Annotation of gforth/gforth.ds, revision 1.40

1.1       anton       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.)
1.4       anton       4: @setfilename
1.17      anton       5: @settitle Gforth Manual
1.4       anton       6: @comment @setchapternewpage odd
1.1       anton       7: @comment %**end of header (This is for running Texinfo on a region.)
                      9: @ifinfo
1.30      anton      10: This file documents Gforth 0.2
1.1       anton      11: 
1.32      anton      12: Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
1.1       anton      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.
1.4       anton      18: @ignore
1.1       anton      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).
1.4       anton      24: @end ignore
1.1       anton      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.
                     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
1.24      anton      39: @finalout
1.1       anton      40: @titlepage
                     41: @sp 10
1.17      anton      42: @center @titlefont{Gforth Manual}
1.1       anton      43: @sp 2
1.30      anton      44: @center for version 0.2
1.1       anton      45: @sp 2
                     46: @center Anton Ertl
1.25      anton      47: @center Bernd Paysan
1.17      anton      48: @sp 3
                     49: @center This manual is under construction
1.1       anton      50: 
                     51: @comment  The following two commands start the copyright page.
                     52: @page
                     53: @vskip 0pt plus 1filll
1.32      anton      54: Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
1.1       anton      55: 
                     56: @comment !! Published by ... or You can get a copy of this manual ...
                     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.
                     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.
                     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
                     77: @node Top, License, (dir), (dir)
                     78: @ifinfo
1.17      anton      79: Gforth is a free implementation of ANS Forth available on many
1.30      anton      80: personal machines. This manual corresponds to version 0.2.
1.1       anton      81: @end ifinfo
                     83: @menu
1.4       anton      84: * License::                     
1.17      anton      85: * Goals::                       About the Gforth Project
1.4       anton      86: * Other Books::                 Things you might want to read
1.17      anton      87: * Invocation::                  Starting Gforth
                     88: * Words::                       Forth words available in Gforth
1.40    ! anton      89: * Tools::                       Programming tools
1.4       anton      90: * ANS conformance::             Implementation-defined options etc.
1.17      anton      91: * Model::                       The abstract machine of Gforth
1.34      anton      92: * Integrating Gforth::          Forth as scripting language for applications.
1.17      anton      93: * Emacs and Gforth::            The Gforth Mode
1.4       anton      94: * Internals::                   Implementation details
                     95: * Bugs::                        How to report them
1.29      anton      96: * Origin::                      Authors and ancestors of Gforth
1.4       anton      97: * Word Index::                  An item for each Forth word
                     98: * Node Index::                  An item for each node
1.1       anton      99: @end menu
                    101: @node License, Goals, Top, Top
1.20      pazsan    102: @unnumbered GNU GENERAL PUBLIC LICENSE
                    103: @center Version 2, June 1991
                    105: @display
                    106: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
                    107: 675 Mass Ave, Cambridge, MA 02139, USA
                    109: Everyone is permitted to copy and distribute verbatim copies
                    110: of this license document, but changing it is not allowed.
                    111: @end display
                    113: @unnumberedsec Preamble
                    115:   The licenses for most software are designed to take away your
                    116: freedom to share and change it.  By contrast, the GNU General Public
                    117: License is intended to guarantee your freedom to share and change free
                    118: software---to make sure the software is free for all its users.  This
                    119: General Public License applies to most of the Free Software
                    120: Foundation's software and to any other program whose authors commit to
                    121: using it.  (Some other Free Software Foundation software is covered by
                    122: the GNU Library General Public License instead.)  You can apply it to
                    123: your programs, too.
                    125:   When we speak of free software, we are referring to freedom, not
                    126: price.  Our General Public Licenses are designed to make sure that you
                    127: have the freedom to distribute copies of free software (and charge for
                    128: this service if you wish), that you receive source code or can get it
                    129: if you want it, that you can change the software or use pieces of it
                    130: in new free programs; and that you know you can do these things.
                    132:   To protect your rights, we need to make restrictions that forbid
                    133: anyone to deny you these rights or to ask you to surrender the rights.
                    134: These restrictions translate to certain responsibilities for you if you
                    135: distribute copies of the software, or if you modify it.
                    137:   For example, if you distribute copies of such a program, whether
                    138: gratis or for a fee, you must give the recipients all the rights that
                    139: you have.  You must make sure that they, too, receive or can get the
                    140: source code.  And you must show them these terms so they know their
                    141: rights.
                    143:   We protect your rights with two steps: (1) copyright the software, and
                    144: (2) offer you this license which gives you legal permission to copy,
                    145: distribute and/or modify the software.
                    147:   Also, for each author's protection and ours, we want to make certain
                    148: that everyone understands that there is no warranty for this free
                    149: software.  If the software is modified by someone else and passed on, we
                    150: want its recipients to know that what they have is not the original, so
                    151: that any problems introduced by others will not reflect on the original
                    152: authors' reputations.
                    154:   Finally, any free program is threatened constantly by software
                    155: patents.  We wish to avoid the danger that redistributors of a free
                    156: program will individually obtain patent licenses, in effect making the
                    157: program proprietary.  To prevent this, we have made it clear that any
                    158: patent must be licensed for everyone's free use or not licensed at all.
                    160:   The precise terms and conditions for copying, distribution and
                    161: modification follow.
                    163: @iftex
                    165: @end iftex
                    166: @ifinfo
                    168: @end ifinfo
                    170: @enumerate 0
                    171: @item
                    172: This License applies to any program or other work which contains
                    173: a notice placed by the copyright holder saying it may be distributed
                    174: under the terms of this General Public License.  The ``Program'', below,
                    175: refers to any such program or work, and a ``work based on the Program''
                    176: means either the Program or any derivative work under copyright law:
                    177: that is to say, a work containing the Program or a portion of it,
                    178: either verbatim or with modifications and/or translated into another
                    179: language.  (Hereinafter, translation is included without limitation in
                    180: the term ``modification''.)  Each licensee is addressed as ``you''.
                    182: Activities other than copying, distribution and modification are not
                    183: covered by this License; they are outside its scope.  The act of
                    184: running the Program is not restricted, and the output from the Program
                    185: is covered only if its contents constitute a work based on the
                    186: Program (independent of having been made by running the Program).
                    187: Whether that is true depends on what the Program does.
                    189: @item
                    190: You may copy and distribute verbatim copies of the Program's
                    191: source code as you receive it, in any medium, provided that you
                    192: conspicuously and appropriately publish on each copy an appropriate
                    193: copyright notice and disclaimer of warranty; keep intact all the
                    194: notices that refer to this License and to the absence of any warranty;
                    195: and give any other recipients of the Program a copy of this License
                    196: along with the Program.
                    198: You may charge a fee for the physical act of transferring a copy, and
                    199: you may at your option offer warranty protection in exchange for a fee.
                    201: @item
                    202: You may modify your copy or copies of the Program or any portion
                    203: of it, thus forming a work based on the Program, and copy and
                    204: distribute such modifications or work under the terms of Section 1
                    205: above, provided that you also meet all of these conditions:
                    207: @enumerate a
                    208: @item
                    209: You must cause the modified files to carry prominent notices
                    210: stating that you changed the files and the date of any change.
                    212: @item
                    213: You must cause any work that you distribute or publish, that in
                    214: whole or in part contains or is derived from the Program or any
                    215: part thereof, to be licensed as a whole at no charge to all third
                    216: parties under the terms of this License.
                    218: @item
                    219: If the modified program normally reads commands interactively
                    220: when run, you must cause it, when started running for such
                    221: interactive use in the most ordinary way, to print or display an
                    222: announcement including an appropriate copyright notice and a
                    223: notice that there is no warranty (or else, saying that you provide
                    224: a warranty) and that users may redistribute the program under
                    225: these conditions, and telling the user how to view a copy of this
                    226: License.  (Exception: if the Program itself is interactive but
                    227: does not normally print such an announcement, your work based on
                    228: the Program is not required to print an announcement.)
                    229: @end enumerate
                    231: These requirements apply to the modified work as a whole.  If
                    232: identifiable sections of that work are not derived from the Program,
                    233: and can be reasonably considered independent and separate works in
                    234: themselves, then this License, and its terms, do not apply to those
                    235: sections when you distribute them as separate works.  But when you
                    236: distribute the same sections as part of a whole which is a work based
                    237: on the Program, the distribution of the whole must be on the terms of
                    238: this License, whose permissions for other licensees extend to the
                    239: entire whole, and thus to each and every part regardless of who wrote it.
                    241: Thus, it is not the intent of this section to claim rights or contest
                    242: your rights to work written entirely by you; rather, the intent is to
                    243: exercise the right to control the distribution of derivative or
                    244: collective works based on the Program.
                    246: In addition, mere aggregation of another work not based on the Program
                    247: with the Program (or with a work based on the Program) on a volume of
                    248: a storage or distribution medium does not bring the other work under
                    249: the scope of this License.
                    251: @item
                    252: You may copy and distribute the Program (or a work based on it,
                    253: under Section 2) in object code or executable form under the terms of
                    254: Sections 1 and 2 above provided that you also do one of the following:
                    256: @enumerate a
                    257: @item
                    258: Accompany it with the complete corresponding machine-readable
                    259: source code, which must be distributed under the terms of Sections
                    260: 1 and 2 above on a medium customarily used for software interchange; or,
                    262: @item
                    263: Accompany it with a written offer, valid for at least three
                    264: years, to give any third party, for a charge no more than your
                    265: cost of physically performing source distribution, a complete
                    266: machine-readable copy of the corresponding source code, to be
                    267: distributed under the terms of Sections 1 and 2 above on a medium
                    268: customarily used for software interchange; or,
                    270: @item
                    271: Accompany it with the information you received as to the offer
                    272: to distribute corresponding source code.  (This alternative is
                    273: allowed only for noncommercial distribution and only if you
                    274: received the program in object code or executable form with such
                    275: an offer, in accord with Subsection b above.)
                    276: @end enumerate
                    278: The source code for a work means the preferred form of the work for
                    279: making modifications to it.  For an executable work, complete source
                    280: code means all the source code for all modules it contains, plus any
                    281: associated interface definition files, plus the scripts used to
                    282: control compilation and installation of the executable.  However, as a
                    283: special exception, the source code distributed need not include
                    284: anything that is normally distributed (in either source or binary
                    285: form) with the major components (compiler, kernel, and so on) of the
                    286: operating system on which the executable runs, unless that component
                    287: itself accompanies the executable.
                    289: If distribution of executable or object code is made by offering
                    290: access to copy from a designated place, then offering equivalent
                    291: access to copy the source code from the same place counts as
                    292: distribution of the source code, even though third parties are not
                    293: compelled to copy the source along with the object code.
                    295: @item
                    296: You may not copy, modify, sublicense, or distribute the Program
                    297: except as expressly provided under this License.  Any attempt
                    298: otherwise to copy, modify, sublicense or distribute the Program is
                    299: void, and will automatically terminate your rights under this License.
                    300: However, parties who have received copies, or rights, from you under
                    301: this License will not have their licenses terminated so long as such
                    302: parties remain in full compliance.
                    304: @item
                    305: You are not required to accept this License, since you have not
                    306: signed it.  However, nothing else grants you permission to modify or
                    307: distribute the Program or its derivative works.  These actions are
                    308: prohibited by law if you do not accept this License.  Therefore, by
                    309: modifying or distributing the Program (or any work based on the
                    310: Program), you indicate your acceptance of this License to do so, and
                    311: all its terms and conditions for copying, distributing or modifying
                    312: the Program or works based on it.
                    314: @item
                    315: Each time you redistribute the Program (or any work based on the
                    316: Program), the recipient automatically receives a license from the
                    317: original licensor to copy, distribute or modify the Program subject to
                    318: these terms and conditions.  You may not impose any further
                    319: restrictions on the recipients' exercise of the rights granted herein.
                    320: You are not responsible for enforcing compliance by third parties to
                    321: this License.
                    323: @item
                    324: If, as a consequence of a court judgment or allegation of patent
                    325: infringement or for any other reason (not limited to patent issues),
                    326: conditions are imposed on you (whether by court order, agreement or
                    327: otherwise) that contradict the conditions of this License, they do not
                    328: excuse you from the conditions of this License.  If you cannot
                    329: distribute so as to satisfy simultaneously your obligations under this
                    330: License and any other pertinent obligations, then as a consequence you
                    331: may not distribute the Program at all.  For example, if a patent
                    332: license would not permit royalty-free redistribution of the Program by
                    333: all those who receive copies directly or indirectly through you, then
                    334: the only way you could satisfy both it and this License would be to
                    335: refrain entirely from distribution of the Program.
                    337: If any portion of this section is held invalid or unenforceable under
                    338: any particular circumstance, the balance of the section is intended to
                    339: apply and the section as a whole is intended to apply in other
                    340: circumstances.
                    342: It is not the purpose of this section to induce you to infringe any
                    343: patents or other property right claims or to contest validity of any
                    344: such claims; this section has the sole purpose of protecting the
                    345: integrity of the free software distribution system, which is
                    346: implemented by public license practices.  Many people have made
                    347: generous contributions to the wide range of software distributed
                    348: through that system in reliance on consistent application of that
                    349: system; it is up to the author/donor to decide if he or she is willing
                    350: to distribute software through any other system and a licensee cannot
                    351: impose that choice.
                    353: This section is intended to make thoroughly clear what is believed to
                    354: be a consequence of the rest of this License.
                    356: @item
                    357: If the distribution and/or use of the Program is restricted in
                    358: certain countries either by patents or by copyrighted interfaces, the
                    359: original copyright holder who places the Program under this License
                    360: may add an explicit geographical distribution limitation excluding
                    361: those countries, so that distribution is permitted only in or among
                    362: countries not thus excluded.  In such case, this License incorporates
                    363: the limitation as if written in the body of this License.
                    365: @item
                    366: The Free Software Foundation may publish revised and/or new versions
                    367: of the General Public License from time to time.  Such new versions will
                    368: be similar in spirit to the present version, but may differ in detail to
                    369: address new problems or concerns.
                    371: Each version is given a distinguishing version number.  If the Program
                    372: specifies a version number of this License which applies to it and ``any
                    373: later version'', you have the option of following the terms and conditions
                    374: either of that version or of any later version published by the Free
                    375: Software Foundation.  If the Program does not specify a version number of
                    376: this License, you may choose any version ever published by the Free Software
                    377: Foundation.
                    379: @item
                    380: If you wish to incorporate parts of the Program into other free
                    381: programs whose distribution conditions are different, write to the author
                    382: to ask for permission.  For software which is copyrighted by the Free
                    383: Software Foundation, write to the Free Software Foundation; we sometimes
                    384: make exceptions for this.  Our decision will be guided by the two goals
                    385: of preserving the free status of all derivatives of our free software and
                    386: of promoting the sharing and reuse of software generally.
                    388: @iftex
                    389: @heading NO WARRANTY
                    390: @end iftex
                    391: @ifinfo
                    392: @center NO WARRANTY
                    393: @end ifinfo
                    395: @item
                    404: REPAIR OR CORRECTION.
                    406: @item
                    415: POSSIBILITY OF SUCH DAMAGES.
                    416: @end enumerate
                    418: @iftex
                    419: @heading END OF TERMS AND CONDITIONS
                    420: @end iftex
                    421: @ifinfo
                    422: @center END OF TERMS AND CONDITIONS
                    423: @end ifinfo
                    425: @page
                    426: @unnumberedsec How to Apply These Terms to Your New Programs
                    428:   If you develop a new program, and you want it to be of the greatest
                    429: possible use to the public, the best way to achieve this is to make it
                    430: free software which everyone can redistribute and change under these terms.
                    432:   To do so, attach the following notices to the program.  It is safest
                    433: to attach them to the start of each source file to most effectively
                    434: convey the exclusion of warranty; and each file should have at least
                    435: the ``copyright'' line and a pointer to where the full notice is found.
                    437: @smallexample
                    438: @var{one line to give the program's name and a brief idea of what it does.}
                    439: Copyright (C) 19@var{yy}  @var{name of author}
                    441: This program is free software; you can redistribute it and/or modify 
                    442: it under the terms of the GNU General Public License as published by 
                    443: the Free Software Foundation; either version 2 of the License, or 
                    444: (at your option) any later version.
                    446: This program is distributed in the hope that it will be useful,
                    447: but WITHOUT ANY WARRANTY; without even the implied warranty of
                    449: GNU General Public License for more details.
                    451: You should have received a copy of the GNU General Public License
                    452: along with this program; if not, write to the Free Software
                    453: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
                    454: @end smallexample
                    456: Also add information on how to contact you by electronic and paper mail.
                    458: If the program is interactive, make it output a short notice like this
                    459: when it starts in an interactive mode:
                    461: @smallexample
                    462: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
                    463: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
                    464: type `show w'.  
                    465: This is free software, and you are welcome to redistribute it 
                    466: under certain conditions; type `show c' for details.
                    467: @end smallexample
                    469: The hypothetical commands @samp{show w} and @samp{show c} should show
                    470: the appropriate parts of the General Public License.  Of course, the
                    471: commands you use may be called something other than @samp{show w} and
                    472: @samp{show c}; they could even be mouse-clicks or menu items---whatever
                    473: suits your program.
                    475: You should also get your employer (if you work as a programmer) or your
                    476: school, if any, to sign a ``copyright disclaimer'' for the program, if
                    477: necessary.  Here is a sample; alter the names:
                    479: @smallexample
                    480: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
                    481: `Gnomovision' (which makes passes at compilers) written by James Hacker.
                    483: @var{signature of Ty Coon}, 1 April 1989
                    484: Ty Coon, President of Vice
                    485: @end smallexample
                    487: This General Public License does not permit incorporating your program into
                    488: proprietary programs.  If your program is a subroutine library, you may
                    489: consider it more useful to permit linking proprietary applications with the
                    490: library.  If this is what you want to do, use the GNU Library General
                    491: Public License instead of this License.
1.1       anton     492: 
                    493: @iftex
1.23      pazsan    494: @node    Preface
                    495: @comment node-name,     next,           previous, up
1.1       anton     496: @unnumbered Preface
1.23      pazsan    497: @cindex Preface
1.17      anton     498: This manual documents Gforth. The reader is expected to know
1.1       anton     499: Forth. This manual is primarily a reference manual. @xref{Other Books}
                    500: for introductory material.
                    501: @end iftex
                    503: @node    Goals, Other Books, License, Top
                    504: @comment node-name,     next,           previous, up
1.17      anton     505: @chapter Goals of Gforth
1.1       anton     506: @cindex Goals
1.17      anton     507: The goal of the Gforth Project is to develop a standard model for
1.1       anton     508: ANSI Forth. This can be split into several subgoals:
                    510: @itemize @bullet
                    511: @item
1.17      anton     512: Gforth should conform to the ANSI Forth standard.
1.1       anton     513: @item
                    514: It should be a model, i.e. it should define all the
                    515: implementation-dependent things.
                    516: @item
                    517: It should become standard, i.e. widely accepted and used. This goal
                    518: is the most difficult one.
                    519: @end itemize
1.17      anton     521: To achieve these goals Gforth should be
1.1       anton     522: @itemize @bullet
                    523: @item
                    524: Similar to previous models (fig-Forth, F83)
                    525: @item
                    526: Powerful. It should provide for all the things that are considered
                    527: necessary today and even some that are not yet considered necessary.
                    528: @item
                    529: Efficient. It should not get the reputation of being exceptionally
                    530: slow.
                    531: @item
                    532: Free.
                    533: @item
                    534: Available on many machines/easy to port.
                    535: @end itemize
1.17      anton     537: Have we achieved these goals? Gforth conforms to the ANS Forth
                    538: standard. It may be considered a model, but we have not yet documented
1.1       anton     539: which parts of the model are stable and which parts we are likely to
1.17      anton     540: change. It certainly has not yet become a de facto standard. It has some
                    541: similarities and some differences to previous models. It has some
                    542: powerful features, but not yet everything that we envisioned. We
                    543: certainly have achieved our execution speed goals (@pxref{Performance}).
                    544: It is free and available on many machines.
1.1       anton     545: 
                    546: @node Other Books, Invocation, Goals, Top
                    547: @chapter Other books on ANS Forth
                    549: As the standard is relatively new, there are not many books out yet. It
1.17      anton     550: is not recommended to learn Forth by using Gforth and a book that is
1.1       anton     551: not written for ANS Forth, as you will not know your mistakes from the
                    552: deviations of the book.
                    554: There is, of course, the standard, the definite reference if you want to
1.19      anton     555: write ANS Forth programs. It is available in printed form from the
                    556: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
                    557: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
                    558: can also get it from Global Engineering Documents (Tel.: USA (800)
                    559: 854-7179; Fax.: (303) 843-9880) for about $300.
                    561: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
                    562: for publication is available electronically and for free in some MS Word
                    563: format, and it has been converted to HTML. Some pointers to these
                    564: versions can be found through
1.24      anton     565: @*@file{}.
1.1       anton     566: 
1.21      anton     567: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
1.1       anton     568: introductory book based on a draft version of the standard. It does not
                    569: cover the whole standard. It also contains interesting background
                    570: information (Jack Woehr was in the ANS Forth Technical Committe). It is
                    571: not appropriate for complete newbies, but programmers experienced in
                    572: other languages should find it ok.
                    574: @node Invocation, Words, Other Books, Top
                    575: @chapter Invocation
                    577: You will usually just say @code{gforth}. In many other cases the default
1.17      anton     578: Gforth image will be invoked like this:
1.1       anton     579: 
                    580: @example
                    581: gforth [files] [-e forth-code]
                    582: @end example
                    584: executing the contents of the files and the Forth code in the order they
                    585: are given.
                    587: In general, the command line looks like this:
                    589: @example
                    590: gforth [initialization options] [image-specific options]
                    591: @end example
                    593: The initialization options must come before the rest of the command
                    594: line. They are:
                    596: @table @code
                    597: @item --image-file @var{file}
1.20      pazsan    598: @item -i @var{file}
1.1       anton     599: Loads the Forth image @var{file} instead of the default
                    600: @file{}.
                    602: @item --path @var{path}
1.20      pazsan    603: @item -p @var{path}
1.39      anton     604: Uses @var{path} for searching the image file and Forth source code files
                    605: instead of the default in the environment variable @code{GFORTHPATH} or
                    606: the path specified at installation time (e.g.,
                    607: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
                    608: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1.1       anton     609: 
                    610: @item --dictionary-size @var{size}
                    611: @item -m @var{size}
                    612: Allocate @var{size} space for the Forth dictionary space instead of
                    613: using the default specified in the image (typically 256K). The
                    614: @var{size} specification consists of an integer and a unit (e.g.,
                    615: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
                    616: size, in this case Cells), @code{k} (kilobytes), and @code{M}
                    617: (Megabytes). If no unit is specified, @code{e} is used.
                    619: @item --data-stack-size @var{size}
                    620: @item -d @var{size}
                    621: Allocate @var{size} space for the data stack instead of using the
                    622: default specified in the image (typically 16K).
                    624: @item --return-stack-size @var{size}
                    625: @item -r @var{size}
                    626: Allocate @var{size} space for the return stack instead of using the
                    627: default specified in the image (typically 16K).
                    629: @item --fp-stack-size @var{size}
                    630: @item -f @var{size}
                    631: Allocate @var{size} space for the floating point stack instead of
                    632: using the default specified in the image (typically 16K). In this case
                    633: the unit specifier @code{e} refers to floating point numbers.
                    635: @item --locals-stack-size @var{size}
                    636: @item -l @var{size}
                    637: Allocate @var{size} space for the locals stack instead of using the
                    638: default specified in the image (typically 16K).
                    640: @end table
                    642: As explained above, the image-specific command-line arguments for the
                    643: default image @file{} consist of a sequence of filenames and
                    644: @code{-e @var{forth-code}} options that are interpreted in the seqence
                    645: in which they are given. The @code{-e @var{forth-code}} or
                    646: @code{--evaluate @var{forth-code}} option evaluates the forth
                    647: code. This option takes only one argument; if you want to evaluate more
                    648: Forth words, you have to quote them or use several @code{-e}s. To exit
                    649: after processing the command line (instead of entering interactive mode)
                    650: append @code{-e bye} to the command line.
1.22      anton     652: If you have several versions of Gforth installed, @code{gforth} will
                    653: invoke the version that was installed last. @code{gforth-@var{version}}
                    654: invokes a specific version. You may want to use the option
                    655: @code{--path}, if your environment contains the variable
                    656: @code{GFORTHPATH}.
1.1       anton     658: Not yet implemented:
                    659: On startup the system first executes the system initialization file
                    660: (unless the option @code{--no-init-file} is given; note that the system
                    661: resulting from using this option may not be ANS Forth conformant). Then
                    662: the user initialization file @file{.gforth.fs} is executed, unless the
                    663: option @code{--no-rc} is given; this file is first searched in @file{.},
                    664: then in @file{~}, then in the normal path (see above).
1.40    ! anton     666: @node Words, Tools, Invocation, Top
1.1       anton     667: @chapter Forth Words
                    669: @menu
1.4       anton     670: * Notation::                    
                    671: * Arithmetic::                  
                    672: * Stack Manipulation::          
                    673: * Memory access::               
                    674: * Control Structures::          
                    675: * Locals::                      
                    676: * Defining Words::              
1.37      anton     677: * Tokens for Words::            
1.4       anton     678: * Wordlists::                   
                    679: * Files::                       
                    680: * Blocks::                      
                    681: * Other I/O::                   
                    682: * Programming Tools::           
1.18      anton     683: * Assembler and Code words::    
1.4       anton     684: * Threading Words::             
1.1       anton     685: @end menu
                    687: @node Notation, Arithmetic, Words, Words
                    688: @section Notation
                    690: The Forth words are described in this section in the glossary notation
                    691: that has become a de-facto standard for Forth texts, i.e.
1.4       anton     693: @format
1.1       anton     694: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
1.4       anton     695: @end format
1.1       anton     696: @var{Description}
                    698: @table @var
                    699: @item word
1.17      anton     700: The name of the word. BTW, Gforth is case insensitive, so you can
1.14      anton     701: type the words in in lower case (However, @pxref{core-idef}).
1.1       anton     702: 
                    703: @item Stack effect
                    704: The stack effect is written in the notation @code{@var{before} --
                    705: @var{after}}, where @var{before} and @var{after} describe the top of
                    706: stack entries before and after the execution of the word. The rest of
                    707: the stack is not touched by the word. The top of stack is rightmost,
1.17      anton     708: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1       anton     709: uses a separate floating point stack, but a unified stack
                    710: notation. Also, return stack effects are not shown in @var{stack
                    711: effect}, but in @var{Description}. The name of a stack item describes
                    712: the type and/or the function of the item. See below for a discussion of
                    713: the types.
1.19      anton     715: All words have two stack effects: A compile-time stack effect and a
                    716: run-time stack effect. The compile-time stack-effect of most words is
                    717: @var{ -- }. If the compile-time stack-effect of a word deviates from
                    718: this standard behaviour, or the word does other unusual things at
                    719: compile time, both stack effects are shown; otherwise only the run-time
                    720: stack effect is shown.
1.1       anton     722: @item pronunciation
                    723: How the word is pronounced
                    725: @item wordset
                    726: The ANS Forth standard is divided into several wordsets. A standard
                    727: system need not support all of them. So, the fewer wordsets your program
                    728: uses the more portable it will be in theory. However, we suspect that
                    729: most ANS Forth systems on personal machines will feature all
                    730: wordsets. Words that are not defined in the ANS standard have
1.19      anton     731: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
                    732: describes words that will work in future releases of Gforth;
                    733: @code{gforth-internal} words are more volatile. Environmental query
                    734: strings are also displayed like words; you can recognize them by the
                    735: @code{environment} in the wordset field.
1.1       anton     736: 
                    737: @item Description
                    738: A description of the behaviour of the word.
                    739: @end table
1.4       anton     741: The type of a stack item is specified by the character(s) the name
                    742: starts with:
1.1       anton     743: 
                    744: @table @code
                    745: @item f
                    746: Bool, i.e. @code{false} or @code{true}.
                    747: @item c
                    748: Char
                    749: @item w
                    750: Cell, can contain an integer or an address
                    751: @item n
                    752: signed integer
                    753: @item u
                    754: unsigned integer
                    755: @item d
                    756: double sized signed integer
                    757: @item ud
                    758: double sized unsigned integer
                    759: @item r
1.36      anton     760: Float (on the FP stack)
1.1       anton     761: @item a_
                    762: Cell-aligned address
                    763: @item c_
1.36      anton     764: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.1       anton     765: @item f_
                    766: Float-aligned address
                    767: @item df_
                    768: Address aligned for IEEE double precision float
                    769: @item sf_
                    770: Address aligned for IEEE single precision float
                    771: @item xt
                    772: Execution token, same size as Cell
                    773: @item wid
                    774: Wordlist ID, same size as Cell
                    775: @item f83name
                    776: Pointer to a name structure
1.36      anton     777: @item "
                    778: string in the input stream (not the stack). The terminating character is
                    779: a blank by default. If it is not a blank, it is shown in @code{<>}
                    780: quotes.
1.1       anton     782: @end table
1.4       anton     784: @node Arithmetic, Stack Manipulation, Notation, Words
1.1       anton     785: @section Arithmetic
                    786: Forth arithmetic is not checked, i.e., you will not hear about integer
                    787: overflow on addition or multiplication, you may hear about division by
                    788: zero if you are lucky. The operator is written after the operands, but
                    789: the operands are still in the original order. I.e., the infix @code{2-1}
                    790: corresponds to @code{2 1 -}. Forth offers a variety of division
                    791: operators. If you perform division with potentially negative operands,
                    792: you do not want to use @code{/} or @code{/mod} with its undefined
                    793: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4       anton     794: former, @pxref{Mixed precision}).
                    796: @menu
                    797: * Single precision::            
                    798: * Bitwise operations::          
                    799: * Mixed precision::             operations with single and double-cell integers
                    800: * Double precision::            Double-cell integer arithmetic
                    801: * Floating Point::              
                    802: @end menu
1.1       anton     803: 
1.4       anton     804: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1       anton     805: @subsection Single precision
                    806: doc-+
                    807: doc--
                    808: doc-*
                    809: doc-/
                    810: doc-mod
                    811: doc-/mod
                    812: doc-negate
                    813: doc-abs
                    814: doc-min
                    815: doc-max
1.4       anton     817: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1       anton     818: @subsection Bitwise operations
                    819: doc-and
                    820: doc-or
                    821: doc-xor
                    822: doc-invert
                    823: doc-2*
                    824: doc-2/
1.4       anton     826: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1       anton     827: @subsection Mixed precision
                    828: doc-m+
                    829: doc-*/
                    830: doc-*/mod
                    831: doc-m*
                    832: doc-um*
                    833: doc-m*/
                    834: doc-um/mod
                    835: doc-fm/mod
                    836: doc-sm/rem
1.4       anton     838: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1       anton     839: @subsection Double precision
1.16      anton     840: 
                    841: The outer (aka text) interpreter converts numbers containing a dot into
                    842: a double precision number. Note that only numbers with the dot as last
                    843: character are standard-conforming.
1.1       anton     845: doc-d+
                    846: doc-d-
                    847: doc-dnegate
                    848: doc-dabs
                    849: doc-dmin
                    850: doc-dmax
1.4       anton     852: @node Floating Point,  , Double precision, Arithmetic
                    853: @subsection Floating Point
1.16      anton     854: 
                    855: The format of floating point numbers recognized by the outer (aka text)
                    856: interpreter is: a signed decimal number, possibly containing a decimal
                    857: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
                    858: by a signed integer (the exponent). E.g., @code{1e} ist the same as
1.35      anton     859: @code{+1.0e+0}. Note that a number without @code{e}
1.16      anton     860: is not interpreted as floating-point number, but as double (if the
                    861: number contains a @code{.}) or single precision integer. Also,
                    862: conversions between string and floating point numbers always use base
                    863: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
                    864: value greater then 14, the @code{E} may be interpreted as digit and the
                    865: number will be interpreted as integer, unless it has a signed exponent
                    866: (both @code{+} and @code{-} are allowed as signs).
1.4       anton     867: 
                    868: Angles in floating point operations are given in radians (a full circle
1.17      anton     869: has 2 pi radians). Note, that Gforth has a separate floating point
1.4       anton     870: stack, but we use the unified notation.
                    872: Floating point numbers have a number of unpleasant surprises for the
                    873: unwary (e.g., floating point addition is not associative) and even a few
                    874: for the wary. You should not use them unless you know what you are doing
                    875: or you don't care that the results you get are totally bogus. If you
                    876: want to learn about the problems of floating point numbers (and how to
1.11      anton     877: avoid them), you might start with @cite{David Goldberg, What Every
1.6       anton     878: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
                    879: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4       anton     880: 
                    881: doc-f+
                    882: doc-f-
                    883: doc-f*
                    884: doc-f/
                    885: doc-fnegate
                    886: doc-fabs
                    887: doc-fmax
                    888: doc-fmin
                    889: doc-floor
                    890: doc-fround
                    891: doc-f**
                    892: doc-fsqrt
                    893: doc-fexp
                    894: doc-fexpm1
                    895: doc-fln
                    896: doc-flnp1
                    897: doc-flog
1.6       anton     898: doc-falog
1.4       anton     899: doc-fsin
                    900: doc-fcos
                    901: doc-fsincos
                    902: doc-ftan
                    903: doc-fasin
                    904: doc-facos
                    905: doc-fatan
                    906: doc-fatan2
                    907: doc-fsinh
                    908: doc-fcosh
                    909: doc-ftanh
                    910: doc-fasinh
                    911: doc-facosh
                    912: doc-fatanh
                    914: @node Stack Manipulation, Memory access, Arithmetic, Words
1.1       anton     915: @section Stack Manipulation
1.17      anton     917: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1       anton     918: addresses, and double cells, a floating point stack for floating point
                    919: numbers, a return stack for storing the return addresses of colon
                    920: definitions and other data, and a locals stack for storing local
                    921: variables. Note that while every sane Forth has a separate floating
                    922: point stack, this is not strictly required; an ANS Forth system could
                    923: theoretically keep floating point numbers on the data stack. As an
                    924: additional difficulty, you don't know how many cells a floating point
                    925: number takes. It is reportedly possible to write words in a way that
                    926: they work also for a unified stack model, but we do not recommend trying
1.4       anton     927: it. Instead, just say that your program has an environmental dependency
                    928: on a separate FP stack.
                    930: Also, a Forth system is allowed to keep the local variables on the
1.1       anton     931: return stack. This is reasonable, as local variables usually eliminate
                    932: the need to use the return stack explicitly. So, if you want to produce
                    933: a standard complying program and if you are using local variables in a
                    934: word, forget about return stack manipulations in that word (see the
                    935: standard document for the exact rules).
1.4       anton     937: @menu
                    938: * Data stack::                  
                    939: * Floating point stack::        
                    940: * Return stack::                
                    941: * Locals stack::                
                    942: * Stack pointer manipulation::  
                    943: @end menu
                    945: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1       anton     946: @subsection Data stack
                    947: doc-drop
                    948: doc-nip
                    949: doc-dup
                    950: doc-over
                    951: doc-tuck
                    952: doc-swap
                    953: doc-rot
                    954: doc--rot
                    955: doc-?dup
                    956: doc-pick
                    957: doc-roll
                    958: doc-2drop
                    959: doc-2nip
                    960: doc-2dup
                    961: doc-2over
                    962: doc-2tuck
                    963: doc-2swap
                    964: doc-2rot
1.4       anton     966: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1       anton     967: @subsection Floating point stack
                    968: doc-fdrop
                    969: doc-fnip
                    970: doc-fdup
                    971: doc-fover
                    972: doc-ftuck
                    973: doc-fswap
                    974: doc-frot
1.4       anton     976: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1       anton     977: @subsection Return stack
                    978: doc->r
                    979: doc-r>
                    980: doc-r@
                    981: doc-rdrop
                    982: doc-2>r
                    983: doc-2r>
                    984: doc-2r@
                    985: doc-2rdrop
1.4       anton     987: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1       anton     988: @subsection Locals stack
1.4       anton     990: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
1.1       anton     991: @subsection Stack pointer manipulation
                    992: doc-sp@
                    993: doc-sp!
                    994: doc-fp@
                    995: doc-fp!
                    996: doc-rp@
                    997: doc-rp!
                    998: doc-lp@
                    999: doc-lp!
1.4       anton    1001: @node Memory access, Control Structures, Stack Manipulation, Words
1.1       anton    1002: @section Memory access
1.4       anton    1004: @menu
                   1005: * Stack-Memory transfers::      
                   1006: * Address arithmetic::          
                   1007: * Memory block access::         
                   1008: @end menu
                   1010: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1.1       anton    1011: @subsection Stack-Memory transfers
                   1013: doc-@
                   1014: doc-!
                   1015: doc-+!
                   1016: doc-c@
                   1017: doc-c!
                   1018: doc-2@
                   1019: doc-2!
                   1020: doc-f@
                   1021: doc-f!
                   1022: doc-sf@
                   1023: doc-sf!
                   1024: doc-df@
                   1025: doc-df!
1.4       anton    1027: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1.1       anton    1028: @subsection Address arithmetic
                   1030: ANS Forth does not specify the sizes of the data types. Instead, it
                   1031: offers a number of words for computing sizes and doing address
                   1032: arithmetic. Basically, address arithmetic is performed in terms of
                   1033: address units (aus); on most systems the address unit is one byte. Note
                   1034: that a character may have more than one au, so @code{chars} is no noop
                   1035: (on systems where it is a noop, it compiles to nothing).
                   1037: ANS Forth also defines words for aligning addresses for specific
                   1038: addresses. Many computers require that accesses to specific data types
                   1039: must only occur at specific addresses; e.g., that cells may only be
                   1040: accessed at addresses divisible by 4. Even if a machine allows unaligned
                   1041: accesses, it can usually perform aligned accesses faster. 
1.17      anton    1043: For the performance-conscious: alignment operations are usually only
1.1       anton    1044: necessary during the definition of a data structure, not during the
                   1045: (more frequent) accesses to it.
                   1047: ANS Forth defines no words for character-aligning addresses. This is not
                   1048: an oversight, but reflects the fact that addresses that are not
                   1049: char-aligned have no use in the standard and therefore will not be
                   1050: created.
                   1052: The standard guarantees that addresses returned by @code{CREATE}d words
1.17      anton    1053: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1       anton    1054: are aligned for all purposes.
1.9       anton    1056: Note that the standard defines a word @code{char}, which has nothing to
                   1057: do with address arithmetic.
1.1       anton    1059: doc-chars
                   1060: doc-char+
                   1061: doc-cells
                   1062: doc-cell+
                   1063: doc-align
                   1064: doc-aligned
                   1065: doc-floats
                   1066: doc-float+
                   1067: doc-falign
                   1068: doc-faligned
                   1069: doc-sfloats
                   1070: doc-sfloat+
                   1071: doc-sfalign
                   1072: doc-sfaligned
                   1073: doc-dfloats
                   1074: doc-dfloat+
                   1075: doc-dfalign
                   1076: doc-dfaligned
1.10      anton    1077: doc-maxalign
                   1078: doc-maxaligned
                   1079: doc-cfalign
                   1080: doc-cfaligned
1.1       anton    1081: doc-address-unit-bits
1.4       anton    1083: @node Memory block access,  , Address arithmetic, Memory access
1.1       anton    1084: @subsection Memory block access
                   1086: doc-move
                   1087: doc-erase
                   1089: While the previous words work on address units, the rest works on
                   1090: characters.
                   1092: doc-cmove
                   1093: doc-cmove>
                   1094: doc-fill
                   1095: doc-blank
1.4       anton    1097: @node Control Structures, Locals, Memory access, Words
1.1       anton    1098: @section Control Structures
                   1100: Control structures in Forth cannot be used in interpret state, only in
                   1101: compile state, i.e., in a colon definition. We do not like this
                   1102: limitation, but have not seen a satisfying way around it yet, although
                   1103: many schemes have been proposed.
1.4       anton    1105: @menu
                   1106: * Selection::                   
                   1107: * Simple Loops::                
                   1108: * Counted Loops::               
                   1109: * Arbitrary control structures::  
                   1110: * Calls and returns::           
                   1111: * Exception Handling::          
                   1112: @end menu
                   1114: @node Selection, Simple Loops, Control Structures, Control Structures
1.1       anton    1115: @subsection Selection
                   1117: @example
                   1118: @var{flag}
                   1119: IF
                   1120:   @var{code}
                   1121: ENDIF
                   1122: @end example
                   1123: or
                   1124: @example
                   1125: @var{flag}
                   1126: IF
                   1127:   @var{code1}
                   1128: ELSE
                   1129:   @var{code2}
                   1130: ENDIF
                   1131: @end example
1.4       anton    1133: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1       anton    1134: standard, and @code{ENDIF} is not, although it is quite popular. We
                   1135: recommend using @code{ENDIF}, because it is less confusing for people
                   1136: who also know other languages (and is not prone to reinforcing negative
                   1137: prejudices against Forth in these people). Adding @code{ENDIF} to a
                   1138: system that only supplies @code{THEN} is simple:
                   1139: @example
                   1140: : endif   POSTPONE then ; immediate
                   1141: @end example
                   1143: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
                   1144: (adv.)}  has the following meanings:
                   1145: @quotation
                   1146: ... 2b: following next after in order ... 3d: as a necessary consequence
                   1147: (if you were there, then you saw them).
                   1148: @end quotation
                   1149: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
                   1150: and many other programming languages has the meaning 3d.]
1.31      anton    1152: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
                   1153: you can avoid using @code{?dup}. Using these alternatives is also more
                   1154: efficient than using @code{?dup}. Definitions in plain standard Forth
                   1155: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
                   1156: @file{compat/control.fs}.
1.1       anton    1157: 
                   1158: @example
                   1159: @var{n}
                   1160: CASE
                   1161:   @var{n1} OF @var{code1} ENDOF
                   1162:   @var{n2} OF @var{code2} ENDOF
1.4       anton    1163:   @dots{}
1.1       anton    1164: ENDCASE
                   1165: @end example
                   1167: Executes the first @var{codei}, where the @var{ni} is equal to
                   1168: @var{n}. A default case can be added by simply writing the code after
                   1169: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
                   1170: but must not consume it.
1.4       anton    1172: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1       anton    1173: @subsection Simple Loops
                   1175: @example
                   1176: BEGIN
                   1177:   @var{code1}
                   1178:   @var{flag}
                   1179: WHILE
                   1180:   @var{code2}
                   1181: REPEAT
                   1182: @end example
                   1184: @var{code1} is executed and @var{flag} is computed. If it is true,
                   1185: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
                   1187: @example
                   1188: BEGIN
                   1189:   @var{code}
                   1190:   @var{flag}
                   1191: UNTIL
                   1192: @end example
                   1194: @var{code} is executed. The loop is restarted if @code{flag} is false.
                   1196: @example
                   1197: BEGIN
                   1198:   @var{code}
                   1199: AGAIN
                   1200: @end example
                   1202: This is an endless loop.
1.4       anton    1204: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1       anton    1205: @subsection Counted Loops
                   1207: The basic counted loop is:
                   1208: @example
                   1209: @var{limit} @var{start}
                   1210: ?DO
                   1211:   @var{body}
                   1212: LOOP
                   1213: @end example
                   1215: This performs one iteration for every integer, starting from @var{start}
                   1216: and up to, but excluding @var{limit}. The counter, aka index, can be
                   1217: accessed with @code{i}. E.g., the loop
                   1218: @example
                   1219: 10 0 ?DO
                   1220:   i .
                   1221: LOOP
                   1222: @end example
                   1223: prints
                   1224: @example
                   1225: 0 1 2 3 4 5 6 7 8 9
                   1226: @end example
                   1227: The index of the innermost loop can be accessed with @code{i}, the index
                   1228: of the next loop with @code{j}, and the index of the third loop with
                   1229: @code{k}.
                   1231: The loop control data are kept on the return stack, so there are some
                   1232: restrictions on mixing return stack accesses and counted loop
                   1233: words. E.g., if you put values on the return stack outside the loop, you
                   1234: cannot read them inside the loop. If you put values on the return stack
                   1235: within a loop, you have to remove them before the end of the loop and
                   1236: before accessing the index of the loop.
                   1238: There are several variations on the counted loop:
                   1240: @code{LEAVE} leaves the innermost counted loop immediately.
1.18      anton    1242: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
                   1243: (and @code{LOOP} iterates until they become equal by wrap-around
                   1244: arithmetic). This behaviour is usually not what you want. Therefore,
                   1245: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
                   1246: @code{?DO}), which do not enter the loop if @var{start} is greater than
                   1247: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.30      anton    1248: unsigned loop parameters.
1.18      anton    1249: 
1.1       anton    1250: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
                   1251: index by @var{n} instead of by 1. The loop is terminated when the border
                   1252: between @var{limit-1} and @var{limit} is crossed. E.g.:
1.18      anton    1254: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
1.1       anton    1255: 
1.18      anton    1256: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
1.1       anton    1257: 
                   1258: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1.2       anton    1260: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
1.1       anton    1261: 
1.2       anton    1262: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
1.1       anton    1263: 
1.18      anton    1264: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
                   1265: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
                   1266: index by @var{u} each iteration. The loop is terminated when the border
                   1267: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
                   1268: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1       anton    1269: 
1.18      anton    1270: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
1.1       anton    1271: 
1.18      anton    1272: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
1.1       anton    1273: 
1.18      anton    1274: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
1.1       anton    1275: 
1.30      anton    1276: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
                   1277: @code{-LOOP} are not in the ANS Forth standard. However, an
                   1278: implementation for these words that uses only standard words is provided
                   1279: in @file{compat/loops.fs}.
1.18      anton    1280: 
                   1281: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
                   1282: the loop, independent of the loop parameters. Do not use @code{DO}, even
                   1283: if you know that the loop is entered in any case. Such knowledge tends
                   1284: to become invalid during maintenance of a program, and then the
                   1285: @code{DO} will make trouble.
1.1       anton    1286: 
                   1287: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
                   1288: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
                   1289: return stack so @code{EXIT} can get to its return address.
                   1291: Another counted loop is
                   1292: @example
                   1293: @var{n}
                   1294: FOR
                   1295:   @var{body}
                   1296: NEXT
                   1297: @end example
                   1298: This is the preferred loop of native code compiler writers who are too
1.17      anton    1299: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1       anton    1300: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
                   1301: and ending with 0. Other Forth systems may behave differently, even if
1.30      anton    1302: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
                   1303: loops.
1.1       anton    1304: 
1.4       anton    1305: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2       anton    1306: @subsection Arbitrary control structures
                   1308: ANS Forth permits and supports using control structures in a non-nested
                   1309: way. Information about incomplete control structures is stored on the
                   1310: control-flow stack. This stack may be implemented on the Forth data
1.17      anton    1311: stack, and this is what we have done in Gforth.
1.2       anton    1312: 
                   1313: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
                   1314: entry represents a backward branch target. A few words are the basis for
                   1315: building any control structure possible (except control structures that
                   1316: need storage, like calls, coroutines, and backtracking).
1.3       anton    1318: doc-if
                   1319: doc-ahead
                   1320: doc-then
                   1321: doc-begin
                   1322: doc-until
                   1323: doc-again
                   1324: doc-cs-pick
                   1325: doc-cs-roll
1.2       anton    1326: 
1.17      anton    1327: On many systems control-flow stack items take one word, in Gforth they
1.2       anton    1328: currently take three (this may change in the future). Therefore it is a
                   1329: really good idea to manipulate the control flow stack with
                   1330: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
                   1331: words.
                   1333: Some standard control structure words are built from these words:
1.3       anton    1335: doc-else
                   1336: doc-while
                   1337: doc-repeat
1.2       anton    1338: 
1.31      anton    1339: Gforth adds some more control-structure words:
                   1341: doc-endif
                   1342: doc-?dup-if
                   1343: doc-?dup-0=-if
1.2       anton    1345: Counted loop words constitute a separate group of words:
1.3       anton    1347: doc-?do
1.18      anton    1348: doc-+do
                   1349: doc-u+do
                   1350: doc--do
                   1351: doc-u-do
1.3       anton    1352: doc-do
                   1353: doc-for
                   1354: doc-loop
                   1355: doc-+loop
1.18      anton    1356: doc--loop
1.3       anton    1357: doc-next
                   1358: doc-leave
                   1359: doc-?leave
                   1360: doc-unloop
1.10      anton    1361: doc-done
1.2       anton    1362: 
                   1363: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
                   1364: @i{do-sys}. Our system allows it, but it's your job to ensure that for
                   1365: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3       anton    1366: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
                   1367: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7       pazsan   1368: resolved (by using one of the loop-ending words or @code{DONE}).
1.2       anton    1369: 
                   1370: Another group of control structure words are
1.3       anton    1372: doc-case
                   1373: doc-endcase
                   1374: doc-of
                   1375: doc-endof
1.2       anton    1376: 
                   1377: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
                   1378: @code{cs-roll}.
1.3       anton    1380: @subsubsection Programming Style
                   1382: In order to ensure readability we recommend that you do not create
                   1383: arbitrary control structures directly, but define new control structure
                   1384: words for the control structure you want and use these words in your
                   1385: program.
                   1387: E.g., instead of writing
                   1389: @example
                   1390: begin
                   1391:   ...
                   1392: if [ 1 cs-roll ]
                   1393:   ...
                   1394: again then
                   1395: @end example
                   1397: we recommend defining control structure words, e.g.,
                   1399: @example
                   1400: : while ( dest -- orig dest )
                   1401:  POSTPONE if
                   1402:  1 cs-roll ; immediate
                   1404: : repeat ( orig dest -- )
                   1405:  POSTPONE again
                   1406:  POSTPONE then ; immediate
                   1407: @end example
                   1409: and then using these to create the control structure:
                   1411: @example
                   1412: begin
                   1413:   ...
                   1414: while
                   1415:   ...
                   1416: repeat
                   1417: @end example
1.30      anton    1419: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1.3       anton    1420: @code{WHILE} are predefined, so in this example it would not be
                   1421: necessary to define them.
1.4       anton    1423: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3       anton    1424: @subsection Calls and returns
                   1426: A definition can be called simply be writing the name of the
1.17      anton    1427: definition. When the end of the definition is reached, it returns. An
                   1428: earlier return can be forced using
1.3       anton    1429: 
                   1430: doc-exit
                   1432: Don't forget to clean up the return stack and @code{UNLOOP} any
                   1433: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
                   1434: primitive compiled by @code{EXIT} is
                   1436: doc-;s
1.4       anton    1438: @node Exception Handling,  , Calls and returns, Control Structures
1.3       anton    1439: @subsection Exception Handling
                   1441: doc-catch
                   1442: doc-throw
1.4       anton    1444: @node Locals, Defining Words, Control Structures, Words
1.1       anton    1445: @section Locals
1.2       anton    1447: Local variables can make Forth programming more enjoyable and Forth
                   1448: programs easier to read. Unfortunately, the locals of ANS Forth are
                   1449: laden with restrictions. Therefore, we provide not only the ANS Forth
                   1450: locals wordset, but also our own, more powerful locals wordset (we
                   1451: implemented the ANS Forth locals wordset through our locals wordset).
1.24      anton    1453: The ideas in this section have also been published in the paper
                   1454: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
                   1455: at EuroForth '94; it is available at
                   1456: @*@file{}.
1.2       anton    1458: @menu
1.17      anton    1459: * Gforth locals::               
1.4       anton    1460: * ANS Forth locals::            
1.2       anton    1461: @end menu
1.17      anton    1463: @node Gforth locals, ANS Forth locals, Locals, Locals
                   1464: @subsection Gforth locals
1.2       anton    1465: 
                   1466: Locals can be defined with
                   1468: @example
                   1469: @{ local1 local2 ... -- comment @}
                   1470: @end example
                   1471: or
                   1472: @example
                   1473: @{ local1 local2 ... @}
                   1474: @end example
                   1476: E.g.,
                   1477: @example
                   1478: : max @{ n1 n2 -- n3 @}
                   1479:  n1 n2 > if
                   1480:    n1
                   1481:  else
                   1482:    n2
                   1483:  endif ;
                   1484: @end example
                   1486: The similarity of locals definitions with stack comments is intended. A
                   1487: locals definition often replaces the stack comment of a word. The order
                   1488: of the locals corresponds to the order in a stack comment and everything
                   1489: after the @code{--} is really a comment.
                   1491: This similarity has one disadvantage: It is too easy to confuse locals
                   1492: declarations with stack comments, causing bugs and making them hard to
                   1493: find. However, this problem can be avoided by appropriate coding
                   1494: conventions: Do not use both notations in the same program. If you do,
                   1495: they should be distinguished using additional means, e.g. by position.
                   1497: The name of the local may be preceded by a type specifier, e.g.,
                   1498: @code{F:} for a floating point value:
                   1500: @example
                   1501: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
                   1502: \ complex multiplication
                   1503:  Ar Br f* Ai Bi f* f-
                   1504:  Ar Bi f* Ai Br f* f+ ;
                   1505: @end example
1.17      anton    1507: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2       anton    1508: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
                   1509: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
                   1510: with @code{W:}, @code{D:} etc.) produces its value and can be changed
                   1511: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
                   1512: produces its address (which becomes invalid when the variable's scope is
                   1513: left). E.g., the standard word @code{emit} can be defined in therms of
                   1514: @code{type} like this:
                   1516: @example
                   1517: : emit @{ C^ char* -- @}
                   1518:     char* 1 type ;
                   1519: @end example
                   1521: A local without type specifier is a @code{W:} local. Both flavours of
                   1522: locals are initialized with values from the data or FP stack.
                   1524: Currently there is no way to define locals with user-defined data
                   1525: structures, but we are working on it.
1.17      anton    1527: Gforth allows defining locals everywhere in a colon definition. This
1.7       pazsan   1528: poses the following questions:
1.2       anton    1529: 
1.4       anton    1530: @menu
                   1531: * Where are locals visible by name?::  
1.14      anton    1532: * How long do locals live?::    
1.4       anton    1533: * Programming Style::           
                   1534: * Implementation::              
                   1535: @end menu
1.17      anton    1537: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2       anton    1538: @subsubsection Where are locals visible by name?
                   1540: Basically, the answer is that locals are visible where you would expect
                   1541: it in block-structured languages, and sometimes a little longer. If you
                   1542: want to restrict the scope of a local, enclose its definition in
                   1543: @code{SCOPE}...@code{ENDSCOPE}.
                   1545: doc-scope
                   1546: doc-endscope
                   1548: These words behave like control structure words, so you can use them
                   1549: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
                   1550: arbitrary ways.
                   1552: If you want a more exact answer to the visibility question, here's the
                   1553: basic principle: A local is visible in all places that can only be
                   1554: reached through the definition of the local@footnote{In compiler
                   1555: construction terminology, all places dominated by the definition of the
                   1556: local.}. In other words, it is not visible in places that can be reached
                   1557: without going through the definition of the local. E.g., locals defined
                   1558: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
                   1559: defined in @code{BEGIN}...@code{UNTIL} are visible after the
                   1560: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
                   1562: The reasoning behind this solution is: We want to have the locals
                   1563: visible as long as it is meaningful. The user can always make the
                   1564: visibility shorter by using explicit scoping. In a place that can
                   1565: only be reached through the definition of a local, the meaning of a
                   1566: local name is clear. In other places it is not: How is the local
                   1567: initialized at the control flow path that does not contain the
                   1568: definition? Which local is meant, if the same name is defined twice in
                   1569: two independent control flow paths?
                   1571: This should be enough detail for nearly all users, so you can skip the
                   1572: rest of this section. If you relly must know all the gory details and
                   1573: options, read on.
                   1575: In order to implement this rule, the compiler has to know which places
                   1576: are unreachable. It knows this automatically after @code{AHEAD},
                   1577: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
                   1578: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
                   1579: compiler that the control flow never reaches that place. If
                   1580: @code{UNREACHABLE} is not used where it could, the only consequence is
                   1581: that the visibility of some locals is more limited than the rule above
                   1582: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
                   1583: lie to the compiler), buggy code will be produced.
                   1585: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3       anton    1586: does not know which locals will be visible on the incoming
                   1587: back-edge. All problems discussed in the following are due to this
                   1588: ignorance of the compiler (we discuss the problems using @code{BEGIN}
                   1589: loops as examples; the discussion also applies to @code{?DO} and other
1.2       anton    1590: loops). Perhaps the most insidious example is:
                   1591: @example
                   1592: AHEAD
                   1593: BEGIN
                   1594:   x
                   1595: [ 1 CS-ROLL ] THEN
1.4       anton    1596:   @{ x @}
1.2       anton    1597:   ...
                   1598: UNTIL
                   1599: @end example
                   1601: This should be legal according to the visibility rule. The use of
                   1602: @code{x} can only be reached through the definition; but that appears
                   1603: textually below the use.
                   1605: From this example it is clear that the visibility rules cannot be fully
                   1606: implemented without major headaches. Our implementation treats common
                   1607: cases as advertised and the exceptions are treated in a safe way: The
                   1608: compiler makes a reasonable guess about the locals visible after a
                   1609: @code{BEGIN}; if it is too pessimistic, the
                   1610: user will get a spurious error about the local not being defined; if the
                   1611: compiler is too optimistic, it will notice this later and issue a
                   1612: warning. In the case above the compiler would complain about @code{x}
                   1613: being undefined at its use. You can see from the obscure examples in
                   1614: this section that it takes quite unusual control structures to get the
                   1615: compiler into trouble, and even then it will often do fine.
                   1617: If the @code{BEGIN} is reachable from above, the most optimistic guess
                   1618: is that all locals visible before the @code{BEGIN} will also be
                   1619: visible after the @code{BEGIN}. This guess is valid for all loops that
                   1620: are entered only through the @code{BEGIN}, in particular, for normal
                   1621: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
                   1622: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
                   1623: compiler. When the branch to the @code{BEGIN} is finally generated by
                   1624: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
                   1625: warns the user if it was too optimisitic:
                   1626: @example
                   1627: IF
1.4       anton    1628:   @{ x @}
1.2       anton    1629: BEGIN
                   1630:   \ x ? 
                   1631: [ 1 cs-roll ] THEN
                   1632:   ...
                   1633: UNTIL
                   1634: @end example
                   1636: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
                   1637: optimistically assumes that it lives until the @code{THEN}. It notices
                   1638: this difference when it compiles the @code{UNTIL} and issues a
                   1639: warning. The user can avoid the warning, and make sure that @code{x}
                   1640: is not used in the wrong area by using explicit scoping:
                   1641: @example
                   1642: IF
                   1643:   SCOPE
1.4       anton    1644:   @{ x @}
1.2       anton    1645:   ENDSCOPE
                   1646: BEGIN
                   1647: [ 1 cs-roll ] THEN
                   1648:   ...
                   1649: UNTIL
                   1650: @end example
                   1652: Since the guess is optimistic, there will be no spurious error messages
                   1653: about undefined locals.
                   1655: If the @code{BEGIN} is not reachable from above (e.g., after
                   1656: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
                   1657: optimistic guess, as the locals visible after the @code{BEGIN} may be
                   1658: defined later. Therefore, the compiler assumes that no locals are
1.17      anton    1659: visible after the @code{BEGIN}. However, the user can use
1.2       anton    1660: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17      anton    1661: visible at the BEGIN as at the point where the top control-flow stack
                   1662: item was created.
1.2       anton    1663: 
                   1664: doc-assume-live
                   1666: E.g.,
                   1667: @example
1.4       anton    1668: @{ x @}
1.2       anton    1669: AHEAD
                   1670: ASSUME-LIVE
                   1671: BEGIN
                   1672:   x
                   1673: [ 1 CS-ROLL ] THEN
                   1674:   ...
                   1675: UNTIL
                   1676: @end example
                   1678: Other cases where the locals are defined before the @code{BEGIN} can be
                   1679: handled by inserting an appropriate @code{CS-ROLL} before the
                   1680: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
                   1681: behind the @code{ASSUME-LIVE}).
                   1683: Cases where locals are defined after the @code{BEGIN} (but should be
                   1684: visible immediately after the @code{BEGIN}) can only be handled by
                   1685: rearranging the loop. E.g., the ``most insidious'' example above can be
                   1686: arranged into:
                   1687: @example
                   1688: BEGIN
1.4       anton    1689:   @{ x @}
1.2       anton    1690:   ... 0=
                   1691: WHILE
                   1692:   x
                   1693: REPEAT
                   1694: @end example
1.17      anton    1696: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2       anton    1697: @subsubsection How long do locals live?
                   1699: The right answer for the lifetime question would be: A local lives at
                   1700: least as long as it can be accessed. For a value-flavoured local this
                   1701: means: until the end of its visibility. However, a variable-flavoured
                   1702: local could be accessed through its address far beyond its visibility
                   1703: scope. Ultimately, this would mean that such locals would have to be
                   1704: garbage collected. Since this entails un-Forth-like implementation
                   1705: complexities, I adopted the same cowardly solution as some other
                   1706: languages (e.g., C): The local lives only as long as it is visible;
                   1707: afterwards its address is invalid (and programs that access it
                   1708: afterwards are erroneous).
1.17      anton    1710: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2       anton    1711: @subsubsection Programming Style
                   1713: The freedom to define locals anywhere has the potential to change
                   1714: programming styles dramatically. In particular, the need to use the
                   1715: return stack for intermediate storage vanishes. Moreover, all stack
                   1716: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
                   1717: determined arguments) can be eliminated: If the stack items are in the
                   1718: wrong order, just write a locals definition for all of them; then
                   1719: write the items in the order you want.
                   1721: This seems a little far-fetched and eliminating stack manipulations is
1.4       anton    1722: unlikely to become a conscious programming objective. Still, the number
                   1723: of stack manipulations will be reduced dramatically if local variables
1.17      anton    1724: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4       anton    1725: a traditional implementation of @code{max}).
1.2       anton    1726: 
                   1727: This shows one potential benefit of locals: making Forth programs more
                   1728: readable. Of course, this benefit will only be realized if the
                   1729: programmers continue to honour the principle of factoring instead of
                   1730: using the added latitude to make the words longer.
                   1732: Using @code{TO} can and should be avoided.  Without @code{TO},
                   1733: every value-flavoured local has only a single assignment and many
                   1734: advantages of functional languages apply to Forth. I.e., programs are
                   1735: easier to analyse, to optimize and to read: It is clear from the
                   1736: definition what the local stands for, it does not turn into something
                   1737: different later.
                   1739: E.g., a definition using @code{TO} might look like this:
                   1740: @example
                   1741: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1742:  u1 u2 min 0
                   1743:  ?do
1.36      anton    1744:    addr1 c@@ addr2 c@@ -
1.31      anton    1745:    ?dup-if
1.2       anton    1746:      unloop exit
                   1747:    then
                   1748:    addr1 char+ TO addr1
                   1749:    addr2 char+ TO addr2
                   1750:  loop
                   1751:  u1 u2 - ;
                   1752: @end example
                   1753: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
                   1754: every loop iteration. @code{strcmp} is a typical example of the
                   1755: readability problems of using @code{TO}. When you start reading
                   1756: @code{strcmp}, you think that @code{addr1} refers to the start of the
                   1757: string. Only near the end of the loop you realize that it is something
                   1758: else.
                   1760: This can be avoided by defining two locals at the start of the loop that
                   1761: are initialized with the right value for the current iteration.
                   1762: @example
                   1763: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1764:  addr1 addr2
                   1765:  u1 u2 min 0 
                   1766:  ?do @{ s1 s2 @}
1.36      anton    1767:    s1 c@@ s2 c@@ -
1.31      anton    1768:    ?dup-if
1.2       anton    1769:      unloop exit
                   1770:    then
                   1771:    s1 char+ s2 char+
                   1772:  loop
                   1773:  2drop
                   1774:  u1 u2 - ;
                   1775: @end example
                   1776: Here it is clear from the start that @code{s1} has a different value
                   1777: in every loop iteration.
1.17      anton    1779: @node Implementation,  , Programming Style, Gforth locals
1.2       anton    1780: @subsubsection Implementation
1.17      anton    1782: Gforth uses an extra locals stack. The most compelling reason for
1.2       anton    1783: this is that the return stack is not float-aligned; using an extra stack
                   1784: also eliminates the problems and restrictions of using the return stack
                   1785: as locals stack. Like the other stacks, the locals stack grows toward
                   1786: lower addresses. A few primitives allow an efficient implementation:
                   1788: doc-@local#
                   1789: doc-f@local#
                   1790: doc-laddr#
                   1791: doc-lp+!#
                   1792: doc-lp!
                   1793: doc->l
                   1794: doc-f>l
                   1796: In addition to these primitives, some specializations of these
                   1797: primitives for commonly occurring inline arguments are provided for
                   1798: efficiency reasons, e.g., @code{@@local0} as specialization of
                   1799: @code{@@local#} for the inline argument 0. The following compiling words
                   1800: compile the right specialized version, or the general version, as
                   1801: appropriate:
1.12      anton    1803: doc-compile-@local
                   1804: doc-compile-f@local
1.2       anton    1805: doc-compile-lp+!
                   1807: Combinations of conditional branches and @code{lp+!#} like
                   1808: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
                   1809: is taken) are provided for efficiency and correctness in loops.
                   1811: A special area in the dictionary space is reserved for keeping the
                   1812: local variable names. @code{@{} switches the dictionary pointer to this
                   1813: area and @code{@}} switches it back and generates the locals
                   1814: initializing code. @code{W:} etc.@ are normal defining words. This
                   1815: special area is cleared at the start of every colon definition.
1.17      anton    1817: A special feature of Gforth's dictionary is used to implement the
1.2       anton    1818: definition of locals without type specifiers: every wordlist (aka
                   1819: vocabulary) has its own methods for searching
1.4       anton    1820: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2       anton    1821: with a special search method: When it is searched for a word, it
                   1822: actually creates that word using @code{W:}. @code{@{} changes the search
                   1823: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
                   1824: and then the wordlist for defining locals without type specifiers.
                   1826: The lifetime rules support a stack discipline within a colon
                   1827: definition: The lifetime of a local is either nested with other locals
                   1828: lifetimes or it does not overlap them.
                   1830: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
                   1831: pointer manipulation is generated. Between control structure words
                   1832: locals definitions can push locals onto the locals stack. @code{AGAIN}
                   1833: is the simplest of the other three control flow words. It has to
                   1834: restore the locals stack depth of the corresponding @code{BEGIN}
                   1835: before branching. The code looks like this:
                   1836: @format
                   1837: @code{lp+!#} current-locals-size @minus{} dest-locals-size
                   1838: @code{branch} <begin>
                   1839: @end format
                   1841: @code{UNTIL} is a little more complicated: If it branches back, it
                   1842: must adjust the stack just like @code{AGAIN}. But if it falls through,
                   1843: the locals stack must not be changed. The compiler generates the
                   1844: following code:
                   1845: @format
                   1846: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
                   1847: @end format
                   1848: The locals stack pointer is only adjusted if the branch is taken.
                   1850: @code{THEN} can produce somewhat inefficient code:
                   1851: @format
                   1852: @code{lp+!#} current-locals-size @minus{} orig-locals-size
                   1853: <orig target>:
                   1854: @code{lp+!#} orig-locals-size @minus{} new-locals-size
                   1855: @end format
                   1856: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4       anton    1857: level at the @var{orig} point to the level after the @code{THEN}. The
1.2       anton    1858: first @code{lp+!#} adjusts the locals stack pointer from the current
                   1859: level to the level at the orig point, so the complete effect is an
                   1860: adjustment from the current level to the right level after the
                   1861: @code{THEN}.
                   1863: In a conventional Forth implementation a dest control-flow stack entry
                   1864: is just the target address and an orig entry is just the address to be
                   1865: patched. Our locals implementation adds a wordlist to every orig or dest
                   1866: item. It is the list of locals visible (or assumed visible) at the point
                   1867: described by the entry. Our implementation also adds a tag to identify
                   1868: the kind of entry, in particular to differentiate between live and dead
                   1869: (reachable and unreachable) orig entries.
                   1871: A few unusual operations have to be performed on locals wordlists:
                   1873: doc-common-list
                   1874: doc-sub-list?
                   1875: doc-list-size
                   1877: Several features of our locals wordlist implementation make these
                   1878: operations easy to implement: The locals wordlists are organised as
                   1879: linked lists; the tails of these lists are shared, if the lists
                   1880: contain some of the same locals; and the address of a name is greater
                   1881: than the address of the names behind it in the list.
                   1883: Another important implementation detail is the variable
                   1884: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
                   1885: determine if they can be reached directly or only through the branch
                   1886: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
                   1887: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
                   1888: definition, by @code{BEGIN} and usually by @code{THEN}.
                   1890: Counted loops are similar to other loops in most respects, but
                   1891: @code{LEAVE} requires special attention: It performs basically the same
                   1892: service as @code{AHEAD}, but it does not create a control-flow stack
                   1893: entry. Therefore the information has to be stored elsewhere;
                   1894: traditionally, the information was stored in the target fields of the
                   1895: branches created by the @code{LEAVE}s, by organizing these fields into a
                   1896: linked list. Unfortunately, this clever trick does not provide enough
                   1897: space for storing our extended control flow information. Therefore, we
                   1898: introduce another stack, the leave stack. It contains the control-flow
                   1899: stack entries for all unresolved @code{LEAVE}s.
                   1901: Local names are kept until the end of the colon definition, even if
                   1902: they are no longer visible in any control-flow path. In a few cases
                   1903: this may lead to increased space needs for the locals name area, but
                   1904: usually less than reclaiming this space would cost in code size.
1.17      anton    1907: @node ANS Forth locals,  , Gforth locals, Locals
1.2       anton    1908: @subsection ANS Forth locals
                   1910: The ANS Forth locals wordset does not define a syntax for locals, but
                   1911: words that make it possible to define various syntaxes. One of the
1.17      anton    1912: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2       anton    1913: wordset, i.e.:
                   1915: @example
                   1916: @{ local1 local2 ... -- comment @}
                   1917: @end example
                   1918: or
                   1919: @example
                   1920: @{ local1 local2 ... @}
                   1921: @end example
                   1923: The order of the locals corresponds to the order in a stack comment. The
                   1924: restrictions are:
1.1       anton    1925: 
1.2       anton    1926: @itemize @bullet
                   1927: @item
1.17      anton    1928: Locals can only be cell-sized values (no type specifiers are allowed).
1.2       anton    1929: @item
                   1930: Locals can be defined only outside control structures.
                   1931: @item
                   1932: Locals can interfere with explicit usage of the return stack. For the
                   1933: exact (and long) rules, see the standard. If you don't use return stack
1.17      anton    1934: accessing words in a definition using locals, you will be all right. The
1.2       anton    1935: purpose of this rule is to make locals implementation on the return
                   1936: stack easier.
                   1937: @item
                   1938: The whole definition must be in one line.
                   1939: @end itemize
1.35      anton    1941: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
                   1942: Defining Words}). I.e., they are initialized from the stack. Using their
1.2       anton    1943: name produces their value. Their value can be changed using @code{TO}.
1.17      anton    1945: Since this syntax is supported by Gforth directly, you need not do
1.2       anton    1946: anything to use it. If you want to port a program using this syntax to
1.30      anton    1947: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
                   1948: syntax on the other system.
1.2       anton    1949: 
                   1950: Note that a syntax shown in the standard, section A.13 looks
                   1951: similar, but is quite different in having the order of locals
                   1952: reversed. Beware!
                   1954: The ANS Forth locals wordset itself consists of the following word
                   1956: doc-(local)
                   1958: The ANS Forth locals extension wordset defines a syntax, but it is so
                   1959: awful that we strongly recommend not to use it. We have implemented this
1.17      anton    1960: syntax to make porting to Gforth easy, but do not document it here. The
1.2       anton    1961: problem with this syntax is that the locals are defined in an order
                   1962: reversed with respect to the standard stack comment notation, making
                   1963: programs harder to read, and easier to misread and miswrite. The only
                   1964: merit of this syntax is that it is easy to implement using the ANS Forth
                   1965: locals wordset.
1.3       anton    1966: 
1.37      anton    1967: @node Defining Words, Tokens for Words, Locals, Words
1.4       anton    1968: @section Defining Words
1.14      anton    1970: @menu
1.35      anton    1971: * Simple Defining Words::       
                   1972: * Colon Definitions::           
                   1973: * User-defined Defining Words::  
                   1974: * Supplying names::             
                   1975: * Interpretation and Compilation Semantics::  
1.14      anton    1976: @end menu
1.35      anton    1978: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
                   1979: @subsection Simple Defining Words
                   1981: doc-constant
                   1982: doc-2constant
                   1983: doc-fconstant
                   1984: doc-variable
                   1985: doc-2variable
                   1986: doc-fvariable
                   1987: doc-create
                   1988: doc-user
                   1989: doc-value
                   1990: doc-to
                   1991: doc-defer
                   1992: doc-is
                   1994: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
                   1995: @subsection Colon Definitions
                   1997: @example
                   1998: : name ( ... -- ... )
                   1999:     word1 word2 word3 ;
                   2000: @end example
                   2002: creates a word called @code{name}, that, upon execution, executes
                   2003: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
                   2005: The explanation above is somewhat superficial. @xref{Interpretation and
                   2006: Compilation Semantics} for an in-depth discussion of some of the issues
                   2007: involved.
                   2009: doc-:
                   2010: doc-;
                   2012: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
                   2013: @subsection User-defined Defining Words
                   2015: You can create new defining words simply by wrapping defining-time code
                   2016: around existing defining words and putting the sequence in a colon
                   2017: definition.
1.36      anton    2019: If you want the words defined with your defining words to behave
                   2020: differently from words defined with standard defining words, you can
1.35      anton    2021: write your defining word like this:
                   2023: @example
                   2024: : def-word ( "name" -- )
                   2025:     Create @var{code1}
                   2026: DOES> ( ... -- ... )
                   2027:     @var{code2} ;
                   2029: def-word name
                   2030: @end example
                   2032: Technically, this fragment defines a defining word @code{def-word}, and
                   2033: a word @code{name}; when you execute @code{name}, the address of the
                   2034: body of @code{name} is put on the data stack and @var{code2} is executed
                   2035: (the address of the body of @code{name} is the address @code{HERE}
1.36      anton    2036: returns immediately after the @code{CREATE}).
                   2038: In other words, if you make the following definitions:
                   2040: @example
                   2041: : def-word1 ( "name" -- )
                   2042:     Create @var{code1} ;
                   2044: : action1 ( ... -- ... )
                   2045:     @var{code2} ;
                   2047: def-word name1
                   2048: @end example
                   2050: Using @code{name1 action1} is equivalent to using @code{name}.
                   2052: E.g., you can implement @code{Constant} in this way:
1.35      anton    2053: 
                   2054: @example
                   2055: : constant ( w "name" -- )
                   2056:     create ,
                   2057: DOES> ( -- w )
1.36      anton    2058:     @@ ;
1.35      anton    2059: @end example
                   2061: When you create a constant with @code{5 constant five}, first a new word
                   2062: @code{five} is created, then the value 5 is laid down in the body of
                   2063: @code{five} with @code{,}. When @code{five} is invoked, the address of
                   2064: the body is put on the stack, and @code{@@} retrieves the value 5.
                   2066: In the example above the stack comment after the @code{DOES>} specifies
                   2067: the stack effect of the defined words, not the stack effect of the
                   2068: following code (the following code expects the address of the body on
                   2069: the top of stack, which is not reflected in the stack comment). This is
                   2070: the convention that I use and recommend (it clashes a bit with using
                   2071: locals declarations for stack effect specification, though).
                   2073: @subsubsection Applications of @code{CREATE..DOES>}
1.36      anton    2075: You may wonder how to use this feature. Here are some usage patterns:
1.35      anton    2076: 
                   2077: When you see a sequence of code occurring several times, and you can
                   2078: identify a meaning, you will factor it out as a colon definition. When
                   2079: you see similar colon definitions, you can factor them using
                   2080: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
                   2081: that look very similar:
                   2082: @example
                   2083: : ori, ( reg-taget reg-source n -- )
                   2084:     0 asm-reg-reg-imm ;
                   2085: : andi, ( reg-taget reg-source n -- )
                   2086:     1 asm-reg-reg-imm ;
                   2087: @end example
                   2089: This could be factored with:
                   2090: @example
                   2091: : reg-reg-imm ( op-code -- )
                   2092:     create ,
                   2093: DOES> ( reg-taget reg-source n -- )
1.36      anton    2094:     @@ asm-reg-reg-imm ;
1.35      anton    2095: 
                   2096: 0 reg-reg-imm ori,
                   2097: 1 reg-reg-imm andi,
                   2098: @end example
                   2100: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
                   2101: supply a part of the parameters for a word (known as @dfn{currying} in
                   2102: the functional language community). E.g., @code{+} needs two
                   2103: parameters. Creating versions of @code{+} with one parameter fixed can
                   2104: be done like this:
                   2105: @example
                   2106: : curry+ ( n1 -- )
                   2107:     create ,
                   2108: DOES> ( n2 -- n1+n2 )
1.36      anton    2109:     @@ + ;
1.35      anton    2110: 
                   2111:  3 curry+ 3+
                   2112: -2 curry+ 2-
                   2113: @end example
                   2115: @subsubsection The gory details of @code{CREATE..DOES>}
                   2117: doc-does>
                   2119: This means that you need not use @code{CREATE} and @code{DOES>} in the
                   2120: same definition; E.g., you can put the @code{DOES>}-part in a separate
                   2121: definition. This allows us to, e.g., select among different DOES>-parts:
                   2122: @example
                   2123: : does1 
                   2124: DOES> ( ... -- ... )
                   2125:     ... ;
                   2127: : does2
                   2128: DOES> ( ... -- ... )
                   2129:     ... ;
                   2131: : def-word ( ... -- ... )
                   2132:     create ...
                   2133:     IF
                   2134:        does1
                   2135:     ELSE
                   2136:        does2
                   2137:     ENDIF ;
                   2138: @end example
                   2140: In a standard program you can apply a @code{DOES>}-part only if the last
                   2141: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
                   2142: will override the behaviour of the last word defined in any case. In a
                   2143: standard program, you can use @code{DOES>} only in a colon
                   2144: definition. In Gforth, you can also use it in interpretation state, in a
                   2145: kind of one-shot mode:
                   2146: @example
                   2147: CREATE name ( ... -- ... )
                   2148:   @var{initialization}
                   2149: DOES>
                   2150:   @var{code} ;
                   2151: @end example
                   2152: This is equivalwent to the standard
                   2153: @example
                   2154: :noname
                   2155: DOES>
                   2156:     @var{code} ;
                   2157: CREATE name EXECUTE ( ... -- ... )
                   2158:     @var{initialization}
                   2159: @end example
                   2161: You can get the address of the body of a word with
                   2163: doc->body
                   2165: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
                   2166: @subsection Supplying names for the defined words
                   2168: By default, defining words take the names for the defined words from the
                   2169: input stream. Sometimes you want to supply the name from a string. You
                   2170: can do this with
                   2172: doc-nextname
                   2174: E.g.,
                   2176: @example
                   2177: s" foo" nextname create
                   2178: @end example
                   2179: is equivalent to
                   2180: @example
                   2181: create foo
                   2182: @end example
                   2184: Sometimes you want to define a word without a name. You can do this with
                   2186: doc-noname
                   2188: To make any use of the newly defined word, you need its execution
                   2189: token. You can get it with
                   2191: doc-lastxt
                   2193: E.g., you can initialize a deferred word with an anonymous colon
                   2194: definition:
                   2195: @example
                   2196: Defer deferred
                   2197: noname : ( ... -- ... )
                   2198:   ... ;
                   2199: lastxt IS deferred
                   2200: @end example
                   2202: @code{lastxt} also works when the last word was not defined as
                   2203: @code{noname}. 
                   2205: The standard has also recognized the need for anonymous words and
                   2206: provides
                   2208: doc-:noname
                   2210: This leaves the execution token for the word on the stack after the
                   2211: closing @code{;}. You can rewrite the last example with @code{:noname}:
                   2212: @example
                   2213: Defer deferred
                   2214: :noname ( ... -- ... )
                   2215:   ... ;
                   2216: IS deferred
                   2217: @end example
                   2219: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
                   2220: @subsection Interpretation and Compilation Semantics
1.36      anton    2222: The @dfn{interpretation semantics} of a word are what the text
                   2223: interpreter does when it encounters the word in interpret state. It also
                   2224: appears in some other contexts, e.g., the execution token returned by
                   2225: @code{' @var{word}} identifies the interpretation semantics of
                   2226: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
                   2227: interpret-state text interpretation of @code{@var{word}}).
                   2229: The @dfn{compilation semantics} of a word are what the text interpreter
                   2230: does when it encounters the word in compile state. It also appears in
                   2231: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
                   2232: standard terminology, ``appends to the current definition''.} the
                   2233: compilation semantics of @var{word}.
                   2235: The standard also talks about @dfn{execution semantics}. They are used
                   2236: only for defining the interpretation and compilation semantics of many
                   2237: words. By default, the interpretation semantics of a word are to
                   2238: @code{execute} its execution semantics, and the compilation semantics of
                   2239: a word are to @code{compile,} its execution semantics.@footnote{In
                   2240: standard terminology: The default interpretation semantics are its
                   2241: execution semantics; the default compilation semantics are to append its
                   2242: execution semantics to the execution semantics of the current
                   2243: definition.}
                   2245: You can change the compilation semantics into @code{execute}ing the
                   2246: execution semantics with
1.35      anton    2248: doc-immediate
1.36      anton    2249: 
                   2250: You can remove the interpretation semantics of a word with
                   2252: doc-compile-only
                   2253: doc-restrict
                   2255: Note that ticking (@code{'}) compile-only words gives an error
                   2256: (``Interpreting a compile-only word'').
                   2258: Gforth also allows you to define words with arbitrary combinations of
                   2259: interpretation and compilation semantics.
1.35      anton    2261: doc-interpret/compile:
1.36      anton    2263: This feature was introduced for implementing @code{TO} and @code{S"}. I
                   2264: recommend that you do not define such words, as cute as they may be:
                   2265: they make it hard to get at both parts of the word in some contexts.
                   2266: E.g., assume you want to get an execution token for the compilation
                   2267: part. Instead, define two words, one that embodies the interpretation
                   2268: part, and one that embodies the compilation part.
                   2270: There is, however, a potentially useful application of this feature:
                   2271: Providing differing implementations for the default semantics. While
                   2272: this introduces redundancy and is therefore usually a bad idea, a
                   2273: performance improvement may be worth the trouble. E.g., consider the
                   2274: word @code{foobar}:
                   2276: @example
                   2277: : foobar
                   2278:     foo bar ;
                   2279: @end example
                   2281: Let us assume that @code{foobar} is called so frequently that the
                   2282: calling overhead would take a significant amount of the run-time. We can
                   2283: optimize it with @code{interpret/compile:}:
1.35      anton    2284: 
1.36      anton    2285: @example
                   2286: :noname
                   2287:    foo bar ;
                   2288: :noname
                   2289:    POSTPONE foo POSTPONE bar ;
                   2290: interpret/compile: foobar
                   2291: @end example
                   2293: This definition has the same interpretation semantics and essentially
                   2294: the same compilation semantics as the simple definition of
                   2295: @code{foobar}, but the implementation of the compilation semantics is
                   2296: more efficient with respect to run-time.
                   2298: Some people try to use state-smart words to emulate the feature provided
                   2299: by @code{interpret/compile:} (words are state-smart if they check
                   2300: @code{STATE} during execution). E.g., they would try to code
                   2301: @code{foobar} like this:
                   2303: @example
                   2304: : foobar
                   2305:   STATE @@
                   2306:   IF ( compilation state )
                   2307:     POSTPONE foo POSTPONE bar
                   2308:   ELSE
                   2309:     foo bar
                   2310:   ENDIF ; immediate
                   2311: @end example
                   2313: While this works if @code{foobar} is processed only by the text
                   2314: interpreter, it does not work in other contexts (like @code{'} or
                   2315: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
                   2316: for a state-smart word, not for the interpretation semantics of the
                   2317: original @code{foobar}; when you execute this execution token (directly
                   2318: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
                   2319: state, the result will not be what you expected (i.e., it will not
                   2320: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
                   2321: write them!
                   2323: It is also possible to write defining words that define words with
                   2324: arbitrary combinations of interpretation and compilation semantics (or,
                   2325: preferably, arbitrary combinations of implementations of the default
                   2326: semantics). In general, this looks like:
                   2328: @example
                   2329: : def-word
                   2330:     create-interpret/compile
                   2331:     @var{code1}
                   2332: interpretation>
                   2333:     @var{code2}
                   2334: <interpretation
                   2335: compilation>
                   2336:     @var{code3}
                   2337: <compilation ;
                   2338: @end example
                   2340: For a @var{word} defined with @code{def-word}, the interpretation
                   2341: semantics are to push the address of the body of @var{word} and perform
                   2342: @var{code2}, and the compilation semantics are to push the address of
                   2343: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
                   2344: can also be defined like this:
                   2346: @example
                   2347: : constant ( n "name" -- )
                   2348:     create-interpret/compile
                   2349:     ,
                   2350: interpretation> ( -- n )
                   2351:     @@
                   2352: <interpretation
                   2353: compilation> ( compilation. -- ; run-time. -- n )
                   2354:     @@ postpone literal
                   2355: <compilation ;
                   2356: @end example
                   2358: doc-create-interpret/compile
                   2359: doc-interpretation>
                   2360: doc-<interpretation
                   2361: doc-compilation>
                   2362: doc-<compilation
                   2364: Note that words defined with @code{interpret/compile:} and
                   2365: @code{create-interpret/compile} have an extended header structure that
                   2366: differs from other words; however, unless you try to access them with
                   2367: plain address arithmetic, you should not notice this. Words for
                   2368: accessing the header structure usually know how to deal with this; e.g.,
                   2369: @code{' word >body} also gives you the body of a word created with
                   2370: @code{create-interpret/compile}.
1.4       anton    2371: 
1.37      anton    2372: @node Tokens for Words, Wordlists, Defining Words, Words
                   2373: @section Tokens for Words
                   2375: This chapter describes the creation and use of tokens that represent
                   2376: words on the stack (and in data space).
                   2378: Named words have interpretation and compilation semantics. Unnamed words
                   2379: just have execution semantics.
                   2381: An @dfn{execution token} represents the execution semantics of an
                   2382: unnamed word. An execution token occupies one cell. As explained in
                   2383: section @ref{Supplying names}, the execution token of the last words
                   2384: defined can be produced with
                   2386: short-lastxt
                   2388: You can perform the semantics represented by an execution token with
                   2389: doc-execute
                   2390: You can compile the word with
                   2391: doc-compile,
                   2393: In Gforth, the abstract data type @emph{execution token} is implemented
                   2394: as CFA (code field address).
                   2396: The interpretation semantics of a named word are also represented by an
                   2397: execution token. You can get it with
                   2399: doc-[']
                   2400: doc-'
                   2402: For literals, you use @code{'} in interpreted code and @code{[']} in
                   2403: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
                   2404: by complaining about compile-only words. To get an execution token for a
                   2405: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
                   2406: @var{X} drop}.
                   2408: The compilation semantics are represented by a @dfn{compilation token}
                   2409: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
                   2410: execution token. The compilation semantics represented by the
                   2411: compilation token can be performed with @code{execute}, which consumes
                   2412: the whole compilation token, with an additional stack effect determined
                   2413: by the represented compilation semantics.
                   2415: doc-[comp']
                   2416: doc-comp'
1.38      anton    2418: You can compile the compilation semantics with @code{postpone,}. I.e.,
                   2419: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
                   2420: @var{word}}.
                   2422: doc-postpone,
1.37      anton    2424: At present, the @var{w} part of a compilation token is an execution
                   2425: token, and the @var{xt} part represents either @code{execute} or
                   2426: @code{compile,}. However, don't rely on that kowledge, unless necessary;
                   2427: we may introduce unusual compilation tokens in the future (e.g.,
                   2428: compilation tokens representing the compilation semantics of literals).
                   2430: Named words are also represented by the @dfn{name token}. The abstract
                   2431: data type @emph{name token} is implemented as NFA (name field address).
                   2433: doc-find-name
                   2434: doc-name>int
                   2435: doc-name?int
                   2436: doc-name>comp
                   2437: doc-name>string
                   2439: @node Wordlists, Files, Tokens for Words, Words
1.4       anton    2440: @section Wordlists
                   2442: @node Files, Blocks, Wordlists, Words
                   2443: @section Files
                   2445: @node Blocks, Other I/O, Files, Words
                   2446: @section Blocks
                   2448: @node Other I/O, Programming Tools, Blocks, Words
                   2449: @section Other I/O
1.18      anton    2451: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4       anton    2452: @section Programming Tools
1.5       anton    2454: @menu
                   2455: * Debugging::                   Simple and quick.
                   2456: * Assertions::                  Making your programs self-checking.
                   2457: @end menu
                   2459: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4       anton    2460: @subsection Debugging
                   2462: The simple debugging aids provided in @file{debugging.fs}
                   2463: are meant to support a different style of debugging than the
                   2464: tracing/stepping debuggers used in languages with long turn-around
                   2465: times.
                   2467: A much better (faster) way in fast-compilig languages is to add
                   2468: printing code at well-selected places, let the program run, look at
                   2469: the output, see where things went wrong, add more printing code, etc.,
                   2470: until the bug is found.
                   2472: The word @code{~~} is easy to insert. It just prints debugging
                   2473: information (by default the source location and the stack contents). It
                   2474: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
                   2475: query-replace them with nothing). The deferred words
                   2476: @code{printdebugdata} and @code{printdebugline} control the output of
                   2477: @code{~~}. The default source location output format works well with
                   2478: Emacs' compilation mode, so you can step through the program at the
1.5       anton    2479: source level using @kbd{C-x `} (the advantage over a stepping debugger
                   2480: is that you can step in any direction and you know where the crash has
                   2481: happened or where the strange data has occurred).
1.4       anton    2482: 
                   2483: Note that the default actions clobber the contents of the pictured
                   2484: numeric output string, so you should not use @code{~~}, e.g., between
                   2485: @code{<#} and @code{#>}.
                   2487: doc-~~
                   2488: doc-printdebugdata
                   2489: doc-printdebugline
1.5       anton    2491: @node Assertions,  , Debugging, Programming Tools
1.4       anton    2492: @subsection Assertions
1.5       anton    2494: It is a good idea to make your programs self-checking, in particular, if
                   2495: you use an assumption (e.g., that a certain field of a data structure is
1.17      anton    2496: never zero) that may become wrong during maintenance. Gforth supports
1.5       anton    2497: assertions for this purpose. They are used like this:
                   2499: @example
                   2500: assert( @var{flag} )
                   2501: @end example
                   2503: The code between @code{assert(} and @code{)} should compute a flag, that
                   2504: should be true if everything is alright and false otherwise. It should
                   2505: not change anything else on the stack. The overall stack effect of the
                   2506: assertion is @code{( -- )}. E.g.
                   2508: @example
                   2509: assert( 1 1 + 2 = ) \ what we learn in school
                   2510: assert( dup 0<> ) \ assert that the top of stack is not zero
                   2511: assert( false ) \ this code should not be reached
                   2512: @end example
                   2514: The need for assertions is different at different times. During
                   2515: debugging, we want more checking, in production we sometimes care more
                   2516: for speed. Therefore, assertions can be turned off, i.e., the assertion
                   2517: becomes a comment. Depending on the importance of an assertion and the
                   2518: time it takes to check it, you may want to turn off some assertions and
1.17      anton    2519: keep others turned on. Gforth provides several levels of assertions for
1.5       anton    2520: this purpose:
                   2522: doc-assert0(
                   2523: doc-assert1(
                   2524: doc-assert2(
                   2525: doc-assert3(
                   2526: doc-assert(
                   2527: doc-)
                   2529: @code{Assert(} is the same as @code{assert1(}. The variable
                   2530: @code{assert-level} specifies the highest assertions that are turned
                   2531: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
                   2532: @code{assert1(} assertions perform checking, while @code{assert2(} and
                   2533: @code{assert3(} assertions are treated as comments.
                   2535: Note that the @code{assert-level} is evaluated at compile-time, not at
                   2536: run-time. I.e., you cannot turn assertions on or off at run-time, you
                   2537: have to set the @code{assert-level} appropriately before compiling a
                   2538: piece of code. You can compile several pieces of code at several
                   2539: @code{assert-level}s (e.g., a trusted library at level 1 and newly
                   2540: written code at level 3).
                   2542: doc-assert-level
                   2544: If an assertion fails, a message compatible with Emacs' compilation mode
                   2545: is produced and the execution is aborted (currently with @code{ABORT"}.
                   2546: If there is interest, we will introduce a special throw code. But if you
                   2547: intend to @code{catch} a specific condition, using @code{throw} is
                   2548: probably more appropriate than an assertion).
1.18      anton    2550: @node Assembler and Code words, Threading Words, Programming Tools, Words
                   2551: @section Assembler and Code words
                   2553: Gforth provides some words for defining primitives (words written in
                   2554: machine code), and for defining the the machine-code equivalent of
                   2555: @code{DOES>}-based defining words. However, the machine-independent
1.40    ! anton    2556: nature of Gforth poses a few problems: First of all, Gforth runs on
1.18      anton    2557: several architectures, so it can provide no standard assembler. What's
                   2558: worse is that the register allocation not only depends on the processor,
1.25      anton    2559: but also on the @code{gcc} version and options used.
1.18      anton    2560: 
1.25      anton    2561: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18      anton    2562: header structure), so a system-independent assembler may be used in
                   2563: Gforth. If you do not have an assembler, you can compile machine code
                   2564: directly with @code{,} and @code{c,}.
                   2566: doc-assembler
                   2567: doc-code
                   2568: doc-end-code
                   2569: doc-;code
                   2570: doc-flush-icache
                   2572: If @code{flush-icache} does not work correctly, @code{code} words
                   2573: etc. will not work (reliably), either.
                   2575: These words are rarely used. Therefore they reside in @code{code.fs},
                   2576: which is usually not loaded (except @code{flush-icache}, which is always
1.19      anton    2577: present). You can load them with @code{require code.fs}.
1.18      anton    2578: 
1.25      anton    2579: In the assembly code you will want to refer to the inner interpreter's
                   2580: registers (e.g., the data stack pointer) and you may want to use other
                   2581: registers for temporary storage. Unfortunately, the register allocation
                   2582: is installation-dependent.
                   2584: The easiest solution is to use explicit register declarations
                   2585: (@pxref{Explicit Reg Vars, , Variables in Specified Registers,,
                   2586: GNU C Manual}) for all of the inner interpreter's registers: You have to
                   2587: compile Gforth with @code{-DFORCE_REG} (configure option
                   2588: @code{--enable-force-reg}) and the appropriate declarations must be
                   2589: present in the @code{machine.h} file (see @code{mips.h} for an example;
                   2590: you can find a full list of all declarable register symbols with
                   2591: @code{grep register engine.c}). If you give explicit registers to all
                   2592: variables that are declared at the beginning of @code{engine()}, you
                   2593: should be able to use the other caller-saved registers for temporary
                   2594: storage. Alternatively, you can use the @code{gcc} option
                   2595: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
                   2596: Generation Conventions,, GNU C Manual}) to reserve a register
                   2597: (however, this restriction on register allocation may slow Gforth
                   2598: significantly).
                   2600: If this solution is not viable (e.g., because @code{gcc} does not allow
                   2601: you to explicitly declare all the registers you need), you have to find
                   2602: out by looking at the code where the inner interpreter's registers
                   2603: reside and which registers can be used for temporary storage. You can
                   2604: get an assembly listing of the engine's code with @code{make engine.s}.
                   2606: In any case, it is good practice to abstract your assembly code from the
                   2607: actual register allocation. E.g., if the data stack pointer resides in
                   2608: register @code{$17}, create an alias for this register called @code{sp},
                   2609: and use that in your assembly code.
1.18      anton    2611: Another option for implementing normal and defining words efficiently
                   2612: is: adding the wanted functionality to the source of Gforth. For normal
1.35      anton    2613: words you just have to edit @file{primitives} (@pxref{Automatic
                   2614: Generation}), defining words (equivalent to @code{;CODE} words, for fast
                   2615: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
                   2616: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.18      anton    2617: 
                   2619: @node Threading Words,  , Assembler and Code words, Words
1.4       anton    2620: @section Threading Words
                   2622: These words provide access to code addresses and other threading stuff
1.17      anton    2623: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4       anton    2624: abstracts away the differences between direct and indirect threading
                   2625: (and, for direct threading, the machine dependences). However, at
                   2626: present this wordset is still inclomplete. It is also pretty low-level;
                   2627: some day it will hopefully be made unnecessary by an internals words set
                   2628: that abstracts implementation details away completely.
                   2630: doc->code-address
                   2631: doc->does-code
                   2632: doc-code-address!
                   2633: doc-does-code!
                   2634: doc-does-handler!
                   2635: doc-/does-handler
1.18      anton    2637: The code addresses produced by various defining words are produced by
                   2638: the following words:
1.14      anton    2639: 
1.18      anton    2640: doc-docol:
                   2641: doc-docon:
                   2642: doc-dovar:
                   2643: doc-douser:
                   2644: doc-dodefer:
                   2645: doc-dofield:
1.35      anton    2647: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
                   2648: with @code{>DOES-CODE}. If the word was defined in that way, the value
                   2649: returned is different from 0 and identifies the @code{DOES>} used by the
                   2650: defining word.
1.14      anton    2651: 
1.40    ! anton    2652: @node Tools, ANS conformance, Words, Top
        !          2653: @chapter Tools
        !          2654: 
        !          2655: @menu
        !          2656: * ANS Report::                  Report the words used, sorted by wordset
        !          2657: @end menu
        !          2658: 
        !          2659: See also @ref{Emacs and Gforth}.
        !          2660: 
        !          2661: @node ANS Report,  , Tools, Tools
        !          2662: @section @file{ans-report.fs}: Report the words used, sorted by wordset
        !          2663: 
        !          2664: If you want to label a Forth program as ANS Forth Program, you must
        !          2665: document which wordsets the program uses; for extension wordsets, it is
        !          2666: helpful to list the words the program requires from these wordsets
        !          2667: (because Forth systems are allowed to provide only some words of them).
        !          2668: 
        !          2669: The @file{ans-report.fs} tool makes it easy for you to determine which
        !          2670: words from which wordset and which non-ANS words your application
        !          2671: uses. You simply have to include @file{ans-report.fs} before loading the
        !          2672: program you want to check. After loading your program, you can get the
        !          2673: report with @code{print-ans-report}. A typical use is to run this as
        !          2674: batch job like this:
        !          2675: @example
        !          2676: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
        !          2677: @end example
        !          2678: 
        !          2679: The output looks like this (for @file{compat/control.fs}):
        !          2680: @example
        !          2681: The program uses the following words
        !          2682: from CORE :
        !          2683: : POSTPONE THEN ; immediate ?dup IF 0= 
        !          2684: from BLOCK-EXT :
        !          2685: \ 
        !          2686: from FILE :
        !          2687: ( 
        !          2688: @end example
        !          2689: 
        !          2690: @subsection Caveats
        !          2691: 
        !          2692: Note that @file{ans-report.fs} just checks which words are used, not whether
        !          2693: they are used in an ANS Forth conforming way!
        !          2694: 
        !          2695: Some words are defined in several wordsets in the
        !          2696: standard. @file{ans-report.fs} reports them for only one of the
        !          2697: wordsets, and not necessarily the one you expect. It depends on usage
        !          2698: which wordset is the right one to specify. E.g., if you only use the
        !          2699: compilation semantics of @code{S"}, it is a Core word; if you also use
        !          2700: its interpretation semantics, it is a File word.
        !          2701: 
        !          2702: 
        !          2703: @node ANS conformance, Model, Tools, Top
1.4       anton    2704: @chapter ANS conformance
1.17      anton    2706: To the best of our knowledge, Gforth is an
1.14      anton    2707: 
1.15      anton    2708: ANS Forth System
1.34      anton    2709: @itemize @bullet
1.15      anton    2710: @item providing the Core Extensions word set
                   2711: @item providing the Block word set
                   2712: @item providing the Block Extensions word set
                   2713: @item providing the Double-Number word set
                   2714: @item providing the Double-Number Extensions word set
                   2715: @item providing the Exception word set
                   2716: @item providing the Exception Extensions word set
                   2717: @item providing the Facility word set
                   2718: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
                   2719: @item providing the File Access word set
                   2720: @item providing the File Access Extensions word set
                   2721: @item providing the Floating-Point word set
                   2722: @item providing the Floating-Point Extensions word set
                   2723: @item providing the Locals word set
                   2724: @item providing the Locals Extensions word set
                   2725: @item providing the Memory-Allocation word set
                   2726: @item providing the Memory-Allocation Extensions word set (that one's easy)
                   2727: @item providing the Programming-Tools word set
1.34      anton    2728: @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
1.15      anton    2729: @item providing the Search-Order word set
                   2730: @item providing the Search-Order Extensions word set
                   2731: @item providing the String word set
                   2732: @item providing the String Extensions word set (another easy one)
                   2733: @end itemize
                   2735: In addition, ANS Forth systems are required to document certain
                   2736: implementation choices. This chapter tries to meet these
                   2737: requirements. In many cases it gives a way to ask the system for the
                   2738: information instead of providing the information directly, in
                   2739: particular, if the information depends on the processor, the operating
                   2740: system or the installation options chosen, or if they are likely to
1.17      anton    2741: change during the maintenance of Gforth.
1.15      anton    2742: 
1.14      anton    2743: @comment The framework for the rest has been taken from pfe.
                   2745: @menu
                   2746: * The Core Words::              
                   2747: * The optional Block word set::  
                   2748: * The optional Double Number word set::  
                   2749: * The optional Exception word set::  
                   2750: * The optional Facility word set::  
                   2751: * The optional File-Access word set::  
                   2752: * The optional Floating-Point word set::  
                   2753: * The optional Locals word set::  
                   2754: * The optional Memory-Allocation word set::  
                   2755: * The optional Programming-Tools word set::  
                   2756: * The optional Search-Order word set::  
                   2757: @end menu
                   2760: @c =====================================================================
                   2761: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
                   2762: @comment  node-name,  next,  previous,  up
                   2763: @section The Core Words
                   2764: @c =====================================================================
                   2766: @menu
1.15      anton    2767: * core-idef::                   Implementation Defined Options                   
                   2768: * core-ambcond::                Ambiguous Conditions                
                   2769: * core-other::                  Other System Documentation                  
1.14      anton    2770: @end menu
                   2772: @c ---------------------------------------------------------------------
                   2773: @node core-idef, core-ambcond, The Core Words, The Core Words
                   2774: @subsection Implementation Defined Options
                   2775: @c ---------------------------------------------------------------------
                   2777: @table @i
                   2779: @item (Cell) aligned addresses:
1.17      anton    2780: processor-dependent. Gforth's alignment words perform natural alignment
1.14      anton    2781: (e.g., an address aligned for a datum of size 8 is divisible by
                   2782: 8). Unaligned accesses usually result in a @code{-23 THROW}.
                   2784: @item @code{EMIT} and non-graphic characters:
                   2785: The character is output using the C library function (actually, macro)
1.36      anton    2786: @code{putc}.
1.14      anton    2787: 
                   2788: @item character editing of @code{ACCEPT} and @code{EXPECT}:
                   2789: This is modeled on the GNU readline library (@pxref{Readline
                   2790: Interaction, , Command Line Editing, readline, The GNU Readline
                   2791: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
                   2792: producing a full word completion every time you type it (instead of
                   2793: producing the common prefix of all completions).
                   2795: @item character set:
                   2796: The character set of your computer and display device. Gforth is
                   2797: 8-bit-clean (but some other component in your system may make trouble).
                   2799: @item Character-aligned address requirements:
                   2800: installation-dependent. Currently a character is represented by a C
                   2801: @code{unsigned char}; in the future we might switch to @code{wchar_t}
                   2802: (Comments on that requested).
                   2804: @item character-set extensions and matching of names:
1.17      anton    2805: Any character except the ASCII NUL charcter can be used in a
1.36      anton    2806: name. Matching is case-insensitive (except in @code{TABLE}s. The
                   2807: matching is performed using the C function @code{strncasecmp}, whose
                   2808: function is probably influenced by the locale. E.g., the @code{C} locale
                   2809: does not know about accents and umlauts, so they are matched
                   2810: case-sensitively in that locale. For portability reasons it is best to
                   2811: write programs such that they work in the @code{C} locale. Then one can
                   2812: use libraries written by a Polish programmer (who might use words
                   2813: containing ISO Latin-2 encoded characters) and by a French programmer
                   2814: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
                   2815: funny results for some of the words (which ones, depends on the font you
                   2816: are using)). Also, the locale you prefer may not be available in other
                   2817: operating systems. Hopefully, Unicode will solve these problems one day.
1.14      anton    2818: 
                   2819: @item conditions under which control characters match a space delimiter:
                   2820: If @code{WORD} is called with the space character as a delimiter, all
                   2821: white-space characters (as identified by the C macro @code{isspace()})
                   2822: are delimiters. @code{PARSE}, on the other hand, treats space like other
                   2823: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
                   2824: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
                   2825: interpreter (aka text interpreter) by default, treats all white-space
                   2826: characters as delimiters.
                   2828: @item format of the control flow stack:
                   2829: The data stack is used as control flow stack. The size of a control flow
                   2830: stack item in cells is given by the constant @code{cs-item-size}. At the
                   2831: time of this writing, an item consists of a (pointer to a) locals list
                   2832: (third), an address in the code (second), and a tag for identifying the
                   2833: item (TOS). The following tags are used: @code{defstart},
                   2834: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
                   2835: @code{scopestart}.
                   2837: @item conversion of digits > 35
                   2838: The characters @code{[\]^_'} are the digits with the decimal value
                   2839: 36@minus{}41. There is no way to input many of the larger digits.
                   2841: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
                   2842: The cursor is moved to the end of the entered string. If the input is
                   2843: terminated using the @kbd{Return} key, a space is typed.
                   2845: @item exception abort sequence of @code{ABORT"}:
                   2846: The error string is stored into the variable @code{"error} and a
                   2847: @code{-2 throw} is performed.
                   2849: @item input line terminator:
1.36      anton    2850: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
                   2851: lines. One of these characters is typically produced when you type the
                   2852: @kbd{Enter} or @kbd{Return} key.
1.14      anton    2853: 
                   2854: @item maximum size of a counted string:
                   2855: @code{s" /counted-string" environment? drop .}. Currently 255 characters
                   2856: on all ports, but this may change.
                   2858: @item maximum size of a parsed string:
                   2859: Given by the constant @code{/line}. Currently 255 characters.
                   2861: @item maximum size of a definition name, in characters:
                   2862: 31
                   2864: @item maximum string length for @code{ENVIRONMENT?}, in characters:
                   2865: 31
                   2867: @item method of selecting the user input device:
1.17      anton    2868: The user input device is the standard input. There is currently no way to
                   2869: change it from within Gforth. However, the input can typically be
                   2870: redirected in the command line that starts Gforth.
1.14      anton    2871: 
                   2872: @item method of selecting the user output device:
1.36      anton    2873: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
                   2874: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
                   2875: output, so output on a terminal does not become visible before the next
                   2876: newline or buffer overflow. Output on non-terminals is invisible until
                   2877: the buffer overflows.
1.14      anton    2878: 
                   2879: @item methods of dictionary compilation:
1.17      anton    2880: What are we expected to document here?
1.14      anton    2881: 
                   2882: @item number of bits in one address unit:
                   2883: @code{s" address-units-bits" environment? drop .}. 8 in all current
                   2884: ports.
                   2886: @item number representation and arithmetic:
                   2887: Processor-dependent. Binary two's complement on all current ports.
                   2889: @item ranges for integer types:
                   2890: Installation-dependent. Make environmental queries for @code{MAX-N},
                   2891: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
                   2892: unsigned (and positive) types is 0. The lower bound for signed types on
                   2893: two's complement and one's complement machines machines can be computed
                   2894: by adding 1 to the upper bound.
                   2896: @item read-only data space regions:
                   2897: The whole Forth data space is writable.
                   2899: @item size of buffer at @code{WORD}:
                   2900: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   2901: shared with the pictured numeric output string. If overwriting
                   2902: @code{PAD} is acceptable, it is as large as the remaining dictionary
                   2903: space, although only as much can be sensibly used as fits in a counted
                   2904: string.
                   2906: @item size of one cell in address units:
                   2907: @code{1 cells .}.
                   2909: @item size of one character in address units:
                   2910: @code{1 chars .}. 1 on all current ports.
                   2912: @item size of the keyboard terminal buffer:
1.36      anton    2913: Varies. You can determine the size at a specific time using @code{lp@@
1.14      anton    2914: tib - .}. It is shared with the locals stack and TIBs of files that
                   2915: include the current file. You can change the amount of space for TIBs
1.17      anton    2916: and locals stack at Gforth startup with the command line option
1.14      anton    2917: @code{-l}.
                   2919: @item size of the pictured numeric output buffer:
                   2920: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   2921: shared with @code{WORD}.
                   2923: @item size of the scratch area returned by @code{PAD}:
                   2924: The remainder of dictionary space. You can even use the unused part of
1.36      anton    2925: the data stack space. The current size can be computed with @code{sp@@
1.14      anton    2926: pad - .}.
                   2928: @item system case-sensitivity characteristics:
1.36      anton    2929: Dictionary searches are case insensitive (except in
                   2930: @code{TABLE}s). However, as explained above under @i{character-set
                   2931: extensions}, the matching for non-ASCII characters is determined by the
                   2932: locale you are using. In the default @code{C} locale all non-ASCII
                   2933: characters are matched case-sensitively.
1.14      anton    2934: 
                   2935: @item system prompt:
                   2936: @code{ ok} in interpret state, @code{ compiled} in compile state.
                   2938: @item division rounding:
                   2939: installation dependent. @code{s" floored" environment? drop .}. We leave
1.25      anton    2940: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
1.14      anton    2941: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
                   2943: @item values of @code{STATE} when true:
                   2944: -1.
                   2946: @item values returned after arithmetic overflow:
                   2947: On two's complement machines, arithmetic is performed modulo
                   2948: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   2949: arithmetic (with appropriate mapping for signed types). Division by zero
1.36      anton    2950: typically results in a @code{-55 throw} (Floating-point unidentified
1.14      anton    2951: fault), although a @code{-10 throw} (divide by zero) would be more
                   2952: appropriate.
                   2954: @item whether the current definition can be found after @t{DOES>}:
                   2955: No.
                   2957: @end table
                   2959: @c ---------------------------------------------------------------------
                   2960: @node core-ambcond, core-other, core-idef, The Core Words
                   2961: @subsection Ambiguous conditions
                   2962: @c ---------------------------------------------------------------------
                   2964: @table @i
                   2966: @item a name is neither a word nor a number:
1.36      anton    2967: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
                   2968: preserves the data and FP stack, so you don't lose more work than
                   2969: necessary.
1.14      anton    2970: 
                   2971: @item a definition name exceeds the maximum length allowed:
                   2972: @code{-19 throw} (Word name too long)
                   2974: @item addressing a region not inside the various data spaces of the forth system:
                   2975: The stacks, code space and name space are accessible. Machine code space is
                   2976: typically readable. Accessing other addresses gives results dependent on
                   2977: the operating system. On decent systems: @code{-9 throw} (Invalid memory
                   2978: address).
                   2980: @item argument type incompatible with parameter:
                   2981: This is usually not caught. Some words perform checks, e.g., the control
                   2982: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
                   2983: mismatch).
                   2985: @item attempting to obtain the execution token of a word with undefined execution semantics:
1.36      anton    2986: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
                   2987: get an execution token for @code{compile-only-error} (which performs a
                   2988: @code{-14 throw} when executed).
1.14      anton    2989: 
                   2990: @item dividing by zero:
                   2991: typically results in a @code{-55 throw} (floating point unidentified
                   2992: fault), although a @code{-10 throw} (divide by zero) would be more
                   2993: appropriate.
                   2995: @item insufficient data stack or return stack space:
                   2996: Not checked. This typically results in mysterious illegal memory
                   2997: accesses, producing @code{-9 throw} (Invalid memory address) or
                   2998: @code{-23 throw} (Address alignment exception).
                   3000: @item insufficient space for loop control parameters:
                   3001: like other return stack overflows.
                   3003: @item insufficient space in the dictionary:
                   3004: Not checked. Similar results as stack overflows. However, typically the
                   3005: error appears at a different place when one inserts or removes code.
                   3007: @item interpreting a word with undefined interpretation semantics:
                   3008: For some words, we defined interpretation semantics. For the others:
1.36      anton    3009: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    3010: 
                   3011: @item modifying the contents of the input buffer or a string literal:
                   3012: These are located in writable memory and can be modified.
                   3014: @item overflow of the pictured numeric output string:
                   3015: Not checked.
                   3017: @item parsed string overflow:
                   3018: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
                   3020: @item producing a result out of range:
                   3021: On two's complement machines, arithmetic is performed modulo
                   3022: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   3023: arithmetic (with appropriate mapping for signed types). Division by zero
                   3024: typically results in a @code{-55 throw} (floatingpoint unidentified
                   3025: fault), although a @code{-10 throw} (divide by zero) would be more
                   3026: appropriate. @code{convert} and @code{>number} currently overflow
                   3027: silently.
                   3029: @item reading from an empty data or return stack:
                   3030: The data stack is checked by the outer (aka text) interpreter after
                   3031: every word executed. If it has underflowed, a @code{-4 throw} (Stack
                   3032: underflow) is performed. Apart from that, the stacks are not checked and
                   3033: underflows can result in similar behaviour as overflows (of adjacent
                   3034: stacks).
1.36      anton    3036: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
1.14      anton    3037: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
                   3038: use zero-length string as a name). Words like @code{'} probably will not
                   3039: find what they search. Note that it is possible to create zero-length
                   3040: names with @code{nextname} (should it not?).
                   3042: @item @code{>IN} greater than input buffer:
                   3043: The next invocation of a parsing word returns a string wih length 0.
                   3045: @item @code{RECURSE} appears after @code{DOES>}:
1.36      anton    3046: Compiles a recursive call to the defining word, not to the defined word.
1.14      anton    3047: 
                   3048: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.27      anton    3049: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
                   3050: the end of the file was reached), its source-id may be
                   3051: reused. Therefore, restoring an input source specification referencing a
                   3052: closed file may lead to unpredictable results instead of a @code{-12
                   3053: THROW}.
1.36      anton    3055: In the future, Gforth may be able to restore input source specifications
1.27      anton    3056: from other than the current input soruce.
1.14      anton    3057: 
                   3058: @item data space containing definitions gets de-allocated:
                   3059: Deallocation with @code{allot} is not checked. This typically resuls in
                   3060: memory access faults or execution of illegal instructions.
                   3062: @item data space read/write with incorrect alignment:
                   3063: Processor-dependent. Typically results in a @code{-23 throw} (Address
                   3064: alignment exception). Under Linux on a 486 or later processor with
                   3065: alignment turned on, incorrect alignment results in a @code{-9 throw}
                   3066: (Invalid memory address). There are reportedly some processors with
                   3067: alignment restrictions that do not report them.
                   3069: @item data space pointer not properly aligned, @code{,}, @code{C,}:
                   3070: Like other alignment errors.
                   3072: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
                   3073: Not checked. May cause an illegal memory access.
                   3075: @item loop control parameters not available:
                   3076: Not checked. The counted loop words simply assume that the top of return
                   3077: stack items are loop control parameters and behave accordingly.
                   3079: @item most recent definition does not have a name (@code{IMMEDIATE}):
                   3080: @code{abort" last word was headerless"}.
                   3082: @item name not defined by @code{VALUE} used by @code{TO}:
1.36      anton    3083: @code{-32 throw} (Invalid name argument) (unless name was defined by
                   3084: @code{CONSTANT}; then it just changes the constant).
1.14      anton    3085: 
1.15      anton    3086: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.14      anton    3087: @code{-13 throw} (Undefined word)
                   3089: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
                   3090: Gforth behaves as if they were of the same type. I.e., you can predict
                   3091: the behaviour by interpreting all parameters as, e.g., signed.
                   3093: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
1.36      anton    3094: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
                   3095: compilation semantics of @code{TO}.
1.14      anton    3096: 
                   3097: @item String longer than a counted string returned by @code{WORD}:
                   3098: Not checked. The string will be ok, but the count will, of course,
                   3099: contain only the least significant bits of the length.
1.15      anton    3101: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.14      anton    3102: Processor-dependent. Typical behaviours are returning 0 and using only
                   3103: the low bits of the shift count.
                   3105: @item word not defined via @code{CREATE}:
                   3106: @code{>BODY} produces the PFA of the word no matter how it was defined.
                   3108: @code{DOES>} changes the execution semantics of the last defined word no
                   3109: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
                   3110: @code{CREATE , DOES>}.
                   3112: @item words improperly used outside @code{<#} and @code{#>}:
                   3113: Not checked. As usual, you can expect memory faults.
                   3115: @end table
                   3118: @c ---------------------------------------------------------------------
                   3119: @node core-other,  , core-ambcond, The Core Words
                   3120: @subsection Other system documentation
                   3121: @c ---------------------------------------------------------------------
                   3123: @table @i
                   3125: @item nonstandard words using @code{PAD}:
                   3126: None.
                   3128: @item operator's terminal facilities available:
1.26      anton    3129: After processing the command line, Gforth goes into interactive mode,
                   3130: and you can give commands to Gforth interactively. The actual facilities
                   3131: available depend on how you invoke Gforth.
1.14      anton    3132: 
                   3133: @item program data space available:
1.36      anton    3134: @code{sp@@ here - .} gives the space remaining for dictionary and data
1.14      anton    3135: stack together.
                   3137: @item return stack space available:
1.26      anton    3138: By default 16 KBytes. The default can be overridden with the @code{-r}
                   3139: switch (@pxref{Invocation}) when Gforth starts up.
1.14      anton    3140: 
                   3141: @item stack space available:
1.36      anton    3142: @code{sp@@ here - .} gives the space remaining for dictionary and data
1.14      anton    3143: stack together.
                   3145: @item system dictionary space required, in address units:
                   3146: Type @code{here forthstart - .} after startup. At the time of this
                   3147: writing, this gives 70108 (bytes) on a 32-bit system.
                   3148: @end table
                   3151: @c =====================================================================
                   3152: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
                   3153: @section The optional Block word set
                   3154: @c =====================================================================
                   3156: @menu
1.15      anton    3157: * block-idef::                  Implementation Defined Options                  
                   3158: * block-ambcond::               Ambiguous Conditions               
                   3159: * block-other::                 Other System Documentation                 
1.14      anton    3160: @end menu
                   3163: @c ---------------------------------------------------------------------
                   3164: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
                   3165: @subsection Implementation Defined Options
                   3166: @c ---------------------------------------------------------------------
                   3168: @table @i
                   3170: @item the format for display by @code{LIST}:
                   3171: First the screen number is displayed, then 16 lines of 64 characters,
                   3172: each line preceded by the line number.
                   3174: @item the length of a line affected by @code{\}:
                   3175: 64 characters.
                   3176: @end table
                   3179: @c ---------------------------------------------------------------------
                   3180: @node block-ambcond, block-other, block-idef, The optional Block word set
                   3181: @subsection Ambiguous conditions
                   3182: @c ---------------------------------------------------------------------
                   3184: @table @i
                   3186: @item correct block read was not possible:
                   3187: Typically results in a @code{throw} of some OS-derived value (between
                   3188: -512 and -2048). If the blocks file was just not long enough, blanks are
                   3189: supplied for the missing portion.
                   3191: @item I/O exception in block transfer:
                   3192: Typically results in a @code{throw} of some OS-derived value (between
                   3193: -512 and -2048).
                   3195: @item invalid block number:
                   3196: @code{-35 throw} (Invalid block number)
                   3198: @item a program directly alters the contents of @code{BLK}:
                   3199: The input stream is switched to that other block, at the same
                   3200: position. If the storing to @code{BLK} happens when interpreting
                   3201: non-block input, the system will get quite confused when the block ends.
                   3203: @item no current block buffer for @code{UPDATE}:
                   3204: @code{UPDATE} has no effect.
                   3206: @end table
                   3209: @c ---------------------------------------------------------------------
                   3210: @node block-other,  , block-ambcond, The optional Block word set
                   3211: @subsection Other system documentation
                   3212: @c ---------------------------------------------------------------------
                   3214: @table @i
                   3216: @item any restrictions a multiprogramming system places on the use of buffer addresses:
                   3217: No restrictions (yet).
                   3219: @item the number of blocks available for source and data:
                   3220: depends on your disk space.
                   3222: @end table
                   3225: @c =====================================================================
                   3226: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
                   3227: @section The optional Double Number word set
                   3228: @c =====================================================================
                   3230: @menu
1.15      anton    3231: * double-ambcond::              Ambiguous Conditions              
1.14      anton    3232: @end menu
                   3235: @c ---------------------------------------------------------------------
1.15      anton    3236: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
1.14      anton    3237: @subsection Ambiguous conditions
                   3238: @c ---------------------------------------------------------------------
                   3240: @table @i
1.15      anton    3242: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.14      anton    3243: The least significant cell of @var{d} is produced.
                   3245: @end table
                   3248: @c =====================================================================
                   3249: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
                   3250: @section The optional Exception word set
                   3251: @c =====================================================================
                   3253: @menu
1.15      anton    3254: * exception-idef::              Implementation Defined Options              
1.14      anton    3255: @end menu
                   3258: @c ---------------------------------------------------------------------
1.15      anton    3259: @node exception-idef,  , The optional Exception word set, The optional Exception word set
1.14      anton    3260: @subsection Implementation Defined Options
                   3261: @c ---------------------------------------------------------------------
                   3263: @table @i
                   3264: @item @code{THROW}-codes used in the system:
                   3265: The codes -256@minus{}-511 are used for reporting signals (see
                   3266: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
                   3267: (for file and memory allocation operations). The mapping from OS error
1.37      anton    3268: numbers to throw code is -512@minus{}@code{errno}. One side effect of
1.14      anton    3269: this mapping is that undefined OS errors produce a message with a
                   3270: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
                   3271: 488} on my system.
                   3272: @end table
                   3274: @c =====================================================================
                   3275: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
                   3276: @section The optional Facility word set
                   3277: @c =====================================================================
                   3279: @menu
1.15      anton    3280: * facility-idef::               Implementation Defined Options               
                   3281: * facility-ambcond::            Ambiguous Conditions            
1.14      anton    3282: @end menu
                   3285: @c ---------------------------------------------------------------------
                   3286: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
                   3287: @subsection Implementation Defined Options
                   3288: @c ---------------------------------------------------------------------
                   3290: @table @i
                   3292: @item encoding of keyboard events (@code{EKEY}):
                   3293: Not yet implemeted.
                   3295: @item duration of a system clock tick
                   3296: System dependent. With respect to @code{MS}, the time is specified in
                   3297: microseconds. How well the OS and the hardware implement this, is
                   3298: another question.
                   3300: @item repeatability to be expected from the execution of @code{MS}:
                   3301: System dependent. On Unix, a lot depends on load. If the system is
1.17      anton    3302: lightly loaded, and the delay is short enough that Gforth does not get
1.14      anton    3303: swapped out, the performance should be acceptable. Under MS-DOS and
                   3304: other single-tasking systems, it should be good.
                   3306: @end table
                   3309: @c ---------------------------------------------------------------------
1.15      anton    3310: @node facility-ambcond,  , facility-idef, The optional Facility word set
1.14      anton    3311: @subsection Ambiguous conditions
                   3312: @c ---------------------------------------------------------------------
                   3314: @table @i
                   3316: @item @code{AT-XY} can't be performed on user output device:
                   3317: Largely terminal dependant. No range checks are done on the arguments.
                   3318: No errors are reported. You may see some garbage appearing, you may see
                   3319: simply nothing happen.
                   3321: @end table
                   3324: @c =====================================================================
                   3325: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
                   3326: @section The optional File-Access word set
                   3327: @c =====================================================================
                   3329: @menu
1.15      anton    3330: * file-idef::                   Implementation Defined Options                   
                   3331: * file-ambcond::                Ambiguous Conditions                
1.14      anton    3332: @end menu
                   3335: @c ---------------------------------------------------------------------
                   3336: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
                   3337: @subsection Implementation Defined Options
                   3338: @c ---------------------------------------------------------------------
                   3340: @table @i
                   3342: @item File access methods used:
                   3343: @code{R/O}, @code{R/W} and @code{BIN} work as you would
                   3344: expect. @code{W/O} translates into the C file opening mode @code{w} (or
                   3345: @code{wb}): The file is cleared, if it exists, and created, if it does
1.15      anton    3346: not (both with @code{open-file} and @code{create-file}).  Under Unix
1.14      anton    3347: @code{create-file} creates a file with 666 permissions modified by your
                   3348: umask.
                   3350: @item file exceptions:
                   3351: The file words do not raise exceptions (except, perhaps, memory access
                   3352: faults when you pass illegal addresses or file-ids).
                   3354: @item file line terminator:
                   3355: System-dependent. Gforth uses C's newline character as line
                   3356: terminator. What the actual character code(s) of this are is
                   3357: system-dependent.
                   3359: @item file name format
                   3360: System dependent. Gforth just uses the file name format of your OS.
                   3362: @item information returned by @code{FILE-STATUS}:
                   3363: @code{FILE-STATUS} returns the most powerful file access mode allowed
                   3364: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
                   3365: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
                   3366: along with the retured mode.
                   3368: @item input file state after an exception when including source:
                   3369: All files that are left via the exception are closed.
                   3371: @item @var{ior} values and meaning:
1.15      anton    3372: The @var{ior}s returned by the file and memory allocation words are
                   3373: intended as throw codes. They typically are in the range
                   3374: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3375: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3376: 
                   3377: @item maximum depth of file input nesting:
                   3378: limited by the amount of return stack, locals/TIB stack, and the number
                   3379: of open files available. This should not give you troubles.
                   3381: @item maximum size of input line:
                   3382: @code{/line}. Currently 255.
                   3384: @item methods of mapping block ranges to files:
1.37      anton    3385: By default, blocks are accessed in the file @file{blocks.fb} in the
                   3386: current working directory. The file can be switched with @code{USE}.
1.14      anton    3387: 
                   3388: @item number of string buffers provided by @code{S"}:
                   3389: 1
                   3391: @item size of string buffer used by @code{S"}:
                   3392: @code{/line}. currently 255.
                   3394: @end table
                   3396: @c ---------------------------------------------------------------------
1.15      anton    3397: @node file-ambcond,  , file-idef, The optional File-Access word set
1.14      anton    3398: @subsection Ambiguous conditions
                   3399: @c ---------------------------------------------------------------------
                   3401: @table @i
                   3403: @item attempting to position a file outside it's boundaries:
                   3404: @code{REPOSITION-FILE} is performed as usual: Afterwards,
                   3405: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
                   3407: @item attempting to read from file positions not yet written:
                   3408: End-of-file, i.e., zero characters are read and no error is reported.
                   3410: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
                   3411: An appropriate exception may be thrown, but a memory fault or other
                   3412: problem is more probable.
                   3414: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
                   3415: The @var{ior} produced by the operation, that discovered the problem, is
                   3416: thrown.
                   3418: @item named file cannot be opened (@code{included}):
                   3419: The @var{ior} produced by @code{open-file} is thrown.
                   3421: @item requesting an unmapped block number:
                   3422: There are no unmapped legal block numbers. On some operating systems,
                   3423: writing a block with a large number may overflow the file system and
                   3424: have an error message as consequence.
                   3426: @item using @code{source-id} when @code{blk} is non-zero:
                   3427: @code{source-id} performs its function. Typically it will give the id of
                   3428: the source which loaded the block. (Better ideas?)
                   3430: @end table
                   3433: @c =====================================================================
                   3434: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15      anton    3435: @section The optional Floating-Point word set
1.14      anton    3436: @c =====================================================================
                   3438: @menu
1.15      anton    3439: * floating-idef::               Implementation Defined Options
                   3440: * floating-ambcond::            Ambiguous Conditions            
1.14      anton    3441: @end menu
                   3444: @c ---------------------------------------------------------------------
                   3445: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
                   3446: @subsection Implementation Defined Options
                   3447: @c ---------------------------------------------------------------------
                   3449: @table @i
1.15      anton    3451: @item format and range of floating point numbers:
                   3452: System-dependent; the @code{double} type of C.
1.14      anton    3453: 
1.15      anton    3454: @item results of @code{REPRESENT} when @var{float} is out of range:
                   3455: System dependent; @code{REPRESENT} is implemented using the C library
                   3456: function @code{ecvt()} and inherits its behaviour in this respect.
1.14      anton    3457: 
1.15      anton    3458: @item rounding or truncation of floating-point numbers:
1.26      anton    3459: System dependent; the rounding behaviour is inherited from the hosting C
                   3460: compiler. IEEE-FP-based (i.e., most) systems by default round to
                   3461: nearest, and break ties by rounding to even (i.e., such that the last
                   3462: bit of the mantissa is 0).
1.14      anton    3463: 
1.15      anton    3464: @item size of floating-point stack:
                   3465: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
                   3466: with the command-line option @code{-f}.
1.14      anton    3467: 
1.15      anton    3468: @item width of floating-point stack:
                   3469: @code{1 floats}.
1.14      anton    3470: 
                   3471: @end table
                   3474: @c ---------------------------------------------------------------------
1.15      anton    3475: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
                   3476: @subsection Ambiguous conditions
1.14      anton    3477: @c ---------------------------------------------------------------------
                   3479: @table @i
1.15      anton    3481: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
1.37      anton    3482: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3483: alignment violations.
1.14      anton    3484: 
1.15      anton    3485: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
1.37      anton    3486: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3487: alignment violations.
1.14      anton    3488: 
1.15      anton    3489: @item Floating-point result out of range:
                   3490: System-dependent. Can result in a @code{-55 THROW} (Floating-point
                   3491: unidentified fault), or can produce a special value representing, e.g.,
                   3492: Infinity.
1.14      anton    3493: 
1.15      anton    3494: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
                   3495: System-dependent. Typically results in an alignment fault like other
                   3496: alignment violations.
1.14      anton    3497: 
1.15      anton    3498: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
                   3499: The floating-point number is converted into decimal nonetheless.
1.14      anton    3500: 
1.15      anton    3501: @item Both arguments are equal to zero (@code{FATAN2}):
                   3502: System-dependent. @code{FATAN2} is implemented using the C library
                   3503: function @code{atan2()}.
1.14      anton    3504: 
1.15      anton    3505: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
                   3506: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
                   3507: because of small errors and the tan will be a very large (or very small)
                   3508: but finite number.
1.14      anton    3509: 
1.15      anton    3510: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
                   3511: The result is rounded to the nearest float.
1.14      anton    3512: 
1.15      anton    3513: @item dividing by zero:
                   3514: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3515: 
1.15      anton    3516: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
                   3517: System dependent. On IEEE-FP based systems the number is converted into
                   3518: an infinity.
1.14      anton    3519: 
1.15      anton    3520: @item @var{float}<1 (@code{facosh}):
                   3521: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3522: 
1.15      anton    3523: @item @var{float}=<-1 (@code{flnp1}):
                   3524: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3525: negative infinity is typically produced for @var{float}=-1.
1.14      anton    3526: 
1.15      anton    3527: @item @var{float}=<0 (@code{fln}, @code{flog}):
                   3528: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3529: negative infinity is typically produced for @var{float}=0.
1.14      anton    3530: 
1.15      anton    3531: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
                   3532: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
                   3533: produces values for these inputs on my Linux box (Bug in the C library?)
1.14      anton    3534: 
1.15      anton    3535: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
                   3536: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    3537: 
1.15      anton    3538: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
                   3539: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    3540: 
1.15      anton    3541: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
                   3542: This does not happen.
                   3543: @end table
1.14      anton    3544: 
                   3547: @c =====================================================================
1.15      anton    3548: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
                   3549: @section The optional Locals word set
1.14      anton    3550: @c =====================================================================
                   3552: @menu
1.15      anton    3553: * locals-idef::                 Implementation Defined Options                 
                   3554: * locals-ambcond::              Ambiguous Conditions              
1.14      anton    3555: @end menu
                   3558: @c ---------------------------------------------------------------------
1.15      anton    3559: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14      anton    3560: @subsection Implementation Defined Options
                   3561: @c ---------------------------------------------------------------------
                   3563: @table @i
1.15      anton    3565: @item maximum number of locals in a definition:
                   3566: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
                   3567: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
                   3568: characters. The number of locals in a definition is bounded by the size
                   3569: of locals-buffer, which contains the names of the locals.
1.14      anton    3570: 
                   3571: @end table
                   3574: @c ---------------------------------------------------------------------
1.15      anton    3575: @node locals-ambcond,  , locals-idef, The optional Locals word set
1.14      anton    3576: @subsection Ambiguous conditions
                   3577: @c ---------------------------------------------------------------------
                   3579: @table @i
1.15      anton    3581: @item executing a named local in interpretation state:
                   3582: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    3583: 
1.15      anton    3584: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
                   3585: @code{-32 throw} (Invalid name argument)
1.14      anton    3586: 
                   3587: @end table
                   3590: @c =====================================================================
1.15      anton    3591: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
                   3592: @section The optional Memory-Allocation word set
1.14      anton    3593: @c =====================================================================
                   3595: @menu
1.15      anton    3596: * memory-idef::                 Implementation Defined Options                 
1.14      anton    3597: @end menu
                   3600: @c ---------------------------------------------------------------------
1.15      anton    3601: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14      anton    3602: @subsection Implementation Defined Options
                   3603: @c ---------------------------------------------------------------------
                   3605: @table @i
1.15      anton    3607: @item values and meaning of @var{ior}:
                   3608: The @var{ior}s returned by the file and memory allocation words are
                   3609: intended as throw codes. They typically are in the range
                   3610: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3611: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3612: 
                   3613: @end table
                   3615: @c =====================================================================
1.15      anton    3616: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
                   3617: @section The optional Programming-Tools word set
1.14      anton    3618: @c =====================================================================
                   3620: @menu
1.15      anton    3621: * programming-idef::            Implementation Defined Options            
                   3622: * programming-ambcond::         Ambiguous Conditions         
1.14      anton    3623: @end menu
                   3626: @c ---------------------------------------------------------------------
1.15      anton    3627: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14      anton    3628: @subsection Implementation Defined Options
                   3629: @c ---------------------------------------------------------------------
                   3631: @table @i
1.15      anton    3633: @item ending sequence for input following @code{;code} and @code{code}:
1.37      anton    3634: @code{end-code}
1.14      anton    3635: 
1.15      anton    3636: @item manner of processing input following @code{;code} and @code{code}:
1.37      anton    3637: The @code{assembler} vocabulary is pushed on the search order stack, and
                   3638: the input is processed by the text interpreter, (starting) in interpret
                   3639: state.
1.15      anton    3640: 
                   3641: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
1.37      anton    3642: The ANS Forth search order word set.
1.15      anton    3643: 
                   3644: @item source and format of display by @code{SEE}:
                   3645: The source for @code{see} is the intermediate code used by the inner
                   3646: interpreter.  The current @code{see} tries to output Forth source code
                   3647: as well as possible.
1.14      anton    3649: @end table
                   3651: @c ---------------------------------------------------------------------
1.15      anton    3652: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
1.14      anton    3653: @subsection Ambiguous conditions
                   3654: @c ---------------------------------------------------------------------
                   3656: @table @i
1.15      anton    3658: @item deleting the compilation wordlist (@code{FORGET}):
                   3659: Not implemented (yet).
1.14      anton    3660: 
1.15      anton    3661: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
                   3662: This typically results in an @code{abort"} with a descriptive error
                   3663: message (may change into a @code{-22 throw} (Control structure mismatch)
                   3664: in the future). You may also get a memory access error. If you are
                   3665: unlucky, this ambiguous condition is not caught.
                   3667: @item @var{name} can't be found (@code{forget}):
                   3668: Not implemented (yet).
1.14      anton    3669: 
1.15      anton    3670: @item @var{name} not defined via @code{CREATE}:
1.37      anton    3671: @code{;code} behaves like @code{DOES>} in this respect, i.e., it changes
                   3672: the execution semantics of the last defined word no matter how it was
                   3673: defined.
1.14      anton    3674: 
1.15      anton    3675: @item @code{POSTPONE} applied to @code{[IF]}:
                   3676: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
                   3677: equivalent to @code{[IF]}.
1.14      anton    3678: 
1.15      anton    3679: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
                   3680: Continue in the same state of conditional compilation in the next outer
                   3681: input source. Currently there is no warning to the user about this.
1.14      anton    3682: 
1.15      anton    3683: @item removing a needed definition (@code{FORGET}):
                   3684: Not implemented (yet).
1.14      anton    3685: 
                   3686: @end table
                   3689: @c =====================================================================
1.15      anton    3690: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
                   3691: @section The optional Search-Order word set
1.14      anton    3692: @c =====================================================================
                   3694: @menu
1.15      anton    3695: * search-idef::                 Implementation Defined Options                 
                   3696: * search-ambcond::              Ambiguous Conditions              
1.14      anton    3697: @end menu
                   3700: @c ---------------------------------------------------------------------
1.15      anton    3701: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14      anton    3702: @subsection Implementation Defined Options
                   3703: @c ---------------------------------------------------------------------
                   3705: @table @i
1.15      anton    3707: @item maximum number of word lists in search order:
                   3708: @code{s" wordlists" environment? drop .}. Currently 16.
                   3710: @item minimum search order:
                   3711: @code{root root}.
1.14      anton    3712: 
                   3713: @end table
                   3715: @c ---------------------------------------------------------------------
1.15      anton    3716: @node search-ambcond,  , search-idef, The optional Search-Order word set
1.14      anton    3717: @subsection Ambiguous conditions
                   3718: @c ---------------------------------------------------------------------
                   3720: @table @i
1.15      anton    3722: @item changing the compilation wordlist (during compilation):
1.33      anton    3723: The word is entered into the wordlist that was the compilation wordlist
                   3724: at the start of the definition. Any changes to the name field (e.g.,
                   3725: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
                   3726: are applied to the latest defined word (as reported by @code{last} or
                   3727: @code{lastxt}), if possible, irrespective of the compilation wordlist.
1.14      anton    3728: 
1.15      anton    3729: @item search order empty (@code{previous}):
                   3730: @code{abort" Vocstack empty"}.
1.14      anton    3731: 
1.15      anton    3732: @item too many word lists in search order (@code{also}):
                   3733: @code{abort" Vocstack full"}.
1.14      anton    3734: 
                   3735: @end table
1.13      anton    3736: 
1.34      anton    3737: @node Model, Integrating Gforth, ANS conformance, Top
                   3738: @chapter Model
                   3740: This chapter has yet to be written. It will contain information, on
                   3741: which internal structures you can rely.
                   3743: @node Integrating Gforth, Emacs and Gforth, Model, Top
                   3744: @chapter Integrating Gforth into C programs
                   3746: This is not yet implemented.
                   3748: Several people like to use Forth as scripting language for applications
                   3749: that are otherwise written in C, C++, or some other language.
                   3751: The Forth system ATLAST provides facilities for embedding it into
                   3752: applications; unfortunately it has several disadvantages: most
1.36      anton    3753: importantly, it is not based on ANS Forth, and it is apparently dead
1.34      anton    3754: (i.e., not developed further and not supported). The facilities
                   3755: provided by Gforth in this area are inspired by ATLASTs facilities, so
                   3756: making the switch should not be hard.
                   3758: We also tried to design the interface such that it can easily be
                   3759: implemented by other Forth systems, so that we may one day arrive at a
                   3760: standardized interface. Such a standard interface would allow you to
                   3761: replace the Forth system without having to rewrite C code.
                   3763: You embed the Gforth interpreter by linking with the library
                   3764: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
                   3765: global symbols in this library that belong to the interface, have the
                   3766: prefix @code{forth_}. (Global symbols that are used internally have the
                   3767: prefix @code{gforth_}).
                   3769: You can include the declarations of Forth types and the functions and
1.36      anton    3770: variables of the interface with @code{#include <forth.h>}.
1.34      anton    3771: 
                   3772: Types.
1.13      anton    3773: 
1.34      anton    3774: Variables.
                   3776: Data and FP Stack pointer. Area sizes.
                   3778: functions.
                   3780: forth_init(imagefile)
                   3781: forth_evaluate(string) exceptions?
                   3782: forth_goto(address) (or forth_execute(xt)?)
                   3783: forth_continue() (a corountining mechanism)
                   3785: Adding primitives.
                   3787: No checking.
                   3789: Signals?
                   3791: Accessing the Stacks
1.4       anton    3792: 
1.34      anton    3793: @node Emacs and Gforth, Internals, Integrating Gforth, Top
1.17      anton    3794: @chapter Emacs and Gforth
1.4       anton    3795: 
1.17      anton    3796: Gforth comes with @file{gforth.el}, an improved version of
1.33      anton    3797: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.4       anton    3798: improvements are a better (but still not perfect) handling of
                   3799: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8       anton    3800: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
                   3801: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
                   3802: stuff I do not use alone, even though some of it only makes sense for
                   3803: TILE. To get a description of these features, enter Forth mode and type
                   3804: @kbd{C-h m}.
1.4       anton    3805: 
1.17      anton    3806: In addition, Gforth supports Emacs quite well: The source code locations
1.4       anton    3807: given in error messages, debugging output (from @code{~~}) and failed
                   3808: assertion messages are in the right format for Emacs' compilation mode
                   3809: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
                   3810: Manual}) so the source location corresponding to an error or other
                   3811: message is only a few keystrokes away (@kbd{C-x `} for the next error,
                   3812: @kbd{C-c C-c} for the error under the cursor).
                   3814: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
                   3815: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
                   3816: contains the definitions of all words defined afterwards. You can then
                   3817: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17      anton    3818: several tags files at the same time (e.g., one for the Gforth sources
1.28      anton    3819: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
                   3820: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
                   3821: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
1.33      anton    3822: @file{/usr/local/share/gforth/0.2.0/TAGS}).
1.4       anton    3823: 
                   3824: To get all these benefits, add the following lines to your @file{.emacs}
                   3825: file:
                   3827: @example
                   3828: (autoload 'forth-mode "gforth.el")
                   3829: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
                   3830: @end example
1.17      anton    3832: @node Internals, Bugs, Emacs and Gforth, Top
1.3       anton    3833: @chapter Internals
1.17      anton    3835: Reading this section is not necessary for programming with Gforth. It
                   3836: should be helpful for finding your way in the Gforth sources.
1.3       anton    3837: 
1.24      anton    3838: The ideas in this section have also been published in the papers
                   3839: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
                   3840: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
                   3841: Ertl, presented at EuroForth '93; the latter is available at
                   3842: @*@file{}.
1.4       anton    3844: @menu
                   3845: * Portability::                 
                   3846: * Threading::                   
                   3847: * Primitives::                  
                   3848: * System Architecture::         
1.17      anton    3849: * Performance::                 
1.4       anton    3850: @end menu
                   3852: @node Portability, Threading, Internals, Internals
1.3       anton    3853: @section Portability
                   3855: One of the main goals of the effort is availability across a wide range
                   3856: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
                   3857: this goal by manually coding the engine in assembly language for several
                   3858: then-popular processors. This approach is very labor-intensive and the
                   3859: results are short-lived due to progress in computer architecture.
                   3861: Others have avoided this problem by coding in C, e.g., Mitch Bradley
                   3862: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
                   3863: particularly popular for UNIX-based Forths due to the large variety of
                   3864: architectures of UNIX machines. Unfortunately an implementation in C
                   3865: does not mix well with the goals of efficiency and with using
                   3866: traditional techniques: Indirect or direct threading cannot be expressed
                   3867: in C, and switch threading, the fastest technique available in C, is
                   3868: significantly slower. Another problem with C is that it's very
                   3869: cumbersome to express double integer arithmetic.
                   3871: Fortunately, there is a portable language that does not have these
                   3872: limitations: GNU C, the version of C processed by the GNU C compiler
                   3873: (@pxref{C Extensions, , Extensions to the C Language Family,,
                   3874: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
                   3875: Labels as Values,, GNU C Manual}) makes direct and indirect
                   3876: threading possible, its @code{long long} type (@pxref{Long Long, ,
1.33      anton    3877: Double-Word Integers,, GNU C Manual}) corresponds to Forth's
1.32      anton    3878: double numbers@footnote{Unfortunately, long longs are not implemented
                   3879: properly on all machines (e.g., on alpha-osf1, long longs are only 64
                   3880: bits, the same size as longs (and pointers), but they should be twice as
                   3881: long according to @ref{Long Long, , Double-Word Integers,, GNU
                   3882: C Manual}). So, we had to implement doubles in C after all. Still, on
                   3883: most machines we can use long longs and achieve better performance than
                   3884: with the emulation package.}. GNU C is available for free on all
                   3885: important (and many unimportant) UNIX machines, VMS, 80386s running
                   3886: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
                   3887: on all these machines.
1.3       anton    3888: 
                   3889: Writing in a portable language has the reputation of producing code that
                   3890: is slower than assembly. For our Forth engine we repeatedly looked at
                   3891: the code produced by the compiler and eliminated most compiler-induced
                   3892: inefficiencies by appropriate changes in the source-code.
                   3894: However, register allocation cannot be portably influenced by the
                   3895: programmer, leading to some inefficiencies on register-starved
                   3896: machines. We use explicit register declarations (@pxref{Explicit Reg
                   3897: Vars, , Variables in Specified Registers,, GNU C Manual}) to
                   3898: improve the speed on some machines. They are turned on by using the
                   3899: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
                   3900: only depends on the machine, but also on the compiler version: On some
                   3901: machines some compiler versions produce incorrect code when certain
                   3902: explicit register declarations are used. So by default
                   3903: @code{-DFORCE_REG} is not used.
1.4       anton    3905: @node Threading, Primitives, Portability, Internals
1.3       anton    3906: @section Threading
                   3908: GNU C's labels as values extension (available since @code{gcc-2.0},
                   3909: @pxref{Labels as Values, , Labels as Values,, GNU C Manual})
                   3910: makes it possible to take the address of @var{label} by writing
                   3911: @code{&&@var{label}}.  This address can then be used in a statement like
                   3912: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
                   3913: @code{goto x}.
                   3915: With this feature an indirect threaded NEXT looks like:
                   3916: @example
                   3917: cfa = *ip++;
                   3918: ca = *cfa;
                   3919: goto *ca;
                   3920: @end example
                   3921: For those unfamiliar with the names: @code{ip} is the Forth instruction
                   3922: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
                   3923: execution token and points to the code field of the next word to be
                   3924: executed; The @code{ca} (code address) fetched from there points to some
                   3925: executable code, e.g., a primitive or the colon definition handler
                   3926: @code{docol}.
                   3928: Direct threading is even simpler:
                   3929: @example
                   3930: ca = *ip++;
                   3931: goto *ca;
                   3932: @end example
                   3934: Of course we have packaged the whole thing neatly in macros called
                   3935: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
1.4       anton    3937: @menu
                   3938: * Scheduling::                  
                   3939: * Direct or Indirect Threaded?::  
                   3940: * DOES>::                       
                   3941: @end menu
                   3943: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3       anton    3944: @subsection Scheduling
                   3946: There is a little complication: Pipelined and superscalar processors,
                   3947: i.e., RISC and some modern CISC machines can process independent
                   3948: instructions while waiting for the results of an instruction. The
                   3949: compiler usually reorders (schedules) the instructions in a way that
                   3950: achieves good usage of these delay slots. However, on our first tries
                   3951: the compiler did not do well on scheduling primitives. E.g., for
                   3952: @code{+} implemented as
                   3953: @example
                   3954: n=sp[0]+sp[1];
                   3955: sp++;
                   3956: sp[0]=n;
                   3957: NEXT;
                   3958: @end example
                   3959: the NEXT comes strictly after the other code, i.e., there is nearly no
                   3960: scheduling. After a little thought the problem becomes clear: The
                   3961: compiler cannot know that sp and ip point to different addresses (and
1.4       anton    3962: the version of @code{gcc} we used would not know it even if it was
                   3963: possible), so it could not move the load of the cfa above the store to
                   3964: the TOS. Indeed the pointers could be the same, if code on or very near
                   3965: the top of stack were executed. In the interest of speed we chose to
                   3966: forbid this probably unused ``feature'' and helped the compiler in
                   3967: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
                   3968: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3       anton    3969: @example
                   3970: n=sp[0]+sp[1];
                   3971: sp++;
                   3972: NEXT_P1;
                   3973: sp[0]=n;
                   3974: NEXT_P2;
                   3975: @end example
1.4       anton    3976: This can be scheduled optimally by the compiler.
1.3       anton    3977: 
                   3978: This division can be turned off with the switch @code{-DCISC_NEXT}. This
                   3979: switch is on by default on machines that do not profit from scheduling
                   3980: (e.g., the 80386), in order to preserve registers.
1.4       anton    3982: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3       anton    3983: @subsection Direct or Indirect Threaded?
                   3985: Both! After packaging the nasty details in macro definitions we
                   3986: realized that we could switch between direct and indirect threading by
                   3987: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
                   3988: defining a few machine-specific macros for the direct-threading case.
                   3989: On the Forth level we also offer access words that hide the
                   3990: differences between the threading methods (@pxref{Threading Words}).
                   3992: Indirect threading is implemented completely
                   3993: machine-independently. Direct threading needs routines for creating
                   3994: jumps to the executable code (e.g. to docol or dodoes). These routines
                   3995: are inherently machine-dependent, but they do not amount to many source
                   3996: lines. I.e., even porting direct threading to a new machine is a small
                   3997: effort.
1.4       anton    3999: @node DOES>,  , Direct or Indirect Threaded?, Threading
1.3       anton    4000: @subsection DOES>
                   4001: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
                   4002: the chunk of code executed by every word defined by a
                   4003: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
                   4004: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
                   4005: DOES-code)? There are two solutions:
                   4007: In fig-Forth the code field points directly to the dodoes and the
                   4008: DOES-code address is stored in the cell after the code address
                   4009: (i.e. at cfa cell+). It may seem that this solution is illegal in the
                   4010: Forth-79 and all later standards, because in fig-Forth this address
                   4011: lies in the body (which is illegal in these standards). However, by
                   4012: making the code field larger for all words this solution becomes legal
                   4013: again. We use this approach for the indirect threaded version. Leaving
                   4014: a cell unused in most words is a bit wasteful, but on the machines we
                   4015: are targetting this is hardly a problem. The other reason for having a
                   4016: code field size of two cells is to avoid having different image files
1.4       anton    4017: for direct and indirect threaded systems (@pxref{System Architecture}).
1.3       anton    4018: 
                   4019: The other approach is that the code field points or jumps to the cell
                   4020: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
                   4021: this address. @code{dodoes} can then get the DOES-code address by
                   4022: computing the code address, i.e., the address of the jump to dodoes,
                   4023: and add the length of that jump field. A variant of this is to have a
                   4024: call to @code{dodoes} after the @code{DOES>}; then the return address
                   4025: (which can be found in the return register on RISCs) is the DOES-code
                   4026: address. Since the two cells available in the code field are usually
                   4027: used up by the jump to the code address in direct threading, we use
                   4028: this approach for direct threading. We did not want to add another
                   4029: cell to the code field.
1.4       anton    4031: @node Primitives, System Architecture, Threading, Internals
1.3       anton    4032: @section Primitives
1.4       anton    4034: @menu
                   4035: * Automatic Generation::        
                   4036: * TOS Optimization::            
                   4037: * Produced code::               
                   4038: @end menu
                   4040: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3       anton    4041: @subsection Automatic Generation
                   4043: Since the primitives are implemented in a portable language, there is no
                   4044: longer any need to minimize the number of primitives. On the contrary,
                   4045: having many primitives is an advantage: speed. In order to reduce the
                   4046: number of errors in primitives and to make programming them easier, we
                   4047: provide a tool, the primitive generator (@file{prims2x.fs}), that
                   4048: automatically generates most (and sometimes all) of the C code for a
                   4049: primitive from the stack effect notation.  The source for a primitive
                   4050: has the following form:
                   4052: @format
                   4053: @var{Forth-name}       @var{stack-effect}      @var{category}  [@var{pronounc.}]
                   4054: [@code{""}@var{glossary entry}@code{""}]
                   4055: @var{C code}
                   4056: [@code{:}
                   4057: @var{Forth code}]
                   4058: @end format
                   4060: The items in brackets are optional. The category and glossary fields
                   4061: are there for generating the documentation, the Forth code is there
                   4062: for manual implementations on machines without GNU C. E.g., the source
                   4063: for the primitive @code{+} is:
                   4064: @example
                   4065: +    n1 n2 -- n    core    plus
                   4066: n = n1+n2;
                   4067: @end example
                   4069: This looks like a specification, but in fact @code{n = n1+n2} is C
                   4070: code. Our primitive generation tool extracts a lot of information from
                   4071: the stack effect notations@footnote{We use a one-stack notation, even
                   4072: though we have separate data and floating-point stacks; The separate
                   4073: notation can be generated easily from the unified notation.}: The number
                   4074: of items popped from and pushed on the stack, their type, and by what
                   4075: name they are referred to in the C code. It then generates a C code
                   4076: prelude and postlude for each primitive. The final C code for @code{+}
                   4077: looks like this:
                   4079: @example
                   4080: I_plus:        /* + ( n1 n2 -- n ) */  /* label, stack effect */
                   4081: /*  */                          /* documentation */
1.4       anton    4082: @{
1.3       anton    4083: DEF_CA                          /* definition of variable ca (indirect threading) */
                   4084: Cell n1;                        /* definitions of variables */
                   4085: Cell n2;
                   4086: Cell n;
                   4087: n1 = (Cell) sp[1];              /* input */
                   4088: n2 = (Cell) TOS;
                   4089: sp += 1;                        /* stack adjustment */
                   4090: NAME("+")                       /* debugging output (with -DDEBUG) */
1.4       anton    4091: @{
1.3       anton    4092: n = n1+n2;                      /* C code taken from the source */
1.4       anton    4093: @}
1.3       anton    4094: NEXT_P1;                        /* NEXT part 1 */
                   4095: TOS = (Cell)n;                  /* output */
                   4096: NEXT_P2;                        /* NEXT part 2 */
1.4       anton    4097: @}
1.3       anton    4098: @end example
                   4100: This looks long and inefficient, but the GNU C compiler optimizes quite
                   4101: well and produces optimal code for @code{+} on, e.g., the R3000 and the
                   4102: HP RISC machines: Defining the @code{n}s does not produce any code, and
                   4103: using them as intermediate storage also adds no cost.
                   4105: There are also other optimizations, that are not illustrated by this
                   4106: example: Assignments between simple variables are usually for free (copy
                   4107: propagation). If one of the stack items is not used by the primitive
                   4108: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
                   4109: (dead code elimination). On the other hand, there are some things that
                   4110: the compiler does not do, therefore they are performed by
                   4111: @file{prims2x.fs}: The compiler does not optimize code away that stores
                   4112: a stack item to the place where it just came from (e.g., @code{over}).
                   4114: While programming a primitive is usually easy, there are a few cases
                   4115: where the programmer has to take the actions of the generator into
                   4116: account, most notably @code{?dup}, but also words that do not (always)
                   4117: fall through to NEXT.
1.4       anton    4119: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3       anton    4120: @subsection TOS Optimization
                   4122: An important optimization for stack machine emulators, e.g., Forth
                   4123: engines, is keeping  one or more of the top stack items in
1.4       anton    4124: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
                   4125: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.34      anton    4126: @itemize @bullet
1.3       anton    4127: @item
                   4128: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
                   4129: due to fewer loads from and stores to the stack.
                   4130: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
                   4131: @var{y<n}, due to additional moves between registers.
                   4132: @end itemize
                   4134: In particular, keeping one item in a register is never a disadvantage,
                   4135: if there are enough registers. Keeping two items in registers is a
                   4136: disadvantage for frequent words like @code{?branch}, constants,
                   4137: variables, literals and @code{i}. Therefore our generator only produces
                   4138: code that keeps zero or one items in registers. The generated C code
                   4139: covers both cases; the selection between these alternatives is made at
                   4140: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
                   4141: code for @code{+} is just a simple variable name in the one-item case,
                   4142: otherwise it is a macro that expands into @code{sp[0]}. Note that the
                   4143: GNU C compiler tries to keep simple variables like @code{TOS} in
                   4144: registers, and it usually succeeds, if there are enough registers.
                   4146: The primitive generator performs the TOS optimization for the
                   4147: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
                   4148: operations the benefit of this optimization is even larger:
                   4149: floating-point operations take quite long on most processors, but can be
                   4150: performed in parallel with other operations as long as their results are
                   4151: not used. If the FP-TOS is kept in a register, this works. If
                   4152: it is kept on the stack, i.e., in memory, the store into memory has to
                   4153: wait for the result of the floating-point operation, lengthening the
                   4154: execution time of the primitive considerably.
                   4156: The TOS optimization makes the automatic generation of primitives a
                   4157: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
                   4158: @code{TOS} is not sufficient. There are some special cases to
                   4159: consider:
1.34      anton    4160: @itemize @bullet
1.3       anton    4161: @item In the case of @code{dup ( w -- w w )} the generator must not
                   4162: eliminate the store to the original location of the item on the stack,
                   4163: if the TOS optimization is turned on.
1.4       anton    4164: @item Primitives with stack effects of the form @code{--}
                   4165: @var{out1}...@var{outy} must store the TOS to the stack at the start.
                   4166: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3       anton    4167: must load the TOS from the stack at the end. But for the null stack
                   4168: effect @code{--} no stores or loads should be generated.
                   4169: @end itemize
1.4       anton    4171: @node Produced code,  , TOS Optimization, Primitives
1.3       anton    4172: @subsection Produced code
                   4174: To see what assembly code is produced for the primitives on your machine
                   4175: with your compiler and your flag settings, type @code{make engine.s} and
1.4       anton    4176: look at the resulting file @file{engine.s}.
1.3       anton    4177: 
1.17      anton    4178: @node System Architecture, Performance, Primitives, Internals
1.3       anton    4179: @section System Architecture
                   4181: Our Forth system consists not only of primitives, but also of
                   4182: definitions written in Forth. Since the Forth compiler itself belongs
                   4183: to those definitions, it is not possible to start the system with the
                   4184: primitives and the Forth source alone. Therefore we provide the Forth
                   4185: code as an image file in nearly executable form. At the start of the
                   4186: system a C routine loads the image file into memory, sets up the
                   4187: memory (stacks etc.) according to information in the image file, and
                   4188: starts executing Forth code.
                   4190: The image file format is a compromise between the goals of making it
                   4191: easy to generate image files and making them portable. The easiest way
                   4192: to generate an image file is to just generate a memory dump. However,
                   4193: this kind of image file cannot be used on a different machine, or on
                   4194: the next version of the engine on the same machine, it even might not
                   4195: work with the same engine compiled by a different version of the C
                   4196: compiler. We would like to have as few versions of the image file as
                   4197: possible, because we do not want to distribute many versions of the
                   4198: same image file, and to make it easy for the users to use their image
                   4199: files on many machines. We currently need to create a different image
                   4200: file for machines with different cell sizes and different byte order
1.17      anton    4201: (little- or big-endian)@footnote{We are considering adding information to the
1.3       anton    4202: image file that enables the loader to change the byte order.}.
                   4204: Forth code that is going to end up in a portable image file has to
1.4       anton    4205: comply to some restrictions: addresses have to be stored in memory with
                   4206: special words (@code{A!}, @code{A,}, etc.) in order to make the code
                   4207: relocatable. Cells, floats, etc., have to be stored at the natural
                   4208: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
                   4209: dividable by~8. This happens automatically in our system when you use
                   4210: the ANS Forth alignment words.}, in order to avoid alignment faults on
                   4211: machines with stricter alignment. The image file is produced by a
                   4212: metacompiler (@file{cross.fs}).
1.3       anton    4213: 
                   4214: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
                   4215: file is not directly executable, but has to undergo some manipulations
                   4216: during loading. Address relocation is performed at image load-time, not
                   4217: at run-time. The loader also has to replace tokens standing for
                   4218: primitive calls with the appropriate code-field addresses (or code
                   4219: addresses in the case of direct threading).
1.4       anton    4220: 
1.17      anton    4221: @node  Performance,  , System Architecture, Internals
                   4222: @section Performance
                   4224: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
                   4225: impossible to write a significantly faster engine.
                   4227: On register-starved machines like the 386 architecture processors
                   4228: improvements are possible, because @code{gcc} does not utilize the
                   4229: registers as well as a human, even with explicit register declarations;
                   4230: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
                   4231: and hand-tuned it for the 486; this system is 1.19 times faster on the
                   4232: Sieve benchmark on a 486DX2/66 than Gforth compiled with
                   4233: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
                   4235: However, this potential advantage of assembly language implementations
                   4236: is not necessarily realized in complete Forth systems: We compared
1.26      anton    4237: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
                   4238: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
                   4239: 1994) and Eforth (with and without peephole (aka pinhole) optimization
                   4240: of the threaded code); all these systems were written in assembly
1.30      anton    4241: language. We also compared Gforth with three systems written in C:
1.32      anton    4242: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
                   4243: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
                   4244: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
                   4245: -fomit-frame-pointer; ThisForth employs peephole optimization of the
                   4246: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
                   4247: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
                   4248: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
                   4249: 486DX2/66 with similar memory performance under Windows NT. Marcel
                   4250: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
                   4251: added the peephole optimizer, ran the benchmarks and reported the
                   4252: results.
1.17      anton    4253:  
                   4254: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
                   4255: matrix multiplication come from the Stanford integer benchmarks and have
                   4256: been translated into Forth by Martin Fraeman; we used the versions
1.30      anton    4257: included in the TILE Forth package, but with bigger data set sizes; and
                   4258: a recursive Fibonacci number computation for benchmarking calling
                   4259: performance. The following table shows the time taken for the benchmarks
                   4260: scaled by the time taken by Gforth (in other words, it shows the speedup
                   4261: factor that Gforth achieved over the other systems).
1.17      anton    4262: 
                   4263: @example
1.30      anton    4264: relative      Win32-    NT       eforth       This-
                   4265:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
1.32      anton    4266: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
                   4267: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
1.38      anton    4268: matmul    1.00  1.47  1.35   1.46  0.74  1.58        4.09
                   4269: fib       1.00  1.52  1.34   1.22  0.86  1.74  2.99  4.30
1.17      anton    4270: @end example
                   4272: You may find the good performance of Gforth compared with the systems
                   4273: written in assembly language quite surprising. One important reason for
                   4274: the disappointing performance of these systems is probably that they are
                   4275: not written optimally for the 486 (e.g., they use the @code{lods}
                   4276: instruction). In addition, Win32Forth uses a comfortable, but costly
                   4277: method for relocating the Forth image: like @code{cforth}, it computes
                   4278: the actual addresses at run time, resulting in two address computations
                   4279: per NEXT (@pxref{System Architecture}).
1.26      anton    4281: Only Eforth with the peephole optimizer performs comparable to
                   4282: Gforth. The speedups achieved with peephole optimization of threaded
                   4283: code are quite remarkable. Adding a peephole optimizer to Gforth should
                   4284: cause similar speedups.
1.30      anton    4286: The speedup of Gforth over PFE, ThisForth and TILE can be easily
                   4287: explained with the self-imposed restriction to standard C, which makes
                   4288: efficient threading impossible (however, the measured implementation of
                   4289: PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
                   4290: Register Variables,, GNU C Manual}).  Moreover, current C
                   4291: compilers have a hard time optimizing other aspects of the ThisForth
                   4292: and the TILE source.
1.17      anton    4293: 
                   4294: Note that the performance of Gforth on 386 architecture processors
                   4295: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
                   4296: failed to allocate any of the virtual machine registers into real
                   4297: machine registers by itself and would not work correctly with explicit
                   4298: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
                   4299: running the Sieve) than the one measured above.
1.26      anton    4301: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
                   4302: Maierhofer (presented at EuroForth '95), an indirect threaded version of
                   4303: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
                   4304: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
                   4305: used here. The paper available at
1.24      anton    4306: @*@file{};
                   4307: it also contains numbers for some native code systems. You can find
                   4308: numbers for Gforth on various machines in @file{Benchres}.
1.29      anton    4310: @node Bugs, Origin, Internals, Top
1.4       anton    4311: @chapter Bugs
1.17      anton    4313: Known bugs are described in the file BUGS in the Gforth distribution.
1.24      anton    4315: If you find a bug, please send a bug report to
1.32      anton    4316: @code{}. A bug report should
1.17      anton    4317: describe the Gforth version used (it is announced at the start of an
                   4318: interactive Gforth session), the machine and operating system (on Unix
                   4319: systems you can use @code{uname -a} to produce this information), the
1.24      anton    4320: installation options (send the @code{config.status} file), and a
                   4321: complete list of changes you (or your installer) have made to the Gforth
                   4322: sources (if any); it should contain a program (or a sequence of keyboard
                   4323: commands) that reproduces the bug and a description of what you think
                   4324: constitutes the buggy behaviour.
1.17      anton    4325: 
                   4326: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
                   4327: to Report Bugs,, GNU C Manual}.
1.29      anton    4330: @node Origin, Word Index, Bugs, Top
                   4331: @chapter Authors and Ancestors of Gforth
                   4333: @section Authors and Contributors
                   4335: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
1.30      anton    4336: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
                   4337: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
                   4338: with their continuous feedback. Lennart Benshop contributed
1.29      anton    4339: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
                   4340: support for calling C libraries. Helpful comments also came from Paul
1.37      anton    4341: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.39      anton    4342: Wavrik, Barrie Stott and Marc de Groot.
1.29      anton    4343: 
1.30      anton    4344: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
                   4345: and autoconf, among others), and to the creators of the Internet: Gforth
                   4346: was developed across the Internet, and its authors have not met
                   4347: physically yet.
1.29      anton    4349: @section Pedigree
1.4       anton    4350: 
1.17      anton    4351: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24      anton    4352: Dirk Zoller) will cross-fertilize each other. Of course, a significant
                   4353: part of the design of Gforth was prescribed by ANS Forth.
1.17      anton    4354: 
1.23      pazsan   4355: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
                   4356: 32 bit native code version of VolksForth for the Atari ST, written
                   4357: mostly by Dietrich Weineck.
                   4359: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
                   4360: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24      anton    4361: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17      anton    4362: 
1.34      anton    4363: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17      anton    4364: Forth-83 standard. !! Pedigree? When?
                   4366: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24      anton    4367: 1979. Robert Selzer and Bill Ragsdale developed the original
                   4368: implementation of fig-Forth for the 6502 based on microForth.
                   4370: The principal architect of microForth was Dean Sanderson. microForth was
                   4371: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
                   4372: the 1802, and subsequently implemented on the 8080, the 6800 and the
                   4373: Z80.
1.17      anton    4374: 
1.24      anton    4375: All earlier Forth systems were custom-made, usually by Charles Moore,
1.30      anton    4376: who discovered (as he puts it) Forth during the late 60s. The first full
                   4377: Forth existed in 1971.
1.17      anton    4378: 
                   4379: A part of the information in this section comes from @cite{The Evolution
                   4380: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
                   4381: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
                   4382: Notices 28(3), 1993.  You can find more historical and genealogical
                   4383: information about Forth there.
1.29      anton    4385: @node Word Index, Node Index, Origin, Top
1.4       anton    4386: @chapter Word Index
1.18      anton    4388: This index is as incomplete as the manual. Each word is listed with
                   4389: stack effect and wordset.
1.17      anton    4390: 
                   4391: @printindex fn
1.4       anton    4393: @node Node Index,  , Word Index, Top
                   4394: @chapter Node Index
1.17      anton    4395: 
                   4396: This index is even less complete than the manual.
1.1       anton    4397: 
                   4398: @contents
                   4399: @bye

FreeBSD-CVSweb <>