File:  [gforth] / gforth / Attic / gforth.ds
Revision 1.31: download - view: text, annotated - select for diffs
Fri Feb 9 17:34:09 1996 UTC (28 years, 1 month ago) by anton
Branches: MAIN
CVS tags: HEAD
?DUP-IF and ?DUP-0=-IF are now supported by primitives
added primitives EMIT-FILE, STDOUT, STDERR
EMIT and TYPE now work through file words
added some code for the BUGGY_LONG_LONG case (not yet complete)
eliminated D! and D@
made DMIN, DMAX, DABS high-level
added compat/control.fs (?DUP-IF etc.)

    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 0.2
   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: @finalout
   40: @titlepage
   41: @sp 10
   42: @center @titlefont{Gforth Manual}
   43: @sp 2
   44: @center for version 0.2
   45: @sp 2
   46: @center Anton Ertl
   47: @center Bernd Paysan
   48: @sp 3
   49: @center This manual is under construction
   50: 
   51: @comment  The following two commands start the copyright page.
   52: @page
   53: @vskip 0pt plus 1filll
   54: Copyright @copyright{} 1995 Free Software Foundation, Inc.
   55: 
   56: @comment !! Published by ... or You can get a copy of this manual ...
   57: 
   58:      Permission is granted to make and distribute verbatim copies of
   59:      this manual provided the copyright notice and this permission notice
   60:      are preserved on all copies.
   61:      
   62:      Permission is granted to copy and distribute modified versions of this
   63:      manual under the conditions for verbatim copying, provided also that the
   64:      sections entitled "Distribution" and "General Public License" are
   65:      included exactly as in the original, and provided that the entire
   66:      resulting derived work is distributed under the terms of a permission
   67:      notice identical to this one.
   68:      
   69:      Permission is granted to copy and distribute translations of this manual
   70:      into another language, under the above conditions for modified versions,
   71:      except that the sections entitled "Distribution" and "General Public
   72:      License" may be included in a translation approved by the author instead
   73:      of in the original English.
   74: @end titlepage
   75: 
   76: 
   77: @node Top, License, (dir), (dir)
   78: @ifinfo
   79: Gforth is a free implementation of ANS Forth available on many
   80: personal machines. This manual corresponds to version 0.2.
   81: @end ifinfo
   82: 
   83: @menu
   84: * License::                     
   85: * Goals::                       About the Gforth Project
   86: * Other Books::                 Things you might want to read
   87: * Invocation::                  Starting Gforth
   88: * Words::                       Forth words available in Gforth
   89: * ANS conformance::             Implementation-defined options etc.
   90: * Model::                       The abstract machine of Gforth
   91: * Emacs and Gforth::            The Gforth Mode
   92: * Internals::                   Implementation details
   93: * Bugs::                        How to report them
   94: * Origin::                      Authors and ancestors of Gforth
   95: * Word Index::                  An item for each Forth word
   96: * Node Index::                  An item for each node
   97: @end menu
   98: 
   99: @node License, Goals, Top, Top
  100: @unnumbered GNU GENERAL PUBLIC LICENSE
  101: @center Version 2, June 1991
  102: 
  103: @display
  104: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
  105: 675 Mass Ave, Cambridge, MA 02139, USA
  106: 
  107: Everyone is permitted to copy and distribute verbatim copies
  108: of this license document, but changing it is not allowed.
  109: @end display
  110: 
  111: @unnumberedsec Preamble
  112: 
  113:   The licenses for most software are designed to take away your
  114: freedom to share and change it.  By contrast, the GNU General Public
  115: License is intended to guarantee your freedom to share and change free
  116: software---to make sure the software is free for all its users.  This
  117: General Public License applies to most of the Free Software
  118: Foundation's software and to any other program whose authors commit to
  119: using it.  (Some other Free Software Foundation software is covered by
  120: the GNU Library General Public License instead.)  You can apply it to
  121: your programs, too.
  122: 
  123:   When we speak of free software, we are referring to freedom, not
  124: price.  Our General Public Licenses are designed to make sure that you
  125: have the freedom to distribute copies of free software (and charge for
  126: this service if you wish), that you receive source code or can get it
  127: if you want it, that you can change the software or use pieces of it
  128: in new free programs; and that you know you can do these things.
  129: 
  130:   To protect your rights, we need to make restrictions that forbid
  131: anyone to deny you these rights or to ask you to surrender the rights.
  132: These restrictions translate to certain responsibilities for you if you
  133: distribute copies of the software, or if you modify it.
  134: 
  135:   For example, if you distribute copies of such a program, whether
  136: gratis or for a fee, you must give the recipients all the rights that
  137: you have.  You must make sure that they, too, receive or can get the
  138: source code.  And you must show them these terms so they know their
  139: rights.
  140: 
  141:   We protect your rights with two steps: (1) copyright the software, and
  142: (2) offer you this license which gives you legal permission to copy,
  143: distribute and/or modify the software.
  144: 
  145:   Also, for each author's protection and ours, we want to make certain
  146: that everyone understands that there is no warranty for this free
  147: software.  If the software is modified by someone else and passed on, we
  148: want its recipients to know that what they have is not the original, so
  149: that any problems introduced by others will not reflect on the original
  150: authors' reputations.
  151: 
  152:   Finally, any free program is threatened constantly by software
  153: patents.  We wish to avoid the danger that redistributors of a free
  154: program will individually obtain patent licenses, in effect making the
  155: program proprietary.  To prevent this, we have made it clear that any
  156: patent must be licensed for everyone's free use or not licensed at all.
  157: 
  158:   The precise terms and conditions for copying, distribution and
  159: modification follow.
  160: 
  161: @iftex
  162: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  163: @end iftex
  164: @ifinfo
  165: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
  166: @end ifinfo
  167: 
  168: @enumerate 0
  169: @item
  170: This License applies to any program or other work which contains
  171: a notice placed by the copyright holder saying it may be distributed
  172: under the terms of this General Public License.  The ``Program'', below,
  173: refers to any such program or work, and a ``work based on the Program''
  174: means either the Program or any derivative work under copyright law:
  175: that is to say, a work containing the Program or a portion of it,
  176: either verbatim or with modifications and/or translated into another
  177: language.  (Hereinafter, translation is included without limitation in
  178: the term ``modification''.)  Each licensee is addressed as ``you''.
  179: 
  180: Activities other than copying, distribution and modification are not
  181: covered by this License; they are outside its scope.  The act of
  182: running the Program is not restricted, and the output from the Program
  183: is covered only if its contents constitute a work based on the
  184: Program (independent of having been made by running the Program).
  185: Whether that is true depends on what the Program does.
  186: 
  187: @item
  188: You may copy and distribute verbatim copies of the Program's
  189: source code as you receive it, in any medium, provided that you
  190: conspicuously and appropriately publish on each copy an appropriate
  191: copyright notice and disclaimer of warranty; keep intact all the
  192: notices that refer to this License and to the absence of any warranty;
  193: and give any other recipients of the Program a copy of this License
  194: along with the Program.
  195: 
  196: You may charge a fee for the physical act of transferring a copy, and
  197: you may at your option offer warranty protection in exchange for a fee.
  198: 
  199: @item
  200: You may modify your copy or copies of the Program or any portion
  201: of it, thus forming a work based on the Program, and copy and
  202: distribute such modifications or work under the terms of Section 1
  203: above, provided that you also meet all of these conditions:
  204: 
  205: @enumerate a
  206: @item
  207: You must cause the modified files to carry prominent notices
  208: stating that you changed the files and the date of any change.
  209: 
  210: @item
  211: You must cause any work that you distribute or publish, that in
  212: whole or in part contains or is derived from the Program or any
  213: part thereof, to be licensed as a whole at no charge to all third
  214: parties under the terms of this License.
  215: 
  216: @item
  217: If the modified program normally reads commands interactively
  218: when run, you must cause it, when started running for such
  219: interactive use in the most ordinary way, to print or display an
  220: announcement including an appropriate copyright notice and a
  221: notice that there is no warranty (or else, saying that you provide
  222: a warranty) and that users may redistribute the program under
  223: these conditions, and telling the user how to view a copy of this
  224: License.  (Exception: if the Program itself is interactive but
  225: does not normally print such an announcement, your work based on
  226: the Program is not required to print an announcement.)
  227: @end enumerate
  228: 
  229: These requirements apply to the modified work as a whole.  If
  230: identifiable sections of that work are not derived from the Program,
  231: and can be reasonably considered independent and separate works in
  232: themselves, then this License, and its terms, do not apply to those
  233: sections when you distribute them as separate works.  But when you
  234: distribute the same sections as part of a whole which is a work based
  235: on the Program, the distribution of the whole must be on the terms of
  236: this License, whose permissions for other licensees extend to the
  237: entire whole, and thus to each and every part regardless of who wrote it.
  238: 
  239: Thus, it is not the intent of this section to claim rights or contest
  240: your rights to work written entirely by you; rather, the intent is to
  241: exercise the right to control the distribution of derivative or
  242: collective works based on the Program.
  243: 
  244: In addition, mere aggregation of another work not based on the Program
  245: with the Program (or with a work based on the Program) on a volume of
  246: a storage or distribution medium does not bring the other work under
  247: the scope of this License.
  248: 
  249: @item
  250: You may copy and distribute the Program (or a work based on it,
  251: under Section 2) in object code or executable form under the terms of
  252: Sections 1 and 2 above provided that you also do one of the following:
  253: 
  254: @enumerate a
  255: @item
  256: Accompany it with the complete corresponding machine-readable
  257: source code, which must be distributed under the terms of Sections
  258: 1 and 2 above on a medium customarily used for software interchange; or,
  259: 
  260: @item
  261: Accompany it with a written offer, valid for at least three
  262: years, to give any third party, for a charge no more than your
  263: cost of physically performing source distribution, a complete
  264: machine-readable copy of the corresponding source code, to be
  265: distributed under the terms of Sections 1 and 2 above on a medium
  266: customarily used for software interchange; or,
  267: 
  268: @item
  269: Accompany it with the information you received as to the offer
  270: to distribute corresponding source code.  (This alternative is
  271: allowed only for noncommercial distribution and only if you
  272: received the program in object code or executable form with such
  273: an offer, in accord with Subsection b above.)
  274: @end enumerate
  275: 
  276: The source code for a work means the preferred form of the work for
  277: making modifications to it.  For an executable work, complete source
  278: code means all the source code for all modules it contains, plus any
  279: associated interface definition files, plus the scripts used to
  280: control compilation and installation of the executable.  However, as a
  281: special exception, the source code distributed need not include
  282: anything that is normally distributed (in either source or binary
  283: form) with the major components (compiler, kernel, and so on) of the
  284: operating system on which the executable runs, unless that component
  285: itself accompanies the executable.
  286: 
  287: If distribution of executable or object code is made by offering
  288: access to copy from a designated place, then offering equivalent
  289: access to copy the source code from the same place counts as
  290: distribution of the source code, even though third parties are not
  291: compelled to copy the source along with the object code.
  292: 
  293: @item
  294: You may not copy, modify, sublicense, or distribute the Program
  295: except as expressly provided under this License.  Any attempt
  296: otherwise to copy, modify, sublicense or distribute the Program is
  297: void, and will automatically terminate your rights under this License.
  298: However, parties who have received copies, or rights, from you under
  299: this License will not have their licenses terminated so long as such
  300: parties remain in full compliance.
  301: 
  302: @item
  303: You are not required to accept this License, since you have not
  304: signed it.  However, nothing else grants you permission to modify or
  305: distribute the Program or its derivative works.  These actions are
  306: prohibited by law if you do not accept this License.  Therefore, by
  307: modifying or distributing the Program (or any work based on the
  308: Program), you indicate your acceptance of this License to do so, and
  309: all its terms and conditions for copying, distributing or modifying
  310: the Program or works based on it.
  311: 
  312: @item
  313: Each time you redistribute the Program (or any work based on the
  314: Program), the recipient automatically receives a license from the
  315: original licensor to copy, distribute or modify the Program subject to
  316: these terms and conditions.  You may not impose any further
  317: restrictions on the recipients' exercise of the rights granted herein.
  318: You are not responsible for enforcing compliance by third parties to
  319: this License.
  320: 
  321: @item
  322: If, as a consequence of a court judgment or allegation of patent
  323: infringement or for any other reason (not limited to patent issues),
  324: conditions are imposed on you (whether by court order, agreement or
  325: otherwise) that contradict the conditions of this License, they do not
  326: excuse you from the conditions of this License.  If you cannot
  327: distribute so as to satisfy simultaneously your obligations under this
  328: License and any other pertinent obligations, then as a consequence you
  329: may not distribute the Program at all.  For example, if a patent
  330: license would not permit royalty-free redistribution of the Program by
  331: all those who receive copies directly or indirectly through you, then
  332: the only way you could satisfy both it and this License would be to
  333: refrain entirely from distribution of the Program.
  334: 
  335: If any portion of this section is held invalid or unenforceable under
  336: any particular circumstance, the balance of the section is intended to
  337: apply and the section as a whole is intended to apply in other
  338: circumstances.
  339: 
  340: It is not the purpose of this section to induce you to infringe any
  341: patents or other property right claims or to contest validity of any
  342: such claims; this section has the sole purpose of protecting the
  343: integrity of the free software distribution system, which is
  344: implemented by public license practices.  Many people have made
  345: generous contributions to the wide range of software distributed
  346: through that system in reliance on consistent application of that
  347: system; it is up to the author/donor to decide if he or she is willing
  348: to distribute software through any other system and a licensee cannot
  349: impose that choice.
  350: 
  351: This section is intended to make thoroughly clear what is believed to
  352: be a consequence of the rest of this License.
  353: 
  354: @item
  355: If the distribution and/or use of the Program is restricted in
  356: certain countries either by patents or by copyrighted interfaces, the
  357: original copyright holder who places the Program under this License
  358: may add an explicit geographical distribution limitation excluding
  359: those countries, so that distribution is permitted only in or among
  360: countries not thus excluded.  In such case, this License incorporates
  361: the limitation as if written in the body of this License.
  362: 
  363: @item
  364: The Free Software Foundation may publish revised and/or new versions
  365: of the General Public License from time to time.  Such new versions will
  366: be similar in spirit to the present version, but may differ in detail to
  367: address new problems or concerns.
  368: 
  369: Each version is given a distinguishing version number.  If the Program
  370: specifies a version number of this License which applies to it and ``any
  371: later version'', you have the option of following the terms and conditions
  372: either of that version or of any later version published by the Free
  373: Software Foundation.  If the Program does not specify a version number of
  374: this License, you may choose any version ever published by the Free Software
  375: Foundation.
  376: 
  377: @item
  378: If you wish to incorporate parts of the Program into other free
  379: programs whose distribution conditions are different, write to the author
  380: to ask for permission.  For software which is copyrighted by the Free
  381: Software Foundation, write to the Free Software Foundation; we sometimes
  382: make exceptions for this.  Our decision will be guided by the two goals
  383: of preserving the free status of all derivatives of our free software and
  384: of promoting the sharing and reuse of software generally.
  385: 
  386: @iftex
  387: @heading NO WARRANTY
  388: @end iftex
  389: @ifinfo
  390: @center NO WARRANTY
  391: @end ifinfo
  392: 
  393: @item
  394: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
  395: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
  396: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
  397: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
  398: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
  399: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
  400: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
  401: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
  402: REPAIR OR CORRECTION.
  403: 
  404: @item
  405: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
  406: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
  407: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
  408: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
  409: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
  410: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
  411: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
  412: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
  413: POSSIBILITY OF SUCH DAMAGES.
  414: @end enumerate
  415: 
  416: @iftex
  417: @heading END OF TERMS AND CONDITIONS
  418: @end iftex
  419: @ifinfo
  420: @center END OF TERMS AND CONDITIONS
  421: @end ifinfo
  422: 
  423: @page
  424: @unnumberedsec How to Apply These Terms to Your New Programs
  425: 
  426:   If you develop a new program, and you want it to be of the greatest
  427: possible use to the public, the best way to achieve this is to make it
  428: free software which everyone can redistribute and change under these terms.
  429: 
  430:   To do so, attach the following notices to the program.  It is safest
  431: to attach them to the start of each source file to most effectively
  432: convey the exclusion of warranty; and each file should have at least
  433: the ``copyright'' line and a pointer to where the full notice is found.
  434: 
  435: @smallexample
  436: @var{one line to give the program's name and a brief idea of what it does.}
  437: Copyright (C) 19@var{yy}  @var{name of author}
  438: 
  439: This program is free software; you can redistribute it and/or modify 
  440: it under the terms of the GNU General Public License as published by 
  441: the Free Software Foundation; either version 2 of the License, or 
  442: (at your option) any later version.
  443: 
  444: This program is distributed in the hope that it will be useful,
  445: but WITHOUT ANY WARRANTY; without even the implied warranty of
  446: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  447: GNU General Public License for more details.
  448: 
  449: You should have received a copy of the GNU General Public License
  450: along with this program; if not, write to the Free Software
  451: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  452: @end smallexample
  453: 
  454: Also add information on how to contact you by electronic and paper mail.
  455: 
  456: If the program is interactive, make it output a short notice like this
  457: when it starts in an interactive mode:
  458: 
  459: @smallexample
  460: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
  461: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
  462: type `show w'.  
  463: This is free software, and you are welcome to redistribute it 
  464: under certain conditions; type `show c' for details.
  465: @end smallexample
  466: 
  467: The hypothetical commands @samp{show w} and @samp{show c} should show
  468: the appropriate parts of the General Public License.  Of course, the
  469: commands you use may be called something other than @samp{show w} and
  470: @samp{show c}; they could even be mouse-clicks or menu items---whatever
  471: suits your program.
  472: 
  473: You should also get your employer (if you work as a programmer) or your
  474: school, if any, to sign a ``copyright disclaimer'' for the program, if
  475: necessary.  Here is a sample; alter the names:
  476: 
  477: @smallexample
  478: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
  479: `Gnomovision' (which makes passes at compilers) written by James Hacker.
  480: 
  481: @var{signature of Ty Coon}, 1 April 1989
  482: Ty Coon, President of Vice
  483: @end smallexample
  484: 
  485: This General Public License does not permit incorporating your program into
  486: proprietary programs.  If your program is a subroutine library, you may
  487: consider it more useful to permit linking proprietary applications with the
  488: library.  If this is what you want to do, use the GNU Library General
  489: Public License instead of this License.
  490: 
  491: @iftex
  492: @node    Preface
  493: @comment node-name,     next,           previous, up
  494: @unnumbered Preface
  495: @cindex Preface
  496: This manual documents Gforth. The reader is expected to know
  497: Forth. This manual is primarily a reference manual. @xref{Other Books}
  498: for introductory material.
  499: @end iftex
  500: 
  501: @node    Goals, Other Books, License, Top
  502: @comment node-name,     next,           previous, up
  503: @chapter Goals of Gforth
  504: @cindex Goals
  505: The goal of the Gforth Project is to develop a standard model for
  506: ANSI Forth. This can be split into several subgoals:
  507: 
  508: @itemize @bullet
  509: @item
  510: Gforth should conform to the ANSI Forth standard.
  511: @item
  512: It should be a model, i.e. it should define all the
  513: implementation-dependent things.
  514: @item
  515: It should become standard, i.e. widely accepted and used. This goal
  516: is the most difficult one.
  517: @end itemize
  518: 
  519: To achieve these goals Gforth should be
  520: @itemize @bullet
  521: @item
  522: Similar to previous models (fig-Forth, F83)
  523: @item
  524: Powerful. It should provide for all the things that are considered
  525: necessary today and even some that are not yet considered necessary.
  526: @item
  527: Efficient. It should not get the reputation of being exceptionally
  528: slow.
  529: @item
  530: Free.
  531: @item
  532: Available on many machines/easy to port.
  533: @end itemize
  534: 
  535: Have we achieved these goals? Gforth conforms to the ANS Forth
  536: standard. It may be considered a model, but we have not yet documented
  537: which parts of the model are stable and which parts we are likely to
  538: change. It certainly has not yet become a de facto standard. It has some
  539: similarities and some differences to previous models. It has some
  540: powerful features, but not yet everything that we envisioned. We
  541: certainly have achieved our execution speed goals (@pxref{Performance}).
  542: It is free and available on many machines.
  543: 
  544: @node Other Books, Invocation, Goals, Top
  545: @chapter Other books on ANS Forth
  546: 
  547: As the standard is relatively new, there are not many books out yet. It
  548: is not recommended to learn Forth by using Gforth and a book that is
  549: not written for ANS Forth, as you will not know your mistakes from the
  550: deviations of the book.
  551: 
  552: There is, of course, the standard, the definite reference if you want to
  553: write ANS Forth programs. It is available in printed form from the
  554: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
  555: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
  556: can also get it from Global Engineering Documents (Tel.: USA (800)
  557: 854-7179; Fax.: (303) 843-9880) for about $300.
  558: 
  559: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
  560: for publication is available electronically and for free in some MS Word
  561: format, and it has been converted to HTML. Some pointers to these
  562: versions can be found through
  563: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
  564: 
  565: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
  566: introductory book based on a draft version of the standard. It does not
  567: cover the whole standard. It also contains interesting background
  568: information (Jack Woehr was in the ANS Forth Technical Committe). It is
  569: not appropriate for complete newbies, but programmers experienced in
  570: other languages should find it ok.
  571: 
  572: @node Invocation, Words, Other Books, Top
  573: @chapter Invocation
  574: 
  575: You will usually just say @code{gforth}. In many other cases the default
  576: Gforth image will be invoked like this:
  577: 
  578: @example
  579: gforth [files] [-e forth-code]
  580: @end example
  581: 
  582: executing the contents of the files and the Forth code in the order they
  583: are given.
  584: 
  585: In general, the command line looks like this:
  586: 
  587: @example
  588: gforth [initialization options] [image-specific options]
  589: @end example
  590: 
  591: The initialization options must come before the rest of the command
  592: line. They are:
  593: 
  594: @table @code
  595: @item --image-file @var{file}
  596: @item -i @var{file}
  597: Loads the Forth image @var{file} instead of the default
  598: @file{gforth.fi}.
  599: 
  600: @item --path @var{path}
  601: @item -p @var{path}
  602: Uses @var{path} for searching the image file and Forth source code
  603: files instead of the default in the environment variable
  604: @code{GFORTHPATH} or the path specified at installation time (typically
  605: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
  606: list.
  607: 
  608: @item --dictionary-size @var{size}
  609: @item -m @var{size}
  610: Allocate @var{size} space for the Forth dictionary space instead of
  611: using the default specified in the image (typically 256K). The
  612: @var{size} specification consists of an integer and a unit (e.g.,
  613: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
  614: size, in this case Cells), @code{k} (kilobytes), and @code{M}
  615: (Megabytes). If no unit is specified, @code{e} is used.
  616: 
  617: @item --data-stack-size @var{size}
  618: @item -d @var{size}
  619: Allocate @var{size} space for the data stack instead of using the
  620: default specified in the image (typically 16K).
  621: 
  622: @item --return-stack-size @var{size}
  623: @item -r @var{size}
  624: Allocate @var{size} space for the return stack instead of using the
  625: default specified in the image (typically 16K).
  626: 
  627: @item --fp-stack-size @var{size}
  628: @item -f @var{size}
  629: Allocate @var{size} space for the floating point stack instead of
  630: using the default specified in the image (typically 16K). In this case
  631: the unit specifier @code{e} refers to floating point numbers.
  632: 
  633: @item --locals-stack-size @var{size}
  634: @item -l @var{size}
  635: Allocate @var{size} space for the locals stack instead of using the
  636: default specified in the image (typically 16K).
  637: 
  638: @end table
  639: 
  640: As explained above, the image-specific command-line arguments for the
  641: default image @file{gforth.fi} consist of a sequence of filenames and
  642: @code{-e @var{forth-code}} options that are interpreted in the seqence
  643: in which they are given. The @code{-e @var{forth-code}} or
  644: @code{--evaluate @var{forth-code}} option evaluates the forth
  645: code. This option takes only one argument; if you want to evaluate more
  646: Forth words, you have to quote them or use several @code{-e}s. To exit
  647: after processing the command line (instead of entering interactive mode)
  648: append @code{-e bye} to the command line.
  649: 
  650: If you have several versions of Gforth installed, @code{gforth} will
  651: invoke the version that was installed last. @code{gforth-@var{version}}
  652: invokes a specific version. You may want to use the option
  653: @code{--path}, if your environment contains the variable
  654: @code{GFORTHPATH}.
  655: 
  656: Not yet implemented:
  657: On startup the system first executes the system initialization file
  658: (unless the option @code{--no-init-file} is given; note that the system
  659: resulting from using this option may not be ANS Forth conformant). Then
  660: the user initialization file @file{.gforth.fs} is executed, unless the
  661: option @code{--no-rc} is given; this file is first searched in @file{.},
  662: then in @file{~}, then in the normal path (see above).
  663: 
  664: @node Words, ANS conformance, Invocation, Top
  665: @chapter Forth Words
  666: 
  667: @menu
  668: * Notation::                    
  669: * Arithmetic::                  
  670: * Stack Manipulation::          
  671: * Memory access::               
  672: * Control Structures::          
  673: * Locals::                      
  674: * Defining Words::              
  675: * Wordlists::                   
  676: * Files::                       
  677: * Blocks::                      
  678: * Other I/O::                   
  679: * Programming Tools::           
  680: * Assembler and Code words::    
  681: * Threading Words::             
  682: @end menu
  683: 
  684: @node Notation, Arithmetic, Words, Words
  685: @section Notation
  686: 
  687: The Forth words are described in this section in the glossary notation
  688: that has become a de-facto standard for Forth texts, i.e.
  689: 
  690: @format
  691: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
  692: @end format
  693: @var{Description}
  694: 
  695: @table @var
  696: @item word
  697: The name of the word. BTW, Gforth is case insensitive, so you can
  698: type the words in in lower case (However, @pxref{core-idef}).
  699: 
  700: @item Stack effect
  701: The stack effect is written in the notation @code{@var{before} --
  702: @var{after}}, where @var{before} and @var{after} describe the top of
  703: stack entries before and after the execution of the word. The rest of
  704: the stack is not touched by the word. The top of stack is rightmost,
  705: i.e., a stack sequence is written as it is typed in. Note that Gforth
  706: uses a separate floating point stack, but a unified stack
  707: notation. Also, return stack effects are not shown in @var{stack
  708: effect}, but in @var{Description}. The name of a stack item describes
  709: the type and/or the function of the item. See below for a discussion of
  710: the types.
  711: 
  712: All words have two stack effects: A compile-time stack effect and a
  713: run-time stack effect. The compile-time stack-effect of most words is
  714: @var{ -- }. If the compile-time stack-effect of a word deviates from
  715: this standard behaviour, or the word does other unusual things at
  716: compile time, both stack effects are shown; otherwise only the run-time
  717: stack effect is shown.
  718: 
  719: @item pronunciation
  720: How the word is pronounced
  721: 
  722: @item wordset
  723: The ANS Forth standard is divided into several wordsets. A standard
  724: system need not support all of them. So, the fewer wordsets your program
  725: uses the more portable it will be in theory. However, we suspect that
  726: most ANS Forth systems on personal machines will feature all
  727: wordsets. Words that are not defined in the ANS standard have
  728: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
  729: describes words that will work in future releases of Gforth;
  730: @code{gforth-internal} words are more volatile. Environmental query
  731: strings are also displayed like words; you can recognize them by the
  732: @code{environment} in the wordset field.
  733: 
  734: @item Description
  735: A description of the behaviour of the word.
  736: @end table
  737: 
  738: The type of a stack item is specified by the character(s) the name
  739: starts with:
  740: 
  741: @table @code
  742: @item f
  743: Bool, i.e. @code{false} or @code{true}.
  744: @item c
  745: Char
  746: @item w
  747: Cell, can contain an integer or an address
  748: @item n
  749: signed integer
  750: @item u
  751: unsigned integer
  752: @item d
  753: double sized signed integer
  754: @item ud
  755: double sized unsigned integer
  756: @item r
  757: Float
  758: @item a_
  759: Cell-aligned address
  760: @item c_
  761: Char-aligned address (note that a Char is two bytes in Windows NT)
  762: @item f_
  763: Float-aligned address
  764: @item df_
  765: Address aligned for IEEE double precision float
  766: @item sf_
  767: Address aligned for IEEE single precision float
  768: @item xt
  769: Execution token, same size as Cell
  770: @item wid
  771: Wordlist ID, same size as Cell
  772: @item f83name
  773: Pointer to a name structure
  774: @end table
  775: 
  776: @node Arithmetic, Stack Manipulation, Notation, Words
  777: @section Arithmetic
  778: Forth arithmetic is not checked, i.e., you will not hear about integer
  779: overflow on addition or multiplication, you may hear about division by
  780: zero if you are lucky. The operator is written after the operands, but
  781: the operands are still in the original order. I.e., the infix @code{2-1}
  782: corresponds to @code{2 1 -}. Forth offers a variety of division
  783: operators. If you perform division with potentially negative operands,
  784: you do not want to use @code{/} or @code{/mod} with its undefined
  785: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
  786: former, @pxref{Mixed precision}).
  787: 
  788: @menu
  789: * Single precision::            
  790: * Bitwise operations::          
  791: * Mixed precision::             operations with single and double-cell integers
  792: * Double precision::            Double-cell integer arithmetic
  793: * Floating Point::              
  794: @end menu
  795: 
  796: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
  797: @subsection Single precision
  798: doc-+
  799: doc--
  800: doc-*
  801: doc-/
  802: doc-mod
  803: doc-/mod
  804: doc-negate
  805: doc-abs
  806: doc-min
  807: doc-max
  808: 
  809: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
  810: @subsection Bitwise operations
  811: doc-and
  812: doc-or
  813: doc-xor
  814: doc-invert
  815: doc-2*
  816: doc-2/
  817: 
  818: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
  819: @subsection Mixed precision
  820: doc-m+
  821: doc-*/
  822: doc-*/mod
  823: doc-m*
  824: doc-um*
  825: doc-m*/
  826: doc-um/mod
  827: doc-fm/mod
  828: doc-sm/rem
  829: 
  830: @node Double precision, Floating Point, Mixed precision, Arithmetic
  831: @subsection Double precision
  832: 
  833: The outer (aka text) interpreter converts numbers containing a dot into
  834: a double precision number. Note that only numbers with the dot as last
  835: character are standard-conforming.
  836: 
  837: doc-d+
  838: doc-d-
  839: doc-dnegate
  840: doc-dabs
  841: doc-dmin
  842: doc-dmax
  843: 
  844: @node Floating Point,  , Double precision, Arithmetic
  845: @subsection Floating Point
  846: 
  847: The format of floating point numbers recognized by the outer (aka text)
  848: interpreter is: a signed decimal number, possibly containing a decimal
  849: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
  850: by a signed integer (the exponent). E.g., @code{1e} ist the same as
  851: @code{+1.0e+1}. Note that a number without @code{e}
  852: is not interpreted as floating-point number, but as double (if the
  853: number contains a @code{.}) or single precision integer. Also,
  854: conversions between string and floating point numbers always use base
  855: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
  856: value greater then 14, the @code{E} may be interpreted as digit and the
  857: number will be interpreted as integer, unless it has a signed exponent
  858: (both @code{+} and @code{-} are allowed as signs).
  859: 
  860: Angles in floating point operations are given in radians (a full circle
  861: has 2 pi radians). Note, that Gforth has a separate floating point
  862: stack, but we use the unified notation.
  863: 
  864: Floating point numbers have a number of unpleasant surprises for the
  865: unwary (e.g., floating point addition is not associative) and even a few
  866: for the wary. You should not use them unless you know what you are doing
  867: or you don't care that the results you get are totally bogus. If you
  868: want to learn about the problems of floating point numbers (and how to
  869: avoid them), you might start with @cite{David Goldberg, What Every
  870: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
  871: Computing Surveys 23(1):5@minus{}48, March 1991}.
  872: 
  873: doc-f+
  874: doc-f-
  875: doc-f*
  876: doc-f/
  877: doc-fnegate
  878: doc-fabs
  879: doc-fmax
  880: doc-fmin
  881: doc-floor
  882: doc-fround
  883: doc-f**
  884: doc-fsqrt
  885: doc-fexp
  886: doc-fexpm1
  887: doc-fln
  888: doc-flnp1
  889: doc-flog
  890: doc-falog
  891: doc-fsin
  892: doc-fcos
  893: doc-fsincos
  894: doc-ftan
  895: doc-fasin
  896: doc-facos
  897: doc-fatan
  898: doc-fatan2
  899: doc-fsinh
  900: doc-fcosh
  901: doc-ftanh
  902: doc-fasinh
  903: doc-facosh
  904: doc-fatanh
  905: 
  906: @node Stack Manipulation, Memory access, Arithmetic, Words
  907: @section Stack Manipulation
  908: 
  909: Gforth has a data stack (aka parameter stack) for characters, cells,
  910: addresses, and double cells, a floating point stack for floating point
  911: numbers, a return stack for storing the return addresses of colon
  912: definitions and other data, and a locals stack for storing local
  913: variables. Note that while every sane Forth has a separate floating
  914: point stack, this is not strictly required; an ANS Forth system could
  915: theoretically keep floating point numbers on the data stack. As an
  916: additional difficulty, you don't know how many cells a floating point
  917: number takes. It is reportedly possible to write words in a way that
  918: they work also for a unified stack model, but we do not recommend trying
  919: it. Instead, just say that your program has an environmental dependency
  920: on a separate FP stack.
  921: 
  922: Also, a Forth system is allowed to keep the local variables on the
  923: return stack. This is reasonable, as local variables usually eliminate
  924: the need to use the return stack explicitly. So, if you want to produce
  925: a standard complying program and if you are using local variables in a
  926: word, forget about return stack manipulations in that word (see the
  927: standard document for the exact rules).
  928: 
  929: @menu
  930: * Data stack::                  
  931: * Floating point stack::        
  932: * Return stack::                
  933: * Locals stack::                
  934: * Stack pointer manipulation::  
  935: @end menu
  936: 
  937: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
  938: @subsection Data stack
  939: doc-drop
  940: doc-nip
  941: doc-dup
  942: doc-over
  943: doc-tuck
  944: doc-swap
  945: doc-rot
  946: doc--rot
  947: doc-?dup
  948: doc-pick
  949: doc-roll
  950: doc-2drop
  951: doc-2nip
  952: doc-2dup
  953: doc-2over
  954: doc-2tuck
  955: doc-2swap
  956: doc-2rot
  957: 
  958: @node Floating point stack, Return stack, Data stack, Stack Manipulation
  959: @subsection Floating point stack
  960: doc-fdrop
  961: doc-fnip
  962: doc-fdup
  963: doc-fover
  964: doc-ftuck
  965: doc-fswap
  966: doc-frot
  967: 
  968: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
  969: @subsection Return stack
  970: doc->r
  971: doc-r>
  972: doc-r@
  973: doc-rdrop
  974: doc-2>r
  975: doc-2r>
  976: doc-2r@
  977: doc-2rdrop
  978: 
  979: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
  980: @subsection Locals stack
  981: 
  982: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
  983: @subsection Stack pointer manipulation
  984: doc-sp@
  985: doc-sp!
  986: doc-fp@
  987: doc-fp!
  988: doc-rp@
  989: doc-rp!
  990: doc-lp@
  991: doc-lp!
  992: 
  993: @node Memory access, Control Structures, Stack Manipulation, Words
  994: @section Memory access
  995: 
  996: @menu
  997: * Stack-Memory transfers::      
  998: * Address arithmetic::          
  999: * Memory block access::         
 1000: @end menu
 1001: 
 1002: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
 1003: @subsection Stack-Memory transfers
 1004: 
 1005: doc-@
 1006: doc-!
 1007: doc-+!
 1008: doc-c@
 1009: doc-c!
 1010: doc-2@
 1011: doc-2!
 1012: doc-f@
 1013: doc-f!
 1014: doc-sf@
 1015: doc-sf!
 1016: doc-df@
 1017: doc-df!
 1018: 
 1019: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
 1020: @subsection Address arithmetic
 1021: 
 1022: ANS Forth does not specify the sizes of the data types. Instead, it
 1023: offers a number of words for computing sizes and doing address
 1024: arithmetic. Basically, address arithmetic is performed in terms of
 1025: address units (aus); on most systems the address unit is one byte. Note
 1026: that a character may have more than one au, so @code{chars} is no noop
 1027: (on systems where it is a noop, it compiles to nothing).
 1028: 
 1029: ANS Forth also defines words for aligning addresses for specific
 1030: addresses. Many computers require that accesses to specific data types
 1031: must only occur at specific addresses; e.g., that cells may only be
 1032: accessed at addresses divisible by 4. Even if a machine allows unaligned
 1033: accesses, it can usually perform aligned accesses faster. 
 1034: 
 1035: For the performance-conscious: alignment operations are usually only
 1036: necessary during the definition of a data structure, not during the
 1037: (more frequent) accesses to it.
 1038: 
 1039: ANS Forth defines no words for character-aligning addresses. This is not
 1040: an oversight, but reflects the fact that addresses that are not
 1041: char-aligned have no use in the standard and therefore will not be
 1042: created.
 1043: 
 1044: The standard guarantees that addresses returned by @code{CREATE}d words
 1045: are cell-aligned; in addition, Gforth guarantees that these addresses
 1046: are aligned for all purposes.
 1047: 
 1048: Note that the standard defines a word @code{char}, which has nothing to
 1049: do with address arithmetic.
 1050: 
 1051: doc-chars
 1052: doc-char+
 1053: doc-cells
 1054: doc-cell+
 1055: doc-align
 1056: doc-aligned
 1057: doc-floats
 1058: doc-float+
 1059: doc-falign
 1060: doc-faligned
 1061: doc-sfloats
 1062: doc-sfloat+
 1063: doc-sfalign
 1064: doc-sfaligned
 1065: doc-dfloats
 1066: doc-dfloat+
 1067: doc-dfalign
 1068: doc-dfaligned
 1069: doc-maxalign
 1070: doc-maxaligned
 1071: doc-cfalign
 1072: doc-cfaligned
 1073: doc-address-unit-bits
 1074: 
 1075: @node Memory block access,  , Address arithmetic, Memory access
 1076: @subsection Memory block access
 1077: 
 1078: doc-move
 1079: doc-erase
 1080: 
 1081: While the previous words work on address units, the rest works on
 1082: characters.
 1083: 
 1084: doc-cmove
 1085: doc-cmove>
 1086: doc-fill
 1087: doc-blank
 1088: 
 1089: @node Control Structures, Locals, Memory access, Words
 1090: @section Control Structures
 1091: 
 1092: Control structures in Forth cannot be used in interpret state, only in
 1093: compile state, i.e., in a colon definition. We do not like this
 1094: limitation, but have not seen a satisfying way around it yet, although
 1095: many schemes have been proposed.
 1096: 
 1097: @menu
 1098: * Selection::                   
 1099: * Simple Loops::                
 1100: * Counted Loops::               
 1101: * Arbitrary control structures::  
 1102: * Calls and returns::           
 1103: * Exception Handling::          
 1104: @end menu
 1105: 
 1106: @node Selection, Simple Loops, Control Structures, Control Structures
 1107: @subsection Selection
 1108: 
 1109: @example
 1110: @var{flag}
 1111: IF
 1112:   @var{code}
 1113: ENDIF
 1114: @end example
 1115: or
 1116: @example
 1117: @var{flag}
 1118: IF
 1119:   @var{code1}
 1120: ELSE
 1121:   @var{code2}
 1122: ENDIF
 1123: @end example
 1124: 
 1125: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
 1126: standard, and @code{ENDIF} is not, although it is quite popular. We
 1127: recommend using @code{ENDIF}, because it is less confusing for people
 1128: who also know other languages (and is not prone to reinforcing negative
 1129: prejudices against Forth in these people). Adding @code{ENDIF} to a
 1130: system that only supplies @code{THEN} is simple:
 1131: @example
 1132: : endif   POSTPONE then ; immediate
 1133: @end example
 1134: 
 1135: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
 1136: (adv.)}  has the following meanings:
 1137: @quotation
 1138: ... 2b: following next after in order ... 3d: as a necessary consequence
 1139: (if you were there, then you saw them).
 1140: @end quotation
 1141: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
 1142: and many other programming languages has the meaning 3d.]
 1143: 
 1144: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
 1145: you can avoid using @code{?dup}. Using these alternatives is also more
 1146: efficient than using @code{?dup}. Definitions in plain standard Forth
 1147: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
 1148: @file{compat/control.fs}.
 1149: 
 1150: @example
 1151: @var{n}
 1152: CASE
 1153:   @var{n1} OF @var{code1} ENDOF
 1154:   @var{n2} OF @var{code2} ENDOF
 1155:   @dots{}
 1156: ENDCASE
 1157: @end example
 1158: 
 1159: Executes the first @var{codei}, where the @var{ni} is equal to
 1160: @var{n}. A default case can be added by simply writing the code after
 1161: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
 1162: but must not consume it.
 1163: 
 1164: @node Simple Loops, Counted Loops, Selection, Control Structures
 1165: @subsection Simple Loops
 1166: 
 1167: @example
 1168: BEGIN
 1169:   @var{code1}
 1170:   @var{flag}
 1171: WHILE
 1172:   @var{code2}
 1173: REPEAT
 1174: @end example
 1175: 
 1176: @var{code1} is executed and @var{flag} is computed. If it is true,
 1177: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
 1178: 
 1179: @example
 1180: BEGIN
 1181:   @var{code}
 1182:   @var{flag}
 1183: UNTIL
 1184: @end example
 1185: 
 1186: @var{code} is executed. The loop is restarted if @code{flag} is false.
 1187: 
 1188: @example
 1189: BEGIN
 1190:   @var{code}
 1191: AGAIN
 1192: @end example
 1193: 
 1194: This is an endless loop.
 1195: 
 1196: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
 1197: @subsection Counted Loops
 1198: 
 1199: The basic counted loop is:
 1200: @example
 1201: @var{limit} @var{start}
 1202: ?DO
 1203:   @var{body}
 1204: LOOP
 1205: @end example
 1206: 
 1207: This performs one iteration for every integer, starting from @var{start}
 1208: and up to, but excluding @var{limit}. The counter, aka index, can be
 1209: accessed with @code{i}. E.g., the loop
 1210: @example
 1211: 10 0 ?DO
 1212:   i .
 1213: LOOP
 1214: @end example
 1215: prints
 1216: @example
 1217: 0 1 2 3 4 5 6 7 8 9
 1218: @end example
 1219: The index of the innermost loop can be accessed with @code{i}, the index
 1220: of the next loop with @code{j}, and the index of the third loop with
 1221: @code{k}.
 1222: 
 1223: The loop control data are kept on the return stack, so there are some
 1224: restrictions on mixing return stack accesses and counted loop
 1225: words. E.g., if you put values on the return stack outside the loop, you
 1226: cannot read them inside the loop. If you put values on the return stack
 1227: within a loop, you have to remove them before the end of the loop and
 1228: before accessing the index of the loop.
 1229: 
 1230: There are several variations on the counted loop:
 1231: 
 1232: @code{LEAVE} leaves the innermost counted loop immediately.
 1233: 
 1234: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
 1235: (and @code{LOOP} iterates until they become equal by wrap-around
 1236: arithmetic). This behaviour is usually not what you want. Therefore,
 1237: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
 1238: @code{?DO}), which do not enter the loop if @var{start} is greater than
 1239: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
 1240: unsigned loop parameters.
 1241: 
 1242: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
 1243: index by @var{n} instead of by 1. The loop is terminated when the border
 1244: between @var{limit-1} and @var{limit} is crossed. E.g.:
 1245: 
 1246: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
 1247: 
 1248: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
 1249: 
 1250: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
 1251: 
 1252: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
 1253: 
 1254: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
 1255: 
 1256: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
 1257: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
 1258: index by @var{u} each iteration. The loop is terminated when the border
 1259: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
 1260: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
 1261: 
 1262: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
 1263: 
 1264: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
 1265: 
 1266: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
 1267: 
 1268: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
 1269: @code{-LOOP} are not in the ANS Forth standard. However, an
 1270: implementation for these words that uses only standard words is provided
 1271: in @file{compat/loops.fs}.
 1272: 
 1273: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
 1274: the loop, independent of the loop parameters. Do not use @code{DO}, even
 1275: if you know that the loop is entered in any case. Such knowledge tends
 1276: to become invalid during maintenance of a program, and then the
 1277: @code{DO} will make trouble.
 1278: 
 1279: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
 1280: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
 1281: return stack so @code{EXIT} can get to its return address.
 1282: 
 1283: Another counted loop is
 1284: @example
 1285: @var{n}
 1286: FOR
 1287:   @var{body}
 1288: NEXT
 1289: @end example
 1290: This is the preferred loop of native code compiler writers who are too
 1291: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
 1292: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
 1293: and ending with 0. Other Forth systems may behave differently, even if
 1294: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
 1295: loops.
 1296: 
 1297: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
 1298: @subsection Arbitrary control structures
 1299: 
 1300: ANS Forth permits and supports using control structures in a non-nested
 1301: way. Information about incomplete control structures is stored on the
 1302: control-flow stack. This stack may be implemented on the Forth data
 1303: stack, and this is what we have done in Gforth.
 1304: 
 1305: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
 1306: entry represents a backward branch target. A few words are the basis for
 1307: building any control structure possible (except control structures that
 1308: need storage, like calls, coroutines, and backtracking).
 1309: 
 1310: doc-if
 1311: doc-ahead
 1312: doc-then
 1313: doc-begin
 1314: doc-until
 1315: doc-again
 1316: doc-cs-pick
 1317: doc-cs-roll
 1318: 
 1319: On many systems control-flow stack items take one word, in Gforth they
 1320: currently take three (this may change in the future). Therefore it is a
 1321: really good idea to manipulate the control flow stack with
 1322: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
 1323: words.
 1324: 
 1325: Some standard control structure words are built from these words:
 1326: 
 1327: doc-else
 1328: doc-while
 1329: doc-repeat
 1330: 
 1331: Gforth adds some more control-structure words:
 1332: 
 1333: doc-endif
 1334: doc-?dup-if
 1335: doc-?dup-0=-if
 1336: 
 1337: Counted loop words constitute a separate group of words:
 1338: 
 1339: doc-?do
 1340: doc-+do
 1341: doc-u+do
 1342: doc--do
 1343: doc-u-do
 1344: doc-do
 1345: doc-for
 1346: doc-loop
 1347: doc-+loop
 1348: doc--loop
 1349: doc-next
 1350: doc-leave
 1351: doc-?leave
 1352: doc-unloop
 1353: doc-done
 1354: 
 1355: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
 1356: @i{do-sys}. Our system allows it, but it's your job to ensure that for
 1357: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
 1358: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
 1359: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
 1360: resolved (by using one of the loop-ending words or @code{DONE}).
 1361: 
 1362: Another group of control structure words are
 1363: 
 1364: doc-case
 1365: doc-endcase
 1366: doc-of
 1367: doc-endof
 1368: 
 1369: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
 1370: @code{cs-roll}.
 1371: 
 1372: @subsubsection Programming Style
 1373: 
 1374: In order to ensure readability we recommend that you do not create
 1375: arbitrary control structures directly, but define new control structure
 1376: words for the control structure you want and use these words in your
 1377: program.
 1378: 
 1379: E.g., instead of writing
 1380: 
 1381: @example
 1382: begin
 1383:   ...
 1384: if [ 1 cs-roll ]
 1385:   ...
 1386: again then
 1387: @end example
 1388: 
 1389: we recommend defining control structure words, e.g.,
 1390: 
 1391: @example
 1392: : while ( dest -- orig dest )
 1393:  POSTPONE if
 1394:  1 cs-roll ; immediate
 1395: 
 1396: : repeat ( orig dest -- )
 1397:  POSTPONE again
 1398:  POSTPONE then ; immediate
 1399: @end example
 1400: 
 1401: and then using these to create the control structure:
 1402: 
 1403: @example
 1404: begin
 1405:   ...
 1406: while
 1407:   ...
 1408: repeat
 1409: @end example
 1410: 
 1411: That's much easier to read, isn't it? Of course, @code{REPEAT} and
 1412: @code{WHILE} are predefined, so in this example it would not be
 1413: necessary to define them.
 1414: 
 1415: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
 1416: @subsection Calls and returns
 1417: 
 1418: A definition can be called simply be writing the name of the
 1419: definition. When the end of the definition is reached, it returns. An
 1420: earlier return can be forced using
 1421: 
 1422: doc-exit
 1423: 
 1424: Don't forget to clean up the return stack and @code{UNLOOP} any
 1425: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
 1426: primitive compiled by @code{EXIT} is
 1427: 
 1428: doc-;s
 1429: 
 1430: @node Exception Handling,  , Calls and returns, Control Structures
 1431: @subsection Exception Handling
 1432: 
 1433: doc-catch
 1434: doc-throw
 1435: 
 1436: @node Locals, Defining Words, Control Structures, Words
 1437: @section Locals
 1438: 
 1439: Local variables can make Forth programming more enjoyable and Forth
 1440: programs easier to read. Unfortunately, the locals of ANS Forth are
 1441: laden with restrictions. Therefore, we provide not only the ANS Forth
 1442: locals wordset, but also our own, more powerful locals wordset (we
 1443: implemented the ANS Forth locals wordset through our locals wordset).
 1444: 
 1445: The ideas in this section have also been published in the paper
 1446: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
 1447: at EuroForth '94; it is available at
 1448: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
 1449: 
 1450: @menu
 1451: * Gforth locals::               
 1452: * ANS Forth locals::            
 1453: @end menu
 1454: 
 1455: @node Gforth locals, ANS Forth locals, Locals, Locals
 1456: @subsection Gforth locals
 1457: 
 1458: Locals can be defined with
 1459: 
 1460: @example
 1461: @{ local1 local2 ... -- comment @}
 1462: @end example
 1463: or
 1464: @example
 1465: @{ local1 local2 ... @}
 1466: @end example
 1467: 
 1468: E.g.,
 1469: @example
 1470: : max @{ n1 n2 -- n3 @}
 1471:  n1 n2 > if
 1472:    n1
 1473:  else
 1474:    n2
 1475:  endif ;
 1476: @end example
 1477: 
 1478: The similarity of locals definitions with stack comments is intended. A
 1479: locals definition often replaces the stack comment of a word. The order
 1480: of the locals corresponds to the order in a stack comment and everything
 1481: after the @code{--} is really a comment.
 1482: 
 1483: This similarity has one disadvantage: It is too easy to confuse locals
 1484: declarations with stack comments, causing bugs and making them hard to
 1485: find. However, this problem can be avoided by appropriate coding
 1486: conventions: Do not use both notations in the same program. If you do,
 1487: they should be distinguished using additional means, e.g. by position.
 1488: 
 1489: The name of the local may be preceded by a type specifier, e.g.,
 1490: @code{F:} for a floating point value:
 1491: 
 1492: @example
 1493: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
 1494: \ complex multiplication
 1495:  Ar Br f* Ai Bi f* f-
 1496:  Ar Bi f* Ai Br f* f+ ;
 1497: @end example
 1498: 
 1499: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
 1500: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
 1501: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
 1502: with @code{W:}, @code{D:} etc.) produces its value and can be changed
 1503: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
 1504: produces its address (which becomes invalid when the variable's scope is
 1505: left). E.g., the standard word @code{emit} can be defined in therms of
 1506: @code{type} like this:
 1507: 
 1508: @example
 1509: : emit @{ C^ char* -- @}
 1510:     char* 1 type ;
 1511: @end example
 1512: 
 1513: A local without type specifier is a @code{W:} local. Both flavours of
 1514: locals are initialized with values from the data or FP stack.
 1515: 
 1516: Currently there is no way to define locals with user-defined data
 1517: structures, but we are working on it.
 1518: 
 1519: Gforth allows defining locals everywhere in a colon definition. This
 1520: poses the following questions:
 1521: 
 1522: @menu
 1523: * Where are locals visible by name?::  
 1524: * How long do locals live?::    
 1525: * Programming Style::           
 1526: * Implementation::              
 1527: @end menu
 1528: 
 1529: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
 1530: @subsubsection Where are locals visible by name?
 1531: 
 1532: Basically, the answer is that locals are visible where you would expect
 1533: it in block-structured languages, and sometimes a little longer. If you
 1534: want to restrict the scope of a local, enclose its definition in
 1535: @code{SCOPE}...@code{ENDSCOPE}.
 1536: 
 1537: doc-scope
 1538: doc-endscope
 1539: 
 1540: These words behave like control structure words, so you can use them
 1541: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
 1542: arbitrary ways.
 1543: 
 1544: If you want a more exact answer to the visibility question, here's the
 1545: basic principle: A local is visible in all places that can only be
 1546: reached through the definition of the local@footnote{In compiler
 1547: construction terminology, all places dominated by the definition of the
 1548: local.}. In other words, it is not visible in places that can be reached
 1549: without going through the definition of the local. E.g., locals defined
 1550: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
 1551: defined in @code{BEGIN}...@code{UNTIL} are visible after the
 1552: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
 1553: 
 1554: The reasoning behind this solution is: We want to have the locals
 1555: visible as long as it is meaningful. The user can always make the
 1556: visibility shorter by using explicit scoping. In a place that can
 1557: only be reached through the definition of a local, the meaning of a
 1558: local name is clear. In other places it is not: How is the local
 1559: initialized at the control flow path that does not contain the
 1560: definition? Which local is meant, if the same name is defined twice in
 1561: two independent control flow paths?
 1562: 
 1563: This should be enough detail for nearly all users, so you can skip the
 1564: rest of this section. If you relly must know all the gory details and
 1565: options, read on.
 1566: 
 1567: In order to implement this rule, the compiler has to know which places
 1568: are unreachable. It knows this automatically after @code{AHEAD},
 1569: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
 1570: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
 1571: compiler that the control flow never reaches that place. If
 1572: @code{UNREACHABLE} is not used where it could, the only consequence is
 1573: that the visibility of some locals is more limited than the rule above
 1574: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
 1575: lie to the compiler), buggy code will be produced.
 1576: 
 1577: Another problem with this rule is that at @code{BEGIN}, the compiler
 1578: does not know which locals will be visible on the incoming
 1579: back-edge. All problems discussed in the following are due to this
 1580: ignorance of the compiler (we discuss the problems using @code{BEGIN}
 1581: loops as examples; the discussion also applies to @code{?DO} and other
 1582: loops). Perhaps the most insidious example is:
 1583: @example
 1584: AHEAD
 1585: BEGIN
 1586:   x
 1587: [ 1 CS-ROLL ] THEN
 1588:   @{ x @}
 1589:   ...
 1590: UNTIL
 1591: @end example
 1592: 
 1593: This should be legal according to the visibility rule. The use of
 1594: @code{x} can only be reached through the definition; but that appears
 1595: textually below the use.
 1596: 
 1597: From this example it is clear that the visibility rules cannot be fully
 1598: implemented without major headaches. Our implementation treats common
 1599: cases as advertised and the exceptions are treated in a safe way: The
 1600: compiler makes a reasonable guess about the locals visible after a
 1601: @code{BEGIN}; if it is too pessimistic, the
 1602: user will get a spurious error about the local not being defined; if the
 1603: compiler is too optimistic, it will notice this later and issue a
 1604: warning. In the case above the compiler would complain about @code{x}
 1605: being undefined at its use. You can see from the obscure examples in
 1606: this section that it takes quite unusual control structures to get the
 1607: compiler into trouble, and even then it will often do fine.
 1608: 
 1609: If the @code{BEGIN} is reachable from above, the most optimistic guess
 1610: is that all locals visible before the @code{BEGIN} will also be
 1611: visible after the @code{BEGIN}. This guess is valid for all loops that
 1612: are entered only through the @code{BEGIN}, in particular, for normal
 1613: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
 1614: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
 1615: compiler. When the branch to the @code{BEGIN} is finally generated by
 1616: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
 1617: warns the user if it was too optimisitic:
 1618: @example
 1619: IF
 1620:   @{ x @}
 1621: BEGIN
 1622:   \ x ? 
 1623: [ 1 cs-roll ] THEN
 1624:   ...
 1625: UNTIL
 1626: @end example
 1627: 
 1628: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
 1629: optimistically assumes that it lives until the @code{THEN}. It notices
 1630: this difference when it compiles the @code{UNTIL} and issues a
 1631: warning. The user can avoid the warning, and make sure that @code{x}
 1632: is not used in the wrong area by using explicit scoping:
 1633: @example
 1634: IF
 1635:   SCOPE
 1636:   @{ x @}
 1637:   ENDSCOPE
 1638: BEGIN
 1639: [ 1 cs-roll ] THEN
 1640:   ...
 1641: UNTIL
 1642: @end example
 1643: 
 1644: Since the guess is optimistic, there will be no spurious error messages
 1645: about undefined locals.
 1646: 
 1647: If the @code{BEGIN} is not reachable from above (e.g., after
 1648: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
 1649: optimistic guess, as the locals visible after the @code{BEGIN} may be
 1650: defined later. Therefore, the compiler assumes that no locals are
 1651: visible after the @code{BEGIN}. However, the user can use
 1652: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
 1653: visible at the BEGIN as at the point where the top control-flow stack
 1654: item was created.
 1655: 
 1656: doc-assume-live
 1657: 
 1658: E.g.,
 1659: @example
 1660: @{ x @}
 1661: AHEAD
 1662: ASSUME-LIVE
 1663: BEGIN
 1664:   x
 1665: [ 1 CS-ROLL ] THEN
 1666:   ...
 1667: UNTIL
 1668: @end example
 1669: 
 1670: Other cases where the locals are defined before the @code{BEGIN} can be
 1671: handled by inserting an appropriate @code{CS-ROLL} before the
 1672: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
 1673: behind the @code{ASSUME-LIVE}).
 1674: 
 1675: Cases where locals are defined after the @code{BEGIN} (but should be
 1676: visible immediately after the @code{BEGIN}) can only be handled by
 1677: rearranging the loop. E.g., the ``most insidious'' example above can be
 1678: arranged into:
 1679: @example
 1680: BEGIN
 1681:   @{ x @}
 1682:   ... 0=
 1683: WHILE
 1684:   x
 1685: REPEAT
 1686: @end example
 1687: 
 1688: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
 1689: @subsubsection How long do locals live?
 1690: 
 1691: The right answer for the lifetime question would be: A local lives at
 1692: least as long as it can be accessed. For a value-flavoured local this
 1693: means: until the end of its visibility. However, a variable-flavoured
 1694: local could be accessed through its address far beyond its visibility
 1695: scope. Ultimately, this would mean that such locals would have to be
 1696: garbage collected. Since this entails un-Forth-like implementation
 1697: complexities, I adopted the same cowardly solution as some other
 1698: languages (e.g., C): The local lives only as long as it is visible;
 1699: afterwards its address is invalid (and programs that access it
 1700: afterwards are erroneous).
 1701: 
 1702: @node Programming Style, Implementation, How long do locals live?, Gforth locals
 1703: @subsubsection Programming Style
 1704: 
 1705: The freedom to define locals anywhere has the potential to change
 1706: programming styles dramatically. In particular, the need to use the
 1707: return stack for intermediate storage vanishes. Moreover, all stack
 1708: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
 1709: determined arguments) can be eliminated: If the stack items are in the
 1710: wrong order, just write a locals definition for all of them; then
 1711: write the items in the order you want.
 1712: 
 1713: This seems a little far-fetched and eliminating stack manipulations is
 1714: unlikely to become a conscious programming objective. Still, the number
 1715: of stack manipulations will be reduced dramatically if local variables
 1716: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
 1717: a traditional implementation of @code{max}).
 1718: 
 1719: This shows one potential benefit of locals: making Forth programs more
 1720: readable. Of course, this benefit will only be realized if the
 1721: programmers continue to honour the principle of factoring instead of
 1722: using the added latitude to make the words longer.
 1723: 
 1724: Using @code{TO} can and should be avoided.  Without @code{TO},
 1725: every value-flavoured local has only a single assignment and many
 1726: advantages of functional languages apply to Forth. I.e., programs are
 1727: easier to analyse, to optimize and to read: It is clear from the
 1728: definition what the local stands for, it does not turn into something
 1729: different later.
 1730: 
 1731: E.g., a definition using @code{TO} might look like this:
 1732: @example
 1733: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1734:  u1 u2 min 0
 1735:  ?do
 1736:    addr1 c@ addr2 c@ -
 1737:    ?dup-if
 1738:      unloop exit
 1739:    then
 1740:    addr1 char+ TO addr1
 1741:    addr2 char+ TO addr2
 1742:  loop
 1743:  u1 u2 - ;
 1744: @end example
 1745: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
 1746: every loop iteration. @code{strcmp} is a typical example of the
 1747: readability problems of using @code{TO}. When you start reading
 1748: @code{strcmp}, you think that @code{addr1} refers to the start of the
 1749: string. Only near the end of the loop you realize that it is something
 1750: else.
 1751: 
 1752: This can be avoided by defining two locals at the start of the loop that
 1753: are initialized with the right value for the current iteration.
 1754: @example
 1755: : strcmp @{ addr1 u1 addr2 u2 -- n @}
 1756:  addr1 addr2
 1757:  u1 u2 min 0 
 1758:  ?do @{ s1 s2 @}
 1759:    s1 c@ s2 c@ -
 1760:    ?dup-if
 1761:      unloop exit
 1762:    then
 1763:    s1 char+ s2 char+
 1764:  loop
 1765:  2drop
 1766:  u1 u2 - ;
 1767: @end example
 1768: Here it is clear from the start that @code{s1} has a different value
 1769: in every loop iteration.
 1770: 
 1771: @node Implementation,  , Programming Style, Gforth locals
 1772: @subsubsection Implementation
 1773: 
 1774: Gforth uses an extra locals stack. The most compelling reason for
 1775: this is that the return stack is not float-aligned; using an extra stack
 1776: also eliminates the problems and restrictions of using the return stack
 1777: as locals stack. Like the other stacks, the locals stack grows toward
 1778: lower addresses. A few primitives allow an efficient implementation:
 1779: 
 1780: doc-@local#
 1781: doc-f@local#
 1782: doc-laddr#
 1783: doc-lp+!#
 1784: doc-lp!
 1785: doc->l
 1786: doc-f>l
 1787: 
 1788: In addition to these primitives, some specializations of these
 1789: primitives for commonly occurring inline arguments are provided for
 1790: efficiency reasons, e.g., @code{@@local0} as specialization of
 1791: @code{@@local#} for the inline argument 0. The following compiling words
 1792: compile the right specialized version, or the general version, as
 1793: appropriate:
 1794: 
 1795: doc-compile-@local
 1796: doc-compile-f@local
 1797: doc-compile-lp+!
 1798: 
 1799: Combinations of conditional branches and @code{lp+!#} like
 1800: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
 1801: is taken) are provided for efficiency and correctness in loops.
 1802: 
 1803: A special area in the dictionary space is reserved for keeping the
 1804: local variable names. @code{@{} switches the dictionary pointer to this
 1805: area and @code{@}} switches it back and generates the locals
 1806: initializing code. @code{W:} etc.@ are normal defining words. This
 1807: special area is cleared at the start of every colon definition.
 1808: 
 1809: A special feature of Gforth's dictionary is used to implement the
 1810: definition of locals without type specifiers: every wordlist (aka
 1811: vocabulary) has its own methods for searching
 1812: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
 1813: with a special search method: When it is searched for a word, it
 1814: actually creates that word using @code{W:}. @code{@{} changes the search
 1815: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
 1816: and then the wordlist for defining locals without type specifiers.
 1817: 
 1818: The lifetime rules support a stack discipline within a colon
 1819: definition: The lifetime of a local is either nested with other locals
 1820: lifetimes or it does not overlap them.
 1821: 
 1822: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
 1823: pointer manipulation is generated. Between control structure words
 1824: locals definitions can push locals onto the locals stack. @code{AGAIN}
 1825: is the simplest of the other three control flow words. It has to
 1826: restore the locals stack depth of the corresponding @code{BEGIN}
 1827: before branching. The code looks like this:
 1828: @format
 1829: @code{lp+!#} current-locals-size @minus{} dest-locals-size
 1830: @code{branch} <begin>
 1831: @end format
 1832: 
 1833: @code{UNTIL} is a little more complicated: If it branches back, it
 1834: must adjust the stack just like @code{AGAIN}. But if it falls through,
 1835: the locals stack must not be changed. The compiler generates the
 1836: following code:
 1837: @format
 1838: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
 1839: @end format
 1840: The locals stack pointer is only adjusted if the branch is taken.
 1841: 
 1842: @code{THEN} can produce somewhat inefficient code:
 1843: @format
 1844: @code{lp+!#} current-locals-size @minus{} orig-locals-size
 1845: <orig target>:
 1846: @code{lp+!#} orig-locals-size @minus{} new-locals-size
 1847: @end format
 1848: The second @code{lp+!#} adjusts the locals stack pointer from the
 1849: level at the @var{orig} point to the level after the @code{THEN}. The
 1850: first @code{lp+!#} adjusts the locals stack pointer from the current
 1851: level to the level at the orig point, so the complete effect is an
 1852: adjustment from the current level to the right level after the
 1853: @code{THEN}.
 1854: 
 1855: In a conventional Forth implementation a dest control-flow stack entry
 1856: is just the target address and an orig entry is just the address to be
 1857: patched. Our locals implementation adds a wordlist to every orig or dest
 1858: item. It is the list of locals visible (or assumed visible) at the point
 1859: described by the entry. Our implementation also adds a tag to identify
 1860: the kind of entry, in particular to differentiate between live and dead
 1861: (reachable and unreachable) orig entries.
 1862: 
 1863: A few unusual operations have to be performed on locals wordlists:
 1864: 
 1865: doc-common-list
 1866: doc-sub-list?
 1867: doc-list-size
 1868: 
 1869: Several features of our locals wordlist implementation make these
 1870: operations easy to implement: The locals wordlists are organised as
 1871: linked lists; the tails of these lists are shared, if the lists
 1872: contain some of the same locals; and the address of a name is greater
 1873: than the address of the names behind it in the list.
 1874: 
 1875: Another important implementation detail is the variable
 1876: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
 1877: determine if they can be reached directly or only through the branch
 1878: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
 1879: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
 1880: definition, by @code{BEGIN} and usually by @code{THEN}.
 1881: 
 1882: Counted loops are similar to other loops in most respects, but
 1883: @code{LEAVE} requires special attention: It performs basically the same
 1884: service as @code{AHEAD}, but it does not create a control-flow stack
 1885: entry. Therefore the information has to be stored elsewhere;
 1886: traditionally, the information was stored in the target fields of the
 1887: branches created by the @code{LEAVE}s, by organizing these fields into a
 1888: linked list. Unfortunately, this clever trick does not provide enough
 1889: space for storing our extended control flow information. Therefore, we
 1890: introduce another stack, the leave stack. It contains the control-flow
 1891: stack entries for all unresolved @code{LEAVE}s.
 1892: 
 1893: Local names are kept until the end of the colon definition, even if
 1894: they are no longer visible in any control-flow path. In a few cases
 1895: this may lead to increased space needs for the locals name area, but
 1896: usually less than reclaiming this space would cost in code size.
 1897: 
 1898: 
 1899: @node ANS Forth locals,  , Gforth locals, Locals
 1900: @subsection ANS Forth locals
 1901: 
 1902: The ANS Forth locals wordset does not define a syntax for locals, but
 1903: words that make it possible to define various syntaxes. One of the
 1904: possible syntaxes is a subset of the syntax we used in the Gforth locals
 1905: wordset, i.e.:
 1906: 
 1907: @example
 1908: @{ local1 local2 ... -- comment @}
 1909: @end example
 1910: or
 1911: @example
 1912: @{ local1 local2 ... @}
 1913: @end example
 1914: 
 1915: The order of the locals corresponds to the order in a stack comment. The
 1916: restrictions are:
 1917: 
 1918: @itemize @bullet
 1919: @item
 1920: Locals can only be cell-sized values (no type specifiers are allowed).
 1921: @item
 1922: Locals can be defined only outside control structures.
 1923: @item
 1924: Locals can interfere with explicit usage of the return stack. For the
 1925: exact (and long) rules, see the standard. If you don't use return stack
 1926: accessing words in a definition using locals, you will be all right. The
 1927: purpose of this rule is to make locals implementation on the return
 1928: stack easier.
 1929: @item
 1930: The whole definition must be in one line.
 1931: @end itemize
 1932: 
 1933: Locals defined in this way behave like @code{VALUE}s
 1934: (@xref{Values}). I.e., they are initialized from the stack. Using their
 1935: name produces their value. Their value can be changed using @code{TO}.
 1936: 
 1937: Since this syntax is supported by Gforth directly, you need not do
 1938: anything to use it. If you want to port a program using this syntax to
 1939: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
 1940: syntax on the other system.
 1941: 
 1942: Note that a syntax shown in the standard, section A.13 looks
 1943: similar, but is quite different in having the order of locals
 1944: reversed. Beware!
 1945: 
 1946: The ANS Forth locals wordset itself consists of the following word
 1947: 
 1948: doc-(local)
 1949: 
 1950: The ANS Forth locals extension wordset defines a syntax, but it is so
 1951: awful that we strongly recommend not to use it. We have implemented this
 1952: syntax to make porting to Gforth easy, but do not document it here. The
 1953: problem with this syntax is that the locals are defined in an order
 1954: reversed with respect to the standard stack comment notation, making
 1955: programs harder to read, and easier to misread and miswrite. The only
 1956: merit of this syntax is that it is easy to implement using the ANS Forth
 1957: locals wordset.
 1958: 
 1959: @node Defining Words, Wordlists, Locals, Words
 1960: @section Defining Words
 1961: 
 1962: @menu
 1963: * Values::                      
 1964: @end menu
 1965: 
 1966: @node Values,  , Defining Words, Defining Words
 1967: @subsection Values
 1968: 
 1969: @node Wordlists, Files, Defining Words, Words
 1970: @section Wordlists
 1971: 
 1972: @node Files, Blocks, Wordlists, Words
 1973: @section Files
 1974: 
 1975: @node Blocks, Other I/O, Files, Words
 1976: @section Blocks
 1977: 
 1978: @node Other I/O, Programming Tools, Blocks, Words
 1979: @section Other I/O
 1980: 
 1981: @node Programming Tools, Assembler and Code words, Other I/O, Words
 1982: @section Programming Tools
 1983: 
 1984: @menu
 1985: * Debugging::                   Simple and quick.
 1986: * Assertions::                  Making your programs self-checking.
 1987: @end menu
 1988: 
 1989: @node Debugging, Assertions, Programming Tools, Programming Tools
 1990: @subsection Debugging
 1991: 
 1992: The simple debugging aids provided in @file{debugging.fs}
 1993: are meant to support a different style of debugging than the
 1994: tracing/stepping debuggers used in languages with long turn-around
 1995: times.
 1996: 
 1997: A much better (faster) way in fast-compilig languages is to add
 1998: printing code at well-selected places, let the program run, look at
 1999: the output, see where things went wrong, add more printing code, etc.,
 2000: until the bug is found.
 2001: 
 2002: The word @code{~~} is easy to insert. It just prints debugging
 2003: information (by default the source location and the stack contents). It
 2004: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
 2005: query-replace them with nothing). The deferred words
 2006: @code{printdebugdata} and @code{printdebugline} control the output of
 2007: @code{~~}. The default source location output format works well with
 2008: Emacs' compilation mode, so you can step through the program at the
 2009: source level using @kbd{C-x `} (the advantage over a stepping debugger
 2010: is that you can step in any direction and you know where the crash has
 2011: happened or where the strange data has occurred).
 2012: 
 2013: Note that the default actions clobber the contents of the pictured
 2014: numeric output string, so you should not use @code{~~}, e.g., between
 2015: @code{<#} and @code{#>}.
 2016: 
 2017: doc-~~
 2018: doc-printdebugdata
 2019: doc-printdebugline
 2020: 
 2021: @node Assertions,  , Debugging, Programming Tools
 2022: @subsection Assertions
 2023: 
 2024: It is a good idea to make your programs self-checking, in particular, if
 2025: you use an assumption (e.g., that a certain field of a data structure is
 2026: never zero) that may become wrong during maintenance. Gforth supports
 2027: assertions for this purpose. They are used like this:
 2028: 
 2029: @example
 2030: assert( @var{flag} )
 2031: @end example
 2032: 
 2033: The code between @code{assert(} and @code{)} should compute a flag, that
 2034: should be true if everything is alright and false otherwise. It should
 2035: not change anything else on the stack. The overall stack effect of the
 2036: assertion is @code{( -- )}. E.g.
 2037: 
 2038: @example
 2039: assert( 1 1 + 2 = ) \ what we learn in school
 2040: assert( dup 0<> ) \ assert that the top of stack is not zero
 2041: assert( false ) \ this code should not be reached
 2042: @end example
 2043: 
 2044: The need for assertions is different at different times. During
 2045: debugging, we want more checking, in production we sometimes care more
 2046: for speed. Therefore, assertions can be turned off, i.e., the assertion
 2047: becomes a comment. Depending on the importance of an assertion and the
 2048: time it takes to check it, you may want to turn off some assertions and
 2049: keep others turned on. Gforth provides several levels of assertions for
 2050: this purpose:
 2051: 
 2052: doc-assert0(
 2053: doc-assert1(
 2054: doc-assert2(
 2055: doc-assert3(
 2056: doc-assert(
 2057: doc-)
 2058: 
 2059: @code{Assert(} is the same as @code{assert1(}. The variable
 2060: @code{assert-level} specifies the highest assertions that are turned
 2061: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
 2062: @code{assert1(} assertions perform checking, while @code{assert2(} and
 2063: @code{assert3(} assertions are treated as comments.
 2064: 
 2065: Note that the @code{assert-level} is evaluated at compile-time, not at
 2066: run-time. I.e., you cannot turn assertions on or off at run-time, you
 2067: have to set the @code{assert-level} appropriately before compiling a
 2068: piece of code. You can compile several pieces of code at several
 2069: @code{assert-level}s (e.g., a trusted library at level 1 and newly
 2070: written code at level 3).
 2071: 
 2072: doc-assert-level
 2073: 
 2074: If an assertion fails, a message compatible with Emacs' compilation mode
 2075: is produced and the execution is aborted (currently with @code{ABORT"}.
 2076: If there is interest, we will introduce a special throw code. But if you
 2077: intend to @code{catch} a specific condition, using @code{throw} is
 2078: probably more appropriate than an assertion).
 2079: 
 2080: @node Assembler and Code words, Threading Words, Programming Tools, Words
 2081: @section Assembler and Code words
 2082: 
 2083: Gforth provides some words for defining primitives (words written in
 2084: machine code), and for defining the the machine-code equivalent of
 2085: @code{DOES>}-based defining words. However, the machine-independent
 2086: nature of Gforth poses a few problems: First of all. Gforth runs on
 2087: several architectures, so it can provide no standard assembler. What's
 2088: worse is that the register allocation not only depends on the processor,
 2089: but also on the @code{gcc} version and options used.
 2090: 
 2091: The words that Gforth offers encapsulate some system dependences (e.g., the
 2092: header structure), so a system-independent assembler may be used in
 2093: Gforth. If you do not have an assembler, you can compile machine code
 2094: directly with @code{,} and @code{c,}.
 2095: 
 2096: doc-assembler
 2097: doc-code
 2098: doc-end-code
 2099: doc-;code
 2100: doc-flush-icache
 2101: 
 2102: If @code{flush-icache} does not work correctly, @code{code} words
 2103: etc. will not work (reliably), either.
 2104: 
 2105: These words are rarely used. Therefore they reside in @code{code.fs},
 2106: which is usually not loaded (except @code{flush-icache}, which is always
 2107: present). You can load them with @code{require code.fs}.
 2108: 
 2109: In the assembly code you will want to refer to the inner interpreter's
 2110: registers (e.g., the data stack pointer) and you may want to use other
 2111: registers for temporary storage. Unfortunately, the register allocation
 2112: is installation-dependent.
 2113: 
 2114: The easiest solution is to use explicit register declarations
 2115: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
 2116: GNU C Manual}) for all of the inner interpreter's registers: You have to
 2117: compile Gforth with @code{-DFORCE_REG} (configure option
 2118: @code{--enable-force-reg}) and the appropriate declarations must be
 2119: present in the @code{machine.h} file (see @code{mips.h} for an example;
 2120: you can find a full list of all declarable register symbols with
 2121: @code{grep register engine.c}). If you give explicit registers to all
 2122: variables that are declared at the beginning of @code{engine()}, you
 2123: should be able to use the other caller-saved registers for temporary
 2124: storage. Alternatively, you can use the @code{gcc} option
 2125: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
 2126: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
 2127: (however, this restriction on register allocation may slow Gforth
 2128: significantly).
 2129: 
 2130: If this solution is not viable (e.g., because @code{gcc} does not allow
 2131: you to explicitly declare all the registers you need), you have to find
 2132: out by looking at the code where the inner interpreter's registers
 2133: reside and which registers can be used for temporary storage. You can
 2134: get an assembly listing of the engine's code with @code{make engine.s}.
 2135: 
 2136: In any case, it is good practice to abstract your assembly code from the
 2137: actual register allocation. E.g., if the data stack pointer resides in
 2138: register @code{$17}, create an alias for this register called @code{sp},
 2139: and use that in your assembly code.
 2140: 
 2141: Another option for implementing normal and defining words efficiently
 2142: is: adding the wanted functionality to the source of Gforth. For normal
 2143: words you just have to edit @file{primitives}, defining words (for fast
 2144: defined words) may require changes in @file{engine.c},
 2145: @file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
 2146: 
 2147: 
 2148: @node Threading Words,  , Assembler and Code words, Words
 2149: @section Threading Words
 2150: 
 2151: These words provide access to code addresses and other threading stuff
 2152: in Gforth (and, possibly, other interpretive Forths). It more or less
 2153: abstracts away the differences between direct and indirect threading
 2154: (and, for direct threading, the machine dependences). However, at
 2155: present this wordset is still inclomplete. It is also pretty low-level;
 2156: some day it will hopefully be made unnecessary by an internals words set
 2157: that abstracts implementation details away completely.
 2158: 
 2159: doc->code-address
 2160: doc->does-code
 2161: doc-code-address!
 2162: doc-does-code!
 2163: doc-does-handler!
 2164: doc-/does-handler
 2165: 
 2166: The code addresses produced by various defining words are produced by
 2167: the following words:
 2168: 
 2169: doc-docol:
 2170: doc-docon:
 2171: doc-dovar:
 2172: doc-douser:
 2173: doc-dodefer:
 2174: doc-dofield:
 2175: 
 2176: Currently there is no installation-independent way for recogizing words
 2177: defined by a @code{CREATE}...@code{DOES>} word; however, once you know
 2178: that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
 2179: use @code{>DOES-CODE}.
 2180: 
 2181: @node ANS conformance, Model, Words, Top
 2182: @chapter ANS conformance
 2183: 
 2184: To the best of our knowledge, Gforth is an
 2185: 
 2186: ANS Forth System
 2187: @itemize
 2188: @item providing the Core Extensions word set
 2189: @item providing the Block word set
 2190: @item providing the Block Extensions word set
 2191: @item providing the Double-Number word set
 2192: @item providing the Double-Number Extensions word set
 2193: @item providing the Exception word set
 2194: @item providing the Exception Extensions word set
 2195: @item providing the Facility word set
 2196: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
 2197: @item providing the File Access word set
 2198: @item providing the File Access Extensions word set
 2199: @item providing the Floating-Point word set
 2200: @item providing the Floating-Point Extensions word set
 2201: @item providing the Locals word set
 2202: @item providing the Locals Extensions word set
 2203: @item providing the Memory-Allocation word set
 2204: @item providing the Memory-Allocation Extensions word set (that one's easy)
 2205: @item providing the Programming-Tools word set
 2206: @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
 2207: @item providing the Search-Order word set
 2208: @item providing the Search-Order Extensions word set
 2209: @item providing the String word set
 2210: @item providing the String Extensions word set (another easy one)
 2211: @end itemize
 2212: 
 2213: In addition, ANS Forth systems are required to document certain
 2214: implementation choices. This chapter tries to meet these
 2215: requirements. In many cases it gives a way to ask the system for the
 2216: information instead of providing the information directly, in
 2217: particular, if the information depends on the processor, the operating
 2218: system or the installation options chosen, or if they are likely to
 2219: change during the maintenance of Gforth.
 2220: 
 2221: @comment The framework for the rest has been taken from pfe.
 2222: 
 2223: @menu
 2224: * The Core Words::              
 2225: * The optional Block word set::  
 2226: * The optional Double Number word set::  
 2227: * The optional Exception word set::  
 2228: * The optional Facility word set::  
 2229: * The optional File-Access word set::  
 2230: * The optional Floating-Point word set::  
 2231: * The optional Locals word set::  
 2232: * The optional Memory-Allocation word set::  
 2233: * The optional Programming-Tools word set::  
 2234: * The optional Search-Order word set::  
 2235: @end menu
 2236: 
 2237: 
 2238: @c =====================================================================
 2239: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
 2240: @comment  node-name,  next,  previous,  up
 2241: @section The Core Words
 2242: @c =====================================================================
 2243: 
 2244: @menu
 2245: * core-idef::                   Implementation Defined Options                   
 2246: * core-ambcond::                Ambiguous Conditions                
 2247: * core-other::                  Other System Documentation                  
 2248: @end menu
 2249: 
 2250: @c ---------------------------------------------------------------------
 2251: @node core-idef, core-ambcond, The Core Words, The Core Words
 2252: @subsection Implementation Defined Options
 2253: @c ---------------------------------------------------------------------
 2254: 
 2255: @table @i
 2256: 
 2257: @item (Cell) aligned addresses:
 2258: processor-dependent. Gforth's alignment words perform natural alignment
 2259: (e.g., an address aligned for a datum of size 8 is divisible by
 2260: 8). Unaligned accesses usually result in a @code{-23 THROW}.
 2261: 
 2262: @item @code{EMIT} and non-graphic characters:
 2263: The character is output using the C library function (actually, macro)
 2264: @code{putchar}.
 2265: 
 2266: @item character editing of @code{ACCEPT} and @code{EXPECT}:
 2267: This is modeled on the GNU readline library (@pxref{Readline
 2268: Interaction, , Command Line Editing, readline, The GNU Readline
 2269: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
 2270: producing a full word completion every time you type it (instead of
 2271: producing the common prefix of all completions).
 2272: 
 2273: @item character set:
 2274: The character set of your computer and display device. Gforth is
 2275: 8-bit-clean (but some other component in your system may make trouble).
 2276: 
 2277: @item Character-aligned address requirements:
 2278: installation-dependent. Currently a character is represented by a C
 2279: @code{unsigned char}; in the future we might switch to @code{wchar_t}
 2280: (Comments on that requested).
 2281: 
 2282: @item character-set extensions and matching of names:
 2283: Any character except the ASCII NUL charcter can be used in a
 2284: name. Matching is case-insensitive. The matching is performed using the
 2285: C function @code{strncasecmp}, whose function is probably influenced by
 2286: the locale. E.g., the @code{C} locale does not know about accents and
 2287: umlauts, so they are matched case-sensitively in that locale. For
 2288: portability reasons it is best to write programs such that they work in
 2289: the @code{C} locale. Then one can use libraries written by a Polish
 2290: programmer (who might use words containing ISO Latin-2 encoded
 2291: characters) and by a French programmer (ISO Latin-1) in the same program
 2292: (of course, @code{WORDS} will produce funny results for some of the
 2293: words (which ones, depends on the font you are using)). Also, the locale
 2294: you prefer may not be available in other operating systems. Hopefully,
 2295: Unicode will solve these problems one day.
 2296: 
 2297: @item conditions under which control characters match a space delimiter:
 2298: If @code{WORD} is called with the space character as a delimiter, all
 2299: white-space characters (as identified by the C macro @code{isspace()})
 2300: are delimiters. @code{PARSE}, on the other hand, treats space like other
 2301: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
 2302: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
 2303: interpreter (aka text interpreter) by default, treats all white-space
 2304: characters as delimiters.
 2305: 
 2306: @item format of the control flow stack:
 2307: The data stack is used as control flow stack. The size of a control flow
 2308: stack item in cells is given by the constant @code{cs-item-size}. At the
 2309: time of this writing, an item consists of a (pointer to a) locals list
 2310: (third), an address in the code (second), and a tag for identifying the
 2311: item (TOS). The following tags are used: @code{defstart},
 2312: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
 2313: @code{scopestart}.
 2314: 
 2315: @item conversion of digits > 35
 2316: The characters @code{[\]^_'} are the digits with the decimal value
 2317: 36@minus{}41. There is no way to input many of the larger digits.
 2318: 
 2319: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
 2320: The cursor is moved to the end of the entered string. If the input is
 2321: terminated using the @kbd{Return} key, a space is typed.
 2322: 
 2323: @item exception abort sequence of @code{ABORT"}:
 2324: The error string is stored into the variable @code{"error} and a
 2325: @code{-2 throw} is performed.
 2326: 
 2327: @item input line terminator:
 2328: For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
 2329: these characters is typically produced when you type the @kbd{Enter} or
 2330: @kbd{Return} key.
 2331: 
 2332: @item maximum size of a counted string:
 2333: @code{s" /counted-string" environment? drop .}. Currently 255 characters
 2334: on all ports, but this may change.
 2335: 
 2336: @item maximum size of a parsed string:
 2337: Given by the constant @code{/line}. Currently 255 characters.
 2338: 
 2339: @item maximum size of a definition name, in characters:
 2340: 31
 2341: 
 2342: @item maximum string length for @code{ENVIRONMENT?}, in characters:
 2343: 31
 2344: 
 2345: @item method of selecting the user input device:
 2346: The user input device is the standard input. There is currently no way to
 2347: change it from within Gforth. However, the input can typically be
 2348: redirected in the command line that starts Gforth.
 2349: 
 2350: @item method of selecting the user output device:
 2351: The user output device is the standard output. It cannot be redirected
 2352: from within Gforth, but typically from the command line that starts
 2353: Gforth. Gforth uses buffered output, so output on a terminal does not
 2354: become visible before the next newline or buffer overflow. Output on
 2355: non-terminals is invisible until the buffer overflows.
 2356: 
 2357: @item methods of dictionary compilation:
 2358: What are we expected to document here?
 2359: 
 2360: @item number of bits in one address unit:
 2361: @code{s" address-units-bits" environment? drop .}. 8 in all current
 2362: ports.
 2363: 
 2364: @item number representation and arithmetic:
 2365: Processor-dependent. Binary two's complement on all current ports.
 2366: 
 2367: @item ranges for integer types:
 2368: Installation-dependent. Make environmental queries for @code{MAX-N},
 2369: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
 2370: unsigned (and positive) types is 0. The lower bound for signed types on
 2371: two's complement and one's complement machines machines can be computed
 2372: by adding 1 to the upper bound.
 2373: 
 2374: @item read-only data space regions:
 2375: The whole Forth data space is writable.
 2376: 
 2377: @item size of buffer at @code{WORD}:
 2378: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 2379: shared with the pictured numeric output string. If overwriting
 2380: @code{PAD} is acceptable, it is as large as the remaining dictionary
 2381: space, although only as much can be sensibly used as fits in a counted
 2382: string.
 2383: 
 2384: @item size of one cell in address units:
 2385: @code{1 cells .}.
 2386: 
 2387: @item size of one character in address units:
 2388: @code{1 chars .}. 1 on all current ports.
 2389: 
 2390: @item size of the keyboard terminal buffer:
 2391: Varies. You can determine the size at a specific time using @code{lp@
 2392: tib - .}. It is shared with the locals stack and TIBs of files that
 2393: include the current file. You can change the amount of space for TIBs
 2394: and locals stack at Gforth startup with the command line option
 2395: @code{-l}.
 2396: 
 2397: @item size of the pictured numeric output buffer:
 2398: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
 2399: shared with @code{WORD}.
 2400: 
 2401: @item size of the scratch area returned by @code{PAD}:
 2402: The remainder of dictionary space. You can even use the unused part of
 2403: the data stack space. The current size can be computed with @code{sp@
 2404: pad - .}.
 2405: 
 2406: @item system case-sensitivity characteristics:
 2407: Dictionary searches are case insensitive. However, as explained above
 2408: under @i{character-set extensions}, the matching for non-ASCII
 2409: characters is determined by the locale you are using. In the default
 2410: @code{C} locale all non-ASCII characters are matched case-sensitively.
 2411: 
 2412: @item system prompt:
 2413: @code{ ok} in interpret state, @code{ compiled} in compile state.
 2414: 
 2415: @item division rounding:
 2416: installation dependent. @code{s" floored" environment? drop .}. We leave
 2417: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
 2418: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
 2419: 
 2420: @item values of @code{STATE} when true:
 2421: -1.
 2422: 
 2423: @item values returned after arithmetic overflow:
 2424: On two's complement machines, arithmetic is performed modulo
 2425: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 2426: arithmetic (with appropriate mapping for signed types). Division by zero
 2427: typically results in a @code{-55 throw} (floatingpoint unidentified
 2428: fault), although a @code{-10 throw} (divide by zero) would be more
 2429: appropriate.
 2430: 
 2431: @item whether the current definition can be found after @t{DOES>}:
 2432: No.
 2433: 
 2434: @end table
 2435: 
 2436: @c ---------------------------------------------------------------------
 2437: @node core-ambcond, core-other, core-idef, The Core Words
 2438: @subsection Ambiguous conditions
 2439: @c ---------------------------------------------------------------------
 2440: 
 2441: @table @i
 2442: 
 2443: @item a name is neither a word nor a number:
 2444: @code{-13 throw} (Undefined word)
 2445: 
 2446: @item a definition name exceeds the maximum length allowed:
 2447: @code{-19 throw} (Word name too long)
 2448: 
 2449: @item addressing a region not inside the various data spaces of the forth system:
 2450: The stacks, code space and name space are accessible. Machine code space is
 2451: typically readable. Accessing other addresses gives results dependent on
 2452: the operating system. On decent systems: @code{-9 throw} (Invalid memory
 2453: address).
 2454: 
 2455: @item argument type incompatible with parameter:
 2456: This is usually not caught. Some words perform checks, e.g., the control
 2457: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
 2458: mismatch).
 2459: 
 2460: @item attempting to obtain the execution token of a word with undefined execution semantics:
 2461: You get an execution token representing the compilation semantics
 2462: instead.
 2463: 
 2464: @item dividing by zero:
 2465: typically results in a @code{-55 throw} (floating point unidentified
 2466: fault), although a @code{-10 throw} (divide by zero) would be more
 2467: appropriate.
 2468: 
 2469: @item insufficient data stack or return stack space:
 2470: Not checked. This typically results in mysterious illegal memory
 2471: accesses, producing @code{-9 throw} (Invalid memory address) or
 2472: @code{-23 throw} (Address alignment exception).
 2473: 
 2474: @item insufficient space for loop control parameters:
 2475: like other return stack overflows.
 2476: 
 2477: @item insufficient space in the dictionary:
 2478: Not checked. Similar results as stack overflows. However, typically the
 2479: error appears at a different place when one inserts or removes code.
 2480: 
 2481: @item interpreting a word with undefined interpretation semantics:
 2482: For some words, we defined interpretation semantics. For the others:
 2483: @code{-14 throw} (Interpreting a compile-only word). Note that this is
 2484: checked only by the outer (aka text) interpreter; if the word is
 2485: @code{execute}d in some other way, it will typically perform it's
 2486: compilation semantics even in interpret state. (We could change @code{'}
 2487: and relatives not to give the xt of such words, but we think that would
 2488: be too restrictive).
 2489: 
 2490: @item modifying the contents of the input buffer or a string literal:
 2491: These are located in writable memory and can be modified.
 2492: 
 2493: @item overflow of the pictured numeric output string:
 2494: Not checked.
 2495: 
 2496: @item parsed string overflow:
 2497: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
 2498: 
 2499: @item producing a result out of range:
 2500: On two's complement machines, arithmetic is performed modulo
 2501: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
 2502: arithmetic (with appropriate mapping for signed types). Division by zero
 2503: typically results in a @code{-55 throw} (floatingpoint unidentified
 2504: fault), although a @code{-10 throw} (divide by zero) would be more
 2505: appropriate. @code{convert} and @code{>number} currently overflow
 2506: silently.
 2507: 
 2508: @item reading from an empty data or return stack:
 2509: The data stack is checked by the outer (aka text) interpreter after
 2510: every word executed. If it has underflowed, a @code{-4 throw} (Stack
 2511: underflow) is performed. Apart from that, the stacks are not checked and
 2512: underflows can result in similar behaviour as overflows (of adjacent
 2513: stacks).
 2514: 
 2515: @item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
 2516: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
 2517: use zero-length string as a name). Words like @code{'} probably will not
 2518: find what they search. Note that it is possible to create zero-length
 2519: names with @code{nextname} (should it not?).
 2520: 
 2521: @item @code{>IN} greater than input buffer:
 2522: The next invocation of a parsing word returns a string wih length 0.
 2523: 
 2524: @item @code{RECURSE} appears after @code{DOES>}:
 2525: Compiles a recursive call to the defining word not to the defined word.
 2526: 
 2527: @item argument input source different than current input source for @code{RESTORE-INPUT}:
 2528: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
 2529: the end of the file was reached), its source-id may be
 2530: reused. Therefore, restoring an input source specification referencing a
 2531: closed file may lead to unpredictable results instead of a @code{-12
 2532: THROW}.
 2533: 
 2534: In the future, Gforth may be able to retore input source specifications
 2535: from other than the current input soruce.
 2536: 
 2537: @item data space containing definitions gets de-allocated:
 2538: Deallocation with @code{allot} is not checked. This typically resuls in
 2539: memory access faults or execution of illegal instructions.
 2540: 
 2541: @item data space read/write with incorrect alignment:
 2542: Processor-dependent. Typically results in a @code{-23 throw} (Address
 2543: alignment exception). Under Linux on a 486 or later processor with
 2544: alignment turned on, incorrect alignment results in a @code{-9 throw}
 2545: (Invalid memory address). There are reportedly some processors with
 2546: alignment restrictions that do not report them.
 2547: 
 2548: @item data space pointer not properly aligned, @code{,}, @code{C,}:
 2549: Like other alignment errors.
 2550: 
 2551: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
 2552: Not checked. May cause an illegal memory access.
 2553: 
 2554: @item loop control parameters not available:
 2555: Not checked. The counted loop words simply assume that the top of return
 2556: stack items are loop control parameters and behave accordingly.
 2557: 
 2558: @item most recent definition does not have a name (@code{IMMEDIATE}):
 2559: @code{abort" last word was headerless"}.
 2560: 
 2561: @item name not defined by @code{VALUE} used by @code{TO}:
 2562: @code{-32 throw} (Invalid name argument)
 2563: 
 2564: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
 2565: @code{-13 throw} (Undefined word)
 2566: 
 2567: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
 2568: Gforth behaves as if they were of the same type. I.e., you can predict
 2569: the behaviour by interpreting all parameters as, e.g., signed.
 2570: 
 2571: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
 2572: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
 2573: @code{TO}.
 2574: 
 2575: @item String longer than a counted string returned by @code{WORD}:
 2576: Not checked. The string will be ok, but the count will, of course,
 2577: contain only the least significant bits of the length.
 2578: 
 2579: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
 2580: Processor-dependent. Typical behaviours are returning 0 and using only
 2581: the low bits of the shift count.
 2582: 
 2583: @item word not defined via @code{CREATE}:
 2584: @code{>BODY} produces the PFA of the word no matter how it was defined.
 2585: 
 2586: @code{DOES>} changes the execution semantics of the last defined word no
 2587: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
 2588: @code{CREATE , DOES>}.
 2589: 
 2590: @item words improperly used outside @code{<#} and @code{#>}:
 2591: Not checked. As usual, you can expect memory faults.
 2592: 
 2593: @end table
 2594: 
 2595: 
 2596: @c ---------------------------------------------------------------------
 2597: @node core-other,  , core-ambcond, The Core Words
 2598: @subsection Other system documentation
 2599: @c ---------------------------------------------------------------------
 2600: 
 2601: @table @i
 2602: 
 2603: @item nonstandard words using @code{PAD}:
 2604: None.
 2605: 
 2606: @item operator's terminal facilities available:
 2607: After processing the command line, Gforth goes into interactive mode,
 2608: and you can give commands to Gforth interactively. The actual facilities
 2609: available depend on how you invoke Gforth.
 2610: 
 2611: @item program data space available:
 2612: @code{sp@ here - .} gives the space remaining for dictionary and data
 2613: stack together.
 2614: 
 2615: @item return stack space available:
 2616: By default 16 KBytes. The default can be overridden with the @code{-r}
 2617: switch (@pxref{Invocation}) when Gforth starts up.
 2618: 
 2619: @item stack space available:
 2620: @code{sp@ here - .} gives the space remaining for dictionary and data
 2621: stack together.
 2622: 
 2623: @item system dictionary space required, in address units:
 2624: Type @code{here forthstart - .} after startup. At the time of this
 2625: writing, this gives 70108 (bytes) on a 32-bit system.
 2626: @end table
 2627: 
 2628: 
 2629: @c =====================================================================
 2630: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
 2631: @section The optional Block word set
 2632: @c =====================================================================
 2633: 
 2634: @menu
 2635: * block-idef::                  Implementation Defined Options                  
 2636: * block-ambcond::               Ambiguous Conditions               
 2637: * block-other::                 Other System Documentation                 
 2638: @end menu
 2639: 
 2640: 
 2641: @c ---------------------------------------------------------------------
 2642: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
 2643: @subsection Implementation Defined Options
 2644: @c ---------------------------------------------------------------------
 2645: 
 2646: @table @i
 2647: 
 2648: @item the format for display by @code{LIST}:
 2649: First the screen number is displayed, then 16 lines of 64 characters,
 2650: each line preceded by the line number.
 2651: 
 2652: @item the length of a line affected by @code{\}:
 2653: 64 characters.
 2654: @end table
 2655: 
 2656: 
 2657: @c ---------------------------------------------------------------------
 2658: @node block-ambcond, block-other, block-idef, The optional Block word set
 2659: @subsection Ambiguous conditions
 2660: @c ---------------------------------------------------------------------
 2661: 
 2662: @table @i
 2663: 
 2664: @item correct block read was not possible:
 2665: Typically results in a @code{throw} of some OS-derived value (between
 2666: -512 and -2048). If the blocks file was just not long enough, blanks are
 2667: supplied for the missing portion.
 2668: 
 2669: @item I/O exception in block transfer:
 2670: Typically results in a @code{throw} of some OS-derived value (between
 2671: -512 and -2048).
 2672: 
 2673: @item invalid block number:
 2674: @code{-35 throw} (Invalid block number)
 2675: 
 2676: @item a program directly alters the contents of @code{BLK}:
 2677: The input stream is switched to that other block, at the same
 2678: position. If the storing to @code{BLK} happens when interpreting
 2679: non-block input, the system will get quite confused when the block ends.
 2680: 
 2681: @item no current block buffer for @code{UPDATE}:
 2682: @code{UPDATE} has no effect.
 2683: 
 2684: @end table
 2685: 
 2686: 
 2687: @c ---------------------------------------------------------------------
 2688: @node block-other,  , block-ambcond, The optional Block word set
 2689: @subsection Other system documentation
 2690: @c ---------------------------------------------------------------------
 2691: 
 2692: @table @i
 2693: 
 2694: @item any restrictions a multiprogramming system places on the use of buffer addresses:
 2695: No restrictions (yet).
 2696: 
 2697: @item the number of blocks available for source and data:
 2698: depends on your disk space.
 2699: 
 2700: @end table
 2701: 
 2702: 
 2703: @c =====================================================================
 2704: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
 2705: @section The optional Double Number word set
 2706: @c =====================================================================
 2707: 
 2708: @menu
 2709: * double-ambcond::              Ambiguous Conditions              
 2710: @end menu
 2711: 
 2712: 
 2713: @c ---------------------------------------------------------------------
 2714: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
 2715: @subsection Ambiguous conditions
 2716: @c ---------------------------------------------------------------------
 2717: 
 2718: @table @i
 2719: 
 2720: @item @var{d} outside of range of @var{n} in @code{D>S}:
 2721: The least significant cell of @var{d} is produced.
 2722: 
 2723: @end table
 2724: 
 2725: 
 2726: @c =====================================================================
 2727: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
 2728: @section The optional Exception word set
 2729: @c =====================================================================
 2730: 
 2731: @menu
 2732: * exception-idef::              Implementation Defined Options              
 2733: @end menu
 2734: 
 2735: 
 2736: @c ---------------------------------------------------------------------
 2737: @node exception-idef,  , The optional Exception word set, The optional Exception word set
 2738: @subsection Implementation Defined Options
 2739: @c ---------------------------------------------------------------------
 2740: 
 2741: @table @i
 2742: @item @code{THROW}-codes used in the system:
 2743: The codes -256@minus{}-511 are used for reporting signals (see
 2744: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
 2745: (for file and memory allocation operations). The mapping from OS error
 2746: numbers to throw code is -512@minus{}@var{errno}. One side effect of
 2747: this mapping is that undefined OS errors produce a message with a
 2748: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
 2749: 488} on my system.
 2750: @end table
 2751: 
 2752: @c =====================================================================
 2753: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
 2754: @section The optional Facility word set
 2755: @c =====================================================================
 2756: 
 2757: @menu
 2758: * facility-idef::               Implementation Defined Options               
 2759: * facility-ambcond::            Ambiguous Conditions            
 2760: @end menu
 2761: 
 2762: 
 2763: @c ---------------------------------------------------------------------
 2764: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
 2765: @subsection Implementation Defined Options
 2766: @c ---------------------------------------------------------------------
 2767: 
 2768: @table @i
 2769: 
 2770: @item encoding of keyboard events (@code{EKEY}):
 2771: Not yet implemeted.
 2772: 
 2773: @item duration of a system clock tick
 2774: System dependent. With respect to @code{MS}, the time is specified in
 2775: microseconds. How well the OS and the hardware implement this, is
 2776: another question.
 2777: 
 2778: @item repeatability to be expected from the execution of @code{MS}:
 2779: System dependent. On Unix, a lot depends on load. If the system is
 2780: lightly loaded, and the delay is short enough that Gforth does not get
 2781: swapped out, the performance should be acceptable. Under MS-DOS and
 2782: other single-tasking systems, it should be good.
 2783: 
 2784: @end table
 2785: 
 2786: 
 2787: @c ---------------------------------------------------------------------
 2788: @node facility-ambcond,  , facility-idef, The optional Facility word set
 2789: @subsection Ambiguous conditions
 2790: @c ---------------------------------------------------------------------
 2791: 
 2792: @table @i
 2793: 
 2794: @item @code{AT-XY} can't be performed on user output device:
 2795: Largely terminal dependant. No range checks are done on the arguments.
 2796: No errors are reported. You may see some garbage appearing, you may see
 2797: simply nothing happen.
 2798: 
 2799: @end table
 2800: 
 2801: 
 2802: @c =====================================================================
 2803: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
 2804: @section The optional File-Access word set
 2805: @c =====================================================================
 2806: 
 2807: @menu
 2808: * file-idef::                   Implementation Defined Options                   
 2809: * file-ambcond::                Ambiguous Conditions                
 2810: @end menu
 2811: 
 2812: 
 2813: @c ---------------------------------------------------------------------
 2814: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
 2815: @subsection Implementation Defined Options
 2816: @c ---------------------------------------------------------------------
 2817: 
 2818: @table @i
 2819: 
 2820: @item File access methods used:
 2821: @code{R/O}, @code{R/W} and @code{BIN} work as you would
 2822: expect. @code{W/O} translates into the C file opening mode @code{w} (or
 2823: @code{wb}): The file is cleared, if it exists, and created, if it does
 2824: not (both with @code{open-file} and @code{create-file}).  Under Unix
 2825: @code{create-file} creates a file with 666 permissions modified by your
 2826: umask.
 2827: 
 2828: @item file exceptions:
 2829: The file words do not raise exceptions (except, perhaps, memory access
 2830: faults when you pass illegal addresses or file-ids).
 2831: 
 2832: @item file line terminator:
 2833: System-dependent. Gforth uses C's newline character as line
 2834: terminator. What the actual character code(s) of this are is
 2835: system-dependent.
 2836: 
 2837: @item file name format
 2838: System dependent. Gforth just uses the file name format of your OS.
 2839: 
 2840: @item information returned by @code{FILE-STATUS}:
 2841: @code{FILE-STATUS} returns the most powerful file access mode allowed
 2842: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
 2843: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
 2844: along with the retured mode.
 2845: 
 2846: @item input file state after an exception when including source:
 2847: All files that are left via the exception are closed.
 2848: 
 2849: @item @var{ior} values and meaning:
 2850: The @var{ior}s returned by the file and memory allocation words are
 2851: intended as throw codes. They typically are in the range
 2852: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 2853: @var{ior}s is -512@minus{}@var{errno}.
 2854: 
 2855: @item maximum depth of file input nesting:
 2856: limited by the amount of return stack, locals/TIB stack, and the number
 2857: of open files available. This should not give you troubles.
 2858: 
 2859: @item maximum size of input line:
 2860: @code{/line}. Currently 255.
 2861: 
 2862: @item methods of mapping block ranges to files:
 2863: Currently, the block words automatically access the file
 2864: @file{blocks.fb} in the currend working directory. More sophisticated
 2865: methods could be implemented if there is demand (and a volunteer).
 2866: 
 2867: @item number of string buffers provided by @code{S"}:
 2868: 1
 2869: 
 2870: @item size of string buffer used by @code{S"}:
 2871: @code{/line}. currently 255.
 2872: 
 2873: @end table
 2874: 
 2875: @c ---------------------------------------------------------------------
 2876: @node file-ambcond,  , file-idef, The optional File-Access word set
 2877: @subsection Ambiguous conditions
 2878: @c ---------------------------------------------------------------------
 2879: 
 2880: @table @i
 2881: 
 2882: @item attempting to position a file outside it's boundaries:
 2883: @code{REPOSITION-FILE} is performed as usual: Afterwards,
 2884: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
 2885: 
 2886: @item attempting to read from file positions not yet written:
 2887: End-of-file, i.e., zero characters are read and no error is reported.
 2888: 
 2889: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
 2890: An appropriate exception may be thrown, but a memory fault or other
 2891: problem is more probable.
 2892: 
 2893: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
 2894: The @var{ior} produced by the operation, that discovered the problem, is
 2895: thrown.
 2896: 
 2897: @item named file cannot be opened (@code{included}):
 2898: The @var{ior} produced by @code{open-file} is thrown.
 2899: 
 2900: @item requesting an unmapped block number:
 2901: There are no unmapped legal block numbers. On some operating systems,
 2902: writing a block with a large number may overflow the file system and
 2903: have an error message as consequence.
 2904: 
 2905: @item using @code{source-id} when @code{blk} is non-zero:
 2906: @code{source-id} performs its function. Typically it will give the id of
 2907: the source which loaded the block. (Better ideas?)
 2908: 
 2909: @end table
 2910: 
 2911: 
 2912: @c =====================================================================
 2913: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
 2914: @section The optional Floating-Point word set
 2915: @c =====================================================================
 2916: 
 2917: @menu
 2918: * floating-idef::               Implementation Defined Options
 2919: * floating-ambcond::            Ambiguous Conditions            
 2920: @end menu
 2921: 
 2922: 
 2923: @c ---------------------------------------------------------------------
 2924: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
 2925: @subsection Implementation Defined Options
 2926: @c ---------------------------------------------------------------------
 2927: 
 2928: @table @i
 2929: 
 2930: @item format and range of floating point numbers:
 2931: System-dependent; the @code{double} type of C.
 2932: 
 2933: @item results of @code{REPRESENT} when @var{float} is out of range:
 2934: System dependent; @code{REPRESENT} is implemented using the C library
 2935: function @code{ecvt()} and inherits its behaviour in this respect.
 2936: 
 2937: @item rounding or truncation of floating-point numbers:
 2938: System dependent; the rounding behaviour is inherited from the hosting C
 2939: compiler. IEEE-FP-based (i.e., most) systems by default round to
 2940: nearest, and break ties by rounding to even (i.e., such that the last
 2941: bit of the mantissa is 0).
 2942: 
 2943: @item size of floating-point stack:
 2944: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
 2945: with the command-line option @code{-f}.
 2946: 
 2947: @item width of floating-point stack:
 2948: @code{1 floats}.
 2949: 
 2950: @end table
 2951: 
 2952: 
 2953: @c ---------------------------------------------------------------------
 2954: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
 2955: @subsection Ambiguous conditions
 2956: @c ---------------------------------------------------------------------
 2957: 
 2958: @table @i
 2959: 
 2960: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
 2961: System-dependent. Typically results in an alignment fault like other
 2962: alignment violations.
 2963: 
 2964: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
 2965: System-dependent. Typically results in an alignment fault like other
 2966: alignment violations.
 2967: 
 2968: @item Floating-point result out of range:
 2969: System-dependent. Can result in a @code{-55 THROW} (Floating-point
 2970: unidentified fault), or can produce a special value representing, e.g.,
 2971: Infinity.
 2972: 
 2973: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
 2974: System-dependent. Typically results in an alignment fault like other
 2975: alignment violations.
 2976: 
 2977: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
 2978: The floating-point number is converted into decimal nonetheless.
 2979: 
 2980: @item Both arguments are equal to zero (@code{FATAN2}):
 2981: System-dependent. @code{FATAN2} is implemented using the C library
 2982: function @code{atan2()}.
 2983: 
 2984: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
 2985: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
 2986: because of small errors and the tan will be a very large (or very small)
 2987: but finite number.
 2988: 
 2989: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
 2990: The result is rounded to the nearest float.
 2991: 
 2992: @item dividing by zero:
 2993: @code{-55 throw} (Floating-point unidentified fault)
 2994: 
 2995: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
 2996: System dependent. On IEEE-FP based systems the number is converted into
 2997: an infinity.
 2998: 
 2999: @item @var{float}<1 (@code{facosh}):
 3000: @code{-55 throw} (Floating-point unidentified fault)
 3001: 
 3002: @item @var{float}=<-1 (@code{flnp1}):
 3003: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 3004: negative infinity is typically produced for @var{float}=-1.
 3005: 
 3006: @item @var{float}=<0 (@code{fln}, @code{flog}):
 3007: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
 3008: negative infinity is typically produced for @var{float}=0.
 3009: 
 3010: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
 3011: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
 3012: produces values for these inputs on my Linux box (Bug in the C library?)
 3013: 
 3014: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
 3015: @code{-55 throw} (Floating-point unidentified fault).
 3016: 
 3017: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
 3018: @code{-55 throw} (Floating-point unidentified fault).
 3019: 
 3020: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
 3021: This does not happen.
 3022: @end table
 3023: 
 3024: 
 3025: 
 3026: @c =====================================================================
 3027: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
 3028: @section The optional Locals word set
 3029: @c =====================================================================
 3030: 
 3031: @menu
 3032: * locals-idef::                 Implementation Defined Options                 
 3033: * locals-ambcond::              Ambiguous Conditions              
 3034: @end menu
 3035: 
 3036: 
 3037: @c ---------------------------------------------------------------------
 3038: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
 3039: @subsection Implementation Defined Options
 3040: @c ---------------------------------------------------------------------
 3041: 
 3042: @table @i
 3043: 
 3044: @item maximum number of locals in a definition:
 3045: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
 3046: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
 3047: characters. The number of locals in a definition is bounded by the size
 3048: of locals-buffer, which contains the names of the locals.
 3049: 
 3050: @end table
 3051: 
 3052: 
 3053: @c ---------------------------------------------------------------------
 3054: @node locals-ambcond,  , locals-idef, The optional Locals word set
 3055: @subsection Ambiguous conditions
 3056: @c ---------------------------------------------------------------------
 3057: 
 3058: @table @i
 3059: 
 3060: @item executing a named local in interpretation state:
 3061: @code{-14 throw} (Interpreting a compile-only word).
 3062: 
 3063: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
 3064: @code{-32 throw} (Invalid name argument)
 3065: 
 3066: @end table
 3067: 
 3068: 
 3069: @c =====================================================================
 3070: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
 3071: @section The optional Memory-Allocation word set
 3072: @c =====================================================================
 3073: 
 3074: @menu
 3075: * memory-idef::                 Implementation Defined Options                 
 3076: @end menu
 3077: 
 3078: 
 3079: @c ---------------------------------------------------------------------
 3080: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
 3081: @subsection Implementation Defined Options
 3082: @c ---------------------------------------------------------------------
 3083: 
 3084: @table @i
 3085: 
 3086: @item values and meaning of @var{ior}:
 3087: The @var{ior}s returned by the file and memory allocation words are
 3088: intended as throw codes. They typically are in the range
 3089: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
 3090: @var{ior}s is -512@minus{}@var{errno}.
 3091: 
 3092: @end table
 3093: 
 3094: @c =====================================================================
 3095: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
 3096: @section The optional Programming-Tools word set
 3097: @c =====================================================================
 3098: 
 3099: @menu
 3100: * programming-idef::            Implementation Defined Options            
 3101: * programming-ambcond::         Ambiguous Conditions         
 3102: @end menu
 3103: 
 3104: 
 3105: @c ---------------------------------------------------------------------
 3106: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
 3107: @subsection Implementation Defined Options
 3108: @c ---------------------------------------------------------------------
 3109: 
 3110: @table @i
 3111: 
 3112: @item ending sequence for input following @code{;code} and @code{code}:
 3113: Not implemented (yet).
 3114: 
 3115: @item manner of processing input following @code{;code} and @code{code}:
 3116: Not implemented (yet).
 3117: 
 3118: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
 3119: Not implemented (yet). If they were implemented, they would use the
 3120: search order wordset.
 3121: 
 3122: @item source and format of display by @code{SEE}:
 3123: The source for @code{see} is the intermediate code used by the inner
 3124: interpreter.  The current @code{see} tries to output Forth source code
 3125: as well as possible.
 3126: 
 3127: @end table
 3128: 
 3129: @c ---------------------------------------------------------------------
 3130: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
 3131: @subsection Ambiguous conditions
 3132: @c ---------------------------------------------------------------------
 3133: 
 3134: @table @i
 3135: 
 3136: @item deleting the compilation wordlist (@code{FORGET}):
 3137: Not implemented (yet).
 3138: 
 3139: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
 3140: This typically results in an @code{abort"} with a descriptive error
 3141: message (may change into a @code{-22 throw} (Control structure mismatch)
 3142: in the future). You may also get a memory access error. If you are
 3143: unlucky, this ambiguous condition is not caught.
 3144: 
 3145: @item @var{name} can't be found (@code{forget}):
 3146: Not implemented (yet).
 3147: 
 3148: @item @var{name} not defined via @code{CREATE}:
 3149: @code{;code} is not implemented (yet). If it were, it would behave like
 3150: @code{DOES>} in this respect, i.e., change the execution semantics of
 3151: the last defined word no matter how it was defined.
 3152: 
 3153: @item @code{POSTPONE} applied to @code{[IF]}:
 3154: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
 3155: equivalent to @code{[IF]}.
 3156: 
 3157: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
 3158: Continue in the same state of conditional compilation in the next outer
 3159: input source. Currently there is no warning to the user about this.
 3160: 
 3161: @item removing a needed definition (@code{FORGET}):
 3162: Not implemented (yet).
 3163: 
 3164: @end table
 3165: 
 3166: 
 3167: @c =====================================================================
 3168: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
 3169: @section The optional Search-Order word set
 3170: @c =====================================================================
 3171: 
 3172: @menu
 3173: * search-idef::                 Implementation Defined Options                 
 3174: * search-ambcond::              Ambiguous Conditions              
 3175: @end menu
 3176: 
 3177: 
 3178: @c ---------------------------------------------------------------------
 3179: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
 3180: @subsection Implementation Defined Options
 3181: @c ---------------------------------------------------------------------
 3182: 
 3183: @table @i
 3184: 
 3185: @item maximum number of word lists in search order:
 3186: @code{s" wordlists" environment? drop .}. Currently 16.
 3187: 
 3188: @item minimum search order:
 3189: @code{root root}.
 3190: 
 3191: @end table
 3192: 
 3193: @c ---------------------------------------------------------------------
 3194: @node search-ambcond,  , search-idef, The optional Search-Order word set
 3195: @subsection Ambiguous conditions
 3196: @c ---------------------------------------------------------------------
 3197: 
 3198: @table @i
 3199: 
 3200: @item changing the compilation wordlist (during compilation):
 3201: The definition is put into the wordlist that is the compilation wordlist
 3202: when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
 3203: @code{RECURSIVE}, etc.).
 3204: 
 3205: @item search order empty (@code{previous}):
 3206: @code{abort" Vocstack empty"}.
 3207: 
 3208: @item too many word lists in search order (@code{also}):
 3209: @code{abort" Vocstack full"}.
 3210: 
 3211: @end table
 3212: 
 3213: 
 3214: @node Model, Emacs and Gforth, ANS conformance, Top
 3215: @chapter Model
 3216: 
 3217: @node Emacs and Gforth, Internals, Model, Top
 3218: @chapter Emacs and Gforth
 3219: 
 3220: Gforth comes with @file{gforth.el}, an improved version of
 3221: @file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
 3222: improvements are a better (but still not perfect) handling of
 3223: indentation. I have also added comment paragraph filling (@kbd{M-q}),
 3224: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
 3225: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
 3226: stuff I do not use alone, even though some of it only makes sense for
 3227: TILE. To get a description of these features, enter Forth mode and type
 3228: @kbd{C-h m}.
 3229: 
 3230: In addition, Gforth supports Emacs quite well: The source code locations
 3231: given in error messages, debugging output (from @code{~~}) and failed
 3232: assertion messages are in the right format for Emacs' compilation mode
 3233: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
 3234: Manual}) so the source location corresponding to an error or other
 3235: message is only a few keystrokes away (@kbd{C-x `} for the next error,
 3236: @kbd{C-c C-c} for the error under the cursor).
 3237: 
 3238: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
 3239: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
 3240: contains the definitions of all words defined afterwards. You can then
 3241: find the source for a word using @kbd{M-.}. Note that emacs can use
 3242: several tags files at the same time (e.g., one for the Gforth sources
 3243: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
 3244: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
 3245: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
 3246: @file{/usr/local/share/gforth/0.2/TAGS}).
 3247: 
 3248: To get all these benefits, add the following lines to your @file{.emacs}
 3249: file:
 3250: 
 3251: @example
 3252: (autoload 'forth-mode "gforth.el")
 3253: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
 3254: @end example
 3255: 
 3256: @node Internals, Bugs, Emacs and Gforth, Top
 3257: @chapter Internals
 3258: 
 3259: Reading this section is not necessary for programming with Gforth. It
 3260: should be helpful for finding your way in the Gforth sources.
 3261: 
 3262: The ideas in this section have also been published in the papers
 3263: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
 3264: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
 3265: Ertl, presented at EuroForth '93; the latter is available at
 3266: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
 3267: 
 3268: @menu
 3269: * Portability::                 
 3270: * Threading::                   
 3271: * Primitives::                  
 3272: * System Architecture::         
 3273: * Performance::                 
 3274: @end menu
 3275: 
 3276: @node Portability, Threading, Internals, Internals
 3277: @section Portability
 3278: 
 3279: One of the main goals of the effort is availability across a wide range
 3280: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
 3281: this goal by manually coding the engine in assembly language for several
 3282: then-popular processors. This approach is very labor-intensive and the
 3283: results are short-lived due to progress in computer architecture.
 3284: 
 3285: Others have avoided this problem by coding in C, e.g., Mitch Bradley
 3286: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
 3287: particularly popular for UNIX-based Forths due to the large variety of
 3288: architectures of UNIX machines. Unfortunately an implementation in C
 3289: does not mix well with the goals of efficiency and with using
 3290: traditional techniques: Indirect or direct threading cannot be expressed
 3291: in C, and switch threading, the fastest technique available in C, is
 3292: significantly slower. Another problem with C is that it's very
 3293: cumbersome to express double integer arithmetic.
 3294: 
 3295: Fortunately, there is a portable language that does not have these
 3296: limitations: GNU C, the version of C processed by the GNU C compiler
 3297: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
 3298: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
 3299: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
 3300: threading possible, its @code{long long} type (@pxref{Long Long, ,
 3301: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
 3302: double numbers. GNU C is available for free on all important (and many
 3303: unimportant) UNIX machines, VMS, 80386s running MS-DOS, the Amiga, and
 3304: the Atari ST, so a Forth written in GNU C can run on all these
 3305: machines.
 3306: 
 3307: Writing in a portable language has the reputation of producing code that
 3308: is slower than assembly. For our Forth engine we repeatedly looked at
 3309: the code produced by the compiler and eliminated most compiler-induced
 3310: inefficiencies by appropriate changes in the source-code.
 3311: 
 3312: However, register allocation cannot be portably influenced by the
 3313: programmer, leading to some inefficiencies on register-starved
 3314: machines. We use explicit register declarations (@pxref{Explicit Reg
 3315: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
 3316: improve the speed on some machines. They are turned on by using the
 3317: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
 3318: only depends on the machine, but also on the compiler version: On some
 3319: machines some compiler versions produce incorrect code when certain
 3320: explicit register declarations are used. So by default
 3321: @code{-DFORCE_REG} is not used.
 3322: 
 3323: @node Threading, Primitives, Portability, Internals
 3324: @section Threading
 3325: 
 3326: GNU C's labels as values extension (available since @code{gcc-2.0},
 3327: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
 3328: makes it possible to take the address of @var{label} by writing
 3329: @code{&&@var{label}}.  This address can then be used in a statement like
 3330: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
 3331: @code{goto x}.
 3332: 
 3333: With this feature an indirect threaded NEXT looks like:
 3334: @example
 3335: cfa = *ip++;
 3336: ca = *cfa;
 3337: goto *ca;
 3338: @end example
 3339: For those unfamiliar with the names: @code{ip} is the Forth instruction
 3340: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
 3341: execution token and points to the code field of the next word to be
 3342: executed; The @code{ca} (code address) fetched from there points to some
 3343: executable code, e.g., a primitive or the colon definition handler
 3344: @code{docol}.
 3345: 
 3346: Direct threading is even simpler:
 3347: @example
 3348: ca = *ip++;
 3349: goto *ca;
 3350: @end example
 3351: 
 3352: Of course we have packaged the whole thing neatly in macros called
 3353: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
 3354: 
 3355: @menu
 3356: * Scheduling::                  
 3357: * Direct or Indirect Threaded?::  
 3358: * DOES>::                       
 3359: @end menu
 3360: 
 3361: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
 3362: @subsection Scheduling
 3363: 
 3364: There is a little complication: Pipelined and superscalar processors,
 3365: i.e., RISC and some modern CISC machines can process independent
 3366: instructions while waiting for the results of an instruction. The
 3367: compiler usually reorders (schedules) the instructions in a way that
 3368: achieves good usage of these delay slots. However, on our first tries
 3369: the compiler did not do well on scheduling primitives. E.g., for
 3370: @code{+} implemented as
 3371: @example
 3372: n=sp[0]+sp[1];
 3373: sp++;
 3374: sp[0]=n;
 3375: NEXT;
 3376: @end example
 3377: the NEXT comes strictly after the other code, i.e., there is nearly no
 3378: scheduling. After a little thought the problem becomes clear: The
 3379: compiler cannot know that sp and ip point to different addresses (and
 3380: the version of @code{gcc} we used would not know it even if it was
 3381: possible), so it could not move the load of the cfa above the store to
 3382: the TOS. Indeed the pointers could be the same, if code on or very near
 3383: the top of stack were executed. In the interest of speed we chose to
 3384: forbid this probably unused ``feature'' and helped the compiler in
 3385: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
 3386: the goto part (@code{NEXT_P2}). @code{+} now looks like:
 3387: @example
 3388: n=sp[0]+sp[1];
 3389: sp++;
 3390: NEXT_P1;
 3391: sp[0]=n;
 3392: NEXT_P2;
 3393: @end example
 3394: This can be scheduled optimally by the compiler.
 3395: 
 3396: This division can be turned off with the switch @code{-DCISC_NEXT}. This
 3397: switch is on by default on machines that do not profit from scheduling
 3398: (e.g., the 80386), in order to preserve registers.
 3399: 
 3400: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
 3401: @subsection Direct or Indirect Threaded?
 3402: 
 3403: Both! After packaging the nasty details in macro definitions we
 3404: realized that we could switch between direct and indirect threading by
 3405: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
 3406: defining a few machine-specific macros for the direct-threading case.
 3407: On the Forth level we also offer access words that hide the
 3408: differences between the threading methods (@pxref{Threading Words}).
 3409: 
 3410: Indirect threading is implemented completely
 3411: machine-independently. Direct threading needs routines for creating
 3412: jumps to the executable code (e.g. to docol or dodoes). These routines
 3413: are inherently machine-dependent, but they do not amount to many source
 3414: lines. I.e., even porting direct threading to a new machine is a small
 3415: effort.
 3416: 
 3417: @node DOES>,  , Direct or Indirect Threaded?, Threading
 3418: @subsection DOES>
 3419: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
 3420: the chunk of code executed by every word defined by a
 3421: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
 3422: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
 3423: DOES-code)? There are two solutions:
 3424: 
 3425: In fig-Forth the code field points directly to the dodoes and the
 3426: DOES-code address is stored in the cell after the code address
 3427: (i.e. at cfa cell+). It may seem that this solution is illegal in the
 3428: Forth-79 and all later standards, because in fig-Forth this address
 3429: lies in the body (which is illegal in these standards). However, by
 3430: making the code field larger for all words this solution becomes legal
 3431: again. We use this approach for the indirect threaded version. Leaving
 3432: a cell unused in most words is a bit wasteful, but on the machines we
 3433: are targetting this is hardly a problem. The other reason for having a
 3434: code field size of two cells is to avoid having different image files
 3435: for direct and indirect threaded systems (@pxref{System Architecture}).
 3436: 
 3437: The other approach is that the code field points or jumps to the cell
 3438: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
 3439: this address. @code{dodoes} can then get the DOES-code address by
 3440: computing the code address, i.e., the address of the jump to dodoes,
 3441: and add the length of that jump field. A variant of this is to have a
 3442: call to @code{dodoes} after the @code{DOES>}; then the return address
 3443: (which can be found in the return register on RISCs) is the DOES-code
 3444: address. Since the two cells available in the code field are usually
 3445: used up by the jump to the code address in direct threading, we use
 3446: this approach for direct threading. We did not want to add another
 3447: cell to the code field.
 3448: 
 3449: @node Primitives, System Architecture, Threading, Internals
 3450: @section Primitives
 3451: 
 3452: @menu
 3453: * Automatic Generation::        
 3454: * TOS Optimization::            
 3455: * Produced code::               
 3456: @end menu
 3457: 
 3458: @node Automatic Generation, TOS Optimization, Primitives, Primitives
 3459: @subsection Automatic Generation
 3460: 
 3461: Since the primitives are implemented in a portable language, there is no
 3462: longer any need to minimize the number of primitives. On the contrary,
 3463: having many primitives is an advantage: speed. In order to reduce the
 3464: number of errors in primitives and to make programming them easier, we
 3465: provide a tool, the primitive generator (@file{prims2x.fs}), that
 3466: automatically generates most (and sometimes all) of the C code for a
 3467: primitive from the stack effect notation.  The source for a primitive
 3468: has the following form:
 3469: 
 3470: @format
 3471: @var{Forth-name}	@var{stack-effect}	@var{category}	[@var{pronounc.}]
 3472: [@code{""}@var{glossary entry}@code{""}]
 3473: @var{C code}
 3474: [@code{:}
 3475: @var{Forth code}]
 3476: @end format
 3477: 
 3478: The items in brackets are optional. The category and glossary fields
 3479: are there for generating the documentation, the Forth code is there
 3480: for manual implementations on machines without GNU C. E.g., the source
 3481: for the primitive @code{+} is:
 3482: @example
 3483: +    n1 n2 -- n    core    plus
 3484: n = n1+n2;
 3485: @end example
 3486: 
 3487: This looks like a specification, but in fact @code{n = n1+n2} is C
 3488: code. Our primitive generation tool extracts a lot of information from
 3489: the stack effect notations@footnote{We use a one-stack notation, even
 3490: though we have separate data and floating-point stacks; The separate
 3491: notation can be generated easily from the unified notation.}: The number
 3492: of items popped from and pushed on the stack, their type, and by what
 3493: name they are referred to in the C code. It then generates a C code
 3494: prelude and postlude for each primitive. The final C code for @code{+}
 3495: looks like this:
 3496: 
 3497: @example
 3498: I_plus:	/* + ( n1 n2 -- n ) */  /* label, stack effect */
 3499: /*  */                          /* documentation */
 3500: @{
 3501: DEF_CA                          /* definition of variable ca (indirect threading) */
 3502: Cell n1;                        /* definitions of variables */
 3503: Cell n2;
 3504: Cell n;
 3505: n1 = (Cell) sp[1];              /* input */
 3506: n2 = (Cell) TOS;
 3507: sp += 1;                        /* stack adjustment */
 3508: NAME("+")                       /* debugging output (with -DDEBUG) */
 3509: @{
 3510: n = n1+n2;                      /* C code taken from the source */
 3511: @}
 3512: NEXT_P1;                        /* NEXT part 1 */
 3513: TOS = (Cell)n;                  /* output */
 3514: NEXT_P2;                        /* NEXT part 2 */
 3515: @}
 3516: @end example
 3517: 
 3518: This looks long and inefficient, but the GNU C compiler optimizes quite
 3519: well and produces optimal code for @code{+} on, e.g., the R3000 and the
 3520: HP RISC machines: Defining the @code{n}s does not produce any code, and
 3521: using them as intermediate storage also adds no cost.
 3522: 
 3523: There are also other optimizations, that are not illustrated by this
 3524: example: Assignments between simple variables are usually for free (copy
 3525: propagation). If one of the stack items is not used by the primitive
 3526: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
 3527: (dead code elimination). On the other hand, there are some things that
 3528: the compiler does not do, therefore they are performed by
 3529: @file{prims2x.fs}: The compiler does not optimize code away that stores
 3530: a stack item to the place where it just came from (e.g., @code{over}).
 3531: 
 3532: While programming a primitive is usually easy, there are a few cases
 3533: where the programmer has to take the actions of the generator into
 3534: account, most notably @code{?dup}, but also words that do not (always)
 3535: fall through to NEXT.
 3536: 
 3537: @node TOS Optimization, Produced code, Automatic Generation, Primitives
 3538: @subsection TOS Optimization
 3539: 
 3540: An important optimization for stack machine emulators, e.g., Forth
 3541: engines, is keeping  one or more of the top stack items in
 3542: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
 3543: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
 3544: @itemize
 3545: @item
 3546: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
 3547: due to fewer loads from and stores to the stack.
 3548: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
 3549: @var{y<n}, due to additional moves between registers.
 3550: @end itemize
 3551: 
 3552: In particular, keeping one item in a register is never a disadvantage,
 3553: if there are enough registers. Keeping two items in registers is a
 3554: disadvantage for frequent words like @code{?branch}, constants,
 3555: variables, literals and @code{i}. Therefore our generator only produces
 3556: code that keeps zero or one items in registers. The generated C code
 3557: covers both cases; the selection between these alternatives is made at
 3558: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
 3559: code for @code{+} is just a simple variable name in the one-item case,
 3560: otherwise it is a macro that expands into @code{sp[0]}. Note that the
 3561: GNU C compiler tries to keep simple variables like @code{TOS} in
 3562: registers, and it usually succeeds, if there are enough registers.
 3563: 
 3564: The primitive generator performs the TOS optimization for the
 3565: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
 3566: operations the benefit of this optimization is even larger:
 3567: floating-point operations take quite long on most processors, but can be
 3568: performed in parallel with other operations as long as their results are
 3569: not used. If the FP-TOS is kept in a register, this works. If
 3570: it is kept on the stack, i.e., in memory, the store into memory has to
 3571: wait for the result of the floating-point operation, lengthening the
 3572: execution time of the primitive considerably.
 3573: 
 3574: The TOS optimization makes the automatic generation of primitives a
 3575: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
 3576: @code{TOS} is not sufficient. There are some special cases to
 3577: consider:
 3578: @itemize
 3579: @item In the case of @code{dup ( w -- w w )} the generator must not
 3580: eliminate the store to the original location of the item on the stack,
 3581: if the TOS optimization is turned on.
 3582: @item Primitives with stack effects of the form @code{--}
 3583: @var{out1}...@var{outy} must store the TOS to the stack at the start.
 3584: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
 3585: must load the TOS from the stack at the end. But for the null stack
 3586: effect @code{--} no stores or loads should be generated.
 3587: @end itemize
 3588: 
 3589: @node Produced code,  , TOS Optimization, Primitives
 3590: @subsection Produced code
 3591: 
 3592: To see what assembly code is produced for the primitives on your machine
 3593: with your compiler and your flag settings, type @code{make engine.s} and
 3594: look at the resulting file @file{engine.s}.
 3595: 
 3596: @node System Architecture, Performance, Primitives, Internals
 3597: @section System Architecture
 3598: 
 3599: Our Forth system consists not only of primitives, but also of
 3600: definitions written in Forth. Since the Forth compiler itself belongs
 3601: to those definitions, it is not possible to start the system with the
 3602: primitives and the Forth source alone. Therefore we provide the Forth
 3603: code as an image file in nearly executable form. At the start of the
 3604: system a C routine loads the image file into memory, sets up the
 3605: memory (stacks etc.) according to information in the image file, and
 3606: starts executing Forth code.
 3607: 
 3608: The image file format is a compromise between the goals of making it
 3609: easy to generate image files and making them portable. The easiest way
 3610: to generate an image file is to just generate a memory dump. However,
 3611: this kind of image file cannot be used on a different machine, or on
 3612: the next version of the engine on the same machine, it even might not
 3613: work with the same engine compiled by a different version of the C
 3614: compiler. We would like to have as few versions of the image file as
 3615: possible, because we do not want to distribute many versions of the
 3616: same image file, and to make it easy for the users to use their image
 3617: files on many machines. We currently need to create a different image
 3618: file for machines with different cell sizes and different byte order
 3619: (little- or big-endian)@footnote{We are considering adding information to the
 3620: image file that enables the loader to change the byte order.}.
 3621: 
 3622: Forth code that is going to end up in a portable image file has to
 3623: comply to some restrictions: addresses have to be stored in memory with
 3624: special words (@code{A!}, @code{A,}, etc.) in order to make the code
 3625: relocatable. Cells, floats, etc., have to be stored at the natural
 3626: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
 3627: dividable by~8. This happens automatically in our system when you use
 3628: the ANS Forth alignment words.}, in order to avoid alignment faults on
 3629: machines with stricter alignment. The image file is produced by a
 3630: metacompiler (@file{cross.fs}).
 3631: 
 3632: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
 3633: file is not directly executable, but has to undergo some manipulations
 3634: during loading. Address relocation is performed at image load-time, not
 3635: at run-time. The loader also has to replace tokens standing for
 3636: primitive calls with the appropriate code-field addresses (or code
 3637: addresses in the case of direct threading).
 3638: 
 3639: @node  Performance,  , System Architecture, Internals
 3640: @section Performance
 3641: 
 3642: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
 3643: impossible to write a significantly faster engine.
 3644: 
 3645: On register-starved machines like the 386 architecture processors
 3646: improvements are possible, because @code{gcc} does not utilize the
 3647: registers as well as a human, even with explicit register declarations;
 3648: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
 3649: and hand-tuned it for the 486; this system is 1.19 times faster on the
 3650: Sieve benchmark on a 486DX2/66 than Gforth compiled with
 3651: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
 3652: 
 3653: However, this potential advantage of assembly language implementations
 3654: is not necessarily realized in complete Forth systems: We compared
 3655: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
 3656: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
 3657: 1994) and Eforth (with and without peephole (aka pinhole) optimization
 3658: of the threaded code); all these systems were written in assembly
 3659: language. We also compared Gforth with three systems written in C:
 3660: PFE-0.9.11 (compiled with @code{gcc-2.6.3} with the default
 3661: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS}),
 3662: ThisForth Beta (compiled with gcc-2.6.3 -O3 -fomit-frame-pointer;
 3663: ThisForth employs peephole optimization of the threaded code) and TILE
 3664: (compiled with @code{make opt}). We benchmarked Gforth, PFE, ThisForth
 3665: and TILE on a 486DX2/66 under Linux. Kenneth O'Heskin kindly provided
 3666: the results for Win32Forth and NT Forth on a 486DX2/66 with similar
 3667: memory performance under Windows NT. Marcel Hendrix ported Eforth to
 3668: Linux, then extended it to run the benchmarks, added the peephole
 3669: optimizer, ran the benchmarks and reported the results.
 3670:  
 3671: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
 3672: matrix multiplication come from the Stanford integer benchmarks and have
 3673: been translated into Forth by Martin Fraeman; we used the versions
 3674: included in the TILE Forth package, but with bigger data set sizes; and
 3675: a recursive Fibonacci number computation for benchmarking calling
 3676: performance. The following table shows the time taken for the benchmarks
 3677: scaled by the time taken by Gforth (in other words, it shows the speedup
 3678: factor that Gforth achieved over the other systems).
 3679: 
 3680: @example
 3681: relative      Win32-    NT       eforth       This-
 3682:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
 3683: sieve     1.00  1.39  1.14   1.39  0.85  1.78  3.18  8.58
 3684: bubble    1.00  1.31  1.41   1.48  0.88  1.67        3.88
 3685: matmul    1.00  1.47  1.35   1.46  1.16  2.36        4.09
 3686: fib       1.00  1.52  1.34   1.22  1.13  1.93  2.99  4.30
 3687: @end example
 3688: 
 3689: You may find the good performance of Gforth compared with the systems
 3690: written in assembly language quite surprising. One important reason for
 3691: the disappointing performance of these systems is probably that they are
 3692: not written optimally for the 486 (e.g., they use the @code{lods}
 3693: instruction). In addition, Win32Forth uses a comfortable, but costly
 3694: method for relocating the Forth image: like @code{cforth}, it computes
 3695: the actual addresses at run time, resulting in two address computations
 3696: per NEXT (@pxref{System Architecture}).
 3697: 
 3698: Only Eforth with the peephole optimizer performs comparable to
 3699: Gforth. The speedups achieved with peephole optimization of threaded
 3700: code are quite remarkable. Adding a peephole optimizer to Gforth should
 3701: cause similar speedups.
 3702: 
 3703: The speedup of Gforth over PFE, ThisForth and TILE can be easily
 3704: explained with the self-imposed restriction to standard C, which makes
 3705: efficient threading impossible (however, the measured implementation of
 3706: PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
 3707: Register Variables, gcc.info, GNU C Manual}).  Moreover, current C
 3708: compilers have a hard time optimizing other aspects of the ThisForth
 3709: and the TILE source.
 3710: 
 3711: Note that the performance of Gforth on 386 architecture processors
 3712: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
 3713: failed to allocate any of the virtual machine registers into real
 3714: machine registers by itself and would not work correctly with explicit
 3715: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
 3716: running the Sieve) than the one measured above.
 3717: 
 3718: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
 3719: Maierhofer (presented at EuroForth '95), an indirect threaded version of
 3720: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
 3721: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
 3722: used here. The paper available at
 3723: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
 3724: it also contains numbers for some native code systems. You can find
 3725: numbers for Gforth on various machines in @file{Benchres}.
 3726: 
 3727: @node Bugs, Origin, Internals, Top
 3728: @chapter Bugs
 3729: 
 3730: Known bugs are described in the file BUGS in the Gforth distribution.
 3731: 
 3732: If you find a bug, please send a bug report to
 3733: @code{gforth-bugs@@mips.complang.tuwien.ac.at}. A bug report should
 3734: describe the Gforth version used (it is announced at the start of an
 3735: interactive Gforth session), the machine and operating system (on Unix
 3736: systems you can use @code{uname -a} to produce this information), the
 3737: installation options (send the @code{config.status} file), and a
 3738: complete list of changes you (or your installer) have made to the Gforth
 3739: sources (if any); it should contain a program (or a sequence of keyboard
 3740: commands) that reproduces the bug and a description of what you think
 3741: constitutes the buggy behaviour.
 3742: 
 3743: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
 3744: to Report Bugs, gcc.info, GNU C Manual}.
 3745: 
 3746: 
 3747: @node Origin, Word Index, Bugs, Top
 3748: @chapter Authors and Ancestors of Gforth
 3749: 
 3750: @section Authors and Contributors
 3751: 
 3752: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
 3753: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
 3754: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
 3755: with their continuous feedback. Lennart Benshop contributed
 3756: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
 3757: support for calling C libraries. Helpful comments also came from Paul
 3758: Kleinrubatscher, Christian Pirker, Dirk Zoller and Marcel Hendrix.
 3759: 
 3760: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
 3761: and autoconf, among others), and to the creators of the Internet: Gforth
 3762: was developed across the Internet, and its authors have not met
 3763: physically yet.
 3764: 
 3765: @section Pedigree
 3766: 
 3767: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
 3768: Dirk Zoller) will cross-fertilize each other. Of course, a significant
 3769: part of the design of Gforth was prescribed by ANS Forth.
 3770: 
 3771: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
 3772: 32 bit native code version of VolksForth for the Atari ST, written
 3773: mostly by Dietrich Weineck.
 3774: 
 3775: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
 3776: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
 3777: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
 3778: 
 3779: Hennry Laxen and Mike Perry wrote F83 as a model implementation of the
 3780: Forth-83 standard. !! Pedigree? When?
 3781: 
 3782: A team led by Bill Ragsdale implemented fig-Forth on many processors in
 3783: 1979. Robert Selzer and Bill Ragsdale developed the original
 3784: implementation of fig-Forth for the 6502 based on microForth.
 3785: 
 3786: The principal architect of microForth was Dean Sanderson. microForth was
 3787: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
 3788: the 1802, and subsequently implemented on the 8080, the 6800 and the
 3789: Z80.
 3790: 
 3791: All earlier Forth systems were custom-made, usually by Charles Moore,
 3792: who discovered (as he puts it) Forth during the late 60s. The first full
 3793: Forth existed in 1971.
 3794: 
 3795: A part of the information in this section comes from @cite{The Evolution
 3796: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
 3797: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
 3798: Notices 28(3), 1993.  You can find more historical and genealogical
 3799: information about Forth there.
 3800: 
 3801: @node Word Index, Node Index, Origin, Top
 3802: @chapter Word Index
 3803: 
 3804: This index is as incomplete as the manual. Each word is listed with
 3805: stack effect and wordset.
 3806: 
 3807: @printindex fn
 3808: 
 3809: @node Node Index,  , Word Index, Top
 3810: @chapter Node Index
 3811: 
 3812: This index is even less complete than the manual.
 3813: 
 3814: @contents
 3815: @bye
 3816: 

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