engine.
 5120: On register-starved machines like the 386 architecture processors
 5121: improvements are possible, because @code{gcc} does not utilize the
 5122: registers as well as a human, even with explicit register declarations;
 5123: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
 5124: and hand-tuned it for the 486; this system is 1.19 times faster on the
 5125: Sieve benchmark on a 486DX2/66 than Gforth compiled with
 5126: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
 5128: @cindex Win32Forth performance
 5129: @cindex NT Forth performance
 5130: @cindex eforth performance
 5131: @cindex ThisForth performance
 5132: @cindex PFE performance
 5133: @cindex TILE performance
 5134: However, this potential advantage of assembly language implementations
 5135: is not necessarily realized in complete Forth systems: We compared
 5136: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
 5137: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
 5138: 1994) and Eforth (with and without peephole (aka pinhole) optimization
 5139: of the threaded code); all these systems were written in assembly
 5140: language. We also compared Gforth with three systems written in C:
 5141: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
 5142: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
 5143: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
 5144: -fomit-frame-pointer; ThisForth employs peephole optimization of the
 5145: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
 5146: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
 5147: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
 5148: 486DX2/66 with similar memory performance under Windows NT. Marcel
 5149: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
 5150: added the peephole optimizer, ran the benchmarks and reported the
 5151: results.
 5153: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
 5154: matrix multiplication come from the Stanford integer benchmarks and have
 5155: been translated into Forth by Martin Fraeman; we used the versions
 5156: included in the TILE Forth package, but with bigger data set sizes; and
 5157: a recursive Fibonacci number computation for benchmarking calling
 5158: performance. The following table shows the time taken for the benchmarks
 5159: scaled by the time taken by Gforth (in other words, it shows the speedup
 5160: factor that Gforth achieved over the other systems).
 5162: @example
 5163: relative      Win32-    NT       eforth       This-
 5164:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
 5165: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
 5166: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
 5167: matmul    1.00  1.47  1.35   1.46  0.74  1.58        4.09
 5168: fib       1.00  1.52  1.34   1.22  0.86  1.74  2.99  4.30
 5169: @end example
 5171: You may find the good performance of Gforth compared with the systems
 5172: written in assembly language quite surprising. One important reason for
 5173: the disappointing performance of these systems is probably that they are
 5174: not written optimally for the 486 (e.g., they use the @code{lods}
 5175: instruction). In addition, Win32Forth uses a comfortable, but costly
 5176: method for relocating the Forth image: like @code{cforth}, it computes
 5177: the actual addresses at run time, resulting in two address computations
 5178: per NEXT (@pxref{Image File Background}).
 5180: Only Eforth with the peephole optimizer performs comparable to
 5181: Gforth. The speedups achieved with peephole optimization of threaded
 5182: code are quite remarkable. Adding a peephole optimizer to Gforth should
 5183: cause similar speedups.
 5185: The speedup of Gforth over PFE, ThisForth and TILE can be easily
 5186: explained with the self-imposed restriction of the latter systems to
 5187: standard C, which makes efficient threading impossible (however, the
 5188: measured implementation of PFE uses a GNU C extension: @ref{Global Reg
 5189: Vars, , Defining Global Register Variables,, GNU C Manual}).
 5190: Moreover, current C compilers have a hard time optimizing other aspects
 5191: of the ThisForth and the TILE source.
 5193: Note that the performance of Gforth on 386 architecture processors
 5194: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
 5195: failed to allocate any of the virtual machine registers into real
 5196: machine registers by itself and would not work correctly with explicit
 5197: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
 5198: running the Sieve) than the one measured above.
 5200: Note also that there have been several releases of Win32Forth since the
 5201: release presented here, so the results presented here may have little
 5202: predictive value for the performance of Win32Forth today.
 5204: @cindex @file{Benchres}
 5205: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
 5206: Maierhofer (presented at EuroForth '95), an indirect threaded version of
 5207: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
 5208: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
 5209: threaded version used here. The paper available at
 5210: @*@url{};
 5211: it also contains numbers for some native code systems. You can find a
 5212: newer version of these measurements at
 5213: @url{}. You can
 5214: find numbers for Gforth on various machines in @file{Benchres}.
 5216: @node Bugs, Origin, Engine, Top
 5217: @chapter Bugs
 5218: @cindex bug reporting
 5220: Known bugs are described in the file BUGS in the Gforth distribution.
 5222: If you find a bug, please send a bug report to
 5223: @email{}. A bug report should
 5224: describe the Gforth version used (it is announced at the start of an
 5225: interactive Gforth session), the machine and operating system (on Unix
 5226: systems you can use @code{uname -a} to produce this information), the
 5227: installation options (send the @file{config.status} file), and a
 5228: complete list of changes you (or your installer) have made to the Gforth
 5229: sources (if any); it should contain a program (or a sequence of keyboard
 5230: commands) that reproduces the bug and a description of what you think
 5231: constitutes the buggy behaviour.
 5233: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
 5234: to Report Bugs,, GNU C Manual}.
 5237: @node Origin, Word Index, Bugs, Top
 5238: @chapter Authors and Ancestors of Gforth
 5240: @section Authors and Contributors
 5241: @cindex authors of Gforth
 5242: @cindex contributors to Gforth
 5244: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
 5245: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
 5246: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
 5247: with their continuous feedback. Lennart Benshop contributed
 5248: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
 5249: support for calling C libraries. Helpful comments also came from Paul
 5250: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
 5251: Wavrik, Barrie Stott and Marc de Groot.
 5253: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
 5254: and autoconf, among others), and to the creators of the Internet: Gforth
 5255: was developed across the Internet, and its authors have not met
 5256: physically yet.
 5258: @section Pedigree
 5259: @cindex Pedigree of Gforth
 5261: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
 5262: Dirk Zoller) will cross-fertilize each other. Of course, a significant
 5263: part of the design of Gforth was prescribed by ANS Forth.
 5265: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
 5266: 32 bit native code version of VolksForth for the Atari ST, written
 5267: mostly by Dietrich Weineck.
 5269: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
 5270: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
 5271: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
 5273: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
 5274: Forth-83 standard. !! Pedigree? When?
 5276: A team led by Bill Ragsdale implemented fig-Forth on many processors in
 5277: 1979. Robert Selzer and Bill Ragsdale developed the original
 5278: implementation of fig-Forth for the 6502 based on microForth.
 5280: The principal architect of microForth was Dean Sanderson. microForth was
 5281: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
 5282: the 1802, and subsequently implemented on the 8080, the 6800 and the
 5283: Z80.
 5285: All earlier Forth systems were custom-made, usually by Charles Moore,
 5286: who discovered (as he puts it) Forth during the late 60s. The first full
 5287: Forth existed in 1971.
 5289: A part of the information in this section comes from @cite{The Evolution
 5290: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
 5291: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
 5292: Notices 28(3), 1993.  You can find more historical and genealogical
 5293: information about Forth there.
 5295: @node Word Index, Concept Index, Origin, Top
 5296: @unnumbered Word Index
 5298: This index is as incomplete as the manual. Each word is listed with
 5299: stack effect and wordset.
 5301: @printindex fn
 5303: @node Concept Index,  , Word Index, Top
 5304: @unnumbered Concept and Word Index
 5306: This index is as incomplete as the manual. Not all entries listed are
 5307: present verbatim in the text. Only the names are listed for the words
 5308: here.
 5310: @printindex cp
 5312: @contents
 5313: @bye

FreeBSD-CVSweb <>