File:  [gforth] / gforth / Attic / gforth.ds
Revision 1.22: download - view: text, annotated - select for diffs
Mon Nov 13 18:37:10 1995 UTC (24 years ago) by anton
Branches: MAIN
CVS tags: HEAD
some documentation changes
added make targets "bench" and "uninstall"

    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 gforth.info
    5: @settitle Gforth Manual
    6: @comment @setchapternewpage odd
    7: @comment %**end of header (This is for running Texinfo on a region.)
    8: 
    9: @ifinfo
   10: This file documents Gforth 0.1
   11: 
   12: Copyright @copyright{} 1995 Free Software Foundation, Inc.
   13: 
   14:      Permission is granted to make and distribute verbatim copies of
   15:      this manual provided the copyright notice and this permission notice
   16:      are preserved on all copies.
   17:      
   18: @ignore
   19:      Permission is granted to process this file through TeX and print the
   20:      results, provided the printed document carries a copying permission
   21:      notice identical to this one except for the removal of this paragraph
   22:      (this paragraph not being relevant to the printed manual).
   23:      
   24: @end ignore
   25:      Permission is granted to copy and distribute modified versions of this
   26:      manual under the conditions for verbatim copying, provided also that the
   27:      sections entitled "Distribution" and "General Public License" are
   28:      included exactly as in the original, and provided that the entire
   29:      resulting derived work is distributed under the terms of a permission
   30:      notice identical to this one.
   31:      
   32:      Permission is granted to copy and distribute translations of this manual
   33:      into another language, under the above conditions for modified versions,
   34:      except that the sections entitled "Distribution" and "General Public
   35:      License" may be included in a translation approved by the author instead
   36:      of in the original English.
   37: @end ifinfo
   38: 
   39: @titlepage
   40: @sp 10
   41: @center @titlefont{Gforth Manual}
   42: @sp 2
   43: @center for version 0.1
   44: @sp 2
   45: @center Anton Ertl
   46: @sp 3
   47: @center This manual is under construction
   48: 
   49: @comment  The following two commands start the copyright page.
   50: @page
   51: @vskip 0pt plus 1filll
   52: Copyright @copyright{} 1995 Free Software Foundation, Inc.
   53: 
   54: @comment !! Published by ... or You can get a copy of this manual ...
   55: 
   56:      Permission is granted to make and distribute verbatim copies of
   57:      this manual provided the copyright notice and this permission notice
   58:      are preserved on all copies.
   59:      
   60:      Permission is granted to copy and distribute modified versions of this
   61:      manual under the conditions for verbatim copying, provided also that the
   62:      sections entitled "Distribution" and "General Public License" are
   63:      included exactly as in the original, and provided that the entire
   64:      resulting derived work is distributed under the terms of a permission
   65:      notice identical to this one.
   66:      
   67:      Permission is granted to copy and distribute translations of this manual
   68:      into another language, under the above conditions for modified versions,
   69:      except that the sections entitled "Distribution" and "General Public
   70:      License" may be included in a translation approved by the author instead
   71:      of in the original English.
   72: @end titlepage
   73: 
   74: 
   75: @node Top, License, (dir), (dir)
   76: @ifinfo
   77: Gforth is a free implementation of ANS Forth available on many
   78: personal machines. This manual corresponds to version 0.1.
   79: @end ifinfo
   80: 
   81: @menu
   82: * License::                     
   83: * Goals::                       About the Gforth Project
   84: * Other Books::                 Things you might want to read
   85: * Invocation::                  Starting Gforth
   86: * Words::                       Forth words available in Gforth
   87: * ANS conformance::             Implementation-defined options etc.
   88: * Model::                       The abstract machine of Gforth
   89: * Emacs and Gforth::            The Gforth Mode
   90: * Internals::                   Implementation details
   91: * Bugs::                        How to report them
   92: * Pedigree::                    Ancestors of Gforth
   93: * Word Index::                  An item for each Forth word
   94: * Node Index::                  An item for each node
   95: @end menu
   96: 
   97: @node License, Goals, Top, Top
   98: @unnumbered GNU GENERAL PUBLIC LICENSE
   99: @center Version 2, June 1991
  100: 
  101: @display
  102: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
  103: 675 Mass Ave, Cambridge, MA 02139, USA
  104: 
  105: Everyone is permitted to copy and distribute verbatim copies
  106: of this license document, but changing it is not allowed.
  107: @end display
  108: 
  109: @unnumberedsec Preamble
  110: 
  111:   The licenses for most software are designed to take away your
  112: freedom to share and change it.  By contrast, the GNU General Public
  113: License is intended to guarantee your freedom to share and change free
  114: software---to make sure the software is free for all its users.  This
  115: General Public License applies to most of the Free Software
  116: Foundation's software and to any other program whose authors commit to
  117: using it.  (Some other Free Software Foundation software is covered by
  118: the GNU Library General Public License instead.)  You can apply it to
  119: your programs, too.
  120: 
  121:   When we speak of free software, we are referring to freedom, not
  122: price.  Our General Public Licenses are designed to make sure that you
  123: have the freedom to distribute copies of free software (and charge for
  124: this service if you wish), that you receive source code or can get it
  125: if you want it, that you can change the software or use pieces of it
  126: in new free programs; and that you know you can do these things.
  127: 
  128:   To protect your rights, we need to make restrictions that forbid
  129: anyone to deny you these rights or to ask you to surrender the rights.
  130: These restrictions translate to certain responsibilities for you if you
  131: distribute copies of the software, or if you modify it.
  132: 
  133:   For example, if you distribute copies of such a program, whether
  134: gratis or for a fee, you must give the recipients all the rights that
  135: you have.  You must make sure that they, too, receive or can get the
  136: source code.  And you must show them these terms so they know their
  137: rights.
  138: 
  139:   We protect your rights with two steps: (1) copyright the software, and
  140: (2) offer you this license which gives you legal permission to copy,
  141: distribute and/or modify the software.
  142: 
  143:   Also, for each author's protection and ours, we want to make certain
  144: that everyone understands that there is no warranty for this free
  145: software.  If the software is modified by someone else and passed on, we
  146: want its recipients to know that what they have is not the original, so
  147: that any problems introduced by others will not reflect on the original
  148: authors' reputations.
  149: 
  150:   Finally, any free program is threatened constantly by software
  151: patents.  We wish to avoid the danger that redistributors of a free
  152: program will individually obtain patent licenses, in effect making the
  153: program proprietary.  To prevent this, we have made it clear that any
  154: patent must be licensed for everyone's free use or not licensed at all.
  155: 
  156:   The precise terms and conditions for copying, distribution and
  157: modification follow.
  158: 
  159: @iftex
  160: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  161: @end iftex
  162: @ifinfo
  163: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  164: @end ifinfo
  165: 
  166: @enumerate 0
  167: @item
  168: This License applies to any program or other work which contains
  169: a notice placed by the copyright holder saying it may be distributed
  170: under the terms of this General Public License.  The ``Program'', below,
  171: refers to any such program or work, and a ``work based on the Program''
  172: means either the Program or any derivative work under copyright law:
  173: that is to say, a work containing the Program or a portion of it,
  174: either verbatim or with modifications and/or translated into another
  175: language.  (Hereinafter, translation is included without limitation in
  176: the term ``modification''.)  Each licensee is addressed as ``you''.
  177: 
  178: Activities other than copying, distribution and modification are not
  179: covered by this License; they are outside its scope.  The act of
  180: running the Program is not restricted, and the output from the Program
  181: is covered only if its contents constitute a work based on the
  182: Program (independent of having been made by running the Program).
  183: Whether that is true depends on what the Program does.
  184: 
  185: @item
  186: You may copy and distribute verbatim copies of the Program's
  187: source code as you receive it, in any medium, provided that you
  188: conspicuously and appropriately publish on each copy an appropriate
  189: copyright notice and disclaimer of warranty; keep intact all the
  190: notices that refer to this License and to the absence of any warranty;
  191: and give any other recipients of the Program a copy of this License
  192: along with the Program.
  193: 
  194: You may charge a fee for the physical act of transferring a copy, and
  195: you may at your option offer warranty protection in exchange for a fee.
  196: 
  197: @item
  198: You may modify your copy or copies of the Program or any portion
  199: of it, thus forming a work based on the Program, and copy and
  200: distribute such modifications or work under the terms of Section 1
  201: above, provided that you also meet all of these conditions:
  202: 
  203: @enumerate a
  204: @item
  205: You must cause the modified files to carry prominent notices
  206: stating that you changed the files and the date of any change.
  207: 
  208: @item
  209: You must cause any work that you distribute or publish, that in
  210: whole or in part contains or is derived from the Program or any
  211: part thereof, to be licensed as a whole at no charge to all third
  212: parties under the terms of this License.
  213: 
  214: @item
  215: If the modified program normally reads commands interactively
  216: when run, you must cause it, when started running for such
  217: interactive use in the most ordinary way, to print or display an
  218: announcement including an appropriate copyright notice and a
  219: notice that there is no warranty (or else, saying that you provide
  220: a warranty) and that users may redistribute the program under
  221: these conditions, and telling the user how to view a copy of this
  222: License.  (Exception: if the Program itself is interactive but
  223: does not normally print such an announcement, your work based on
  224: the Program is not required to print an announcement.)
  225: @end enumerate
  226: 
  227: These requirements apply to the modified work as a whole.  If
  228: identifiable sections of that work are not derived from the Program,
  229: and can be reasonably considered independent and separate works in
  230: themselves, then this License, and its terms, do not apply to those
  231: sections when you distribute them as separate works.  But when you
  232: distribute the same sections as part of a whole which is a work based
  233: on the Program, the distribution of the whole must be on the terms of
  234: this License, whose permissions for other licensees extend to the
  235: entire whole, and thus to each and every part regardless of who wrote it.
  236: 
  237: Thus, it is not the intent of this section to claim rights or contest
  238: your rights to work written entirely by you; rather, the intent is to
  239: exercise the right to control the distribution of derivative or
  240: collective works based on the Program.
  241: 
  242: In addition, mere aggregation of another work not based on the Program
  243: with the Program (or with a work based on the Program) on a volume of
  244: a storage or distribution medium does not bring the other work under
  245: the scope of this License.
  246: 
  247: @item
  248: You may copy and distribute the Program (or a work based on it,
  249: under Section 2) in object code or executable form under the terms of
  250: Sections 1 and 2 above provided that you also do one of the following:
  251: 
  252: @enumerate a
  253: @item
  254: Accompany it with the complete corresponding machine-readable
  255: source code, which must be distributed under the terms of Sections
  256: 1 and 2 above on a medium customarily used for software interchange; or,
  257: 
  258: @item
  259: Accompany it with a written offer, valid for at least three
  260: years, to give any third party, for a charge no more than your
  261: cost of physically performing source distribution, a complete
  262: machine-readable copy of the corresponding source code, to be
  263: distributed under the terms of Sections 1 and 2 above on a medium
  264: customarily used for software interchange; or,
  265: 
  266: @item
  267: Accompany it with the information you received as to the offer
  268: to distribute corresponding source code.  (This alternative is
  269: allowed only for noncommercial distribution and only if you
  270: received the program in object code or executable form with such
  271: an offer, in accord with Subsection b above.)
  272: @end enumerate
  273: 
  274: The source code for a work means the preferred form of the work for
  275: making modifications to it.  For an executable work, complete source
  276: code means all the source code for all modules it contains, plus any
  277: associated interface definition files, plus the scripts used to
  278: control compilation and installation of the executable.  However, as a
  279: special exception, the source code distributed need not include
  280: anything that is normally distributed (in either source or binary
  281: form) with the major components (compiler, kernel, and so on) of the
  282: operating system on which the executable runs, unless that component
  283: itself accompanies the executable.
  284: 
  285: If distribution of executable or object code is made by offering
  286: access to copy from a designated place, then offering equivalent
  287: access to copy the source code from the same place counts as
  288: distribution of the source code, even though third parties are not
  289: compelled to copy the source along with the object code.
  290: 
  291: @item
  292: You may not copy, modify, sublicense, or distribute the Program
  293: except as expressly provided under this License.  Any attempt
  294: otherwise to copy, modify, sublicense or distribute the Program is
  295: void, and will automatically terminate your rights under this License.
  296: However, parties who have received copies, or rights, from you under
  297: this License will not have their licenses terminated so long as such
  298: parties remain in full compliance.
  299: 
  300: @item
  301: You are not required to accept this License, since you have not
  302: signed it.  However, nothing else grants you permission to modify or
  303: distribute the Program or its derivative works.  These actions are
  304: prohibited by law if you do not accept this License.  Therefore, by
  305: modifying or distributing the Program (or any work based on the
  306: Program), you indicate your acceptance of this License to do so, and
  307: all its terms and conditions for copying, distributing or modifying
  308: the Program or works based on it.
  309: 
  310: @item
  311: Each time you redistribute the Program (or any work based on the
  312: Program), the recipient automatically receives a license from the
  313: original licensor to copy, distribute or modify the Program subject to
  314: these terms and conditions.  You may not impose any further
  315: restrictions on the recipients' exercise of the rights granted herein.
  316: You are not responsible for enforcing compliance by third parties to
  317: this License.
  318: 
  319: @item
  320: If, as a consequence of a court judgment or allegation of patent
  321: infringement or for any other reason (not limited to patent issues),
  322: conditions are imposed on you (whether by court order, agreement or
  323: otherwise) that contradict the conditions of this License, they do not
  324: excuse you from the conditions of this License.  If you cannot
  325: distribute so as to satisfy simultaneously your obligations under this
  326: License and any other pertinent obligations, then as a consequence you
  327: may not distribute the Program at all.  For example, if a patent
  328: license would not permit royalty-free redistribution of the Program by
  329: all those who receive copies directly or indirectly through you, then
  330: the only way you could satisfy both it and this License would be to
  331: refrain entirely from distribution of the Program.
  332: 
  333: If any portion of this section is held invalid or unenforceable under
  334: any particular circumstance, the balance of the section is intended to
  335: apply and the section as a whole is intended to apply in other
  336: circumstances.
  337: 
  338: It is not the purpose of this section to induce you to infringe any
  339: patents or other property right claims or to contest validity of any
  340: such claims; this section has the sole purpose of protecting the
  341: integrity of the free software distribution system, which is
  342: implemented by public license practices.  Many people have made
  343: generous contributions to the wide range of software distributed
  344: through that system in reliance on consistent application of that
  345: system; it is up to the author/donor to decide if he or she is willing
  346: to distribute software through any other system and a licensee cannot
  347: impose that choice.
  348: 
  349: This section is intended to make thoroughly clear what is believed to
  350: be a consequence of the rest of this License.
  351: 
  352: @item
  353: If the distribution and/or use of the Program is restricted in
  354: certain countries either by patents or by copyrighted interfaces, the
  355: original copyright holder who places the Program under this License
  356: may add an explicit geographical distribution limitation excluding
  357: those countries, so that distribution is permitted only in or among
  358: countries not thus excluded.  In such case, this License incorporates
  359: the limitation as if written in the body of this License.
  360: 
  361: @item
  362: The Free Software Foundation may publish revised and/or new versions
  363: of the General Public License from time to time.  Such new versions will
  364: be similar in spirit to the present version, but may differ in detail to
  365: address new problems or concerns.
  366: 
  367: Each version is given a distinguishing version number.  If the Program
  368: specifies a version number of this License which applies to it and ``any
  369: later version'', you have the option of following the terms and conditions
  370: either of that version or of any later version published by the Free
  371: Software Foundation.  If the Program does not specify a version number of
  372: this License, you may choose any version ever published by the Free Software
  373: Foundation.
  374: 
  375: @item
  376: If you wish to incorporate parts of the Program into other free
  377: programs whose distribution conditions are different, write to the author
  378: to ask for permission.  For software which is copyrighted by the Free
  379: Software Foundation, write to the Free Software Foundation; we sometimes
  380: make exceptions for this.  Our decision will be guided by the two goals
  381: of preserving the free status of all derivatives of our free software and
  382: of promoting the sharing and reuse of software generally.
  383: 
  384: @iftex
  385: @heading NO WARRANTY
  386: @end iftex
  387: @ifinfo
  388: @center NO WARRANTY
  389: @end ifinfo
  390: 
  391: @item
  392: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
  393: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
  394: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
  395: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
  396: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  397: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
  398: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
  399: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
  400: REPAIR OR CORRECTION.
  401: 
  402: @item
  403: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
  404: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
  405: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
  406: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
  407: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
  408: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
  409: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
  410: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
  411: POSSIBILITY OF SUCH DAMAGES.
  412: @end enumerate
  413: 
  414: @iftex
  415: @heading END OF TERMS AND CONDITIONS
  416: @end iftex
  417: @ifinfo
  418: @center END OF TERMS AND CONDITIONS
  419: @end ifinfo
  420: 
  421: @page
  422: @unnumberedsec How to Apply These Terms to Your New Programs
  423: 
  424:   If you develop a new program, and you want it to be of the greatest
  425: possible use to the public, the best way to achieve this is to make it
  426: free software which everyone can redistribute and change under these terms.
  427: 
  428:   To do so, attach the following notices to the program.  It is safest
  429: to attach them to the start of each source file to most effectively
  430: convey the exclusion of warranty; and each file should have at least
  431: the ``copyright'' line and a pointer to where the full notice is found.
  432: 
  433: @smallexample
  434: @var{one line to give the program's name and a brief idea of what it does.}
  435: Copyright (C) 19@var{yy}  @var{name of author}
  436: 
  437: This program is free software; you can redistribute it and/or modify 
  438: it under the terms of the GNU General Public License as published by 
  439: the Free Software Foundation; either version 2 of the License, or 
  440: (at your option) any later version.
  441: 
  442: This program is distributed in the hope that it will be useful,
  443: but WITHOUT ANY WARRANTY; without even the implied warranty of
  444: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  445: GNU General Public License for more details.
  446: 
  447: You should have received a copy of the GNU General Public License
  448: along with this program; if not, write to the Free Software
  449: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  450: @end smallexample
  451: 
  452: Also add information on how to contact you by electronic and paper mail.
  453: 
  454: If the program is interactive, make it output a short notice like this
  455: when it starts in an interactive mode:
  456: 
  457: @smallexample
  458: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
  459: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
  460: type `show w'.  
  461: This is free software, and you are welcome to redistribute it 
  462: under certain conditions; type `show c' for details.
  463: @end smallexample
  464: 
  465: The hypothetical commands @samp{show w} and @samp{show c} should show
  466: the appropriate parts of the General Public License.  Of course, the
  467: commands you use may be called something other than @samp{show w} and
  468: @samp{show c}; they could even be mouse-clicks or menu items---whatever
  469: suits your program.
  470: 
  471: You should also get your employer (if you work as a programmer) or your
  472: school, if any, to sign a ``copyright disclaimer'' for the program, if
  473: necessary.  Here is a sample; alter the names:
  474: 
  475: @smallexample
  476: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
  477: `Gnomovision' (which makes passes at compilers) written by James Hacker.
  478: 
  479: @var{signature of Ty Coon}, 1 April 1989
  480: Ty Coon, President of Vice
  481: @end smallexample
  482: 
  483: This General Public License does not permit incorporating your program into
  484: proprietary programs.  If your program is a subroutine library, you may
  485: consider it more useful to permit linking proprietary applications with the
  486: library.  If this is what you want to do, use the GNU Library General
  487: Public License instead of this License.
  488: 
  489: @iftex
  490: @unnumbered Preface
  491: This manual documents Gforth. The reader is expected to know
  492: Forth. This manual is primarily a reference manual. @xref{Other Books}
  493: for introductory material.
  494: @end iftex
  495: 
  496: @node    Goals, Other Books, License, Top
  497: @comment node-name,     next,           previous, up
  498: @chapter Goals of Gforth
  499: @cindex Goals
  500: The goal of the Gforth Project is to develop a standard model for
  501: ANSI Forth. This can be split into several subgoals:
  502: 
  503: @itemize @bullet
  504: @item
  505: Gforth should conform to the ANSI Forth standard.
  506: @item
  507: It should be a model, i.e. it should define all the
  508: implementation-dependent things.
  509: @item
  510: It should become standard, i.e. widely accepted and used. This goal
  511: is the most difficult one.
  512: @end itemize
  513: 
  514: To achieve these goals Gforth should be
  515: @itemize @bullet
  516: @item
  517: Similar to previous models (fig-Forth, F83)
  518: @item
  519: Powerful. It should provide for all the things that are considered
  520: necessary today and even some that are not yet considered necessary.
  521: @item
  522: Efficient. It should not get the reputation of being exceptionally
  523: slow.
  524: @item
  525: Free.
  526: @item
  527: Available on many machines/easy to port.
  528: @end itemize
  529: 
  530: Have we achieved these goals? Gforth conforms to the ANS Forth
  531: standard. It may be considered a model, but we have not yet documented
  532: which parts of the model are stable and which parts we are likely to
  533: change. It certainly has not yet become a de facto standard. It has some
  534: similarities and some differences to previous models. It has some
  535: powerful features, but not yet everything that we envisioned. We
  536: certainly have achieved our execution speed goals (@pxref{Performance}).
  537: It is free and available on many machines.
  538: 
  539: @node Other Books, Invocation, Goals, Top
  540: @chapter Other books on ANS Forth
  541: 
  542: As the standard is relatively new, there are not many books out yet. It
  543: is not recommended to learn Forth by using Gforth and a book that is
  544: not written for ANS Forth, as you will not know your mistakes from the
  545: deviations of the book.
  546: 
  547: There is, of course, the standard, the definite reference if you want to
  548: write ANS Forth programs. It is available in printed form from the
  549: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
  550: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
  551: can also get it from Global Engineering Documents (Tel.: USA (800)
  552: 854-7179; Fax.: (303) 843-9880) for about $300.
  553: 
  554: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
  555: for publication is available electronically and for free in some MS Word
  556: format, and it has been converted to HTML. Some pointers to these
  557: versions can be found through
  558: http://www.complang.tuwien.ac.at/projects/forth.html.
  559: 
  560: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
  561: introductory book based on a draft version of the standard. It does not
  562: cover the whole standard. It also contains interesting background
  563: information (Jack Woehr was in the ANS Forth Technical Committe). It is
  564: not appropriate for complete newbies, but programmers experienced in
  565: other languages should find it ok.
  566: 
  567: @node Invocation, Words, Other Books, Top
  568: @chapter Invocation
  569: 
  570: You will usually just say @code{gforth}. In many other cases the default
  571: Gforth image will be invoked like this:
  572: 
  573: @example
  574: gforth [files] [-e forth-code]
  575: @end example
  576: 
  577: executing the contents of the files and the Forth code in the order they
  578: are given.
  579: 
  580: In general, the command line looks like this:
  581: 
  582: @example
  583: gforth [initialization options] [image-specific options]
  584: @end example
  585: 
  586: The initialization options must come before the rest of the command
  587: line. They are:
  588: 
  589: @table @code
  590: @item --image-file @var{file}
  591: @item -i @var{file}
  592: Loads the Forth image @var{file} instead of the default
  593: @file{gforth.fi}.
  594: 
  595: @item --path @var{path}
  596: @item -p @var{path}
  597: Uses @var{path} for searching the image file and Forth source code
  598: files instead of the default in the environment variable
  599: @code{GFORTHPATH} or the path specified at installation time (typically
  600: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
  601: list.
  602: 
  603: @item --dictionary-size @var{size}
  604: @item -m @var{size}
  605: Allocate @var{size} space for the Forth dictionary space instead of
  606: using the default specified in the image (typically 256K). The
  607: @var{size} specification consists of an integer and a unit (e.g.,
  608: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
  609: size, in this case Cells), @code{k} (kilobytes), and @code{M}
  610: (Megabytes). If no unit is specified, @code{e} is used.
  611: 
  612: @item --data-stack-size @var{size}
  613: @item -d @var{size}
  614: Allocate @var{size} space for the data stack instead of using the
  615: default specified in the image (typically 16K).
  616: 
  617: @item --return-stack-size @var{size}
  618: @item -r @var{size}
  619: Allocate @var{size} space for the return stack instead of using the
  620: default specified in the image (typically 16K).
  621: 
  622: @item --fp-stack-size @var{size}
  623: @item -f @var{size}
  624: Allocate @var{size} space for the floating point stack instead of
  625: using the default specified in the image (typically 16K). In this case
  626: the unit specifier @code{e} refers to floating point numbers.
  627: 
  628: @item --locals-stack-size @var{size}
  629: @item -l @var{size}
  630: Allocate @var{size} space for the locals stack instead of using the
  631: default specified in the image (typically 16K).
  632: 
  633: @end table
  634: 
  635: As explained above, the image-specific command-line arguments for the
  636: default image @file{gforth.fi} consist of a sequence of filenames and
  637: @code{-e @var{forth-code}} options that are interpreted in the seqence
  638: in which they are given. The @code{-e @var{forth-code}} or
  639: @code{--evaluate @var{forth-code}} option evaluates the forth
  640: code. This option takes only one argument; if you want to evaluate more
  641: Forth words, you have to quote them or use several @code{-e}s. To exit
  642: after processing the command line (instead of entering interactive mode)
  643: append @code{-e bye} to the command line.
  644: 
  645: If you have several versions of Gforth installed, @code{gforth} will
  646: invoke the version that was installed last. @code{gforth-@var{version}}
  647: invokes a specific version. You may want to use the option
  648: @code{--path}, if your environment contains the variable
  649: @code{GFORTHPATH}.
  650: 
  651: Not yet implemented:
  652: On startup the system first executes the system initialization file
  653: (unless the option @code{--no-init-file} is given; note that the system
  654: resulting from using this option may not be ANS Forth conformant). Then
  655: the user initialization file @file{.gforth.fs} is executed, unless the
  656: option @code{--no-rc} is given; this file is first searched in @file{.},
  657: then in @file{~}, then in the normal path (see above).
  658: 
  659: @node Words, ANS conformance, Invocation, Top
  660: @chapter Forth Words
  661: 
  662: @menu
  663: * Notation::                    
  664: * Arithmetic::                  
  665: * Stack Manipulation::          
  666: * Memory access::               
  667: * Control Structures::          
  668: * Locals::                      
  669: * Defining Words::              
  670: * Wordlists::                   
  671: * Files::                       
  672: * Blocks::                      
  673: * Other I/O::                   
  674: * Programming Tools::           
  675: * Assembler and Code words::    
  676: * Threading Words::             
  677: @end menu
  678: 
  679: @node Notation, Arithmetic, Words, Words
  680: @section Notation
  681: 
  682: The Forth words are described in this section in the glossary notation
  683: that has become a de-facto standard for Forth texts, i.e.
  684: 
  685: @format
  686: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
  687: @end format
  688: @var{Description}
  689: 
  690: @table @var
  691: @item word
  692: The name of the word. BTW, Gforth is case insensitive, so you can
  693: type the words in in lower case (However, @pxref{core-idef}).
  694: 
  695: @item Stack effect
  696: The stack effect is written in the notation @code{@var{before} --
  697: @var{after}}, where @var{before} and @var{after} describe the top of
  698: stack entries before and after the execution of the word. The rest of
  699: the stack is not touched by the word. The top of stack is rightmost,
  700: i.e., a stack sequence is written as it is typed in. Note that Gforth
  701: uses a separate floating point stack, but a unified stack
  702: notation. Also, return stack effects are not shown in @var{stack
  703: effect}, but in @var{Description}. The name of a stack item describes
  704: the type and/or the function of the item. See below for a discussion of
  705: the types.
  706: 
  707: All words have two stack effects: A compile-time stack effect and a
  708: run-time stack effect. The compile-time stack-effect of most words is
  709: @var{ -- }. If the compile-time stack-effect of a word deviates from
  710: this standard behaviour, or the word does other unusual things at
  711: compile time, both stack effects are shown; otherwise only the run-time
  712: stack effect is shown.
  713: 
  714: @item pronunciation
  715: How the word is pronounced
  716: 
  717: @item wordset
  718: The ANS Forth standard is divided into several wordsets. A standard
  719: system need not support all of them. So, the fewer wordsets your program
  720: uses the more portable it will be in theory. However, we suspect that
  721: most ANS Forth systems on personal machines will feature all
  722: wordsets. Words that are not defined in the ANS standard have
  723: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
  724: describes words that will work in future releases of Gforth;
  725: @code{gforth-internal} words are more volatile. Environmental query
  726: strings are also displayed like words; you can recognize them by the
  727: @code{environment} in the wordset field.
  728: 
  729: @item Description
  730: A description of the behaviour of the word.
  731: @end table
  732: 
  733: The type of a stack item is specified by the character(s) the name
  734: starts with:
  735: 
  736: @table @code
  737: @item f
  738: Bool, i.e. @code{false} or @code{true}.
  739: @item c
  740: Char
  741: @item w
  742: Cell, can contain an integer or an address
  743: @item n
  744: signed integer
  745: @item u
  746: unsigned integer
  747: @item d
  748: double sized signed integer
  749: @item ud
  750: double sized unsigned integer
  751: @item r
  752: Float
  753: @item a_
  754: Cell-aligned address
  755: @item c_
  756: Char-aligned address (note that a Char is two bytes in Windows NT)
  757: @item f_
  758: Float-aligned address
  759: @item df_
  760: Address aligned for IEEE double precision float
  761: @item sf_
  762: Address aligned for IEEE single precision float
  763: @item xt
  764: Execution token, same size as Cell
  765: @item wid
  766: Wordlist ID, same size as Cell
  767: @item f83name
  768: Pointer to a name structure
  769: @end table
  770: 
  771: @node Arithmetic, Stack Manipulation, Notation, Words
  772: @section Arithmetic
  773: Forth arithmetic is not checked, i.e., you will not hear about integer
  774: overflow on addition or multiplication, you may hear about division by
  775: zero if you are lucky. The operator is written after the operands, but
  776: the operands are still in the original order. I.e., the infix @code{2-1}
  777: corresponds to @code{2 1 -}. Forth offers a variety of division
  778: operators. If you perform division with potentially negative operands,
  779: you do not want to use @code{/} or @code{/mod} with its undefined
  780: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
  781: former, @pxref{Mixed precision}).
  782: 
  783: @menu
  784: * Single precision::            
  785: * Bitwise operations::          
  786: * Mixed precision::             operations with single and double-cell integers
  787: * Double precision::            Double-cell integer arithmetic
  788: * Floating Point::              
  789: @end menu
  790: 
  791: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
  792: @subsection Single precision
  793: doc-+
  794: doc--
  795: doc-*
  796: doc-/
  797: doc-mod
  798: doc-/mod
  799: doc-negate
  800: doc-abs
  801: doc-min
  802: doc-max
  803: 
  804: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
  805: @subsection Bitwise operations
  806: doc-and
  807: doc-or
  808: doc-xor
  809: doc-invert
  810: doc-2*
  811: doc-2/
  812: 
  813: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
  814: @subsection Mixed precision
  815: doc-m+
  816: doc-*/
  817: doc-*/mod
  818: doc-m*
  819: doc-um*
  820: doc-m*/
  821: doc-um/mod
  822: doc-fm/mod
  823: doc-sm/rem
  824: 
  825: @node Double precision, Floating Point, Mixed precision, Arithmetic
  826: @subsection Double precision
  827: 
  828: The outer (aka text) interpreter converts numbers containing a dot into
  829: a double precision number. Note that only numbers with the dot as last
  830: character are standard-conforming.
  831: 
  832: doc-d+
  833: doc-d-
  834: doc-dnegate
  835: doc-dabs
  836: doc-dmin
  837: doc-dmax
  838: 
  839: @node Floating Point,  , Double precision, Arithmetic
  840: @subsection Floating Point
  841: 
  842: The format of floating point numbers recognized by the outer (aka text)
  843: interpreter is: a signed decimal number, possibly containing a decimal
  844: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
  845: by a signed integer (the exponent). E.g., @code{1e} ist the same as
  846: @code{+1.0e+1}. Note that a number without @code{e}
  847: is not interpreted as floating-point number, but as double (if the
  848: number contains a @code{.}) or single precision integer. Also,
  849: conversions between string and floating point numbers always use base
  850: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
  851: value greater then 14, the @code{E} may be interpreted as digit and the
  852: number will be interpreted as integer, unless it has a signed exponent
  853: (both @code{+} and @code{-} are allowed as signs).
  854: 
  855: Angles in floating point operations are given in radians (a full circle
  856: has 2 pi radians). Note, that Gforth has a separate floating point
  857: stack, but we use the unified notation.
  858: 
  859: Floating point numbers have a number of unpleasant surprises for the
  860: unwary (e.g., floating point addition is not associative) and even a few
  861: for the wary. You should not use them unless you know what you are doing
  862: or you don't care that the results you get are totally bogus. If you
  863: want to learn about the problems of floating point numbers (and how to
  864: avoid them), you might start with @cite{David Goldberg, What Every
  865: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
  866: Computing Surveys 23(1):5@minus{}48, March 1991}.
  867: 
  868: doc-f+
  869: doc-f-
  870: doc-f*
  871: doc-f/
  872: doc-fnegate
  873: doc-fabs
  874: doc-fmax
  875: doc-fmin
  876: doc-floor
  877: doc-fround
  878: doc-f**
  879: doc-fsqrt
  880: doc-fexp
  881: doc-fexpm1
  882: doc-fln
  883: doc-flnp1
  884: doc-flog
  885: doc-falog
  886: doc-fsin
  887: doc-fcos
  888: doc-fsincos
  889: doc-ftan
  890: doc-fasin
  891: doc-facos
  892: doc-fatan
  893: doc-fatan2
  894: doc-fsinh
  895: doc-fcosh
  896: doc-ftanh
  897: doc-fasinh
  898: doc-facosh
  899: doc-fatanh
  900: 
  901: @node Stack Manipulation, Memory access, Arithmetic, Words
  902: @section Stack Manipulation
  903: 
  904: Gforth has a data stack (aka parameter stack) for characters, cells,
  905: addresses, and double cells, a floating point stack for floating point
  906: numbers, a return stack for storing the return addresses of colon
  907: definitions and other data, and a locals stack for storing local
  908: variables. Note that while every sane Forth has a separate floating
  909: point stack, this is not strictly required; an ANS Forth system could
  910: theoretically keep floating point numbers on the data stack. As an
  911: additional difficulty, you don't know how many cells a floating point
  912: number takes. It is reportedly possible to write words in a way that
  913: they work also for a unified stack model, but we do not recommend trying
  914: it. Instead, just say that your program has an environmental dependency
  915: on a separate FP stack.
  916: 
  917: Also, a Forth system is allowed to keep the local variables on the
  918: return stack. This is reasonable, as local variables usually eliminate
  919: the need to use the return stack explicitly. So, if you want to produce
  920: a standard complying program and if you are using local variables in a
  921: word, forget about return stack manipulations in that word (see the
  922: standard document for the exact rules).
  923: 
  924: @menu
  925: * Data stack::                  
  926: * Floating point stack::        
  927: * Return stack::                
  928: * Locals stack::                
  929: * Stack pointer manipulation::  
  930: @end menu
  931: 
  932: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
  933: @subsection Data stack
  934: doc-drop
  935: doc-nip
  936: doc-dup
  937: doc-over
  938: doc-tuck
  939: doc-swap
  940: doc-rot
  941: doc--rot
  942: doc-?dup
  943: doc-pick
  944: doc-roll
  945: doc-2drop
  946: doc-2nip
  947: doc-2dup
  948: doc-2over
  949: doc-2tuck
  950: doc-2swap
  951: doc-2rot
  952: 
  953: @node Floating point stack, Return stack, Data stack, Stack Manipulation
  954: @subsection Floating point stack
  955: doc-fdrop
  956: doc-fnip
  957: doc-fdup
  958: doc-fover
  959: doc-ftuck
  960: doc-fswap
  961: doc-frot
  962: 
  963: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
  964: @subsection Return stack
  965: doc->r
  966: doc-r>
  967: doc-r@
  968: doc-rdrop
  969: doc-2>r
  970: doc-2r>
  971: doc-2r@
  972: doc-2rdrop
  973: 
  974: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
  975: @subsection Locals stack
  976: 
  977: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
  978: @subsection Stack pointer manipulation
  979: doc-sp@
  980: doc-sp!
  981: doc-fp@
  982: doc-fp!
  983: doc-rp@
  984: doc-rp!
  985: doc-lp@
  986: doc-lp!
  987: 
  988: @node Memory access, Control Structures, Stack Manipulation, Words
  989: @section Memory access
  990: 
  991: @menu
  992: * Stack-Memory transfers::      
  993: * Address arithmetic::          
  994: * Memory block access::         
  995: @end menu
  996: 
  997: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
  998: @subsection Stack-Memory transfers
  999: 
 1000: doc-@
 1001: doc-!
 1002: doc-+!
 1003: doc-c@
 1004: doc-c!
 1005: doc-2@
 1006: doc-2!
 1007: doc-f@
 1008: doc-f!
 1009: doc-sf@
 1010: doc-sf!
 1011: doc-df@
 1012: doc-df!
 1013: 
 1014: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
 1015: @subsection Address arithmetic
 1016: 
 1017: ANS Forth does not specify the sizes of the data types. Instead, it
 1018: offers a number of words for computing sizes and doing address
 1019: arithmetic. Basically, address arithmetic is performed in terms of
 1020: address units (aus); on most systems the address unit is one byte. Note
 1021: that a character may have more than one au, so @code{chars} is no noop
 1022: (on systems where it is a noop, it compiles to nothing).
 1023: 
 1024: ANS Forth also defines words for aligning addresses for specific
 1025: addresses. Many computers require that accesses to specific data types
 1026: must only occur at specific addresses; e.g., that cells may only be
 1027: accessed at addresses divisible by 4. Even if a machine allows unaligned
 1028: accesses, it can usually perform aligned accesses faster. 
 1029: 
 1030: For the performance-conscious: alignment operations are usually only
 1031: necessary during the definition of a data structure, not during the
 1032: (more frequent) accesses to it.
 1033: 
 1034: ANS Forth defines no words for character-aligning addresses. This is not
 1035: an oversight, but reflects the fact that addresses that are not
 1036: char-aligned have no use in the standard and therefore will not be
 1037: created.
 1038: 
 1039: The standard guarantees that addresses returned by @code{CREATE}d words
 1040: are cell-aligned; in addition, Gforth guarantees that these addresses
 1041: are aligned for all purposes.
 1042: 
 1043: Note that the standard defines a word @code{char}, which has nothing to
 1044: do with address arithmetic.
 1045: 
 1046: doc-chars
 1047: doc-char+
 1048: doc-cells
 1049: doc-cell+
 1050: doc-align
 1051: doc-aligned
 1052: doc-floats
 1053: doc-float+
 1054: doc-falign
 1055: doc-faligned
 1056: doc-sfloats
 1057: doc-sfloat+
 1058: doc-sfalign
 1059: doc-sfaligned
 1060: doc-dfloats
 1061: doc-dfloat+
 1062: doc-dfalign
 1063: doc-dfaligned
 1064: doc-maxalign
 1065: doc-maxaligned
 1066: doc-cfalign
 1067: doc-cfaligned
 1068: doc-address-unit-bits
 1069: 
 1070: @node Memory block access,  , Address arithmetic, Memory access
 1071: @subsection Memory block access
 1072: 
 1073: doc-move
 1074: doc-erase
 1075: 
 1076: While the previous words work on address units, the rest works on
 1077: characters.
 1078: 
 1079: doc-cmove
 1080: doc-cmove>
 1081: doc-fill
 1082: doc-blank
 1083: 
 1084: @node Control Structures, Locals, Memory access, Words
 1085: @section Control Structures
 1086: 
 1087: Control structures in Forth cannot be used in interpret state, only in
 1088: compile state, i.e., in a colon definition. We do not like this
 1089: limitation, but have not seen a satisfying way around it yet, although
 1090: many schemes have been proposed.
 1091: 
 1092: @menu
 1093: * Selection::                   
 1094: * Simple Loops::                
 1095: * Counted Loops::               
 1096: * Arbitrary control structures::  
 1097: * Calls and returns::           
 1098: * Exception Handling::          
 1099: @end menu
 1100: 
 1101: @node Selection, Simple Loops, Control Structures, Control Structures
 1102: @subsection Selection
 1103: 
 1104: @example
 1105: @var{flag}
 1106: IF
 1107:   @var{code}
 1108: ENDIF
 1109: @end example
 1110: or
 1111: @example
 1112: @var{flag}
 1113: IF
 1114:   @var{code1}
 1115: ELSE
 1116:   @var{code2}
 1117: ENDIF
 1118: @end example
 1119: 
 1120: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
 1121: standard, and @code{ENDIF} is not, although it is quite popular. We
 1122: recommend using @code{ENDIF}, because it is less confusing for people
 1123: who also know other languages (and is not prone to reinforcing negative
 1124: prejudices against Forth in these people). Adding @code{ENDIF} to a
 1125: system that only supplies @code{THEN} is simple:
 1126: @example
 1127: : endif   POSTPONE then ; immediate
 1128: @end example
 1129: 
 1130: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
 1131: (adv.)}  has the following meanings:
 1132: @quotation
 1133: ... 2b: following next after in order ... 3d: as a necessary consequence
 1134: (if you were there, then you saw them).
 1135: @end quotation
 1136: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
 1137: and many other programming languages has the meaning 3d.]
 1138: 
 1139: We also provide the words @code{?dup-if} and @code{?dup-0=-if}, so you
 1140: can avoid using @code{?dup}.
 1141: 
 1142: @example
 1143: @var{n}
 1144: CASE
 1145:   @var{n1} OF @var{code1} ENDOF
 1146:   @var{n2} OF @var{code2} ENDOF
 1147:   @dots{}
 1148: ENDCASE
 1149: @end example
 1150: 
 1151: Executes the first @var{codei}, where the @var{ni} is equal to
 1152: @var{n}. A default case can be added by simply writing the code after
 1153: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
 1154: but must not consume it.
 1155: 
 1156: @node Simple Loops, Counted Loops, Selection, Control Structures
 1157: @subsection Simple Loops
 1158: 
 1159: @example
 1160: BEGIN
 1161:   @var{code1}
 1162:   @var{flag}
 1163: WHILE
 1164:   @var{code2}
 1165: REPEAT
 1166: @end example
 1167: 
 1168: @var{code1} is executed and @var{flag} is computed. If it is true,
 1169: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
 1170: 
 1171: @example
 1172: BEGIN
 1173:   @var{code}
 1174:   @var{flag}
 1175: UNTIL
 1176: @end example
 1177: 
 1178: @var{code} is executed. The loop is restarted if @code{flag} is false.
 1179: 
 1180: @example
 1181: BEGIN
 1182:   @var{code}
 1183: AGAIN
 1184: @end example
 1185: 
 1186: This is an endless loop.
 1187: 
 1188: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
 1189: @subsection Counted Loops
 1190: 
 1191: The basic counted loop is:
 1192: @example
 1193: @var{limit} @var{start}
 1194: ?DO
 1195:   @var{body}
 1196: LOOP
 1197: @end example
 1198: 
 1199: This performs one iteration for every integer, starting from @var{start}
 1200: and up to, but excluding @var{limit}. The counter, aka index, can be
 1201: accessed with @code{i}. E.g., the loop
 1202: @example
 1203: 10 0 ?DO
 1204:   i .
 1205: LOOP
 1206: @end example
 1207: prints
 1208: @example
 1209: 0 1 2 3 4 5 6 7 8 9
 1210: @end example
 1211: The index of the innermost loop can be accessed with @code{i}, the index
 1212: of the next loop with @code{j}, and the index of the third loop with
 1213: @code{k}.
 1214: 
 1215: The loop control data are kept on the return stack, so there are some
 1216: restrictions on mixing return stack accesses and counted loop
 1217: words. E.g., if you put values on the return stack outside the loop, you
 1218: cannot read them inside the loop. If you put values on the return stack
 1219: within a loop, you have to remove them before the end of the loop and
 1220: before accessing the index of the loop.
 1221: 
 1222: There are several variations on the counted loop:
 1223: 
 1224: @code{LEAVE} leaves the innermost counted loop immediately.
 1225: 
 1226: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
 1227: (and @code{LOOP} iterates until they become equal by wrap-around
 1228: arithmetic). This behaviour is usually not what you want. Therefore,
 1229: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
 1230: @code{?DO}), which do not enter the loop if @var{start} is greater than
 1231: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
 1232: unsigned loop parameters. These words can be implemented easily on
 1233: standard systems, so using them does not make your programs hard to
 1234: port; e.g.:
 1235: @example
 1236: : +DO ( compile-time: -- do-sys; run-time: n1 n2 -- )
 1237:     POSTPONE over POSTPONE min POSTPONE ?DO ; immediate
 1238: @end example
 1239: 
 1240: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
 1241: index by @var{n} instead of by 1. The loop is terminated when the border
 1242: between @var{limit-1} and @var{limit} is crossed. E.g.:
 1243: 
 1244: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
 1245: 
 1246: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
 1247: 
 1248: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
 1249: 
 1250: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
 1251: 
 1252: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
 1253: 
 1254: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
 1255: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
 1256: index by @var{u} each iteration. The loop is terminated when the border
 1257: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
 1258: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
 1259: 
 1260: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
 1261: 
 1262: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
 1263: 
 1264: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
 1265: 
 1266: Another alternative is @code{@var{n} S+LOOP}, where the negative
 1267: case behaves symmetrical to the positive case:
 1268: 
 1269: @code{-2 0 -DO  i .  -1 S+LOOP}  prints @code{0 -1}
 1270: 
 1271: The loop is terminated when the border between @var{limit@minus{}sgn(n)}
 1272: and @var{limit} is crossed. Unfortunately, neither @code{-LOOP} nor
 1273: @code{S+LOOP} are part of the ANS Forth standard, and they are not easy
 1274: to implement using standard words. If you want to write standard
 1275: programs, just avoid counting down.
 1276: 
 1277: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
 1278: the loop, independent of the loop parameters. Do not use @code{DO}, even
 1279: if you know that the loop is entered in any case. Such knowledge tends
 1280: to become invalid during maintenance of a program, and then the
 1281: @code{DO} will make trouble.
 1282: 
 1283: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
 1284: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
 1285: return stack so @code{EXIT} can get to its return address.
 1286: 
 1287: Another counted loop is
 1288: @example
 1289: @var{n}
 1290: FOR
 1291:   @var{body}
 1292: NEXT
 1293: @end example
 1294: This is the preferred loop of native code compiler writers who are too
 1295: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
 1296: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
 1297: and ending with 0. Other Forth systems may behave differently, even if
 1298: they support @code{FOR} loops.
 1299: 
 1300: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
 1301: @subsection Arbitrary control structures
 1302: 
 1303: ANS Forth permits and supports using control structures in a non-nested
 1304: way. Information about incomplete control structures is stored on the
 1305: control-flow stack. This stack may be implemented on the Forth data
 1306: stack, and this is what we have done in Gforth.
 1307: 
 1308: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
 1309: entry represents a backward branch target. A few words are the basis for
 1310: building any control structure possible (except control structures that
 1311: need storage, like calls, coroutines, and backtracking).
 1312: 
 1313: doc-if
 1314: doc-ahead
 1315: doc-then
 1316: doc-begin
 1317: doc-until
 1318: doc-again
 1319: doc-cs-pick
 1320: doc-cs-roll
 1321: 
 1322: On many systems control-flow stack items take one word, in Gforth they
 1323: currently take three (this may change in the future). Therefore it is a
 1324: really good idea to manipulate the control flow stack with
 1325: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
 1326: words.
 1327: 
 1328: Some standard control structure words are built from these words:
 1329: 
 1330: doc-else
 1331: doc-while
 1332: doc-repeat
 1333: 
 1334: Counted loop words constitute a separate group of words:
 1335: 
 1336: doc-?do
 1337: doc-+do
 1338: doc-u+do
 1339: doc--do
 1340: doc-u-do
 1341: doc-do
 1342: doc-for
 1343: doc-loop
 1344: doc-s+loop
 1345: doc-+loop
 1346: doc--loop
 1347: doc-next
 1348: doc-leave
 1349: doc-?leave
 1350: doc-unloop
 1351: doc-done
 1352: 
 1353: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
 1354: @i{do-sys}. Our system allows it, but it's your job to ensure that for
 1355: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
 1356: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
 1357: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
 1358: resolved (by using one of the loop-ending words or @code{DONE}).
 1359: 
 1360: Another group of control structure words are
 1361: 
 1362: doc-case
 1363: doc-endcase
 1364: doc-of
 1365: doc-endof
 1366: 
 1367: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
 1368: @code{cs-roll}.
 1369: 
 1370: @subsubsection Programming Style
 1371: 
 1372: In order to ensure readability we recommend that you do not create
 1373: arbitrary control structures directly, but define new control structure
 1374: words for the control structure you want and use these words in your
 1375: program.
 1376: 
 1377: E.g., instead of writing
 1378: 
 1379: @example
 1380: begin
 1381:   ...
 1382: if [ 1 cs-roll ]
 1383:   ...
 1384: again then
 1385: @end example
 1386: 
 1387: we recommend defining control structure words, e.g.,
 1388: 
 1389: @example
 1390: : while ( dest -- orig dest )
 1391:  POSTPONE if
 1392:  1 cs-roll ; immediate
 1393: 
 1394: : repeat ( orig dest -- )
 1395:  POSTPONE again
 1396:  POSTPONE then ; immediate
 1397: @end example
 1398: 
 1399: and then using these to create the control structure:
 1400: 
 1401: @example
 1402: begin
 1403:   ...
 1404: while
 1405:   ...
 1406: repeat
 1407: @end example
 1408: 
 1409: That's much easier to read, isn't it? Of course, @code{BEGIN} and
 1410: @code{WHILE} are predefined, so in this example it would not be
 1411: necessary to define them.
 1412: 
 1413: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
 1414: @subsection Calls and returns
 1415: 
 1416: A definition can be called simply be writing the name of the
 1417: definition. When the end of the definition is reached, it returns. An
 1418: earlier return can be forced using
 1419: 
 1420: doc-exit
 1421: 
 1422: Don't forget to clean up the return stack and @code{UNLOOP} any
 1423: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
 1424: primitive compiled by @code{EXIT} is
 1425: 
 1426: doc-;s
 1427: 
 1428: @node Exception Handling,  , Calls and returns, Control Structures
 1429: @subsection Exception Handling
 1430: 
 1431: doc-catch
 1432: doc-throw
 1433: 
 1434: @node Locals, Defining Words, Control Structures, Words
 1435: @section Locals
 1436: 
 1437: Local variables can make Forth programming more enjoyable and Forth
 1438: programs easier to read. Unfortunately, the locals of ANS Forth are
 1439: laden with restrictions. Therefore, we provide not only the ANS Forth
 1440: locals wordset, but also our own, more powerful locals wordset (we
 1441: implemented the ANS Forth locals wordset through our locals wordset).
 1442: 
 1443: @menu
 1444: * Gforth locals::               
 1445: * ANS Forth locals::            
 1446: @end menu
 1447: 
 1448: @node Gforth locals, ANS Forth locals, Locals, Locals
 1449: @subsection Gforth locals
 1450: 
 1451: Locals can be defined with
 1452: 
 1453: @example
 1454: @{ local1 local2 ... -- comment @}
 1455: @end example
 1456: or
 1457: @example
 1458: @{ local1 local2 ... @}
 1459: @end example
 1460: 
 1461: E.g.,
 1462: @example
 1463: : max @{ n1 n2 -- n3 @}
 1464:  n1 n2 > if
 1465:    n1
 1466:  else
 1467:    n2
 1468:  endif ;
 1469: @end example
 1470: 
 1471: The similarity of locals definitions with stack comments is intended. A
 1472: locals definition often replaces the stack comment of a word. The order
 1473: of the locals corresponds to the order in a stack comment and everything
 1474: after the @code{--} is really a comment.
 1475: 
 1476: This similarity has one disadvantage: It is too easy to confuse locals
 1477: declarations with stack comments, causing bugs and making them hard to
 1478: find. However, this problem can be avoided by appropriate coding
 1479: conventions: Do not use both notations in the same program. If you do,
 1480: they should be distinguished using additional means, e.g. by position.
 1481: 
 1482: The name of the local may be preceded by a type specifier, e.g.,
 1483: @code{F:} for a floating point value:
 1484: 
 1485: @example
 1486: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
 1487: \ complex multiplication
 1488:  Ar Br f* Ai Bi f* f-
 1489:  Ar Bi f* Ai Br f* f+ ;
 1490: @end example
 1491: 
 1492: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
 1493: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
 1494: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
 1495: with @code{W:}, @code{D:} etc.) produces its value and can be changed
 1496: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
 1497: produces its address (which becomes invalid when the variable's scope is
 1498: left). E.g., the standard word @code{emit} can be defined in therms of
 1499: @code{type} like this:
 1500: 
 1501: @example
 1502: : emit @{ C^ char* -- @}
 1503:     char* 1 type ;
 1504: @end example
 1505: 
 1506: A local without type specifier is a @code{W:} local. Both flavours of
 1507: locals are initialized with values from the data or FP stack.
 1508: 
 1509: Currently there is no way to define locals with user-defined data
 1510: structures, but we are working on it.
 1511: 
 1512: Gforth allows defining locals everywhere in a colon definition. This
 1513: poses the following questions:
 1514: 
 1515: @menu
 1516: * Where are locals visible by name?::  
 1517: * How long do locals live?::    
 1518: * Programming Style::           
 1519: * Implementation::              
 1520: @end menu
 1521: 
 1522: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
 1523: @subsubsection Where are locals visible by name?
 1524: 
 1525: Basically, the answer is that locals are visible where you would expect
 1526: it in block-structured languages, and sometimes a little longer. If you
 1527: want to restrict the scope of a local, enclose its definition in
 1528: @code{SCOPE}...@code{ENDSCOPE}.
 1529: 
 1530: doc-scope
 1531: doc-endscope
 1532: 
 1533: These words behave like control structure words, so you can use them
 1534: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
 1535: arbitrary ways.
 1536: 
 1537: If you want a more exact answer to the visibility question, here's the
 1538: basic principle: A local is visible in all places that can only be
 1539: reached through the definition of the local@footnote{In compiler
 1540: construction terminology, all places dominated by the definition of the
 1541: local.}. In other words, it is not visible in places that can be reached
 1542: without going through the definition of the local. E.g., locals defined
 1543: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
 1544: defined in @code{BEGIN}...@code{UNTIL} are visible after the
 1545: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
 1546: 
 1547: The reasoning behind this solution is: We want to have the locals
 1548: visible as long as it is meaningful. The user can always make the
 1549: visibility shorter by using explicit scoping. In a place that can
 1550: only be reached through the definition of a local, the meaning of a
 1551: local name is clear. In other places it is not: How is the local
 1552: initialized at the control flow path that does not contain the
 1553: definition? Which local is meant, if the same name is defined twice in
 1554: two independent control flow paths?
 1555: 
 1556: This should be enough detail for nearly all users, so you can skip the
 1557: rest of this section. If you relly must know all the gory details and
 1558: options, read on.
 1559: 
 1560: In order to implement this rule, the compiler has to know which places
 1561: are unreachable. It knows this automatically after @code{AHEAD},
 1562: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
 1563: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
 1564: compiler that the control flow never reaches that place. If
 1565: @code{UNREACHABLE} is not used where it could, the only consequence is
 1566: that the visibility of some locals is more limited than the rule above
 1567: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
 1568: lie to the compiler), buggy code will be produced.
 1569: 
 1570: Another problem with this rule is that at @code{BEGIN}, the compiler
 1571: does not know which locals will be visible on the incoming
 1572: back-edge. All problems discussed in the following are due to this
 1573: ignorance of the compiler (we discuss the problems using @code{BEGIN}
 1574: loops as examples; the discussion also applies to @code{?DO} and other
 1575: loops). Perhaps the most insidious example is:
 1576: @example
 1577: AHEAD
 1578: BEGIN
 1579:   x
 1580: [ 1 CS-ROLL ] THEN
 1581:   @{ x @}
 1582:   ...
 1583: UNTIL
 1584: @end example
 1585: 
 1586: This should be legal according to the visibility rule. The use of
 1587: @code{x} can only be reached through the definition; but that appears
 1588: textually below the use.
 1589: 
 1590: From this example it is clear that the visibility rules cannot be fully
 1591: implemented without major headaches. Our implementation treats common
 1592: cases as advertised and the exceptions are treated in a safe way: The
 1593: compiler makes a reasonable guess about the locals visible after a
 1594: @code{BEGIN}; if it is too pessimistic, the
 1595: user will get a spurious error about the local not being defined; if the
 1596: compiler is too optimistic, it will notice this later and issue a
 1597: warning. In the case above the compiler would complain about @code{x}
 1598: being undefined at its use. You can see from the obscure examples in
 1599: this section that it takes quite unusual control structures to get the
 1600: compiler into trouble, and even then it will often do fine.
 1601: 
 1602: If the @code{BEGIN} is reachable from above, the most optimistic guess
 1603: is that all locals visible before the @code{BEGIN} will also be
 1604: visible after the @code{BEGIN}. This guess is valid for all loops that
 1605: are entered only through the @code{BEGIN}, in particular, for normal
 1606: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
 1607: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
 1608: compiler. When the branch to the @code{BEGIN} is finally generated by
 1609: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
 1610: warns the user if it was too optimisitic:
 1611: @example
 1612: IF
 1613:   @{ x @}
 1614: BEGIN
 1615:   \ x ? 
 1616: [ 1 cs-roll ] THEN
 1617:   ...
 1618: UNTIL
 1619: @end example
 1620: 
 1621: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
 1622: optimistically assumes that it lives until the @code{THEN}. It notices
 1623: this difference when it compiles the @code{UNTIL} and issues a
 1624: warning. The user can avoid the warning, and make sure that @code{x}
 1625: is not used in the wrong area by using explicit scoping:
 1626: @example
 1627: IF
 1628:   SCOPE
 1629:   @{ x @}
 1630:   ENDSCOPE
 1631: BEGIN
 1632: [ 1 cs-roll ] THEN
 1633:   ...
 1634: UNTIL
 1635: @end example
 1636: 
 1637: Since the guess is optimistic, there will be no spurious error messages
 1638: about undefined locals.
 1639: 
 1640: If the @code{BEGIN} is not reachable from above (e.g., after
 1641: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
 1642: optimistic guess, as the locals visible after the @code{BEGIN} may be
 1643: defined later. Therefore, the compiler assumes that no locals are
 1644: visible after the @code{BEGIN}. However, the user can use
 1645: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
 1646: visible at the BEGIN as at the point where the top control-flow stack
 1647: item was created.
 1648: 
 1649: doc-assume-live
 1650: 
 1651: E.g.,
 1652: @example
 1653: @{ x @}
 1654: AHEAD
 1655: ASSUME-LIVE
 1656: BEGIN
 1657:   x
 1658: [ 1 CS-ROLL ] THEN
 1659:   ...
 1660: UNTIL
 1661: @end example
 1662: 
 1663: Other cases where the locals are defined before the @code{BEGIN} can be
 1664: handled by inserting an appropriate @code{CS-ROLL} before the
 1665: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
 1666: behind the @code{ASSUME-LIVE}).
 1667: 
 1668: Cases where locals are defined after the @code{BEGIN} (but should be
 1669: visible immediately after the @code{BEGIN}) can only be handled by
 1670: rearranging the loop. E.g., the ``most insidious'' example above can be
 1671: arranged into:
 1672: @example
 1673: BEGIN
 1674:   @{ x @}
 1675:   ... 0=
 1676: WHILE
 1677:   x
 1678: REPEAT
 1679: @end example
 1680: 
 1681: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
 1682: @subsubsection How long do locals live?
 1683: 
 1684: The right answer for the lifetime question would be: A local lives at
 1685: least as long as it can be accessed. For a value-flavoured local this
 1686: means: until the end of its visibility. However, a variable-flavoured
 1687: local could be accessed through its address far beyond its visibility
 1688: scope. Ultimately, this would mean that such locals would have to be
 1689: garbage collected. Since this entails un-Forth-like implementation
 1690: complexities, I adopted the same cowardly solution as some other
 1691: languages (e.g., C): The local lives only as long as it is visible;
 1692: afterwards its address is invalid (and programs that access it
 1693: afterwards are erroneous).
 1694: 
 1695: @node Programming Style, Implementation, How long do locals live?, Gforth locals
 1696: @subsubsection Programming Style
 1697: 
 1698: The freedom to define locals anywhere has the potential to change
 1699: programming styles dramatically. In particular, the need to use the
 1700: return stack for intermediate storage vanishes. Moreover, all stack
 1701: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
 1702: determined arguments) can be eliminated: If the stack items are in the
 1703: wrong order, just write a locals definition for all of them; then
 1704: write the items in the order you want.
 1705: 
 1706: This seems a little far-fetched and eliminating stack manipulations is
 1707: unlikely to become a conscious programming objective. Still, the number
 1708: of stack manipulations will be reduced dramatically if local variables
 1709: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
 1710: a traditional implementation of @code{max}).
 1711: 
 1712: This shows one potential benefit of locals: making Forth programs more
 1713: readable. Of course, this benefit will only be realized if the
 1714: programmers continue to honour the principle of factoring instead of
 1715: using the added latitude to make the words longer.
 1716: 
 1717: Using @code{TO} can and should be avoided.  Without @code{TO},
 1718: every value-flavoured local has only a single assignment and many
 1719: advantages of functional languages apply to Forth. I.e., programs are
 1720: easier to analyse, to optimize and to read: It is clear from the
 1721: definition what the local stands for, it does not turn into something
 1722: different later.
 1723: 
 1724: E.g., a definition using @code{TO} might look like this:
 1725: @example
 1726: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1727:  u1 u2 min 0
 1728:  ?do
 1729:    addr1 c@ addr2 c@ - ?dup
 1730:    if
 1731:      unloop exit
 1732:    then
 1733:    addr1 char+ TO addr1
 1734:    addr2 char+ TO addr2
 1735:  loop
 1736:  u1 u2 - ;
 1737: @end example
 1738: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
 1739: every loop iteration. @code{strcmp} is a typical example of the
 1740: readability problems of using @code{TO}. When you start reading
 1741: @code{strcmp}, you think that @code{addr1} refers to the start of the
 1742: string. Only near the end of the loop you realize that it is something
 1743: else.
 1744: 
 1745: This can be avoided by defining two locals at the start of the loop that
 1746: are initialized with the right value for the current iteration.
 1747: @example
 1748: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1749:  addr1 addr2
 1750:  u1 u2 min 0 
 1751:  ?do @{ s1 s2 @}
 1752:    s1 c@ s2 c@ - ?dup 
 1753:    if
 1754:      unloop exit
 1755:    then
 1756:    s1 char+ s2 char+
 1757:  loop
 1758:  2drop
 1759:  u1 u2 - ;
 1760: @end example
 1761: Here it is clear from the start that @code{s1} has a different value
 1762: in every loop iteration.
 1763: 
 1764: @node Implementation,  , Programming Style, Gforth locals
 1765: @subsubsection Implementation
 1766: 
 1767: Gforth uses an extra locals stack. The most compelling reason for
 1768: this is that the return stack is not float-aligned; using an extra stack
 1769: also eliminates the problems and restrictions of using the return stack
 1770: as locals stack. Like the other stacks, the locals stack grows toward
 1771: lower addresses. A few primitives allow an efficient implementation:
 1772: 
 1773: doc-@local#
 1774: doc-f@local#
 1775: doc-laddr#
 1776: doc-lp+!#
 1777: doc-lp!
 1778: doc->l
 1779: doc-f>l
 1780: 
 1781: In addition to these primitives, some specializations of these
 1782: primitives for commonly occurring inline arguments are provided for
 1783: efficiency reasons, e.g., @code{@@local0} as specialization of
 1784: @code{@@local#} for the inline argument 0. The following compiling words
 1785: compile the right specialized version, or the general version, as
 1786: appropriate:
 1787: 
 1788: doc-compile-@local
 1789: doc-compile-f@local
 1790: doc-compile-lp+!
 1791: 
 1792: Combinations of conditional branches and @code{lp+!#} like
 1793: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
 1794: is taken) are provided for efficiency and correctness in loops.
 1795: 
 1796: A special area in the dictionary space is reserved for keeping the
 1797: local variable names. @code{@{} switches the dictionary pointer to this
 1798: area and @code{@}} switches it back and generates the locals
 1799: initializing code. @code{W:} etc.@ are normal defining words. This
 1800: special area is cleared at the start of every colon definition.
 1801: 
 1802: A special feature of Gforth's dictionary is used to implement the
 1803: definition of locals without type specifiers: every wordlist (aka
 1804: vocabulary) has its own methods for searching
 1805: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
 1806: with a special search method: When it is searched for a word, it
 1807: actually creates that word using @code{W:}. @code{@{} changes the search
 1808: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
 1809: and then the wordlist for defining locals without type specifiers.
 1810: 
 1811: The lifetime rules support a stack discipline within a colon
 1812: definition: The lifetime of a local is either nested with other locals
 1813: lifetimes or it does not overlap them.
 1814: 
 1815: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
 1816: pointer manipulation is generated. Between control structure words
 1817: locals definitions can push locals onto the locals stack. @code{AGAIN}
 1818: is the simplest of the other three control flow words. It has to
 1819: restore the locals stack depth of the corresponding @code{BEGIN}
 1820: before branching. The code looks like this:
 1821: @format
 1822: @code{lp+!#} current-locals-size @minus{} dest-locals-size
 1823: @code{branch} <begin>
 1824: @end format
 1825: 
 1826: @code{UNTIL} is a little more complicated: If it branches back, it
 1827: must adjust the stack just like @code{AGAIN}. But if it falls through,
 1828: the locals stack must not be changed. The compiler generates the
 1829: following code:
 1830: @format
 1831: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
 1832: @end format
 1833: The locals stack pointer is only adjusted if the branch is taken.
 1834: 
 1835: @code{THEN} can produce somewhat inefficient code:
 1836: @format
 1837: @code{lp+!#} current-locals-size @minus{} orig-locals-size
 1838: <orig target>:
 1839: @code{lp+!#} orig-locals-size @minus{} new-locals-size
 1840: @end format
 1841: The second @code{lp+!#} adjusts the locals stack pointer from the
 1842: level at the @var{orig} point to the level after the @code{THEN}. The
 1843: first @code{lp+!#} adjusts the locals stack pointer from the current
 1844: level to the level at the orig point, so the complete effect is an
 1845: adjustment from the current level to the right level after the
 1846: @code{THEN}.
 1847: 
 1848: In a conventional Forth implementation a dest control-flow stack entry
 1849: is just the target address and an orig entry is just the address to be
 1850: patched. Our locals implementation adds a wordlist to every orig or dest
 1851: item. It is the list of locals visible (or assumed visible) at the point
 1852: described by the entry. Our implementation also adds a tag to identify
 1853: the kind of entry, in particular to differentiate between live and dead
 1854: (reachable and unreachable) orig entries.
 1855: 
 1856: A few unusual operations have to be performed on locals wordlists:
 1857: 
 1858: doc-common-list
 1859: doc-sub-list?
 1860: doc-list-size
 1861: 
 1862: Several features of our locals wordlist implementation make these
 1863: operations easy to implement: The locals wordlists are organised as
 1864: linked lists; the tails of these lists are shared, if the lists
 1865: contain some of the same locals; and the address of a name is greater
 1866: than the address of the names behind it in the list.
 1867: 
 1868: Another important implementation detail is the variable
 1869: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
 1870: determine if they can be reached directly or only through the branch
 1871: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
 1872: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
 1873: definition, by @code{BEGIN} and usually by @code{THEN}.
 1874: 
 1875: Counted loops are similar to other loops in most respects, but
 1876: @code{LEAVE} requires special attention: It performs basically the same
 1877: service as @code{AHEAD}, but it does not create a control-flow stack
 1878: entry. Therefore the information has to be stored elsewhere;
 1879: traditionally, the information was stored in the target fields of the
 1880: branches created by the @code{LEAVE}s, by organizing these fields into a
 1881: linked list. Unfortunately, this clever trick does not provide enough
 1882: space for storing our extended control flow information. Therefore, we
 1883: introduce another stack, the leave stack. It contains the control-flow
 1884: stack entries for all unresolved @code{LEAVE}s.
 1885: 
 1886: Local names are kept until the end of the colon definition, even if
 1887: they are no longer visible in any control-flow path. In a few cases
 1888: this may lead to increased space needs for the locals name area, but
 1889: usually less than reclaiming this space would cost in code size.
 1890: 
 1891: 
 1892: @node ANS Forth locals,  , Gforth locals, Locals
 1893: @subsection ANS Forth locals
 1894: 
 1895: The ANS Forth locals wordset does not define a syntax for locals, but
 1896: words that make it possible to define various syntaxes. One of the
 1897: possible syntaxes is a subset of the syntax we used in the Gforth locals
 1898: wordset, i.e.:
 1899: 
 1900: @example
 1901: @{ local1 local2 ... -- comment @}
 1902: @end example
 1903: or
 1904: @example
 1905: @{ local1 local2 ... @}
 1906: @end example
 1907: 
 1908: The order of the locals corresponds to the order in a stack comment. The
 1909: restrictions are:
 1910: 
 1911: @itemize @bullet
 1912: @item
 1913: Locals can only be cell-sized values (no type specifiers are allowed).
 1914: @item
 1915: Locals can be defined only outside control structures.
 1916: @item
 1917: Locals can interfere with explicit usage of the return stack. For the
 1918: exact (and long) rules, see the standard. If you don't use return stack
 1919: accessing words in a definition using locals, you will be all right. The
 1920: purpose of this rule is to make locals implementation on the return
 1921: stack easier.
 1922: @item
 1923: The whole definition must be in one line.
 1924: @end itemize
 1925: 
 1926: Locals defined in this way behave like @code{VALUE}s
 1927: (@xref{Values}). I.e., they are initialized from the stack. Using their
 1928: name produces their value. Their value can be changed using @code{TO}.
 1929: 
 1930: Since this syntax is supported by Gforth directly, you need not do
 1931: anything to use it. If you want to port a program using this syntax to
 1932: another ANS Forth system, use @file{anslocal.fs} to implement the syntax
 1933: on the other system.
 1934: 
 1935: Note that a syntax shown in the standard, section A.13 looks
 1936: similar, but is quite different in having the order of locals
 1937: reversed. Beware!
 1938: 
 1939: The ANS Forth locals wordset itself consists of the following word
 1940: 
 1941: doc-(local)
 1942: 
 1943: The ANS Forth locals extension wordset defines a syntax, but it is so
 1944: awful that we strongly recommend not to use it. We have implemented this
 1945: syntax to make porting to Gforth easy, but do not document it here. The
 1946: problem with this syntax is that the locals are defined in an order
 1947: reversed with respect to the standard stack comment notation, making
 1948: programs harder to read, and easier to misread and miswrite. The only
 1949: merit of this syntax is that it is easy to implement using the ANS Forth
 1950: locals wordset.
 1951: 
 1952: @node Defining Words, Wordlists, Locals, Words
 1953: @section Defining Words
 1954: 
 1955: @menu
 1956: * Values::                      
 1957: @end menu
 1958: 
 1959: @node Values,  , Defining Words, Defining Words
 1960: @subsection Values
 1961: 
 1962: @node Wordlists, Files, Defining Words, Words
 1963: @section Wordlists
 1964: 
 1965: @node Files, Blocks, Wordlists, Words
 1966: @section Files
 1967: 
 1968: @node Blocks, Other I/O, Files, Words
 1969: @section Blocks
 1970: 
 1971: @node Other I/O, Programming Tools, Blocks, Words
 1972: @section Other I/O
 1973: 
 1974: @node Programming Tools, Assembler and Code words, Other I/O, Words
 1975: @section Programming Tools
 1976: 
 1977: @menu
 1978: * Debugging::                   Simple and quick.
 1979: * Assertions::                  Making your programs self-checking.
 1980: @end menu
 1981: 
 1982: @node Debugging, Assertions, Programming Tools, Programming Tools
 1983: @subsection Debugging
 1984: 
 1985: The simple debugging aids provided in @file{debugging.fs}
 1986: are meant to support a different style of debugging than the
 1987: tracing/stepping debuggers used in languages with long turn-around
 1988: times.
 1989: 
 1990: A much better (faster) way in fast-compilig languages is to add
 1991: printing code at well-selected places, let the program run, look at
 1992: the output, see where things went wrong, add more printing code, etc.,
 1993: until the bug is found.
 1994: 
 1995: The word @code{~~} is easy to insert. It just prints debugging
 1996: information (by default the source location and the stack contents). It
 1997: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
 1998: query-replace them with nothing). The deferred words
 1999: @code{printdebugdata} and @code{printdebugline} control the output of
 2000: @code{~~}. The default source location output format works well with
 2001: Emacs' compilation mode, so you can step through the program at the
 2002: source level using @kbd{C-x `} (the advantage over a stepping debugger
 2003: is that you can step in any direction and you know where the crash has
 2004: happened or where the strange data has occurred).
 2005: 
 2006: Note that the default actions clobber the contents of the pictured
 2007: numeric output string, so you should not use @code{~~}, e.g., between
 2008: @code{<#} and @code{#>}.
 2009: 
 2010: doc-~~
 2011: doc-printdebugdata
 2012: doc-printdebugline
 2013: 
 2014: @node Assertions,  , Debugging, Programming Tools
 2015: @subsection Assertions
 2016: 
 2017: It is a good idea to make your programs self-checking, in particular, if
 2018: you use an assumption (e.g., that a certain field of a data structure is
 2019: never zero) that may become wrong during maintenance. Gforth supports
 2020: assertions for this purpose. They are used like this:
 2021: 
 2022: @example
 2023: assert( @var{flag} )
 2024: @end example
 2025: 
 2026: The code between @code{assert(} and @code{)} should compute a flag, that
 2027: should be true if everything is alright and false otherwise. It should
 2028: not change anything else on the stack. The overall stack effect of the
 2029: assertion is @code{( -- )}. E.g.
 2030: 
 2031: @example
 2032: assert( 1 1 + 2 = ) \ what we learn in school
 2033: assert( dup 0<> ) \ assert that the top of stack is not zero
 2034: assert( false ) \ this code should not be reached
 2035: @end example
 2036: 
 2037: The need for assertions is different at different times. During
 2038: debugging, we want more checking, in production we sometimes care more
 2039: for speed. Therefore, assertions can be turned off, i.e., the assertion
 2040: becomes a comment. Depending on the importance of an assertion and the
 2041: time it takes to check it, you may want to turn off some assertions and
 2042: keep others turned on. Gforth provides several levels of assertions for
 2043: this purpose:
 2044: 
 2045: doc-assert0(
 2046: doc-assert1(
 2047: doc-assert2(
 2048: doc-assert3(
 2049: doc-assert(
 2050: doc-)
 2051: 
 2052: @code{Assert(} is the same as @code{assert1(}. The variable
 2053: @code{assert-level} specifies the highest assertions that are turned
 2054: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
 2055: @code{assert1(} assertions perform checking, while @code{assert2(} and
 2056: @code{assert3(} assertions are treated as comments.
 2057: 
 2058: Note that the @code{assert-level} is evaluated at compile-time, not at
 2059: run-time. I.e., you cannot turn assertions on or off at run-time, you
 2060: have to set the @code{assert-level} appropriately before compiling a
 2061: piece of code. You can compile several pieces of code at several
 2062: @code{assert-level}s (e.g., a trusted library at level 1 and newly
 2063: written code at level 3).
 2064: 
 2065: doc-assert-level
 2066: 
 2067: If an assertion fails, a message compatible with Emacs' compilation mode
 2068: is produced and the execution is aborted (currently with @code{ABORT"}.
 2069: If there is interest, we will introduce a special throw code. But if you
 2070: intend to @code{catch} a specific condition, using @code{throw} is
 2071: probably more appropriate than an assertion).
 2072: 
 2073: @node Assembler and Code words, Threading Words, Programming Tools, Words
 2074: @section Assembler and Code words
 2075: 
 2076: Gforth provides some words for defining primitives (words written in
 2077: machine code), and for defining the the machine-code equivalent of
 2078: @code{DOES>}-based defining words. However, the machine-independent
 2079: nature of Gforth poses a few problems: First of all. Gforth runs on
 2080: several architectures, so it can provide no standard assembler. What's
 2081: worse is that the register allocation not only depends on the processor,
 2082: but also on the gcc version and options used.
 2083: 
 2084: The words Gforth offers encapsulate some system dependences (e.g., the
 2085: header structure), so a system-independent assembler may be used in
 2086: Gforth. If you do not have an assembler, you can compile machine code
 2087: directly with @code{,} and @code{c,}.
 2088: 
 2089: doc-assembler
 2090: doc-code
 2091: doc-end-code
 2092: doc-;code
 2093: doc-flush-icache
 2094: 
 2095: If @code{flush-icache} does not work correctly, @code{code} words
 2096: etc. will not work (reliably), either.
 2097: 
 2098: These words are rarely used. Therefore they reside in @code{code.fs},
 2099: which is usually not loaded (except @code{flush-icache}, which is always
 2100: present). You can load them with @code{require code.fs}.
 2101: 
 2102: Another option for implementing normal and defining words efficiently
 2103: is: adding the wanted functionality to the source of Gforth. For normal
 2104: words you just have to edit @file{primitives}, defining words (for fast
 2105: defined words) probably require changes in @file{engine.c},
 2106: @file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
 2107: 
 2108: 
 2109: @node Threading Words,  , Assembler and Code words, Words
 2110: @section Threading Words
 2111: 
 2112: These words provide access to code addresses and other threading stuff
 2113: in Gforth (and, possibly, other interpretive Forths). It more or less
 2114: abstracts away the differences between direct and indirect threading
 2115: (and, for direct threading, the machine dependences). However, at
 2116: present this wordset is still inclomplete. It is also pretty low-level;
 2117: some day it will hopefully be made unnecessary by an internals words set
 2118: that abstracts implementation details away completely.
 2119: 
 2120: doc->code-address
 2121: doc->does-code
 2122: doc-code-address!
 2123: doc-does-code!
 2124: doc-does-handler!
 2125: doc-/does-handler
 2126: 
 2127: The code addresses produced by various defining words are produced by
 2128: the following words:
 2129: 
 2130: doc-docol:
 2131: doc-docon:
 2132: doc-dovar:
 2133: doc-douser:
 2134: doc-dodefer:
 2135: doc-dofield:
 2136: 
 2137: Currently there is no installation-independent way for recogizing words
 2138: defined by a @code{CREATE}...@code{DOES>} word; however, once you know
 2139: that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
 2140: use @code{>DOES-CODE}.
 2141: 
 2142: @node ANS conformance, Model, Words, Top
 2143: @chapter ANS conformance
 2144: 
 2145: To the best of our knowledge, Gforth is an
 2146: 
 2147: ANS Forth System
 2148: @itemize
 2149: @item providing the Core Extensions word set
 2150: @item providing the Block word set
 2151: @item providing the Block Extensions word set
 2152: @item providing the Double-Number word set
 2153: @item providing the Double-Number Extensions word set
 2154: @item providing the Exception word set
 2155: @item providing the Exception Extensions word set
 2156: @item providing the Facility word set
 2157: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
 2158: @item providing the File Access word set
 2159: @item providing the File Access Extensions word set
 2160: @item providing the Floating-Point word set
 2161: @item providing the Floating-Point Extensions word set
 2162: @item providing the Locals word set
 2163: @item providing the Locals Extensions word set
 2164: @item providing the Memory-Allocation word set
 2165: @item providing the Memory-Allocation Extensions word set (that one's easy)
 2166: @item providing the Programming-Tools word set
 2167: @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
 2168: @item providing the Search-Order word set
 2169: @item providing the Search-Order Extensions word set
 2170: @item providing the String word set
 2171: @item providing the String Extensions word set (another easy one)
 2172: @end itemize
 2173: 
 2174: In addition, ANS Forth systems are required to document certain
 2175: implementation choices. This chapter tries to meet these
 2176: requirements. In many cases it gives a way to ask the system for the
 2177: information instead of providing the information directly, in
 2178: particular, if the information depends on the processor, the operating
 2179: system or the installation options chosen, or if they are likely to
 2180: change during the maintenance of Gforth.
 2181: 
 2182: @comment The framework for the rest has been taken from pfe.
 2183: 
 2184: @menu
 2185: * The Core Words::              
 2186: * The optional Block word set::  
 2187: * The optional Double Number word set::  
 2188: * The optional Exception word set::  
 2189: * The optional Facility word set::  
 2190: * The optional File-Access word set::  
 2191: * The optional Floating-Point word set::  
 2192: * The optional Locals word set::  
 2193: * The optional Memory-Allocation word set::  
 2194: * The optional Programming-Tools word set::  
 2195: * The optional Search-Order word set::  
 2196: @end menu
 2197: 
 2198: 
 2199: @c =====================================================================
 2200: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
 2201: @comment  node-name,  next,  previous,  up
 2202: @section The Core Words
 2203: @c =====================================================================
 2204: 
 2205: @menu
 2206: * core-idef::                   Implementation Defined Options                   
 2207: * core-ambcond::                Ambiguous Conditions                
 2208: * core-other::                  Other System Documentation                  
 2209: @end menu
 2210: 
 2211: @c ---------------------------------------------------------------------
 2212: @node core-idef, core-ambcond, The Core Words, The Core Words
 2213: @subsection Implementation Defined Options
 2214: @c ---------------------------------------------------------------------
 2215: 
 2216: @table @i
 2217: 
 2218: @item (Cell) aligned addresses:
 2219: processor-dependent. Gforth's alignment words perform natural alignment
 2220: (e.g., an address aligned for a datum of size 8 is divisible by
 2221: 8). Unaligned accesses usually result in a @code{-23 THROW}.
 2222: 
 2223: @item @code{EMIT} and non-graphic characters:
 2224: The character is output using the C library function (actually, macro)
 2225: @code{putchar}.
 2226: 
 2227: @item character editing of @code{ACCEPT} and @code{EXPECT}:
 2228: This is modeled on the GNU readline library (@pxref{Readline
 2229: Interaction, , Command Line Editing, readline, The GNU Readline
 2230: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
 2231: producing a full word completion every time you type it (instead of
 2232: producing the common prefix of all completions).
 2233: 
 2234: @item character set:
 2235: The character set of your computer and display device. Gforth is
 2236: 8-bit-clean (but some other component in your system may make trouble).
 2237: 
 2238: @item Character-aligned address requirements:
 2239: installation-dependent. Currently a character is represented by a C
 2240: @code{unsigned char}; in the future we might switch to @code{wchar_t}
 2241: (Comments on that requested).
 2242: 
 2243: @item character-set extensions and matching of names:
 2244: Any character except the ASCII NUL charcter can be used in a
 2245: name. Matching is case-insensitive. The matching is performed using the
 2246: C function @code{strncasecmp}, whose function is probably influenced by
 2247: the locale. E.g., the @code{C} locale does not know about accents and
 2248: umlauts, so they are matched case-sensitively in that locale. For
 2249: portability reasons it is best to write programs such that they work in
 2250: the @code{C} locale. Then one can use libraries written by a Polish
 2251: programmer (who might use words containing ISO Latin-2 encoded
 2252: characters) and by a French programmer (ISO Latin-1) in the same program
 2253: (of course, @code{WORDS} will produce funny results for some of the
 2254: words (which ones, depends on the font you are using)). Also, the locale
 2255: you prefer may not be available in other operating systems. Hopefully,
 2256: Unicode will solve these problems one day.
 2257: 
 2258: @item conditions under which control characters match a space delimiter:
 2259: If @code{WORD} is called with the space character as a delimiter, all
 2260: white-space characters (as identified by the C macro @code{isspace()})
 2261: are delimiters. @code{PARSE}, on the other hand, treats space like other
 2262: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
 2263: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
 2264: interpreter (aka text interpreter) by default, treats all white-space
 2265: characters as delimiters.
 2266: 
 2267: @item format of the control flow stack:
 2268: The data stack is used as control flow stack. The size of a control flow
 2269: stack item in cells is given by the constant @code{cs-item-size}. At the
 2270: time of this writing, an item consists of a (pointer to a) locals list
 2271: (third), an address in the code (second), and a tag for identifying the
 2272: item (TOS). The following tags are used: @code{defstart},
 2273: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
 2274: @code{scopestart}.
 2275: 
 2276: @item conversion of digits > 35
 2277: The characters @code{[\]^_'} are the digits with the decimal value
 2278: 36@minus{}41. There is no way to input many of the larger digits.
 2279: 
 2280: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
 2281: The cursor is moved to the end of the entered string. If the input is
 2282: terminated using the @kbd{Return} key, a space is typed.
 2283: 
 2284: @item exception abort sequence of @code{ABORT"}:
 2285: The error string is stored into the variable @code{"error} and a
 2286: @code{-2 throw} is performed.
 2287: 
 2288: @item input line terminator:
 2289: For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
 2290: these characters is typically produced when you type the @kbd{Enter} or
 2291: @kbd{Return} key.
 2292: 
 2293: @item maximum size of a counted string:
 2294: @code{s" /counted-string" environment? drop .}. Currently 255 characters
 2295: on all ports, but this may change.
 2296: 
 2297: @item maximum size of a parsed string:
 2298: Given by the constant @code{/line}. Currently 255 characters.
 2299: 
 2300: @item maximum size of a definition name, in characters:
 2301: 31
 2302: 
 2303: @item maximum string length for @code{ENVIRONMENT?}, in characters:
 2304: 31
 2305: 
 2306: @item method of selecting the user input device:
 2307: The user input device is the standard input. There is currently no way to
 2308: change it from within Gforth. However, the input can typically be
 2309: redirected in the command line that starts Gforth.
 2310: 
 2311: @item method of selecting the user output device:
 2312: The user output device is the standard output. It cannot be redirected
 2313: from within Gforth, but typically from the command line that starts
 2314: Gforth. Gforth uses buffered output, so output on a terminal does not
 2315: become visible before the next newline or buffer overflow. Output on
 2316: non-terminals is invisible until the buffer overflows.
 2317: 
 2318: @item methods of dictionary compilation:
 2319: What are we expected to document here?
 2320: 
 2321: @item number of bits in one address unit:
 2322: @code{s" address-units-bits" environment? drop .}. 8 in all current
 2323: ports.
 2324: 
 2325: @item number representation and arithmetic:
 2326: Processor-dependent. Binary two's complement on all current ports.
 2327: 
 2328: @item ranges for integer types:
 2329: Installation-dependent. Make environmental queries for @code{MAX-N},
 2330: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
 2331: unsigned (and positive) types is 0. The lower bound for signed types on
 2332: two's complement and one's complement machines machines can be computed
 2333: by adding 1 to the upper bound.
 2334: 
 2335: @item read-only data space regions:
 2336: The whole Forth data space is writable.
 2337: 
 2338: @item size of buffer at @code{WORD}:
 2339: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 2340: shared with the pictured numeric output string. If overwriting
 2341: @code{PAD} is acceptable, it is as large as the remaining dictionary
 2342: space, although only as much can be sensibly used as fits in a counted
 2343: string.
 2344: 
 2345: @item size of one cell in address units:
 2346: @code{1 cells .}.
 2347: 
 2348: @item size of one character in address units:
 2349: @code{1 chars .}. 1 on all current ports.
 2350: 
 2351: @item size of the keyboard terminal buffer:
 2352: Varies. You can determine the size at a specific time using @code{lp@
 2353: tib - .}. It is shared with the locals stack and TIBs of files that
 2354: include the current file. You can change the amount of space for TIBs
 2355: and locals stack at Gforth startup with the command line option
 2356: @code{-l}.
 2357: 
 2358: @item size of the pictured numeric output buffer:
 2359: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 2360: shared with @code{WORD}.
 2361: 
 2362: @item size of the scratch area returned by @code{PAD}:
 2363: The remainder of dictionary space. You can even use the unused part of
 2364: the data stack space. The current size can be computed with @code{sp@
 2365: pad - .}.
 2366: 
 2367: @item system case-sensitivity characteristics:
 2368: Dictionary searches are case insensitive. However, as explained above
 2369: under @i{character-set extensions}, the matching for non-ASCII
 2370: characters is determined by the locale you are using. In the default
 2371: @code{C} locale all non-ASCII characters are matched case-sensitively.
 2372: 
 2373: @item system prompt:
 2374: @code{ ok} in interpret state, @code{ compiled} in compile state.
 2375: 
 2376: @item division rounding:
 2377: installation dependent. @code{s" floored" environment? drop .}. We leave
 2378: the choice to gcc (what to use for @code{/}) and to you (whether to use
 2379: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
 2380: 
 2381: @item values of @code{STATE} when true:
 2382: -1.
 2383: 
 2384: @item values returned after arithmetic overflow:
 2385: On two's complement machines, arithmetic is performed modulo
 2386: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 2387: arithmetic (with appropriate mapping for signed types). Division by zero
 2388: typically results in a @code{-55 throw} (floatingpoint unidentified
 2389: fault), although a @code{-10 throw} (divide by zero) would be more
 2390: appropriate.
 2391: 
 2392: @item whether the current definition can be found after @t{DOES>}:
 2393: No.
 2394: 
 2395: @end table
 2396: 
 2397: @c ---------------------------------------------------------------------
 2398: @node core-ambcond, core-other, core-idef, The Core Words
 2399: @subsection Ambiguous conditions
 2400: @c ---------------------------------------------------------------------
 2401: 
 2402: @table @i
 2403: 
 2404: @item a name is neither a word nor a number:
 2405: @code{-13 throw} (Undefined word)
 2406: 
 2407: @item a definition name exceeds the maximum length allowed:
 2408: @code{-19 throw} (Word name too long)
 2409: 
 2410: @item addressing a region not inside the various data spaces of the forth system:
 2411: The stacks, code space and name space are accessible. Machine code space is
 2412: typically readable. Accessing other addresses gives results dependent on
 2413: the operating system. On decent systems: @code{-9 throw} (Invalid memory
 2414: address).
 2415: 
 2416: @item argument type incompatible with parameter:
 2417: This is usually not caught. Some words perform checks, e.g., the control
 2418: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
 2419: mismatch).
 2420: 
 2421: @item attempting to obtain the execution token of a word with undefined execution semantics:
 2422: You get an execution token representing the compilation semantics
 2423: instead.
 2424: 
 2425: @item dividing by zero:
 2426: typically results in a @code{-55 throw} (floating point unidentified
 2427: fault), although a @code{-10 throw} (divide by zero) would be more
 2428: appropriate.
 2429: 
 2430: @item insufficient data stack or return stack space:
 2431: Not checked. This typically results in mysterious illegal memory
 2432: accesses, producing @code{-9 throw} (Invalid memory address) or
 2433: @code{-23 throw} (Address alignment exception).
 2434: 
 2435: @item insufficient space for loop control parameters:
 2436: like other return stack overflows.
 2437: 
 2438: @item insufficient space in the dictionary:
 2439: Not checked. Similar results as stack overflows. However, typically the
 2440: error appears at a different place when one inserts or removes code.
 2441: 
 2442: @item interpreting a word with undefined interpretation semantics:
 2443: For some words, we defined interpretation semantics. For the others:
 2444: @code{-14 throw} (Interpreting a compile-only word). Note that this is
 2445: checked only by the outer (aka text) interpreter; if the word is
 2446: @code{execute}d in some other way, it will typically perform it's
 2447: compilation semantics even in interpret state. (We could change @code{'}
 2448: and relatives not to give the xt of such words, but we think that would
 2449: be too restrictive).
 2450: 
 2451: @item modifying the contents of the input buffer or a string literal:
 2452: These are located in writable memory and can be modified.
 2453: 
 2454: @item overflow of the pictured numeric output string:
 2455: Not checked.
 2456: 
 2457: @item parsed string overflow:
 2458: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
 2459: 
 2460: @item producing a result out of range:
 2461: On two's complement machines, arithmetic is performed modulo
 2462: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 2463: arithmetic (with appropriate mapping for signed types). Division by zero
 2464: typically results in a @code{-55 throw} (floatingpoint unidentified
 2465: fault), although a @code{-10 throw} (divide by zero) would be more
 2466: appropriate. @code{convert} and @code{>number} currently overflow
 2467: silently.
 2468: 
 2469: @item reading from an empty data or return stack:
 2470: The data stack is checked by the outer (aka text) interpreter after
 2471: every word executed. If it has underflowed, a @code{-4 throw} (Stack
 2472: underflow) is performed. Apart from that, the stacks are not checked and
 2473: underflows can result in similar behaviour as overflows (of adjacent
 2474: stacks).
 2475: 
 2476: @item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
 2477: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
 2478: use zero-length string as a name). Words like @code{'} probably will not
 2479: find what they search. Note that it is possible to create zero-length
 2480: names with @code{nextname} (should it not?).
 2481: 
 2482: @item @code{>IN} greater than input buffer:
 2483: The next invocation of a parsing word returns a string wih length 0.
 2484: 
 2485: @item @code{RECURSE} appears after @code{DOES>}:
 2486: Compiles a recursive call to the defining word not to the defined word.
 2487: 
 2488: @item argument input source different than current input source for @code{RESTORE-INPUT}:
 2489: !!???If the argument input source is a valid input source then it gets
 2490: restored. Otherwise causes @code{-12 THROW}, which, unless caught, issues
 2491: the message "argument type mismatch" and aborts.
 2492: 
 2493: @item data space containing definitions gets de-allocated:
 2494: Deallocation with @code{allot} is not checked. This typically resuls in
 2495: memory access faults or execution of illegal instructions.
 2496: 
 2497: @item data space read/write with incorrect alignment:
 2498: Processor-dependent. Typically results in a @code{-23 throw} (Address
 2499: alignment exception). Under Linux on a 486 or later processor with
 2500: alignment turned on, incorrect alignment results in a @code{-9 throw}
 2501: (Invalid memory address). There are reportedly some processors with
 2502: alignment restrictions that do not report them.
 2503: 
 2504: @item data space pointer not properly aligned, @code{,}, @code{C,}:
 2505: Like other alignment errors.
 2506: 
 2507: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
 2508: Not checked. May cause an illegal memory access.
 2509: 
 2510: @item loop control parameters not available:
 2511: Not checked. The counted loop words simply assume that the top of return
 2512: stack items are loop control parameters and behave accordingly.
 2513: 
 2514: @item most recent definition does not have a name (@code{IMMEDIATE}):
 2515: @code{abort" last word was headerless"}.
 2516: 
 2517: @item name not defined by @code{VALUE} used by @code{TO}:
 2518: @code{-32 throw} (Invalid name argument)
 2519: 
 2520: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
 2521: @code{-13 throw} (Undefined word)
 2522: 
 2523: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
 2524: Gforth behaves as if they were of the same type. I.e., you can predict
 2525: the behaviour by interpreting all parameters as, e.g., signed.
 2526: 
 2527: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
 2528: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
 2529: @code{TO}.
 2530: 
 2531: @item String longer than a counted string returned by @code{WORD}:
 2532: Not checked. The string will be ok, but the count will, of course,
 2533: contain only the least significant bits of the length.
 2534: 
 2535: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
 2536: Processor-dependent. Typical behaviours are returning 0 and using only
 2537: the low bits of the shift count.
 2538: 
 2539: @item word not defined via @code{CREATE}:
 2540: @code{>BODY} produces the PFA of the word no matter how it was defined.
 2541: 
 2542: @code{DOES>} changes the execution semantics of the last defined word no
 2543: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
 2544: @code{CREATE , DOES>}.
 2545: 
 2546: @item words improperly used outside @code{<#} and @code{#>}:
 2547: Not checked. As usual, you can expect memory faults.
 2548: 
 2549: @end table
 2550: 
 2551: 
 2552: @c ---------------------------------------------------------------------
 2553: @node core-other,  , core-ambcond, The Core Words
 2554: @subsection Other system documentation
 2555: @c ---------------------------------------------------------------------
 2556: 
 2557: @table @i
 2558: 
 2559: @item nonstandard words using @code{PAD}:
 2560: None.
 2561: 
 2562: @item operator's terminal facilities available:
 2563: !!??
 2564: 
 2565: @item program data space available:
 2566: @code{sp@ here - .} gives the space remaining for dictionary and data
 2567: stack together.
 2568: 
 2569: @item return stack space available:
 2570: !!??
 2571: 
 2572: @item stack space available:
 2573: @code{sp@ here - .} gives the space remaining for dictionary and data
 2574: stack together.
 2575: 
 2576: @item system dictionary space required, in address units:
 2577: Type @code{here forthstart - .} after startup. At the time of this
 2578: writing, this gives 70108 (bytes) on a 32-bit system.
 2579: @end table
 2580: 
 2581: 
 2582: @c =====================================================================
 2583: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
 2584: @section The optional Block word set
 2585: @c =====================================================================
 2586: 
 2587: @menu
 2588: * block-idef::                  Implementation Defined Options                  
 2589: * block-ambcond::               Ambiguous Conditions               
 2590: * block-other::                 Other System Documentation                 
 2591: @end menu
 2592: 
 2593: 
 2594: @c ---------------------------------------------------------------------
 2595: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
 2596: @subsection Implementation Defined Options
 2597: @c ---------------------------------------------------------------------
 2598: 
 2599: @table @i
 2600: 
 2601: @item the format for display by @code{LIST}:
 2602: First the screen number is displayed, then 16 lines of 64 characters,
 2603: each line preceded by the line number.
 2604: 
 2605: @item the length of a line affected by @code{\}:
 2606: 64 characters.
 2607: @end table
 2608: 
 2609: 
 2610: @c ---------------------------------------------------------------------
 2611: @node block-ambcond, block-other, block-idef, The optional Block word set
 2612: @subsection Ambiguous conditions
 2613: @c ---------------------------------------------------------------------
 2614: 
 2615: @table @i
 2616: 
 2617: @item correct block read was not possible:
 2618: Typically results in a @code{throw} of some OS-derived value (between
 2619: -512 and -2048). If the blocks file was just not long enough, blanks are
 2620: supplied for the missing portion.
 2621: 
 2622: @item I/O exception in block transfer:
 2623: Typically results in a @code{throw} of some OS-derived value (between
 2624: -512 and -2048).
 2625: 
 2626: @item invalid block number:
 2627: @code{-35 throw} (Invalid block number)
 2628: 
 2629: @item a program directly alters the contents of @code{BLK}:
 2630: The input stream is switched to that other block, at the same
 2631: position. If the storing to @code{BLK} happens when interpreting
 2632: non-block input, the system will get quite confused when the block ends.
 2633: 
 2634: @item no current block buffer for @code{UPDATE}:
 2635: @code{UPDATE} has no effect.
 2636: 
 2637: @end table
 2638: 
 2639: 
 2640: @c ---------------------------------------------------------------------
 2641: @node block-other,  , block-ambcond, The optional Block word set
 2642: @subsection Other system documentation
 2643: @c ---------------------------------------------------------------------
 2644: 
 2645: @table @i
 2646: 
 2647: @item any restrictions a multiprogramming system places on the use of buffer addresses:
 2648: No restrictions (yet).
 2649: 
 2650: @item the number of blocks available for source and data:
 2651: depends on your disk space.
 2652: 
 2653: @end table
 2654: 
 2655: 
 2656: @c =====================================================================
 2657: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
 2658: @section The optional Double Number word set
 2659: @c =====================================================================
 2660: 
 2661: @menu
 2662: * double-ambcond::              Ambiguous Conditions              
 2663: @end menu
 2664: 
 2665: 
 2666: @c ---------------------------------------------------------------------
 2667: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
 2668: @subsection Ambiguous conditions
 2669: @c ---------------------------------------------------------------------
 2670: 
 2671: @table @i
 2672: 
 2673: @item @var{d} outside of range of @var{n} in @code{D>S}:
 2674: The least significant cell of @var{d} is produced.
 2675: 
 2676: @end table
 2677: 
 2678: 
 2679: @c =====================================================================
 2680: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
 2681: @section The optional Exception word set
 2682: @c =====================================================================
 2683: 
 2684: @menu
 2685: * exception-idef::              Implementation Defined Options              
 2686: @end menu
 2687: 
 2688: 
 2689: @c ---------------------------------------------------------------------
 2690: @node exception-idef,  , The optional Exception word set, The optional Exception word set
 2691: @subsection Implementation Defined Options
 2692: @c ---------------------------------------------------------------------
 2693: 
 2694: @table @i
 2695: @item @code{THROW}-codes used in the system:
 2696: The codes -256@minus{}-511 are used for reporting signals (see
 2697: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
 2698: (for file and memory allocation operations). The mapping from OS error
 2699: numbers to throw code is -512@minus{}@var{errno}. One side effect of
 2700: this mapping is that undefined OS errors produce a message with a
 2701: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
 2702: 488} on my system.
 2703: @end table
 2704: 
 2705: @c =====================================================================
 2706: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
 2707: @section The optional Facility word set
 2708: @c =====================================================================
 2709: 
 2710: @menu
 2711: * facility-idef::               Implementation Defined Options               
 2712: * facility-ambcond::            Ambiguous Conditions            
 2713: @end menu
 2714: 
 2715: 
 2716: @c ---------------------------------------------------------------------
 2717: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
 2718: @subsection Implementation Defined Options
 2719: @c ---------------------------------------------------------------------
 2720: 
 2721: @table @i
 2722: 
 2723: @item encoding of keyboard events (@code{EKEY}):
 2724: Not yet implemeted.
 2725: 
 2726: @item duration of a system clock tick
 2727: System dependent. With respect to @code{MS}, the time is specified in
 2728: microseconds. How well the OS and the hardware implement this, is
 2729: another question.
 2730: 
 2731: @item repeatability to be expected from the execution of @code{MS}:
 2732: System dependent. On Unix, a lot depends on load. If the system is
 2733: lightly loaded, and the delay is short enough that Gforth does not get
 2734: swapped out, the performance should be acceptable. Under MS-DOS and
 2735: other single-tasking systems, it should be good.
 2736: 
 2737: @end table
 2738: 
 2739: 
 2740: @c ---------------------------------------------------------------------
 2741: @node facility-ambcond,  , facility-idef, The optional Facility word set
 2742: @subsection Ambiguous conditions
 2743: @c ---------------------------------------------------------------------
 2744: 
 2745: @table @i
 2746: 
 2747: @item @code{AT-XY} can't be performed on user output device:
 2748: Largely terminal dependant. No range checks are done on the arguments.
 2749: No errors are reported. You may see some garbage appearing, you may see
 2750: simply nothing happen.
 2751: 
 2752: @end table
 2753: 
 2754: 
 2755: @c =====================================================================
 2756: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
 2757: @section The optional File-Access word set
 2758: @c =====================================================================
 2759: 
 2760: @menu
 2761: * file-idef::                   Implementation Defined Options                   
 2762: * file-ambcond::                Ambiguous Conditions                
 2763: @end menu
 2764: 
 2765: 
 2766: @c ---------------------------------------------------------------------
 2767: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
 2768: @subsection Implementation Defined Options
 2769: @c ---------------------------------------------------------------------
 2770: 
 2771: @table @i
 2772: 
 2773: @item File access methods used:
 2774: @code{R/O}, @code{R/W} and @code{BIN} work as you would
 2775: expect. @code{W/O} translates into the C file opening mode @code{w} (or
 2776: @code{wb}): The file is cleared, if it exists, and created, if it does
 2777: not (both with @code{open-file} and @code{create-file}).  Under Unix
 2778: @code{create-file} creates a file with 666 permissions modified by your
 2779: umask.
 2780: 
 2781: @item file exceptions:
 2782: The file words do not raise exceptions (except, perhaps, memory access
 2783: faults when you pass illegal addresses or file-ids).
 2784: 
 2785: @item file line terminator:
 2786: System-dependent. Gforth uses C's newline character as line
 2787: terminator. What the actual character code(s) of this are is
 2788: system-dependent.
 2789: 
 2790: @item file name format
 2791: System dependent. Gforth just uses the file name format of your OS.
 2792: 
 2793: @item information returned by @code{FILE-STATUS}:
 2794: @code{FILE-STATUS} returns the most powerful file access mode allowed
 2795: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
 2796: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
 2797: along with the retured mode.
 2798: 
 2799: @item input file state after an exception when including source:
 2800: All files that are left via the exception are closed.
 2801: 
 2802: @item @var{ior} values and meaning:
 2803: The @var{ior}s returned by the file and memory allocation words are
 2804: intended as throw codes. They typically are in the range
 2805: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 2806: @var{ior}s is -512@minus{}@var{errno}.
 2807: 
 2808: @item maximum depth of file input nesting:
 2809: limited by the amount of return stack, locals/TIB stack, and the number
 2810: of open files available. This should not give you troubles.
 2811: 
 2812: @item maximum size of input line:
 2813: @code{/line}. Currently 255.
 2814: 
 2815: @item methods of mapping block ranges to files:
 2816: Currently, the block words automatically access the file
 2817: @file{blocks.fb} in the currend working directory. More sophisticated
 2818: methods could be implemented if there is demand (and a volunteer).
 2819: 
 2820: @item number of string buffers provided by @code{S"}:
 2821: 1
 2822: 
 2823: @item size of string buffer used by @code{S"}:
 2824: @code{/line}. currently 255.
 2825: 
 2826: @end table
 2827: 
 2828: @c ---------------------------------------------------------------------
 2829: @node file-ambcond,  , file-idef, The optional File-Access word set
 2830: @subsection Ambiguous conditions
 2831: @c ---------------------------------------------------------------------
 2832: 
 2833: @table @i
 2834: 
 2835: @item attempting to position a file outside it's boundaries:
 2836: @code{REPOSITION-FILE} is performed as usual: Afterwards,
 2837: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
 2838: 
 2839: @item attempting to read from file positions not yet written:
 2840: End-of-file, i.e., zero characters are read and no error is reported.
 2841: 
 2842: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
 2843: An appropriate exception may be thrown, but a memory fault or other
 2844: problem is more probable.
 2845: 
 2846: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
 2847: The @var{ior} produced by the operation, that discovered the problem, is
 2848: thrown.
 2849: 
 2850: @item named file cannot be opened (@code{included}):
 2851: The @var{ior} produced by @code{open-file} is thrown.
 2852: 
 2853: @item requesting an unmapped block number:
 2854: There are no unmapped legal block numbers. On some operating systems,
 2855: writing a block with a large number may overflow the file system and
 2856: have an error message as consequence.
 2857: 
 2858: @item using @code{source-id} when @code{blk} is non-zero:
 2859: @code{source-id} performs its function. Typically it will give the id of
 2860: the source which loaded the block. (Better ideas?)
 2861: 
 2862: @end table
 2863: 
 2864: 
 2865: @c =====================================================================
 2866: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
 2867: @section The optional Floating-Point word set
 2868: @c =====================================================================
 2869: 
 2870: @menu
 2871: * floating-idef::               Implementation Defined Options
 2872: * floating-ambcond::            Ambiguous Conditions            
 2873: @end menu
 2874: 
 2875: 
 2876: @c ---------------------------------------------------------------------
 2877: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
 2878: @subsection Implementation Defined Options
 2879: @c ---------------------------------------------------------------------
 2880: 
 2881: @table @i
 2882: 
 2883: @item format and range of floating point numbers:
 2884: System-dependent; the @code{double} type of C.
 2885: 
 2886: @item results of @code{REPRESENT} when @var{float} is out of range:
 2887: System dependent; @code{REPRESENT} is implemented using the C library
 2888: function @code{ecvt()} and inherits its behaviour in this respect.
 2889: 
 2890: @item rounding or truncation of floating-point numbers:
 2891: What's the question?!!
 2892: 
 2893: @item size of floating-point stack:
 2894: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
 2895: with the command-line option @code{-f}.
 2896: 
 2897: @item width of floating-point stack:
 2898: @code{1 floats}.
 2899: 
 2900: @end table
 2901: 
 2902: 
 2903: @c ---------------------------------------------------------------------
 2904: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
 2905: @subsection Ambiguous conditions
 2906: @c ---------------------------------------------------------------------
 2907: 
 2908: @table @i
 2909: 
 2910: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
 2911: System-dependent. Typically results in an alignment fault like other
 2912: alignment violations.
 2913: 
 2914: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
 2915: System-dependent. Typically results in an alignment fault like other
 2916: alignment violations.
 2917: 
 2918: @item Floating-point result out of range:
 2919: System-dependent. Can result in a @code{-55 THROW} (Floating-point
 2920: unidentified fault), or can produce a special value representing, e.g.,
 2921: Infinity.
 2922: 
 2923: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
 2924: System-dependent. Typically results in an alignment fault like other
 2925: alignment violations.
 2926: 
 2927: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
 2928: The floating-point number is converted into decimal nonetheless.
 2929: 
 2930: @item Both arguments are equal to zero (@code{FATAN2}):
 2931: System-dependent. @code{FATAN2} is implemented using the C library
 2932: function @code{atan2()}.
 2933: 
 2934: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
 2935: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
 2936: because of small errors and the tan will be a very large (or very small)
 2937: but finite number.
 2938: 
 2939: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
 2940: The result is rounded to the nearest float.
 2941: 
 2942: @item dividing by zero:
 2943: @code{-55 throw} (Floating-point unidentified fault)
 2944: 
 2945: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
 2946: System dependent. On IEEE-FP based systems the number is converted into
 2947: an infinity.
 2948: 
 2949: @item @var{float}<1 (@code{facosh}):
 2950: @code{-55 throw} (Floating-point unidentified fault)
 2951: 
 2952: @item @var{float}=<-1 (@code{flnp1}):
 2953: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 2954: negative infinity is typically produced for @var{float}=-1.
 2955: 
 2956: @item @var{float}=<0 (@code{fln}, @code{flog}):
 2957: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 2958: negative infinity is typically produced for @var{float}=0.
 2959: 
 2960: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
 2961: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
 2962: produces values for these inputs on my Linux box (Bug in the C library?)
 2963: 
 2964: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
 2965: @code{-55 throw} (Floating-point unidentified fault).
 2966: 
 2967: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
 2968: @code{-55 throw} (Floating-point unidentified fault).
 2969: 
 2970: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
 2971: This does not happen.
 2972: @end table
 2973: 
 2974: 
 2975: 
 2976: @c =====================================================================
 2977: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
 2978: @section The optional Locals word set
 2979: @c =====================================================================
 2980: 
 2981: @menu
 2982: * locals-idef::                 Implementation Defined Options                 
 2983: * locals-ambcond::              Ambiguous Conditions              
 2984: @end menu
 2985: 
 2986: 
 2987: @c ---------------------------------------------------------------------
 2988: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
 2989: @subsection Implementation Defined Options
 2990: @c ---------------------------------------------------------------------
 2991: 
 2992: @table @i
 2993: 
 2994: @item maximum number of locals in a definition:
 2995: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
 2996: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
 2997: characters. The number of locals in a definition is bounded by the size
 2998: of locals-buffer, which contains the names of the locals.
 2999: 
 3000: @end table
 3001: 
 3002: 
 3003: @c ---------------------------------------------------------------------
 3004: @node locals-ambcond,  , locals-idef, The optional Locals word set
 3005: @subsection Ambiguous conditions
 3006: @c ---------------------------------------------------------------------
 3007: 
 3008: @table @i
 3009: 
 3010: @item executing a named local in interpretation state:
 3011: @code{-14 throw} (Interpreting a compile-only word).
 3012: 
 3013: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
 3014: @code{-32 throw} (Invalid name argument)
 3015: 
 3016: @end table
 3017: 
 3018: 
 3019: @c =====================================================================
 3020: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
 3021: @section The optional Memory-Allocation word set
 3022: @c =====================================================================
 3023: 
 3024: @menu
 3025: * memory-idef::                 Implementation Defined Options                 
 3026: @end menu
 3027: 
 3028: 
 3029: @c ---------------------------------------------------------------------
 3030: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
 3031: @subsection Implementation Defined Options
 3032: @c ---------------------------------------------------------------------
 3033: 
 3034: @table @i
 3035: 
 3036: @item values and meaning of @var{ior}:
 3037: The @var{ior}s returned by the file and memory allocation words are
 3038: intended as throw codes. They typically are in the range
 3039: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 3040: @var{ior}s is -512@minus{}@var{errno}.
 3041: 
 3042: @end table
 3043: 
 3044: @c =====================================================================
 3045: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
 3046: @section The optional Programming-Tools word set
 3047: @c =====================================================================
 3048: 
 3049: @menu
 3050: * programming-idef::            Implementation Defined Options            
 3051: * programming-ambcond::         Ambiguous Conditions         
 3052: @end menu
 3053: 
 3054: 
 3055: @c ---------------------------------------------------------------------
 3056: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
 3057: @subsection Implementation Defined Options
 3058: @c ---------------------------------------------------------------------
 3059: 
 3060: @table @i
 3061: 
 3062: @item ending sequence for input following @code{;code} and @code{code}:
 3063: Not implemented (yet).
 3064: 
 3065: @item manner of processing input following @code{;code} and @code{code}:
 3066: Not implemented (yet).
 3067: 
 3068: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
 3069: Not implemented (yet). If they were implemented, they would use the
 3070: search order wordset.
 3071: 
 3072: @item source and format of display by @code{SEE}:
 3073: The source for @code{see} is the intermediate code used by the inner
 3074: interpreter.  The current @code{see} tries to output Forth source code
 3075: as well as possible.
 3076: 
 3077: @end table
 3078: 
 3079: @c ---------------------------------------------------------------------
 3080: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
 3081: @subsection Ambiguous conditions
 3082: @c ---------------------------------------------------------------------
 3083: 
 3084: @table @i
 3085: 
 3086: @item deleting the compilation wordlist (@code{FORGET}):
 3087: Not implemented (yet).
 3088: 
 3089: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
 3090: This typically results in an @code{abort"} with a descriptive error
 3091: message (may change into a @code{-22 throw} (Control structure mismatch)
 3092: in the future). You may also get a memory access error. If you are
 3093: unlucky, this ambiguous condition is not caught.
 3094: 
 3095: @item @var{name} can't be found (@code{forget}):
 3096: Not implemented (yet).
 3097: 
 3098: @item @var{name} not defined via @code{CREATE}:
 3099: @code{;code} is not implemented (yet). If it were, it would behave like
 3100: @code{DOES>} in this respect, i.e., change the execution semantics of
 3101: the last defined word no matter how it was defined.
 3102: 
 3103: @item @code{POSTPONE} applied to @code{[IF]}:
 3104: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
 3105: equivalent to @code{[IF]}.
 3106: 
 3107: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
 3108: Continue in the same state of conditional compilation in the next outer
 3109: input source. Currently there is no warning to the user about this.
 3110: 
 3111: @item removing a needed definition (@code{FORGET}):
 3112: Not implemented (yet).
 3113: 
 3114: @end table
 3115: 
 3116: 
 3117: @c =====================================================================
 3118: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
 3119: @section The optional Search-Order word set
 3120: @c =====================================================================
 3121: 
 3122: @menu
 3123: * search-idef::                 Implementation Defined Options                 
 3124: * search-ambcond::              Ambiguous Conditions              
 3125: @end menu
 3126: 
 3127: 
 3128: @c ---------------------------------------------------------------------
 3129: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
 3130: @subsection Implementation Defined Options
 3131: @c ---------------------------------------------------------------------
 3132: 
 3133: @table @i
 3134: 
 3135: @item maximum number of word lists in search order:
 3136: @code{s" wordlists" environment? drop .}. Currently 16.
 3137: 
 3138: @item minimum search order:
 3139: @code{root root}.
 3140: 
 3141: @end table
 3142: 
 3143: @c ---------------------------------------------------------------------
 3144: @node search-ambcond,  , search-idef, The optional Search-Order word set
 3145: @subsection Ambiguous conditions
 3146: @c ---------------------------------------------------------------------
 3147: 
 3148: @table @i
 3149: 
 3150: @item changing the compilation wordlist (during compilation):
 3151: The definition is put into the wordlist that is the compilation wordlist
 3152: when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
 3153: @code{RECURSIVE}, etc.).
 3154: 
 3155: @item search order empty (@code{previous}):
 3156: @code{abort" Vocstack empty"}.
 3157: 
 3158: @item too many word lists in search order (@code{also}):
 3159: @code{abort" Vocstack full"}.
 3160: 
 3161: @end table
 3162: 
 3163: 
 3164: @node Model, Emacs and Gforth, ANS conformance, Top
 3165: @chapter Model
 3166: 
 3167: @node Emacs and Gforth, Internals, Model, Top
 3168: @chapter Emacs and Gforth
 3169: 
 3170: Gforth comes with @file{gforth.el}, an improved version of
 3171: @file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
 3172: improvements are a better (but still not perfect) handling of
 3173: indentation. I have also added comment paragraph filling (@kbd{M-q}),
 3174: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
 3175: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
 3176: stuff I do not use alone, even though some of it only makes sense for
 3177: TILE. To get a description of these features, enter Forth mode and type
 3178: @kbd{C-h m}.
 3179: 
 3180: In addition, Gforth supports Emacs quite well: The source code locations
 3181: given in error messages, debugging output (from @code{~~}) and failed
 3182: assertion messages are in the right format for Emacs' compilation mode
 3183: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
 3184: Manual}) so the source location corresponding to an error or other
 3185: message is only a few keystrokes away (@kbd{C-x `} for the next error,
 3186: @kbd{C-c C-c} for the error under the cursor).
 3187: 
 3188: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
 3189: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
 3190: contains the definitions of all words defined afterwards. You can then
 3191: find the source for a word using @kbd{M-.}. Note that emacs can use
 3192: several tags files at the same time (e.g., one for the Gforth sources
 3193: and one for your program).
 3194: 
 3195: To get all these benefits, add the following lines to your @file{.emacs}
 3196: file:
 3197: 
 3198: @example
 3199: (autoload 'forth-mode "gforth.el")
 3200: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
 3201: @end example
 3202: 
 3203: @node Internals, Bugs, Emacs and Gforth, Top
 3204: @chapter Internals
 3205: 
 3206: Reading this section is not necessary for programming with Gforth. It
 3207: should be helpful for finding your way in the Gforth sources.
 3208: 
 3209: @menu
 3210: * Portability::                 
 3211: * Threading::                   
 3212: * Primitives::                  
 3213: * System Architecture::         
 3214: * Performance::                 
 3215: @end menu
 3216: 
 3217: @node Portability, Threading, Internals, Internals
 3218: @section Portability
 3219: 
 3220: One of the main goals of the effort is availability across a wide range
 3221: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
 3222: this goal by manually coding the engine in assembly language for several
 3223: then-popular processors. This approach is very labor-intensive and the
 3224: results are short-lived due to progress in computer architecture.
 3225: 
 3226: Others have avoided this problem by coding in C, e.g., Mitch Bradley
 3227: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
 3228: particularly popular for UNIX-based Forths due to the large variety of
 3229: architectures of UNIX machines. Unfortunately an implementation in C
 3230: does not mix well with the goals of efficiency and with using
 3231: traditional techniques: Indirect or direct threading cannot be expressed
 3232: in C, and switch threading, the fastest technique available in C, is
 3233: significantly slower. Another problem with C is that it's very
 3234: cumbersome to express double integer arithmetic.
 3235: 
 3236: Fortunately, there is a portable language that does not have these
 3237: limitations: GNU C, the version of C processed by the GNU C compiler
 3238: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
 3239: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
 3240: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
 3241: threading possible, its @code{long long} type (@pxref{Long Long, ,
 3242: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
 3243: double numbers. GNU C is available for free on all important (and many
 3244: unimportant) UNIX machines, VMS, 80386s running MS-DOS, the Amiga, and
 3245: the Atari ST, so a Forth written in GNU C can run on all these
 3246: machines.
 3247: 
 3248: Writing in a portable language has the reputation of producing code that
 3249: is slower than assembly. For our Forth engine we repeatedly looked at
 3250: the code produced by the compiler and eliminated most compiler-induced
 3251: inefficiencies by appropriate changes in the source-code.
 3252: 
 3253: However, register allocation cannot be portably influenced by the
 3254: programmer, leading to some inefficiencies on register-starved
 3255: machines. We use explicit register declarations (@pxref{Explicit Reg
 3256: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
 3257: improve the speed on some machines. They are turned on by using the
 3258: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
 3259: only depends on the machine, but also on the compiler version: On some
 3260: machines some compiler versions produce incorrect code when certain
 3261: explicit register declarations are used. So by default
 3262: @code{-DFORCE_REG} is not used.
 3263: 
 3264: @node Threading, Primitives, Portability, Internals
 3265: @section Threading
 3266: 
 3267: GNU C's labels as values extension (available since @code{gcc-2.0},
 3268: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
 3269: makes it possible to take the address of @var{label} by writing
 3270: @code{&&@var{label}}.  This address can then be used in a statement like
 3271: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
 3272: @code{goto x}.
 3273: 
 3274: With this feature an indirect threaded NEXT looks like:
 3275: @example
 3276: cfa = *ip++;
 3277: ca = *cfa;
 3278: goto *ca;
 3279: @end example
 3280: For those unfamiliar with the names: @code{ip} is the Forth instruction
 3281: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
 3282: execution token and points to the code field of the next word to be
 3283: executed; The @code{ca} (code address) fetched from there points to some
 3284: executable code, e.g., a primitive or the colon definition handler
 3285: @code{docol}.
 3286: 
 3287: Direct threading is even simpler:
 3288: @example
 3289: ca = *ip++;
 3290: goto *ca;
 3291: @end example
 3292: 
 3293: Of course we have packaged the whole thing neatly in macros called
 3294: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
 3295: 
 3296: @menu
 3297: * Scheduling::                  
 3298: * Direct or Indirect Threaded?::  
 3299: * DOES>::                       
 3300: @end menu
 3301: 
 3302: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
 3303: @subsection Scheduling
 3304: 
 3305: There is a little complication: Pipelined and superscalar processors,
 3306: i.e., RISC and some modern CISC machines can process independent
 3307: instructions while waiting for the results of an instruction. The
 3308: compiler usually reorders (schedules) the instructions in a way that
 3309: achieves good usage of these delay slots. However, on our first tries
 3310: the compiler did not do well on scheduling primitives. E.g., for
 3311: @code{+} implemented as
 3312: @example
 3313: n=sp[0]+sp[1];
 3314: sp++;
 3315: sp[0]=n;
 3316: NEXT;
 3317: @end example
 3318: the NEXT comes strictly after the other code, i.e., there is nearly no
 3319: scheduling. After a little thought the problem becomes clear: The
 3320: compiler cannot know that sp and ip point to different addresses (and
 3321: the version of @code{gcc} we used would not know it even if it was
 3322: possible), so it could not move the load of the cfa above the store to
 3323: the TOS. Indeed the pointers could be the same, if code on or very near
 3324: the top of stack were executed. In the interest of speed we chose to
 3325: forbid this probably unused ``feature'' and helped the compiler in
 3326: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
 3327: the goto part (@code{NEXT_P2}). @code{+} now looks like:
 3328: @example
 3329: n=sp[0]+sp[1];
 3330: sp++;
 3331: NEXT_P1;
 3332: sp[0]=n;
 3333: NEXT_P2;
 3334: @end example
 3335: This can be scheduled optimally by the compiler.
 3336: 
 3337: This division can be turned off with the switch @code{-DCISC_NEXT}. This
 3338: switch is on by default on machines that do not profit from scheduling
 3339: (e.g., the 80386), in order to preserve registers.
 3340: 
 3341: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
 3342: @subsection Direct or Indirect Threaded?
 3343: 
 3344: Both! After packaging the nasty details in macro definitions we
 3345: realized that we could switch between direct and indirect threading by
 3346: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
 3347: defining a few machine-specific macros for the direct-threading case.
 3348: On the Forth level we also offer access words that hide the
 3349: differences between the threading methods (@pxref{Threading Words}).
 3350: 
 3351: Indirect threading is implemented completely
 3352: machine-independently. Direct threading needs routines for creating
 3353: jumps to the executable code (e.g. to docol or dodoes). These routines
 3354: are inherently machine-dependent, but they do not amount to many source
 3355: lines. I.e., even porting direct threading to a new machine is a small
 3356: effort.
 3357: 
 3358: @node DOES>,  , Direct or Indirect Threaded?, Threading
 3359: @subsection DOES>
 3360: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
 3361: the chunk of code executed by every word defined by a
 3362: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
 3363: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
 3364: DOES-code)? There are two solutions:
 3365: 
 3366: In fig-Forth the code field points directly to the dodoes and the
 3367: DOES-code address is stored in the cell after the code address
 3368: (i.e. at cfa cell+). It may seem that this solution is illegal in the
 3369: Forth-79 and all later standards, because in fig-Forth this address
 3370: lies in the body (which is illegal in these standards). However, by
 3371: making the code field larger for all words this solution becomes legal
 3372: again. We use this approach for the indirect threaded version. Leaving
 3373: a cell unused in most words is a bit wasteful, but on the machines we
 3374: are targetting this is hardly a problem. The other reason for having a
 3375: code field size of two cells is to avoid having different image files
 3376: for direct and indirect threaded systems (@pxref{System Architecture}).
 3377: 
 3378: The other approach is that the code field points or jumps to the cell
 3379: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
 3380: this address. @code{dodoes} can then get the DOES-code address by
 3381: computing the code address, i.e., the address of the jump to dodoes,
 3382: and add the length of that jump field. A variant of this is to have a
 3383: call to @code{dodoes} after the @code{DOES>}; then the return address
 3384: (which can be found in the return register on RISCs) is the DOES-code
 3385: address. Since the two cells available in the code field are usually
 3386: used up by the jump to the code address in direct threading, we use
 3387: this approach for direct threading. We did not want to add another
 3388: cell to the code field.
 3389: 
 3390: @node Primitives, System Architecture, Threading, Internals
 3391: @section Primitives
 3392: 
 3393: @menu
 3394: * Automatic Generation::        
 3395: * TOS Optimization::            
 3396: * Produced code::               
 3397: @end menu
 3398: 
 3399: @node Automatic Generation, TOS Optimization, Primitives, Primitives
 3400: @subsection Automatic Generation
 3401: 
 3402: Since the primitives are implemented in a portable language, there is no
 3403: longer any need to minimize the number of primitives. On the contrary,
 3404: having many primitives is an advantage: speed. In order to reduce the
 3405: number of errors in primitives and to make programming them easier, we
 3406: provide a tool, the primitive generator (@file{prims2x.fs}), that
 3407: automatically generates most (and sometimes all) of the C code for a
 3408: primitive from the stack effect notation.  The source for a primitive
 3409: has the following form:
 3410: 
 3411: @format
 3412: @var{Forth-name}	@var{stack-effect}	@var{category}	[@var{pronounc.}]
 3413: [@code{""}@var{glossary entry}@code{""}]
 3414: @var{C code}
 3415: [@code{:}
 3416: @var{Forth code}]
 3417: @end format
 3418: 
 3419: The items in brackets are optional. The category and glossary fields
 3420: are there for generating the documentation, the Forth code is there
 3421: for manual implementations on machines without GNU C. E.g., the source
 3422: for the primitive @code{+} is:
 3423: @example
 3424: +    n1 n2 -- n    core    plus
 3425: n = n1+n2;
 3426: @end example
 3427: 
 3428: This looks like a specification, but in fact @code{n = n1+n2} is C
 3429: code. Our primitive generation tool extracts a lot of information from
 3430: the stack effect notations@footnote{We use a one-stack notation, even
 3431: though we have separate data and floating-point stacks; The separate
 3432: notation can be generated easily from the unified notation.}: The number
 3433: of items popped from and pushed on the stack, their type, and by what
 3434: name they are referred to in the C code. It then generates a C code
 3435: prelude and postlude for each primitive. The final C code for @code{+}
 3436: looks like this:
 3437: 
 3438: @example
 3439: I_plus:	/* + ( n1 n2 -- n ) */  /* label, stack effect */
 3440: /*  */                          /* documentation */
 3441: @{
 3442: DEF_CA                          /* definition of variable ca (indirect threading) */
 3443: Cell n1;                        /* definitions of variables */
 3444: Cell n2;
 3445: Cell n;
 3446: n1 = (Cell) sp[1];              /* input */
 3447: n2 = (Cell) TOS;
 3448: sp += 1;                        /* stack adjustment */
 3449: NAME("+")                       /* debugging output (with -DDEBUG) */
 3450: @{
 3451: n = n1+n2;                      /* C code taken from the source */
 3452: @}
 3453: NEXT_P1;                        /* NEXT part 1 */
 3454: TOS = (Cell)n;                  /* output */
 3455: NEXT_P2;                        /* NEXT part 2 */
 3456: @}
 3457: @end example
 3458: 
 3459: This looks long and inefficient, but the GNU C compiler optimizes quite
 3460: well and produces optimal code for @code{+} on, e.g., the R3000 and the
 3461: HP RISC machines: Defining the @code{n}s does not produce any code, and
 3462: using them as intermediate storage also adds no cost.
 3463: 
 3464: There are also other optimizations, that are not illustrated by this
 3465: example: Assignments between simple variables are usually for free (copy
 3466: propagation). If one of the stack items is not used by the primitive
 3467: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
 3468: (dead code elimination). On the other hand, there are some things that
 3469: the compiler does not do, therefore they are performed by
 3470: @file{prims2x.fs}: The compiler does not optimize code away that stores
 3471: a stack item to the place where it just came from (e.g., @code{over}).
 3472: 
 3473: While programming a primitive is usually easy, there are a few cases
 3474: where the programmer has to take the actions of the generator into
 3475: account, most notably @code{?dup}, but also words that do not (always)
 3476: fall through to NEXT.
 3477: 
 3478: @node TOS Optimization, Produced code, Automatic Generation, Primitives
 3479: @subsection TOS Optimization
 3480: 
 3481: An important optimization for stack machine emulators, e.g., Forth
 3482: engines, is keeping  one or more of the top stack items in
 3483: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
 3484: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
 3485: @itemize
 3486: @item
 3487: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
 3488: due to fewer loads from and stores to the stack.
 3489: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
 3490: @var{y<n}, due to additional moves between registers.
 3491: @end itemize
 3492: 
 3493: In particular, keeping one item in a register is never a disadvantage,
 3494: if there are enough registers. Keeping two items in registers is a
 3495: disadvantage for frequent words like @code{?branch}, constants,
 3496: variables, literals and @code{i}. Therefore our generator only produces
 3497: code that keeps zero or one items in registers. The generated C code
 3498: covers both cases; the selection between these alternatives is made at
 3499: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
 3500: code for @code{+} is just a simple variable name in the one-item case,
 3501: otherwise it is a macro that expands into @code{sp[0]}. Note that the
 3502: GNU C compiler tries to keep simple variables like @code{TOS} in
 3503: registers, and it usually succeeds, if there are enough registers.
 3504: 
 3505: The primitive generator performs the TOS optimization for the
 3506: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
 3507: operations the benefit of this optimization is even larger:
 3508: floating-point operations take quite long on most processors, but can be
 3509: performed in parallel with other operations as long as their results are
 3510: not used. If the FP-TOS is kept in a register, this works. If
 3511: it is kept on the stack, i.e., in memory, the store into memory has to
 3512: wait for the result of the floating-point operation, lengthening the
 3513: execution time of the primitive considerably.
 3514: 
 3515: The TOS optimization makes the automatic generation of primitives a
 3516: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
 3517: @code{TOS} is not sufficient. There are some special cases to
 3518: consider:
 3519: @itemize
 3520: @item In the case of @code{dup ( w -- w w )} the generator must not
 3521: eliminate the store to the original location of the item on the stack,
 3522: if the TOS optimization is turned on.
 3523: @item Primitives with stack effects of the form @code{--}
 3524: @var{out1}...@var{outy} must store the TOS to the stack at the start.
 3525: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
 3526: must load the TOS from the stack at the end. But for the null stack
 3527: effect @code{--} no stores or loads should be generated.
 3528: @end itemize
 3529: 
 3530: @node Produced code,  , TOS Optimization, Primitives
 3531: @subsection Produced code
 3532: 
 3533: To see what assembly code is produced for the primitives on your machine
 3534: with your compiler and your flag settings, type @code{make engine.s} and
 3535: look at the resulting file @file{engine.s}.
 3536: 
 3537: @node System Architecture, Performance, Primitives, Internals
 3538: @section System Architecture
 3539: 
 3540: Our Forth system consists not only of primitives, but also of
 3541: definitions written in Forth. Since the Forth compiler itself belongs
 3542: to those definitions, it is not possible to start the system with the
 3543: primitives and the Forth source alone. Therefore we provide the Forth
 3544: code as an image file in nearly executable form. At the start of the
 3545: system a C routine loads the image file into memory, sets up the
 3546: memory (stacks etc.) according to information in the image file, and
 3547: starts executing Forth code.
 3548: 
 3549: The image file format is a compromise between the goals of making it
 3550: easy to generate image files and making them portable. The easiest way
 3551: to generate an image file is to just generate a memory dump. However,
 3552: this kind of image file cannot be used on a different machine, or on
 3553: the next version of the engine on the same machine, it even might not
 3554: work with the same engine compiled by a different version of the C
 3555: compiler. We would like to have as few versions of the image file as
 3556: possible, because we do not want to distribute many versions of the
 3557: same image file, and to make it easy for the users to use their image
 3558: files on many machines. We currently need to create a different image
 3559: file for machines with different cell sizes and different byte order
 3560: (little- or big-endian)@footnote{We are considering adding information to the
 3561: image file that enables the loader to change the byte order.}.
 3562: 
 3563: Forth code that is going to end up in a portable image file has to
 3564: comply to some restrictions: addresses have to be stored in memory with
 3565: special words (@code{A!}, @code{A,}, etc.) in order to make the code
 3566: relocatable. Cells, floats, etc., have to be stored at the natural
 3567: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
 3568: dividable by~8. This happens automatically in our system when you use
 3569: the ANS Forth alignment words.}, in order to avoid alignment faults on
 3570: machines with stricter alignment. The image file is produced by a
 3571: metacompiler (@file{cross.fs}).
 3572: 
 3573: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
 3574: file is not directly executable, but has to undergo some manipulations
 3575: during loading. Address relocation is performed at image load-time, not
 3576: at run-time. The loader also has to replace tokens standing for
 3577: primitive calls with the appropriate code-field addresses (or code
 3578: addresses in the case of direct threading).
 3579: 
 3580: @node  Performance,  , System Architecture, Internals
 3581: @section Performance
 3582: 
 3583: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
 3584: impossible to write a significantly faster engine.
 3585: 
 3586: On register-starved machines like the 386 architecture processors
 3587: improvements are possible, because @code{gcc} does not utilize the
 3588: registers as well as a human, even with explicit register declarations;
 3589: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
 3590: and hand-tuned it for the 486; this system is 1.19 times faster on the
 3591: Sieve benchmark on a 486DX2/66 than Gforth compiled with
 3592: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
 3593: 
 3594: However, this potential advantage of assembly language implementations
 3595: is not necessarily realized in complete Forth systems: We compared
 3596: Gforth (compiled with @code{gcc-2.6.3} and @code{-DFORCE_REG}) with
 3597: Win32Forth 1.2093 and LMI's NT Forth (Beta, May 1994), two systems
 3598: written in assembly, and with two systems written in C: PFE-0.9.11
 3599: (compiled with @code{gcc-2.6.3} with the default configuration for
 3600: Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS}) and ThisForth Beta
 3601: (compiled with gcc-2.6.3 -O3 -fomit-frame-pointer). We benchmarked
 3602: Gforth, PFE and ThisForth on a 486DX2/66 under Linux. Kenneth O'Heskin
 3603: kindly provided the results for Win32Forth and NT Forth on a 486DX2/66
 3604: with similar memory performance under Windows NT.
 3605:  
 3606: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
 3607: matrix multiplication come from the Stanford integer benchmarks and have
 3608: been translated into Forth by Martin Fraeman; we used the versions
 3609: included in the TILE Forth package; and a recursive Fibonacci number
 3610: computation for benchmark calling performance. The following table shows
 3611: the time taken for the benchmarks scaled by the time taken by Gforth (in
 3612: other words, it shows the speedup factor that Gforth achieved over the
 3613: other systems).
 3614: 
 3615: @example
 3616: relative             Win32-        NT               This-
 3617:   time     Gforth     Forth     Forth       PFE     Forth
 3618: sieve        1.00      1.30      1.07      1.67      2.98
 3619: bubble       1.00      1.30      1.40      1.66
 3620: matmul       1.00      1.40      1.29      2.24
 3621: fib          1.00      1.44      1.26      1.82      2.82
 3622: @end example
 3623: 
 3624: You may find the good performance of Gforth compared with the systems
 3625: written in assembly language quite surprising. One important reason for
 3626: the disappointing performance of these systems is probably that they are
 3627: not written optimally for the 486 (e.g., they use the @code{lods}
 3628: instruction). In addition, Win32Forth uses a comfortable, but costly
 3629: method for relocating the Forth image: like @code{cforth}, it computes
 3630: the actual addresses at run time, resulting in two address computations
 3631: per NEXT (@pxref{System Architecture}).
 3632: 
 3633: The speedup of Gforth over PFE and ThisForth can be easily explained
 3634: with the self-imposed restriction to standard C (although the measured
 3635: implementation of PFE uses a GNU C extension: global register
 3636: variables), which makes efficient threading impossible.  Moreover,
 3637: current C compilers have a hard time optimizing other aspects of the
 3638: ThisForth source.
 3639: 
 3640: Note that the performance of Gforth on 386 architecture processors
 3641: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
 3642: failed to allocate any of the virtual machine registers into real
 3643: machine registers by itself and would not work correctly with explicit
 3644: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
 3645: running the Sieve) than the one measured above.
 3646: 
 3647: @node Bugs, Pedigree, Internals, Top
 3648: @chapter Bugs
 3649: 
 3650: Known bugs are described in the file BUGS in the Gforth distribution.
 3651: 
 3652: If you find a bug, please send a bug report to !!. A bug report should
 3653: describe the Gforth version used (it is announced at the start of an
 3654: interactive Gforth session), the machine and operating system (on Unix
 3655: systems you can use @code{uname -a} to produce this information), the
 3656: installation options (!! a way to find them out), and a complete list of
 3657: changes you (or your installer) have made to the Gforth sources (if
 3658: any); it should contain a program (or a sequence of keyboard commands)
 3659: that reproduces the bug and a description of what you think constitutes
 3660: the buggy behaviour.
 3661: 
 3662: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
 3663: to Report Bugs, gcc.info, GNU C Manual}.
 3664: 
 3665: 
 3666: @node Pedigree, Word Index, Bugs, Top
 3667: @chapter Pedigree
 3668: 
 3669: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
 3670: Dirk Zoller) will cross-fertilize each other. Of course, a significant part of the design of Gforth was prescribed by ANS Forth.
 3671: 
 3672: Bernd Paysan wrote BigForth, a child of VolksForth.
 3673: 
 3674: VolksForth descends from F83. !! Authors? When?
 3675: 
 3676: Laxen and Perry wrote F83 as a model implementation of the
 3677: Forth-83 standard. !! Pedigree? When?
 3678: 
 3679: A team led by Bill Ragsdale implemented fig-Forth on many processors in
 3680: 1979. Dean Sanderson and Bill Ragsdale developed the original
 3681: implementation of fig-Forth based on microForth.
 3682: 
 3683: !! microForth pedigree
 3684: 
 3685: A part of the information in this section comes from @cite{The Evolution
 3686: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
 3687: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
 3688: Notices 28(3), 1993.  You can find more historical and genealogical
 3689: information about Forth there.
 3690: 
 3691: @node Word Index, Node Index, Pedigree, Top
 3692: @chapter Word Index
 3693: 
 3694: This index is as incomplete as the manual. Each word is listed with
 3695: stack effect and wordset.
 3696: 
 3697: @printindex fn
 3698: 
 3699: @node Node Index,  , Word Index, Top
 3700: @chapter Node Index
 3701: 
 3702: This index is even less complete than the manual.
 3703: 
 3704: @contents
 3705: @bye
 3706: 

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