Annotation of gforth/gforth.ds, revision 1.38

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
1.4       anton      89: * ANS conformance::             Implementation-defined options etc.
1.17      anton      90: * Model::                       The abstract machine of Gforth
1.34      anton      91: * Integrating Gforth::          Forth as scripting language for applications.
1.17      anton      92: * Emacs and Gforth::            The Gforth Mode
1.4       anton      93: * Internals::                   Implementation details
                     94: * Bugs::                        How to report them
1.29      anton      95: * Origin::                      Authors and ancestors of Gforth
1.4       anton      96: * Word Index::                  An item for each Forth word
                     97: * Node Index::                  An item for each node
1.1       anton      98: @end menu
                    100: @node License, Goals, Top, Top
1.20      pazsan    101: @unnumbered GNU GENERAL PUBLIC LICENSE
                    102: @center Version 2, June 1991
                    104: @display
                    105: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
                    106: 675 Mass Ave, Cambridge, MA 02139, USA
                    108: Everyone is permitted to copy and distribute verbatim copies
                    109: of this license document, but changing it is not allowed.
                    110: @end display
                    112: @unnumberedsec Preamble
                    114:   The licenses for most software are designed to take away your
                    115: freedom to share and change it.  By contrast, the GNU General Public
                    116: License is intended to guarantee your freedom to share and change free
                    117: software---to make sure the software is free for all its users.  This
                    118: General Public License applies to most of the Free Software
                    119: Foundation's software and to any other program whose authors commit to
                    120: using it.  (Some other Free Software Foundation software is covered by
                    121: the GNU Library General Public License instead.)  You can apply it to
                    122: your programs, too.
                    124:   When we speak of free software, we are referring to freedom, not
                    125: price.  Our General Public Licenses are designed to make sure that you
                    126: have the freedom to distribute copies of free software (and charge for
                    127: this service if you wish), that you receive source code or can get it
                    128: if you want it, that you can change the software or use pieces of it
                    129: in new free programs; and that you know you can do these things.
                    131:   To protect your rights, we need to make restrictions that forbid
                    132: anyone to deny you these rights or to ask you to surrender the rights.
                    133: These restrictions translate to certain responsibilities for you if you
                    134: distribute copies of the software, or if you modify it.
                    136:   For example, if you distribute copies of such a program, whether
                    137: gratis or for a fee, you must give the recipients all the rights that
                    138: you have.  You must make sure that they, too, receive or can get the
                    139: source code.  And you must show them these terms so they know their
                    140: rights.
                    142:   We protect your rights with two steps: (1) copyright the software, and
                    143: (2) offer you this license which gives you legal permission to copy,
                    144: distribute and/or modify the software.
                    146:   Also, for each author's protection and ours, we want to make certain
                    147: that everyone understands that there is no warranty for this free
                    148: software.  If the software is modified by someone else and passed on, we
                    149: want its recipients to know that what they have is not the original, so
                    150: that any problems introduced by others will not reflect on the original
                    151: authors' reputations.
                    153:   Finally, any free program is threatened constantly by software
                    154: patents.  We wish to avoid the danger that redistributors of a free
                    155: program will individually obtain patent licenses, in effect making the
                    156: program proprietary.  To prevent this, we have made it clear that any
                    157: patent must be licensed for everyone's free use or not licensed at all.
                    159:   The precise terms and conditions for copying, distribution and
                    160: modification follow.
                    162: @iftex
                    164: @end iftex
                    165: @ifinfo
                    167: @end ifinfo
                    169: @enumerate 0
                    170: @item
                    171: This License applies to any program or other work which contains
                    172: a notice placed by the copyright holder saying it may be distributed
                    173: under the terms of this General Public License.  The ``Program'', below,
                    174: refers to any such program or work, and a ``work based on the Program''
                    175: means either the Program or any derivative work under copyright law:
                    176: that is to say, a work containing the Program or a portion of it,
                    177: either verbatim or with modifications and/or translated into another
                    178: language.  (Hereinafter, translation is included without limitation in
                    179: the term ``modification''.)  Each licensee is addressed as ``you''.
                    181: Activities other than copying, distribution and modification are not
                    182: covered by this License; they are outside its scope.  The act of
                    183: running the Program is not restricted, and the output from the Program
                    184: is covered only if its contents constitute a work based on the
                    185: Program (independent of having been made by running the Program).
                    186: Whether that is true depends on what the Program does.
                    188: @item
                    189: You may copy and distribute verbatim copies of the Program's
                    190: source code as you receive it, in any medium, provided that you
                    191: conspicuously and appropriately publish on each copy an appropriate
                    192: copyright notice and disclaimer of warranty; keep intact all the
                    193: notices that refer to this License and to the absence of any warranty;
                    194: and give any other recipients of the Program a copy of this License
                    195: along with the Program.
                    197: You may charge a fee for the physical act of transferring a copy, and
                    198: you may at your option offer warranty protection in exchange for a fee.
                    200: @item
                    201: You may modify your copy or copies of the Program or any portion
                    202: of it, thus forming a work based on the Program, and copy and
                    203: distribute such modifications or work under the terms of Section 1
                    204: above, provided that you also meet all of these conditions:
                    206: @enumerate a
                    207: @item
                    208: You must cause the modified files to carry prominent notices
                    209: stating that you changed the files and the date of any change.
                    211: @item
                    212: You must cause any work that you distribute or publish, that in
                    213: whole or in part contains or is derived from the Program or any
                    214: part thereof, to be licensed as a whole at no charge to all third
                    215: parties under the terms of this License.
                    217: @item
                    218: If the modified program normally reads commands interactively
                    219: when run, you must cause it, when started running for such
                    220: interactive use in the most ordinary way, to print or display an
                    221: announcement including an appropriate copyright notice and a
                    222: notice that there is no warranty (or else, saying that you provide
                    223: a warranty) and that users may redistribute the program under
                    224: these conditions, and telling the user how to view a copy of this
                    225: License.  (Exception: if the Program itself is interactive but
                    226: does not normally print such an announcement, your work based on
                    227: the Program is not required to print an announcement.)
                    228: @end enumerate
                    230: These requirements apply to the modified work as a whole.  If
                    231: identifiable sections of that work are not derived from the Program,
                    232: and can be reasonably considered independent and separate works in
                    233: themselves, then this License, and its terms, do not apply to those
                    234: sections when you distribute them as separate works.  But when you
                    235: distribute the same sections as part of a whole which is a work based
                    236: on the Program, the distribution of the whole must be on the terms of
                    237: this License, whose permissions for other licensees extend to the
                    238: entire whole, and thus to each and every part regardless of who wrote it.
                    240: Thus, it is not the intent of this section to claim rights or contest
                    241: your rights to work written entirely by you; rather, the intent is to
                    242: exercise the right to control the distribution of derivative or
                    243: collective works based on the Program.
                    245: In addition, mere aggregation of another work not based on the Program
                    246: with the Program (or with a work based on the Program) on a volume of
                    247: a storage or distribution medium does not bring the other work under
                    248: the scope of this License.
                    250: @item
                    251: You may copy and distribute the Program (or a work based on it,
                    252: under Section 2) in object code or executable form under the terms of
                    253: Sections 1 and 2 above provided that you also do one of the following:
                    255: @enumerate a
                    256: @item
                    257: Accompany it with the complete corresponding machine-readable
                    258: source code, which must be distributed under the terms of Sections
                    259: 1 and 2 above on a medium customarily used for software interchange; or,
                    261: @item
                    262: Accompany it with a written offer, valid for at least three
                    263: years, to give any third party, for a charge no more than your
                    264: cost of physically performing source distribution, a complete
                    265: machine-readable copy of the corresponding source code, to be
                    266: distributed under the terms of Sections 1 and 2 above on a medium
                    267: customarily used for software interchange; or,
                    269: @item
                    270: Accompany it with the information you received as to the offer
                    271: to distribute corresponding source code.  (This alternative is
                    272: allowed only for noncommercial distribution and only if you
                    273: received the program in object code or executable form with such
                    274: an offer, in accord with Subsection b above.)
                    275: @end enumerate
                    277: The source code for a work means the preferred form of the work for
                    278: making modifications to it.  For an executable work, complete source
                    279: code means all the source code for all modules it contains, plus any
                    280: associated interface definition files, plus the scripts used to
                    281: control compilation and installation of the executable.  However, as a
                    282: special exception, the source code distributed need not include
                    283: anything that is normally distributed (in either source or binary
                    284: form) with the major components (compiler, kernel, and so on) of the
                    285: operating system on which the executable runs, unless that component
                    286: itself accompanies the executable.
                    288: If distribution of executable or object code is made by offering
                    289: access to copy from a designated place, then offering equivalent
                    290: access to copy the source code from the same place counts as
                    291: distribution of the source code, even though third parties are not
                    292: compelled to copy the source along with the object code.
                    294: @item
                    295: You may not copy, modify, sublicense, or distribute the Program
                    296: except as expressly provided under this License.  Any attempt
                    297: otherwise to copy, modify, sublicense or distribute the Program is
                    298: void, and will automatically terminate your rights under this License.
                    299: However, parties who have received copies, or rights, from you under
                    300: this License will not have their licenses terminated so long as such
                    301: parties remain in full compliance.
                    303: @item
                    304: You are not required to accept this License, since you have not
                    305: signed it.  However, nothing else grants you permission to modify or
                    306: distribute the Program or its derivative works.  These actions are
                    307: prohibited by law if you do not accept this License.  Therefore, by
                    308: modifying or distributing the Program (or any work based on the
                    309: Program), you indicate your acceptance of this License to do so, and
                    310: all its terms and conditions for copying, distributing or modifying
                    311: the Program or works based on it.
                    313: @item
                    314: Each time you redistribute the Program (or any work based on the
                    315: Program), the recipient automatically receives a license from the
                    316: original licensor to copy, distribute or modify the Program subject to
                    317: these terms and conditions.  You may not impose any further
                    318: restrictions on the recipients' exercise of the rights granted herein.
                    319: You are not responsible for enforcing compliance by third parties to
                    320: this License.
                    322: @item
                    323: If, as a consequence of a court judgment or allegation of patent
                    324: infringement or for any other reason (not limited to patent issues),
                    325: conditions are imposed on you (whether by court order, agreement or
                    326: otherwise) that contradict the conditions of this License, they do not
                    327: excuse you from the conditions of this License.  If you cannot
                    328: distribute so as to satisfy simultaneously your obligations under this
                    329: License and any other pertinent obligations, then as a consequence you
                    330: may not distribute the Program at all.  For example, if a patent
                    331: license would not permit royalty-free redistribution of the Program by
                    332: all those who receive copies directly or indirectly through you, then
                    333: the only way you could satisfy both it and this License would be to
                    334: refrain entirely from distribution of the Program.
                    336: If any portion of this section is held invalid or unenforceable under
                    337: any particular circumstance, the balance of the section is intended to
                    338: apply and the section as a whole is intended to apply in other
                    339: circumstances.
                    341: It is not the purpose of this section to induce you to infringe any
                    342: patents or other property right claims or to contest validity of any
                    343: such claims; this section has the sole purpose of protecting the
                    344: integrity of the free software distribution system, which is
                    345: implemented by public license practices.  Many people have made
                    346: generous contributions to the wide range of software distributed
                    347: through that system in reliance on consistent application of that
                    348: system; it is up to the author/donor to decide if he or she is willing
                    349: to distribute software through any other system and a licensee cannot
                    350: impose that choice.
                    352: This section is intended to make thoroughly clear what is believed to
                    353: be a consequence of the rest of this License.
                    355: @item
                    356: If the distribution and/or use of the Program is restricted in
                    357: certain countries either by patents or by copyrighted interfaces, the
                    358: original copyright holder who places the Program under this License
                    359: may add an explicit geographical distribution limitation excluding
                    360: those countries, so that distribution is permitted only in or among
                    361: countries not thus excluded.  In such case, this License incorporates
                    362: the limitation as if written in the body of this License.
                    364: @item
                    365: The Free Software Foundation may publish revised and/or new versions
                    366: of the General Public License from time to time.  Such new versions will
                    367: be similar in spirit to the present version, but may differ in detail to
                    368: address new problems or concerns.
                    370: Each version is given a distinguishing version number.  If the Program
                    371: specifies a version number of this License which applies to it and ``any
                    372: later version'', you have the option of following the terms and conditions
                    373: either of that version or of any later version published by the Free
                    374: Software Foundation.  If the Program does not specify a version number of
                    375: this License, you may choose any version ever published by the Free Software
                    376: Foundation.
                    378: @item
                    379: If you wish to incorporate parts of the Program into other free
                    380: programs whose distribution conditions are different, write to the author
                    381: to ask for permission.  For software which is copyrighted by the Free
                    382: Software Foundation, write to the Free Software Foundation; we sometimes
                    383: make exceptions for this.  Our decision will be guided by the two goals
                    384: of preserving the free status of all derivatives of our free software and
                    385: of promoting the sharing and reuse of software generally.
                    387: @iftex
                    388: @heading NO WARRANTY
                    389: @end iftex
                    390: @ifinfo
                    391: @center NO WARRANTY
                    392: @end ifinfo
                    394: @item
                    403: REPAIR OR CORRECTION.
                    405: @item
                    414: POSSIBILITY OF SUCH DAMAGES.
                    415: @end enumerate
                    417: @iftex
                    418: @heading END OF TERMS AND CONDITIONS
                    419: @end iftex
                    420: @ifinfo
                    421: @center END OF TERMS AND CONDITIONS
                    422: @end ifinfo
                    424: @page
                    425: @unnumberedsec How to Apply These Terms to Your New Programs
                    427:   If you develop a new program, and you want it to be of the greatest
                    428: possible use to the public, the best way to achieve this is to make it
                    429: free software which everyone can redistribute and change under these terms.
                    431:   To do so, attach the following notices to the program.  It is safest
                    432: to attach them to the start of each source file to most effectively
                    433: convey the exclusion of warranty; and each file should have at least
                    434: the ``copyright'' line and a pointer to where the full notice is found.
                    436: @smallexample
                    437: @var{one line to give the program's name and a brief idea of what it does.}
                    438: Copyright (C) 19@var{yy}  @var{name of author}
                    440: This program is free software; you can redistribute it and/or modify 
                    441: it under the terms of the GNU General Public License as published by 
                    442: the Free Software Foundation; either version 2 of the License, or 
                    443: (at your option) any later version.
                    445: This program is distributed in the hope that it will be useful,
                    446: but WITHOUT ANY WARRANTY; without even the implied warranty of
                    448: GNU General Public License for more details.
                    450: You should have received a copy of the GNU General Public License
                    451: along with this program; if not, write to the Free Software
                    452: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
                    453: @end smallexample
                    455: Also add information on how to contact you by electronic and paper mail.
                    457: If the program is interactive, make it output a short notice like this
                    458: when it starts in an interactive mode:
                    460: @smallexample
                    461: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
                    462: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
                    463: type `show w'.  
                    464: This is free software, and you are welcome to redistribute it 
                    465: under certain conditions; type `show c' for details.
                    466: @end smallexample
                    468: The hypothetical commands @samp{show w} and @samp{show c} should show
                    469: the appropriate parts of the General Public License.  Of course, the
                    470: commands you use may be called something other than @samp{show w} and
                    471: @samp{show c}; they could even be mouse-clicks or menu items---whatever
                    472: suits your program.
                    474: You should also get your employer (if you work as a programmer) or your
                    475: school, if any, to sign a ``copyright disclaimer'' for the program, if
                    476: necessary.  Here is a sample; alter the names:
                    478: @smallexample
                    479: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
                    480: `Gnomovision' (which makes passes at compilers) written by James Hacker.
                    482: @var{signature of Ty Coon}, 1 April 1989
                    483: Ty Coon, President of Vice
                    484: @end smallexample
                    486: This General Public License does not permit incorporating your program into
                    487: proprietary programs.  If your program is a subroutine library, you may
                    488: consider it more useful to permit linking proprietary applications with the
                    489: library.  If this is what you want to do, use the GNU Library General
                    490: Public License instead of this License.
1.1       anton     491: 
                    492: @iftex
1.23      pazsan    493: @node    Preface
                    494: @comment node-name,     next,           previous, up
1.1       anton     495: @unnumbered Preface
1.23      pazsan    496: @cindex Preface
1.17      anton     497: This manual documents Gforth. The reader is expected to know
1.1       anton     498: Forth. This manual is primarily a reference manual. @xref{Other Books}
                    499: for introductory material.
                    500: @end iftex
                    502: @node    Goals, Other Books, License, Top
                    503: @comment node-name,     next,           previous, up
1.17      anton     504: @chapter Goals of Gforth
1.1       anton     505: @cindex Goals
1.17      anton     506: The goal of the Gforth Project is to develop a standard model for
1.1       anton     507: ANSI Forth. This can be split into several subgoals:
                    509: @itemize @bullet
                    510: @item
1.17      anton     511: Gforth should conform to the ANSI Forth standard.
1.1       anton     512: @item
                    513: It should be a model, i.e. it should define all the
                    514: implementation-dependent things.
                    515: @item
                    516: It should become standard, i.e. widely accepted and used. This goal
                    517: is the most difficult one.
                    518: @end itemize
1.17      anton     520: To achieve these goals Gforth should be
1.1       anton     521: @itemize @bullet
                    522: @item
                    523: Similar to previous models (fig-Forth, F83)
                    524: @item
                    525: Powerful. It should provide for all the things that are considered
                    526: necessary today and even some that are not yet considered necessary.
                    527: @item
                    528: Efficient. It should not get the reputation of being exceptionally
                    529: slow.
                    530: @item
                    531: Free.
                    532: @item
                    533: Available on many machines/easy to port.
                    534: @end itemize
1.17      anton     536: Have we achieved these goals? Gforth conforms to the ANS Forth
                    537: standard. It may be considered a model, but we have not yet documented
1.1       anton     538: which parts of the model are stable and which parts we are likely to
1.17      anton     539: change. It certainly has not yet become a de facto standard. It has some
                    540: similarities and some differences to previous models. It has some
                    541: powerful features, but not yet everything that we envisioned. We
                    542: certainly have achieved our execution speed goals (@pxref{Performance}).
                    543: It is free and available on many machines.
1.1       anton     544: 
                    545: @node Other Books, Invocation, Goals, Top
                    546: @chapter Other books on ANS Forth
                    548: As the standard is relatively new, there are not many books out yet. It
1.17      anton     549: is not recommended to learn Forth by using Gforth and a book that is
1.1       anton     550: not written for ANS Forth, as you will not know your mistakes from the
                    551: deviations of the book.
                    553: There is, of course, the standard, the definite reference if you want to
1.19      anton     554: write ANS Forth programs. It is available in printed form from the
                    555: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
                    556: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
                    557: can also get it from Global Engineering Documents (Tel.: USA (800)
                    558: 854-7179; Fax.: (303) 843-9880) for about $300.
                    560: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
                    561: for publication is available electronically and for free in some MS Word
                    562: format, and it has been converted to HTML. Some pointers to these
                    563: versions can be found through
1.24      anton     564: @*@file{}.
1.1       anton     565: 
1.21      anton     566: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
1.1       anton     567: introductory book based on a draft version of the standard. It does not
                    568: cover the whole standard. It also contains interesting background
                    569: information (Jack Woehr was in the ANS Forth Technical Committe). It is
                    570: not appropriate for complete newbies, but programmers experienced in
                    571: other languages should find it ok.
                    573: @node Invocation, Words, Other Books, Top
                    574: @chapter Invocation
                    576: You will usually just say @code{gforth}. In many other cases the default
1.17      anton     577: Gforth image will be invoked like this:
1.1       anton     578: 
                    579: @example
                    580: gforth [files] [-e forth-code]
                    581: @end example
                    583: executing the contents of the files and the Forth code in the order they
                    584: are given.
                    586: In general, the command line looks like this:
                    588: @example
                    589: gforth [initialization options] [image-specific options]
                    590: @end example
                    592: The initialization options must come before the rest of the command
                    593: line. They are:
                    595: @table @code
                    596: @item --image-file @var{file}
1.20      pazsan    597: @item -i @var{file}
1.1       anton     598: Loads the Forth image @var{file} instead of the default
                    599: @file{}.
                    601: @item --path @var{path}
1.20      pazsan    602: @item -p @var{path}
1.1       anton     603: Uses @var{path} for searching the image file and Forth source code
                    604: files instead of the default in the environment variable
                    605: @code{GFORTHPATH} or the path specified at installation time (typically
                    606: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
                    607: list.
                    609: @item --dictionary-size @var{size}
                    610: @item -m @var{size}
                    611: Allocate @var{size} space for the Forth dictionary space instead of
                    612: using the default specified in the image (typically 256K). The
                    613: @var{size} specification consists of an integer and a unit (e.g.,
                    614: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
                    615: size, in this case Cells), @code{k} (kilobytes), and @code{M}
                    616: (Megabytes). If no unit is specified, @code{e} is used.
                    618: @item --data-stack-size @var{size}
                    619: @item -d @var{size}
                    620: Allocate @var{size} space for the data stack instead of using the
                    621: default specified in the image (typically 16K).
                    623: @item --return-stack-size @var{size}
                    624: @item -r @var{size}
                    625: Allocate @var{size} space for the return stack instead of using the
                    626: default specified in the image (typically 16K).
                    628: @item --fp-stack-size @var{size}
                    629: @item -f @var{size}
                    630: Allocate @var{size} space for the floating point stack instead of
                    631: using the default specified in the image (typically 16K). In this case
                    632: the unit specifier @code{e} refers to floating point numbers.
                    634: @item --locals-stack-size @var{size}
                    635: @item -l @var{size}
                    636: Allocate @var{size} space for the locals stack instead of using the
                    637: default specified in the image (typically 16K).
                    639: @end table
                    641: As explained above, the image-specific command-line arguments for the
                    642: default image @file{} consist of a sequence of filenames and
                    643: @code{-e @var{forth-code}} options that are interpreted in the seqence
                    644: in which they are given. The @code{-e @var{forth-code}} or
                    645: @code{--evaluate @var{forth-code}} option evaluates the forth
                    646: code. This option takes only one argument; if you want to evaluate more
                    647: Forth words, you have to quote them or use several @code{-e}s. To exit
                    648: after processing the command line (instead of entering interactive mode)
                    649: append @code{-e bye} to the command line.
1.22      anton     651: If you have several versions of Gforth installed, @code{gforth} will
                    652: invoke the version that was installed last. @code{gforth-@var{version}}
                    653: invokes a specific version. You may want to use the option
                    654: @code{--path}, if your environment contains the variable
                    655: @code{GFORTHPATH}.
1.1       anton     657: Not yet implemented:
                    658: On startup the system first executes the system initialization file
                    659: (unless the option @code{--no-init-file} is given; note that the system
                    660: resulting from using this option may not be ANS Forth conformant). Then
                    661: the user initialization file @file{.gforth.fs} is executed, unless the
                    662: option @code{--no-rc} is given; this file is first searched in @file{.},
                    663: then in @file{~}, then in the normal path (see above).
1.4       anton     665: @node Words, ANS conformance, Invocation, Top
1.1       anton     666: @chapter Forth Words
                    668: @menu
1.4       anton     669: * Notation::                    
                    670: * Arithmetic::                  
                    671: * Stack Manipulation::          
                    672: * Memory access::               
                    673: * Control Structures::          
                    674: * Locals::                      
                    675: * Defining Words::              
1.37      anton     676: * Tokens for Words::            
1.4       anton     677: * Wordlists::                   
                    678: * Files::                       
                    679: * Blocks::                      
                    680: * Other I/O::                   
                    681: * Programming Tools::           
1.18      anton     682: * Assembler and Code words::    
1.4       anton     683: * Threading Words::             
1.1       anton     684: @end menu
                    686: @node Notation, Arithmetic, Words, Words
                    687: @section Notation
                    689: The Forth words are described in this section in the glossary notation
                    690: that has become a de-facto standard for Forth texts, i.e.
1.4       anton     692: @format
1.1       anton     693: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
1.4       anton     694: @end format
1.1       anton     695: @var{Description}
                    697: @table @var
                    698: @item word
1.17      anton     699: The name of the word. BTW, Gforth is case insensitive, so you can
1.14      anton     700: type the words in in lower case (However, @pxref{core-idef}).
1.1       anton     701: 
                    702: @item Stack effect
                    703: The stack effect is written in the notation @code{@var{before} --
                    704: @var{after}}, where @var{before} and @var{after} describe the top of
                    705: stack entries before and after the execution of the word. The rest of
                    706: the stack is not touched by the word. The top of stack is rightmost,
1.17      anton     707: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1       anton     708: uses a separate floating point stack, but a unified stack
                    709: notation. Also, return stack effects are not shown in @var{stack
                    710: effect}, but in @var{Description}. The name of a stack item describes
                    711: the type and/or the function of the item. See below for a discussion of
                    712: the types.
1.19      anton     714: All words have two stack effects: A compile-time stack effect and a
                    715: run-time stack effect. The compile-time stack-effect of most words is
                    716: @var{ -- }. If the compile-time stack-effect of a word deviates from
                    717: this standard behaviour, or the word does other unusual things at
                    718: compile time, both stack effects are shown; otherwise only the run-time
                    719: stack effect is shown.
1.1       anton     721: @item pronunciation
                    722: How the word is pronounced
                    724: @item wordset
                    725: The ANS Forth standard is divided into several wordsets. A standard
                    726: system need not support all of them. So, the fewer wordsets your program
                    727: uses the more portable it will be in theory. However, we suspect that
                    728: most ANS Forth systems on personal machines will feature all
                    729: wordsets. Words that are not defined in the ANS standard have
1.19      anton     730: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
                    731: describes words that will work in future releases of Gforth;
                    732: @code{gforth-internal} words are more volatile. Environmental query
                    733: strings are also displayed like words; you can recognize them by the
                    734: @code{environment} in the wordset field.
1.1       anton     735: 
                    736: @item Description
                    737: A description of the behaviour of the word.
                    738: @end table
1.4       anton     740: The type of a stack item is specified by the character(s) the name
                    741: starts with:
1.1       anton     742: 
                    743: @table @code
                    744: @item f
                    745: Bool, i.e. @code{false} or @code{true}.
                    746: @item c
                    747: Char
                    748: @item w
                    749: Cell, can contain an integer or an address
                    750: @item n
                    751: signed integer
                    752: @item u
                    753: unsigned integer
                    754: @item d
                    755: double sized signed integer
                    756: @item ud
                    757: double sized unsigned integer
                    758: @item r
1.36      anton     759: Float (on the FP stack)
1.1       anton     760: @item a_
                    761: Cell-aligned address
                    762: @item c_
1.36      anton     763: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.1       anton     764: @item f_
                    765: Float-aligned address
                    766: @item df_
                    767: Address aligned for IEEE double precision float
                    768: @item sf_
                    769: Address aligned for IEEE single precision float
                    770: @item xt
                    771: Execution token, same size as Cell
                    772: @item wid
                    773: Wordlist ID, same size as Cell
                    774: @item f83name
                    775: Pointer to a name structure
1.36      anton     776: @item "
                    777: string in the input stream (not the stack). The terminating character is
                    778: a blank by default. If it is not a blank, it is shown in @code{<>}
                    779: quotes.
1.1       anton     781: @end table
1.4       anton     783: @node Arithmetic, Stack Manipulation, Notation, Words
1.1       anton     784: @section Arithmetic
                    785: Forth arithmetic is not checked, i.e., you will not hear about integer
                    786: overflow on addition or multiplication, you may hear about division by
                    787: zero if you are lucky. The operator is written after the operands, but
                    788: the operands are still in the original order. I.e., the infix @code{2-1}
                    789: corresponds to @code{2 1 -}. Forth offers a variety of division
                    790: operators. If you perform division with potentially negative operands,
                    791: you do not want to use @code{/} or @code{/mod} with its undefined
                    792: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4       anton     793: former, @pxref{Mixed precision}).
                    795: @menu
                    796: * Single precision::            
                    797: * Bitwise operations::          
                    798: * Mixed precision::             operations with single and double-cell integers
                    799: * Double precision::            Double-cell integer arithmetic
                    800: * Floating Point::              
                    801: @end menu
1.1       anton     802: 
1.4       anton     803: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1       anton     804: @subsection Single precision
                    805: doc-+
                    806: doc--
                    807: doc-*
                    808: doc-/
                    809: doc-mod
                    810: doc-/mod
                    811: doc-negate
                    812: doc-abs
                    813: doc-min
                    814: doc-max
1.4       anton     816: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1       anton     817: @subsection Bitwise operations
                    818: doc-and
                    819: doc-or
                    820: doc-xor
                    821: doc-invert
                    822: doc-2*
                    823: doc-2/
1.4       anton     825: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1       anton     826: @subsection Mixed precision
                    827: doc-m+
                    828: doc-*/
                    829: doc-*/mod
                    830: doc-m*
                    831: doc-um*
                    832: doc-m*/
                    833: doc-um/mod
                    834: doc-fm/mod
                    835: doc-sm/rem
1.4       anton     837: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1       anton     838: @subsection Double precision
1.16      anton     839: 
                    840: The outer (aka text) interpreter converts numbers containing a dot into
                    841: a double precision number. Note that only numbers with the dot as last
                    842: character are standard-conforming.
1.1       anton     844: doc-d+
                    845: doc-d-
                    846: doc-dnegate
                    847: doc-dabs
                    848: doc-dmin
                    849: doc-dmax
1.4       anton     851: @node Floating Point,  , Double precision, Arithmetic
                    852: @subsection Floating Point
1.16      anton     853: 
                    854: The format of floating point numbers recognized by the outer (aka text)
                    855: interpreter is: a signed decimal number, possibly containing a decimal
                    856: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
                    857: by a signed integer (the exponent). E.g., @code{1e} ist the same as
1.35      anton     858: @code{+1.0e+0}. Note that a number without @code{e}
1.16      anton     859: is not interpreted as floating-point number, but as double (if the
                    860: number contains a @code{.}) or single precision integer. Also,
                    861: conversions between string and floating point numbers always use base
                    862: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
                    863: value greater then 14, the @code{E} may be interpreted as digit and the
                    864: number will be interpreted as integer, unless it has a signed exponent
                    865: (both @code{+} and @code{-} are allowed as signs).
1.4       anton     866: 
                    867: Angles in floating point operations are given in radians (a full circle
1.17      anton     868: has 2 pi radians). Note, that Gforth has a separate floating point
1.4       anton     869: stack, but we use the unified notation.
                    871: Floating point numbers have a number of unpleasant surprises for the
                    872: unwary (e.g., floating point addition is not associative) and even a few
                    873: for the wary. You should not use them unless you know what you are doing
                    874: or you don't care that the results you get are totally bogus. If you
                    875: want to learn about the problems of floating point numbers (and how to
1.11      anton     876: avoid them), you might start with @cite{David Goldberg, What Every
1.6       anton     877: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
                    878: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4       anton     879: 
                    880: doc-f+
                    881: doc-f-
                    882: doc-f*
                    883: doc-f/
                    884: doc-fnegate
                    885: doc-fabs
                    886: doc-fmax
                    887: doc-fmin
                    888: doc-floor
                    889: doc-fround
                    890: doc-f**
                    891: doc-fsqrt
                    892: doc-fexp
                    893: doc-fexpm1
                    894: doc-fln
                    895: doc-flnp1
                    896: doc-flog
1.6       anton     897: doc-falog
1.4       anton     898: doc-fsin
                    899: doc-fcos
                    900: doc-fsincos
                    901: doc-ftan
                    902: doc-fasin
                    903: doc-facos
                    904: doc-fatan
                    905: doc-fatan2
                    906: doc-fsinh
                    907: doc-fcosh
                    908: doc-ftanh
                    909: doc-fasinh
                    910: doc-facosh
                    911: doc-fatanh
                    913: @node Stack Manipulation, Memory access, Arithmetic, Words
1.1       anton     914: @section Stack Manipulation
1.17      anton     916: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1       anton     917: addresses, and double cells, a floating point stack for floating point
                    918: numbers, a return stack for storing the return addresses of colon
                    919: definitions and other data, and a locals stack for storing local
                    920: variables. Note that while every sane Forth has a separate floating
                    921: point stack, this is not strictly required; an ANS Forth system could
                    922: theoretically keep floating point numbers on the data stack. As an
                    923: additional difficulty, you don't know how many cells a floating point
                    924: number takes. It is reportedly possible to write words in a way that
                    925: they work also for a unified stack model, but we do not recommend trying
1.4       anton     926: it. Instead, just say that your program has an environmental dependency
                    927: on a separate FP stack.
                    929: Also, a Forth system is allowed to keep the local variables on the
1.1       anton     930: return stack. This is reasonable, as local variables usually eliminate
                    931: the need to use the return stack explicitly. So, if you want to produce
                    932: a standard complying program and if you are using local variables in a
                    933: word, forget about return stack manipulations in that word (see the
                    934: standard document for the exact rules).
1.4       anton     936: @menu
                    937: * Data stack::                  
                    938: * Floating point stack::        
                    939: * Return stack::                
                    940: * Locals stack::                
                    941: * Stack pointer manipulation::  
                    942: @end menu
                    944: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1       anton     945: @subsection Data stack
                    946: doc-drop
                    947: doc-nip
                    948: doc-dup
                    949: doc-over
                    950: doc-tuck
                    951: doc-swap
                    952: doc-rot
                    953: doc--rot
                    954: doc-?dup
                    955: doc-pick
                    956: doc-roll
                    957: doc-2drop
                    958: doc-2nip
                    959: doc-2dup
                    960: doc-2over
                    961: doc-2tuck
                    962: doc-2swap
                    963: doc-2rot
1.4       anton     965: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1       anton     966: @subsection Floating point stack
                    967: doc-fdrop
                    968: doc-fnip
                    969: doc-fdup
                    970: doc-fover
                    971: doc-ftuck
                    972: doc-fswap
                    973: doc-frot
1.4       anton     975: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1       anton     976: @subsection Return stack
                    977: doc->r
                    978: doc-r>
                    979: doc-r@
                    980: doc-rdrop
                    981: doc-2>r
                    982: doc-2r>
                    983: doc-2r@
                    984: doc-2rdrop
1.4       anton     986: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1       anton     987: @subsection Locals stack
1.4       anton     989: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
1.1       anton     990: @subsection Stack pointer manipulation
                    991: doc-sp@
                    992: doc-sp!
                    993: doc-fp@
                    994: doc-fp!
                    995: doc-rp@
                    996: doc-rp!
                    997: doc-lp@
                    998: doc-lp!
1.4       anton    1000: @node Memory access, Control Structures, Stack Manipulation, Words
1.1       anton    1001: @section Memory access
1.4       anton    1003: @menu
                   1004: * Stack-Memory transfers::      
                   1005: * Address arithmetic::          
                   1006: * Memory block access::         
                   1007: @end menu
                   1009: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1.1       anton    1010: @subsection Stack-Memory transfers
                   1012: doc-@
                   1013: doc-!
                   1014: doc-+!
                   1015: doc-c@
                   1016: doc-c!
                   1017: doc-2@
                   1018: doc-2!
                   1019: doc-f@
                   1020: doc-f!
                   1021: doc-sf@
                   1022: doc-sf!
                   1023: doc-df@
                   1024: doc-df!
1.4       anton    1026: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1.1       anton    1027: @subsection Address arithmetic
                   1029: ANS Forth does not specify the sizes of the data types. Instead, it
                   1030: offers a number of words for computing sizes and doing address
                   1031: arithmetic. Basically, address arithmetic is performed in terms of
                   1032: address units (aus); on most systems the address unit is one byte. Note
                   1033: that a character may have more than one au, so @code{chars} is no noop
                   1034: (on systems where it is a noop, it compiles to nothing).
                   1036: ANS Forth also defines words for aligning addresses for specific
                   1037: addresses. Many computers require that accesses to specific data types
                   1038: must only occur at specific addresses; e.g., that cells may only be
                   1039: accessed at addresses divisible by 4. Even if a machine allows unaligned
                   1040: accesses, it can usually perform aligned accesses faster. 
1.17      anton    1042: For the performance-conscious: alignment operations are usually only
1.1       anton    1043: necessary during the definition of a data structure, not during the
                   1044: (more frequent) accesses to it.
                   1046: ANS Forth defines no words for character-aligning addresses. This is not
                   1047: an oversight, but reflects the fact that addresses that are not
                   1048: char-aligned have no use in the standard and therefore will not be
                   1049: created.
                   1051: The standard guarantees that addresses returned by @code{CREATE}d words
1.17      anton    1052: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1       anton    1053: are aligned for all purposes.
1.9       anton    1055: Note that the standard defines a word @code{char}, which has nothing to
                   1056: do with address arithmetic.
1.1       anton    1058: doc-chars
                   1059: doc-char+
                   1060: doc-cells
                   1061: doc-cell+
                   1062: doc-align
                   1063: doc-aligned
                   1064: doc-floats
                   1065: doc-float+
                   1066: doc-falign
                   1067: doc-faligned
                   1068: doc-sfloats
                   1069: doc-sfloat+
                   1070: doc-sfalign
                   1071: doc-sfaligned
                   1072: doc-dfloats
                   1073: doc-dfloat+
                   1074: doc-dfalign
                   1075: doc-dfaligned
1.10      anton    1076: doc-maxalign
                   1077: doc-maxaligned
                   1078: doc-cfalign
                   1079: doc-cfaligned
1.1       anton    1080: doc-address-unit-bits
1.4       anton    1082: @node Memory block access,  , Address arithmetic, Memory access
1.1       anton    1083: @subsection Memory block access
                   1085: doc-move
                   1086: doc-erase
                   1088: While the previous words work on address units, the rest works on
                   1089: characters.
                   1091: doc-cmove
                   1092: doc-cmove>
                   1093: doc-fill
                   1094: doc-blank
1.4       anton    1096: @node Control Structures, Locals, Memory access, Words
1.1       anton    1097: @section Control Structures
                   1099: Control structures in Forth cannot be used in interpret state, only in
                   1100: compile state, i.e., in a colon definition. We do not like this
                   1101: limitation, but have not seen a satisfying way around it yet, although
                   1102: many schemes have been proposed.
1.4       anton    1104: @menu
                   1105: * Selection::                   
                   1106: * Simple Loops::                
                   1107: * Counted Loops::               
                   1108: * Arbitrary control structures::  
                   1109: * Calls and returns::           
                   1110: * Exception Handling::          
                   1111: @end menu
                   1113: @node Selection, Simple Loops, Control Structures, Control Structures
1.1       anton    1114: @subsection Selection
                   1116: @example
                   1117: @var{flag}
                   1118: IF
                   1119:   @var{code}
                   1120: ENDIF
                   1121: @end example
                   1122: or
                   1123: @example
                   1124: @var{flag}
                   1125: IF
                   1126:   @var{code1}
                   1127: ELSE
                   1128:   @var{code2}
                   1129: ENDIF
                   1130: @end example
1.4       anton    1132: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1       anton    1133: standard, and @code{ENDIF} is not, although it is quite popular. We
                   1134: recommend using @code{ENDIF}, because it is less confusing for people
                   1135: who also know other languages (and is not prone to reinforcing negative
                   1136: prejudices against Forth in these people). Adding @code{ENDIF} to a
                   1137: system that only supplies @code{THEN} is simple:
                   1138: @example
                   1139: : endif   POSTPONE then ; immediate
                   1140: @end example
                   1142: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
                   1143: (adv.)}  has the following meanings:
                   1144: @quotation
                   1145: ... 2b: following next after in order ... 3d: as a necessary consequence
                   1146: (if you were there, then you saw them).
                   1147: @end quotation
                   1148: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
                   1149: and many other programming languages has the meaning 3d.]
1.31      anton    1151: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
                   1152: you can avoid using @code{?dup}. Using these alternatives is also more
                   1153: efficient than using @code{?dup}. Definitions in plain standard Forth
                   1154: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
                   1155: @file{compat/control.fs}.
1.1       anton    1156: 
                   1157: @example
                   1158: @var{n}
                   1159: CASE
                   1160:   @var{n1} OF @var{code1} ENDOF
                   1161:   @var{n2} OF @var{code2} ENDOF
1.4       anton    1162:   @dots{}
1.1       anton    1163: ENDCASE
                   1164: @end example
                   1166: Executes the first @var{codei}, where the @var{ni} is equal to
                   1167: @var{n}. A default case can be added by simply writing the code after
                   1168: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
                   1169: but must not consume it.
1.4       anton    1171: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1       anton    1172: @subsection Simple Loops
                   1174: @example
                   1175: BEGIN
                   1176:   @var{code1}
                   1177:   @var{flag}
                   1178: WHILE
                   1179:   @var{code2}
                   1180: REPEAT
                   1181: @end example
                   1183: @var{code1} is executed and @var{flag} is computed. If it is true,
                   1184: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
                   1186: @example
                   1187: BEGIN
                   1188:   @var{code}
                   1189:   @var{flag}
                   1190: UNTIL
                   1191: @end example
                   1193: @var{code} is executed. The loop is restarted if @code{flag} is false.
                   1195: @example
                   1196: BEGIN
                   1197:   @var{code}
                   1198: AGAIN
                   1199: @end example
                   1201: This is an endless loop.
1.4       anton    1203: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1       anton    1204: @subsection Counted Loops
                   1206: The basic counted loop is:
                   1207: @example
                   1208: @var{limit} @var{start}
                   1209: ?DO
                   1210:   @var{body}
                   1211: LOOP
                   1212: @end example
                   1214: This performs one iteration for every integer, starting from @var{start}
                   1215: and up to, but excluding @var{limit}. The counter, aka index, can be
                   1216: accessed with @code{i}. E.g., the loop
                   1217: @example
                   1218: 10 0 ?DO
                   1219:   i .
                   1220: LOOP
                   1221: @end example
                   1222: prints
                   1223: @example
                   1224: 0 1 2 3 4 5 6 7 8 9
                   1225: @end example
                   1226: The index of the innermost loop can be accessed with @code{i}, the index
                   1227: of the next loop with @code{j}, and the index of the third loop with
                   1228: @code{k}.
                   1230: The loop control data are kept on the return stack, so there are some
                   1231: restrictions on mixing return stack accesses and counted loop
                   1232: words. E.g., if you put values on the return stack outside the loop, you
                   1233: cannot read them inside the loop. If you put values on the return stack
                   1234: within a loop, you have to remove them before the end of the loop and
                   1235: before accessing the index of the loop.
                   1237: There are several variations on the counted loop:
                   1239: @code{LEAVE} leaves the innermost counted loop immediately.
1.18      anton    1241: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
                   1242: (and @code{LOOP} iterates until they become equal by wrap-around
                   1243: arithmetic). This behaviour is usually not what you want. Therefore,
                   1244: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
                   1245: @code{?DO}), which do not enter the loop if @var{start} is greater than
                   1246: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.30      anton    1247: unsigned loop parameters.
1.18      anton    1248: 
1.1       anton    1249: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
                   1250: index by @var{n} instead of by 1. The loop is terminated when the border
                   1251: between @var{limit-1} and @var{limit} is crossed. E.g.:
1.18      anton    1253: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
1.1       anton    1254: 
1.18      anton    1255: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
1.1       anton    1256: 
                   1257: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1.2       anton    1259: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
1.1       anton    1260: 
1.2       anton    1261: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
1.1       anton    1262: 
1.18      anton    1263: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
                   1264: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
                   1265: index by @var{u} each iteration. The loop is terminated when the border
                   1266: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
                   1267: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1       anton    1268: 
1.18      anton    1269: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
1.1       anton    1270: 
1.18      anton    1271: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
1.1       anton    1272: 
1.18      anton    1273: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
1.1       anton    1274: 
1.30      anton    1275: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
                   1276: @code{-LOOP} are not in the ANS Forth standard. However, an
                   1277: implementation for these words that uses only standard words is provided
                   1278: in @file{compat/loops.fs}.
1.18      anton    1279: 
                   1280: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
                   1281: the loop, independent of the loop parameters. Do not use @code{DO}, even
                   1282: if you know that the loop is entered in any case. Such knowledge tends
                   1283: to become invalid during maintenance of a program, and then the
                   1284: @code{DO} will make trouble.
1.1       anton    1285: 
                   1286: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
                   1287: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
                   1288: return stack so @code{EXIT} can get to its return address.
                   1290: Another counted loop is
                   1291: @example
                   1292: @var{n}
                   1293: FOR
                   1294:   @var{body}
                   1295: NEXT
                   1296: @end example
                   1297: This is the preferred loop of native code compiler writers who are too
1.17      anton    1298: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1       anton    1299: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
                   1300: and ending with 0. Other Forth systems may behave differently, even if
1.30      anton    1301: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
                   1302: loops.
1.1       anton    1303: 
1.4       anton    1304: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2       anton    1305: @subsection Arbitrary control structures
                   1307: ANS Forth permits and supports using control structures in a non-nested
                   1308: way. Information about incomplete control structures is stored on the
                   1309: control-flow stack. This stack may be implemented on the Forth data
1.17      anton    1310: stack, and this is what we have done in Gforth.
1.2       anton    1311: 
                   1312: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
                   1313: entry represents a backward branch target. A few words are the basis for
                   1314: building any control structure possible (except control structures that
                   1315: need storage, like calls, coroutines, and backtracking).
1.3       anton    1317: doc-if
                   1318: doc-ahead
                   1319: doc-then
                   1320: doc-begin
                   1321: doc-until
                   1322: doc-again
                   1323: doc-cs-pick
                   1324: doc-cs-roll
1.2       anton    1325: 
1.17      anton    1326: On many systems control-flow stack items take one word, in Gforth they
1.2       anton    1327: currently take three (this may change in the future). Therefore it is a
                   1328: really good idea to manipulate the control flow stack with
                   1329: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
                   1330: words.
                   1332: Some standard control structure words are built from these words:
1.3       anton    1334: doc-else
                   1335: doc-while
                   1336: doc-repeat
1.2       anton    1337: 
1.31      anton    1338: Gforth adds some more control-structure words:
                   1340: doc-endif
                   1341: doc-?dup-if
                   1342: doc-?dup-0=-if
1.2       anton    1344: Counted loop words constitute a separate group of words:
1.3       anton    1346: doc-?do
1.18      anton    1347: doc-+do
                   1348: doc-u+do
                   1349: doc--do
                   1350: doc-u-do
1.3       anton    1351: doc-do
                   1352: doc-for
                   1353: doc-loop
                   1354: doc-+loop
1.18      anton    1355: doc--loop
1.3       anton    1356: doc-next
                   1357: doc-leave
                   1358: doc-?leave
                   1359: doc-unloop
1.10      anton    1360: doc-done
1.2       anton    1361: 
                   1362: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
                   1363: @i{do-sys}. Our system allows it, but it's your job to ensure that for
                   1364: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3       anton    1365: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
                   1366: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7       pazsan   1367: resolved (by using one of the loop-ending words or @code{DONE}).
1.2       anton    1368: 
                   1369: Another group of control structure words are
1.3       anton    1371: doc-case
                   1372: doc-endcase
                   1373: doc-of
                   1374: doc-endof
1.2       anton    1375: 
                   1376: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
                   1377: @code{cs-roll}.
1.3       anton    1379: @subsubsection Programming Style
                   1381: In order to ensure readability we recommend that you do not create
                   1382: arbitrary control structures directly, but define new control structure
                   1383: words for the control structure you want and use these words in your
                   1384: program.
                   1386: E.g., instead of writing
                   1388: @example
                   1389: begin
                   1390:   ...
                   1391: if [ 1 cs-roll ]
                   1392:   ...
                   1393: again then
                   1394: @end example
                   1396: we recommend defining control structure words, e.g.,
                   1398: @example
                   1399: : while ( dest -- orig dest )
                   1400:  POSTPONE if
                   1401:  1 cs-roll ; immediate
                   1403: : repeat ( orig dest -- )
                   1404:  POSTPONE again
                   1405:  POSTPONE then ; immediate
                   1406: @end example
                   1408: and then using these to create the control structure:
                   1410: @example
                   1411: begin
                   1412:   ...
                   1413: while
                   1414:   ...
                   1415: repeat
                   1416: @end example
1.30      anton    1418: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1.3       anton    1419: @code{WHILE} are predefined, so in this example it would not be
                   1420: necessary to define them.
1.4       anton    1422: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3       anton    1423: @subsection Calls and returns
                   1425: A definition can be called simply be writing the name of the
1.17      anton    1426: definition. When the end of the definition is reached, it returns. An
                   1427: earlier return can be forced using
1.3       anton    1428: 
                   1429: doc-exit
                   1431: Don't forget to clean up the return stack and @code{UNLOOP} any
                   1432: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
                   1433: primitive compiled by @code{EXIT} is
                   1435: doc-;s
1.4       anton    1437: @node Exception Handling,  , Calls and returns, Control Structures
1.3       anton    1438: @subsection Exception Handling
                   1440: doc-catch
                   1441: doc-throw
1.4       anton    1443: @node Locals, Defining Words, Control Structures, Words
1.1       anton    1444: @section Locals
1.2       anton    1446: Local variables can make Forth programming more enjoyable and Forth
                   1447: programs easier to read. Unfortunately, the locals of ANS Forth are
                   1448: laden with restrictions. Therefore, we provide not only the ANS Forth
                   1449: locals wordset, but also our own, more powerful locals wordset (we
                   1450: implemented the ANS Forth locals wordset through our locals wordset).
1.24      anton    1452: The ideas in this section have also been published in the paper
                   1453: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
                   1454: at EuroForth '94; it is available at
                   1455: @*@file{}.
1.2       anton    1457: @menu
1.17      anton    1458: * Gforth locals::               
1.4       anton    1459: * ANS Forth locals::            
1.2       anton    1460: @end menu
1.17      anton    1462: @node Gforth locals, ANS Forth locals, Locals, Locals
                   1463: @subsection Gforth locals
1.2       anton    1464: 
                   1465: Locals can be defined with
                   1467: @example
                   1468: @{ local1 local2 ... -- comment @}
                   1469: @end example
                   1470: or
                   1471: @example
                   1472: @{ local1 local2 ... @}
                   1473: @end example
                   1475: E.g.,
                   1476: @example
                   1477: : max @{ n1 n2 -- n3 @}
                   1478:  n1 n2 > if
                   1479:    n1
                   1480:  else
                   1481:    n2
                   1482:  endif ;
                   1483: @end example
                   1485: The similarity of locals definitions with stack comments is intended. A
                   1486: locals definition often replaces the stack comment of a word. The order
                   1487: of the locals corresponds to the order in a stack comment and everything
                   1488: after the @code{--} is really a comment.
                   1490: This similarity has one disadvantage: It is too easy to confuse locals
                   1491: declarations with stack comments, causing bugs and making them hard to
                   1492: find. However, this problem can be avoided by appropriate coding
                   1493: conventions: Do not use both notations in the same program. If you do,
                   1494: they should be distinguished using additional means, e.g. by position.
                   1496: The name of the local may be preceded by a type specifier, e.g.,
                   1497: @code{F:} for a floating point value:
                   1499: @example
                   1500: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
                   1501: \ complex multiplication
                   1502:  Ar Br f* Ai Bi f* f-
                   1503:  Ar Bi f* Ai Br f* f+ ;
                   1504: @end example
1.17      anton    1506: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2       anton    1507: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
                   1508: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
                   1509: with @code{W:}, @code{D:} etc.) produces its value and can be changed
                   1510: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
                   1511: produces its address (which becomes invalid when the variable's scope is
                   1512: left). E.g., the standard word @code{emit} can be defined in therms of
                   1513: @code{type} like this:
                   1515: @example
                   1516: : emit @{ C^ char* -- @}
                   1517:     char* 1 type ;
                   1518: @end example
                   1520: A local without type specifier is a @code{W:} local. Both flavours of
                   1521: locals are initialized with values from the data or FP stack.
                   1523: Currently there is no way to define locals with user-defined data
                   1524: structures, but we are working on it.
1.17      anton    1526: Gforth allows defining locals everywhere in a colon definition. This
1.7       pazsan   1527: poses the following questions:
1.2       anton    1528: 
1.4       anton    1529: @menu
                   1530: * Where are locals visible by name?::  
1.14      anton    1531: * How long do locals live?::    
1.4       anton    1532: * Programming Style::           
                   1533: * Implementation::              
                   1534: @end menu
1.17      anton    1536: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2       anton    1537: @subsubsection Where are locals visible by name?
                   1539: Basically, the answer is that locals are visible where you would expect
                   1540: it in block-structured languages, and sometimes a little longer. If you
                   1541: want to restrict the scope of a local, enclose its definition in
                   1542: @code{SCOPE}...@code{ENDSCOPE}.
                   1544: doc-scope
                   1545: doc-endscope
                   1547: These words behave like control structure words, so you can use them
                   1548: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
                   1549: arbitrary ways.
                   1551: If you want a more exact answer to the visibility question, here's the
                   1552: basic principle: A local is visible in all places that can only be
                   1553: reached through the definition of the local@footnote{In compiler
                   1554: construction terminology, all places dominated by the definition of the
                   1555: local.}. In other words, it is not visible in places that can be reached
                   1556: without going through the definition of the local. E.g., locals defined
                   1557: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
                   1558: defined in @code{BEGIN}...@code{UNTIL} are visible after the
                   1559: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
                   1561: The reasoning behind this solution is: We want to have the locals
                   1562: visible as long as it is meaningful. The user can always make the
                   1563: visibility shorter by using explicit scoping. In a place that can
                   1564: only be reached through the definition of a local, the meaning of a
                   1565: local name is clear. In other places it is not: How is the local
                   1566: initialized at the control flow path that does not contain the
                   1567: definition? Which local is meant, if the same name is defined twice in
                   1568: two independent control flow paths?
                   1570: This should be enough detail for nearly all users, so you can skip the
                   1571: rest of this section. If you relly must know all the gory details and
                   1572: options, read on.
                   1574: In order to implement this rule, the compiler has to know which places
                   1575: are unreachable. It knows this automatically after @code{AHEAD},
                   1576: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
                   1577: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
                   1578: compiler that the control flow never reaches that place. If
                   1579: @code{UNREACHABLE} is not used where it could, the only consequence is
                   1580: that the visibility of some locals is more limited than the rule above
                   1581: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
                   1582: lie to the compiler), buggy code will be produced.
                   1584: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3       anton    1585: does not know which locals will be visible on the incoming
                   1586: back-edge. All problems discussed in the following are due to this
                   1587: ignorance of the compiler (we discuss the problems using @code{BEGIN}
                   1588: loops as examples; the discussion also applies to @code{?DO} and other
1.2       anton    1589: loops). Perhaps the most insidious example is:
                   1590: @example
                   1591: AHEAD
                   1592: BEGIN
                   1593:   x
                   1594: [ 1 CS-ROLL ] THEN
1.4       anton    1595:   @{ x @}
1.2       anton    1596:   ...
                   1597: UNTIL
                   1598: @end example
                   1600: This should be legal according to the visibility rule. The use of
                   1601: @code{x} can only be reached through the definition; but that appears
                   1602: textually below the use.
                   1604: From this example it is clear that the visibility rules cannot be fully
                   1605: implemented without major headaches. Our implementation treats common
                   1606: cases as advertised and the exceptions are treated in a safe way: The
                   1607: compiler makes a reasonable guess about the locals visible after a
                   1608: @code{BEGIN}; if it is too pessimistic, the
                   1609: user will get a spurious error about the local not being defined; if the
                   1610: compiler is too optimistic, it will notice this later and issue a
                   1611: warning. In the case above the compiler would complain about @code{x}
                   1612: being undefined at its use. You can see from the obscure examples in
                   1613: this section that it takes quite unusual control structures to get the
                   1614: compiler into trouble, and even then it will often do fine.
                   1616: If the @code{BEGIN} is reachable from above, the most optimistic guess
                   1617: is that all locals visible before the @code{BEGIN} will also be
                   1618: visible after the @code{BEGIN}. This guess is valid for all loops that
                   1619: are entered only through the @code{BEGIN}, in particular, for normal
                   1620: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
                   1621: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
                   1622: compiler. When the branch to the @code{BEGIN} is finally generated by
                   1623: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
                   1624: warns the user if it was too optimisitic:
                   1625: @example
                   1626: IF
1.4       anton    1627:   @{ x @}
1.2       anton    1628: BEGIN
                   1629:   \ x ? 
                   1630: [ 1 cs-roll ] THEN
                   1631:   ...
                   1632: UNTIL
                   1633: @end example
                   1635: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
                   1636: optimistically assumes that it lives until the @code{THEN}. It notices
                   1637: this difference when it compiles the @code{UNTIL} and issues a
                   1638: warning. The user can avoid the warning, and make sure that @code{x}
                   1639: is not used in the wrong area by using explicit scoping:
                   1640: @example
                   1641: IF
                   1642:   SCOPE
1.4       anton    1643:   @{ x @}
1.2       anton    1644:   ENDSCOPE
                   1645: BEGIN
                   1646: [ 1 cs-roll ] THEN
                   1647:   ...
                   1648: UNTIL
                   1649: @end example
                   1651: Since the guess is optimistic, there will be no spurious error messages
                   1652: about undefined locals.
                   1654: If the @code{BEGIN} is not reachable from above (e.g., after
                   1655: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
                   1656: optimistic guess, as the locals visible after the @code{BEGIN} may be
                   1657: defined later. Therefore, the compiler assumes that no locals are
1.17      anton    1658: visible after the @code{BEGIN}. However, the user can use
1.2       anton    1659: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17      anton    1660: visible at the BEGIN as at the point where the top control-flow stack
                   1661: item was created.
1.2       anton    1662: 
                   1663: doc-assume-live
                   1665: E.g.,
                   1666: @example
1.4       anton    1667: @{ x @}
1.2       anton    1668: AHEAD
                   1669: ASSUME-LIVE
                   1670: BEGIN
                   1671:   x
                   1672: [ 1 CS-ROLL ] THEN
                   1673:   ...
                   1674: UNTIL
                   1675: @end example
                   1677: Other cases where the locals are defined before the @code{BEGIN} can be
                   1678: handled by inserting an appropriate @code{CS-ROLL} before the
                   1679: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
                   1680: behind the @code{ASSUME-LIVE}).
                   1682: Cases where locals are defined after the @code{BEGIN} (but should be
                   1683: visible immediately after the @code{BEGIN}) can only be handled by
                   1684: rearranging the loop. E.g., the ``most insidious'' example above can be
                   1685: arranged into:
                   1686: @example
                   1687: BEGIN
1.4       anton    1688:   @{ x @}
1.2       anton    1689:   ... 0=
                   1690: WHILE
                   1691:   x
                   1692: REPEAT
                   1693: @end example
1.17      anton    1695: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2       anton    1696: @subsubsection How long do locals live?
                   1698: The right answer for the lifetime question would be: A local lives at
                   1699: least as long as it can be accessed. For a value-flavoured local this
                   1700: means: until the end of its visibility. However, a variable-flavoured
                   1701: local could be accessed through its address far beyond its visibility
                   1702: scope. Ultimately, this would mean that such locals would have to be
                   1703: garbage collected. Since this entails un-Forth-like implementation
                   1704: complexities, I adopted the same cowardly solution as some other
                   1705: languages (e.g., C): The local lives only as long as it is visible;
                   1706: afterwards its address is invalid (and programs that access it
                   1707: afterwards are erroneous).
1.17      anton    1709: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2       anton    1710: @subsubsection Programming Style
                   1712: The freedom to define locals anywhere has the potential to change
                   1713: programming styles dramatically. In particular, the need to use the
                   1714: return stack for intermediate storage vanishes. Moreover, all stack
                   1715: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
                   1716: determined arguments) can be eliminated: If the stack items are in the
                   1717: wrong order, just write a locals definition for all of them; then
                   1718: write the items in the order you want.
                   1720: This seems a little far-fetched and eliminating stack manipulations is
1.4       anton    1721: unlikely to become a conscious programming objective. Still, the number
                   1722: of stack manipulations will be reduced dramatically if local variables
1.17      anton    1723: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4       anton    1724: a traditional implementation of @code{max}).
1.2       anton    1725: 
                   1726: This shows one potential benefit of locals: making Forth programs more
                   1727: readable. Of course, this benefit will only be realized if the
                   1728: programmers continue to honour the principle of factoring instead of
                   1729: using the added latitude to make the words longer.
                   1731: Using @code{TO} can and should be avoided.  Without @code{TO},
                   1732: every value-flavoured local has only a single assignment and many
                   1733: advantages of functional languages apply to Forth. I.e., programs are
                   1734: easier to analyse, to optimize and to read: It is clear from the
                   1735: definition what the local stands for, it does not turn into something
                   1736: different later.
                   1738: E.g., a definition using @code{TO} might look like this:
                   1739: @example
                   1740: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1741:  u1 u2 min 0
                   1742:  ?do
1.36      anton    1743:    addr1 c@@ addr2 c@@ -
1.31      anton    1744:    ?dup-if
1.2       anton    1745:      unloop exit
                   1746:    then
                   1747:    addr1 char+ TO addr1
                   1748:    addr2 char+ TO addr2
                   1749:  loop
                   1750:  u1 u2 - ;
                   1751: @end example
                   1752: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
                   1753: every loop iteration. @code{strcmp} is a typical example of the
                   1754: readability problems of using @code{TO}. When you start reading
                   1755: @code{strcmp}, you think that @code{addr1} refers to the start of the
                   1756: string. Only near the end of the loop you realize that it is something
                   1757: else.
                   1759: This can be avoided by defining two locals at the start of the loop that
                   1760: are initialized with the right value for the current iteration.
                   1761: @example
                   1762: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1763:  addr1 addr2
                   1764:  u1 u2 min 0 
                   1765:  ?do @{ s1 s2 @}
1.36      anton    1766:    s1 c@@ s2 c@@ -
1.31      anton    1767:    ?dup-if
1.2       anton    1768:      unloop exit
                   1769:    then
                   1770:    s1 char+ s2 char+
                   1771:  loop
                   1772:  2drop
                   1773:  u1 u2 - ;
                   1774: @end example
                   1775: Here it is clear from the start that @code{s1} has a different value
                   1776: in every loop iteration.
1.17      anton    1778: @node Implementation,  , Programming Style, Gforth locals
1.2       anton    1779: @subsubsection Implementation
1.17      anton    1781: Gforth uses an extra locals stack. The most compelling reason for
1.2       anton    1782: this is that the return stack is not float-aligned; using an extra stack
                   1783: also eliminates the problems and restrictions of using the return stack
                   1784: as locals stack. Like the other stacks, the locals stack grows toward
                   1785: lower addresses. A few primitives allow an efficient implementation:
                   1787: doc-@local#
                   1788: doc-f@local#
                   1789: doc-laddr#
                   1790: doc-lp+!#
                   1791: doc-lp!
                   1792: doc->l
                   1793: doc-f>l
                   1795: In addition to these primitives, some specializations of these
                   1796: primitives for commonly occurring inline arguments are provided for
                   1797: efficiency reasons, e.g., @code{@@local0} as specialization of
                   1798: @code{@@local#} for the inline argument 0. The following compiling words
                   1799: compile the right specialized version, or the general version, as
                   1800: appropriate:
1.12      anton    1802: doc-compile-@local
                   1803: doc-compile-f@local
1.2       anton    1804: doc-compile-lp+!
                   1806: Combinations of conditional branches and @code{lp+!#} like
                   1807: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
                   1808: is taken) are provided for efficiency and correctness in loops.
                   1810: A special area in the dictionary space is reserved for keeping the
                   1811: local variable names. @code{@{} switches the dictionary pointer to this
                   1812: area and @code{@}} switches it back and generates the locals
                   1813: initializing code. @code{W:} etc.@ are normal defining words. This
                   1814: special area is cleared at the start of every colon definition.
1.17      anton    1816: A special feature of Gforth's dictionary is used to implement the
1.2       anton    1817: definition of locals without type specifiers: every wordlist (aka
                   1818: vocabulary) has its own methods for searching
1.4       anton    1819: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2       anton    1820: with a special search method: When it is searched for a word, it
                   1821: actually creates that word using @code{W:}. @code{@{} changes the search
                   1822: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
                   1823: and then the wordlist for defining locals without type specifiers.
                   1825: The lifetime rules support a stack discipline within a colon
                   1826: definition: The lifetime of a local is either nested with other locals
                   1827: lifetimes or it does not overlap them.
                   1829: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
                   1830: pointer manipulation is generated. Between control structure words
                   1831: locals definitions can push locals onto the locals stack. @code{AGAIN}
                   1832: is the simplest of the other three control flow words. It has to
                   1833: restore the locals stack depth of the corresponding @code{BEGIN}
                   1834: before branching. The code looks like this:
                   1835: @format
                   1836: @code{lp+!#} current-locals-size @minus{} dest-locals-size
                   1837: @code{branch} <begin>
                   1838: @end format
                   1840: @code{UNTIL} is a little more complicated: If it branches back, it
                   1841: must adjust the stack just like @code{AGAIN}. But if it falls through,
                   1842: the locals stack must not be changed. The compiler generates the
                   1843: following code:
                   1844: @format
                   1845: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
                   1846: @end format
                   1847: The locals stack pointer is only adjusted if the branch is taken.
                   1849: @code{THEN} can produce somewhat inefficient code:
                   1850: @format
                   1851: @code{lp+!#} current-locals-size @minus{} orig-locals-size
                   1852: <orig target>:
                   1853: @code{lp+!#} orig-locals-size @minus{} new-locals-size
                   1854: @end format
                   1855: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4       anton    1856: level at the @var{orig} point to the level after the @code{THEN}. The
1.2       anton    1857: first @code{lp+!#} adjusts the locals stack pointer from the current
                   1858: level to the level at the orig point, so the complete effect is an
                   1859: adjustment from the current level to the right level after the
                   1860: @code{THEN}.
                   1862: In a conventional Forth implementation a dest control-flow stack entry
                   1863: is just the target address and an orig entry is just the address to be
                   1864: patched. Our locals implementation adds a wordlist to every orig or dest
                   1865: item. It is the list of locals visible (or assumed visible) at the point
                   1866: described by the entry. Our implementation also adds a tag to identify
                   1867: the kind of entry, in particular to differentiate between live and dead
                   1868: (reachable and unreachable) orig entries.
                   1870: A few unusual operations have to be performed on locals wordlists:
                   1872: doc-common-list
                   1873: doc-sub-list?
                   1874: doc-list-size
                   1876: Several features of our locals wordlist implementation make these
                   1877: operations easy to implement: The locals wordlists are organised as
                   1878: linked lists; the tails of these lists are shared, if the lists
                   1879: contain some of the same locals; and the address of a name is greater
                   1880: than the address of the names behind it in the list.
                   1882: Another important implementation detail is the variable
                   1883: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
                   1884: determine if they can be reached directly or only through the branch
                   1885: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
                   1886: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
                   1887: definition, by @code{BEGIN} and usually by @code{THEN}.
                   1889: Counted loops are similar to other loops in most respects, but
                   1890: @code{LEAVE} requires special attention: It performs basically the same
                   1891: service as @code{AHEAD}, but it does not create a control-flow stack
                   1892: entry. Therefore the information has to be stored elsewhere;
                   1893: traditionally, the information was stored in the target fields of the
                   1894: branches created by the @code{LEAVE}s, by organizing these fields into a
                   1895: linked list. Unfortunately, this clever trick does not provide enough
                   1896: space for storing our extended control flow information. Therefore, we
                   1897: introduce another stack, the leave stack. It contains the control-flow
                   1898: stack entries for all unresolved @code{LEAVE}s.
                   1900: Local names are kept until the end of the colon definition, even if
                   1901: they are no longer visible in any control-flow path. In a few cases
                   1902: this may lead to increased space needs for the locals name area, but
                   1903: usually less than reclaiming this space would cost in code size.
1.17      anton    1906: @node ANS Forth locals,  , Gforth locals, Locals
1.2       anton    1907: @subsection ANS Forth locals
                   1909: The ANS Forth locals wordset does not define a syntax for locals, but
                   1910: words that make it possible to define various syntaxes. One of the
1.17      anton    1911: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2       anton    1912: wordset, i.e.:
                   1914: @example
                   1915: @{ local1 local2 ... -- comment @}
                   1916: @end example
                   1917: or
                   1918: @example
                   1919: @{ local1 local2 ... @}
                   1920: @end example
                   1922: The order of the locals corresponds to the order in a stack comment. The
                   1923: restrictions are:
1.1       anton    1924: 
1.2       anton    1925: @itemize @bullet
                   1926: @item
1.17      anton    1927: Locals can only be cell-sized values (no type specifiers are allowed).
1.2       anton    1928: @item
                   1929: Locals can be defined only outside control structures.
                   1930: @item
                   1931: Locals can interfere with explicit usage of the return stack. For the
                   1932: exact (and long) rules, see the standard. If you don't use return stack
1.17      anton    1933: accessing words in a definition using locals, you will be all right. The
1.2       anton    1934: purpose of this rule is to make locals implementation on the return
                   1935: stack easier.
                   1936: @item
                   1937: The whole definition must be in one line.
                   1938: @end itemize
1.35      anton    1940: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
                   1941: Defining Words}). I.e., they are initialized from the stack. Using their
1.2       anton    1942: name produces their value. Their value can be changed using @code{TO}.
1.17      anton    1944: Since this syntax is supported by Gforth directly, you need not do
1.2       anton    1945: anything to use it. If you want to port a program using this syntax to
1.30      anton    1946: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
                   1947: syntax on the other system.
1.2       anton    1948: 
                   1949: Note that a syntax shown in the standard, section A.13 looks
                   1950: similar, but is quite different in having the order of locals
                   1951: reversed. Beware!
                   1953: The ANS Forth locals wordset itself consists of the following word
                   1955: doc-(local)
                   1957: The ANS Forth locals extension wordset defines a syntax, but it is so
                   1958: awful that we strongly recommend not to use it. We have implemented this
1.17      anton    1959: syntax to make porting to Gforth easy, but do not document it here. The
1.2       anton    1960: problem with this syntax is that the locals are defined in an order
                   1961: reversed with respect to the standard stack comment notation, making
                   1962: programs harder to read, and easier to misread and miswrite. The only
                   1963: merit of this syntax is that it is easy to implement using the ANS Forth
                   1964: locals wordset.
1.3       anton    1965: 
1.37      anton    1966: @node Defining Words, Tokens for Words, Locals, Words
1.4       anton    1967: @section Defining Words
1.14      anton    1969: @menu
1.35      anton    1970: * Simple Defining Words::       
                   1971: * Colon Definitions::           
                   1972: * User-defined Defining Words::  
                   1973: * Supplying names::             
                   1974: * Interpretation and Compilation Semantics::  
1.14      anton    1975: @end menu
1.35      anton    1977: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
                   1978: @subsection Simple Defining Words
                   1980: doc-constant
                   1981: doc-2constant
                   1982: doc-fconstant
                   1983: doc-variable
                   1984: doc-2variable
                   1985: doc-fvariable
                   1986: doc-create
                   1987: doc-user
                   1988: doc-value
                   1989: doc-to
                   1990: doc-defer
                   1991: doc-is
                   1993: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
                   1994: @subsection Colon Definitions
                   1996: @example
                   1997: : name ( ... -- ... )
                   1998:     word1 word2 word3 ;
                   1999: @end example
                   2001: creates a word called @code{name}, that, upon execution, executes
                   2002: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
                   2004: The explanation above is somewhat superficial. @xref{Interpretation and
                   2005: Compilation Semantics} for an in-depth discussion of some of the issues
                   2006: involved.
                   2008: doc-:
                   2009: doc-;
                   2011: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
                   2012: @subsection User-defined Defining Words
                   2014: You can create new defining words simply by wrapping defining-time code
                   2015: around existing defining words and putting the sequence in a colon
                   2016: definition.
1.36      anton    2018: If you want the words defined with your defining words to behave
                   2019: differently from words defined with standard defining words, you can
1.35      anton    2020: write your defining word like this:
                   2022: @example
                   2023: : def-word ( "name" -- )
                   2024:     Create @var{code1}
                   2025: DOES> ( ... -- ... )
                   2026:     @var{code2} ;
                   2028: def-word name
                   2029: @end example
                   2031: Technically, this fragment defines a defining word @code{def-word}, and
                   2032: a word @code{name}; when you execute @code{name}, the address of the
                   2033: body of @code{name} is put on the data stack and @var{code2} is executed
                   2034: (the address of the body of @code{name} is the address @code{HERE}
1.36      anton    2035: returns immediately after the @code{CREATE}).
                   2037: In other words, if you make the following definitions:
                   2039: @example
                   2040: : def-word1 ( "name" -- )
                   2041:     Create @var{code1} ;
                   2043: : action1 ( ... -- ... )
                   2044:     @var{code2} ;
                   2046: def-word name1
                   2047: @end example
                   2049: Using @code{name1 action1} is equivalent to using @code{name}.
                   2051: E.g., you can implement @code{Constant} in this way:
1.35      anton    2052: 
                   2053: @example
                   2054: : constant ( w "name" -- )
                   2055:     create ,
                   2056: DOES> ( -- w )
1.36      anton    2057:     @@ ;
1.35      anton    2058: @end example
                   2060: When you create a constant with @code{5 constant five}, first a new word
                   2061: @code{five} is created, then the value 5 is laid down in the body of
                   2062: @code{five} with @code{,}. When @code{five} is invoked, the address of
                   2063: the body is put on the stack, and @code{@@} retrieves the value 5.
                   2065: In the example above the stack comment after the @code{DOES>} specifies
                   2066: the stack effect of the defined words, not the stack effect of the
                   2067: following code (the following code expects the address of the body on
                   2068: the top of stack, which is not reflected in the stack comment). This is
                   2069: the convention that I use and recommend (it clashes a bit with using
                   2070: locals declarations for stack effect specification, though).
                   2072: @subsubsection Applications of @code{CREATE..DOES>}
1.36      anton    2074: You may wonder how to use this feature. Here are some usage patterns:
1.35      anton    2075: 
                   2076: When you see a sequence of code occurring several times, and you can
                   2077: identify a meaning, you will factor it out as a colon definition. When
                   2078: you see similar colon definitions, you can factor them using
                   2079: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
                   2080: that look very similar:
                   2081: @example
                   2082: : ori, ( reg-taget reg-source n -- )
                   2083:     0 asm-reg-reg-imm ;
                   2084: : andi, ( reg-taget reg-source n -- )
                   2085:     1 asm-reg-reg-imm ;
                   2086: @end example
                   2088: This could be factored with:
                   2089: @example
                   2090: : reg-reg-imm ( op-code -- )
                   2091:     create ,
                   2092: DOES> ( reg-taget reg-source n -- )
1.36      anton    2093:     @@ asm-reg-reg-imm ;
1.35      anton    2094: 
                   2095: 0 reg-reg-imm ori,
                   2096: 1 reg-reg-imm andi,
                   2097: @end example
                   2099: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
                   2100: supply a part of the parameters for a word (known as @dfn{currying} in
                   2101: the functional language community). E.g., @code{+} needs two
                   2102: parameters. Creating versions of @code{+} with one parameter fixed can
                   2103: be done like this:
                   2104: @example
                   2105: : curry+ ( n1 -- )
                   2106:     create ,
                   2107: DOES> ( n2 -- n1+n2 )
1.36      anton    2108:     @@ + ;
1.35      anton    2109: 
                   2110:  3 curry+ 3+
                   2111: -2 curry+ 2-
                   2112: @end example
                   2114: @subsubsection The gory details of @code{CREATE..DOES>}
                   2116: doc-does>
                   2118: This means that you need not use @code{CREATE} and @code{DOES>} in the
                   2119: same definition; E.g., you can put the @code{DOES>}-part in a separate
                   2120: definition. This allows us to, e.g., select among different DOES>-parts:
                   2121: @example
                   2122: : does1 
                   2123: DOES> ( ... -- ... )
                   2124:     ... ;
                   2126: : does2
                   2127: DOES> ( ... -- ... )
                   2128:     ... ;
                   2130: : def-word ( ... -- ... )
                   2131:     create ...
                   2132:     IF
                   2133:        does1
                   2134:     ELSE
                   2135:        does2
                   2136:     ENDIF ;
                   2137: @end example
                   2139: In a standard program you can apply a @code{DOES>}-part only if the last
                   2140: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
                   2141: will override the behaviour of the last word defined in any case. In a
                   2142: standard program, you can use @code{DOES>} only in a colon
                   2143: definition. In Gforth, you can also use it in interpretation state, in a
                   2144: kind of one-shot mode:
                   2145: @example
                   2146: CREATE name ( ... -- ... )
                   2147:   @var{initialization}
                   2148: DOES>
                   2149:   @var{code} ;
                   2150: @end example
                   2151: This is equivalwent to the standard
                   2152: @example
                   2153: :noname
                   2154: DOES>
                   2155:     @var{code} ;
                   2156: CREATE name EXECUTE ( ... -- ... )
                   2157:     @var{initialization}
                   2158: @end example
                   2160: You can get the address of the body of a word with
                   2162: doc->body
                   2164: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
                   2165: @subsection Supplying names for the defined words
                   2167: By default, defining words take the names for the defined words from the
                   2168: input stream. Sometimes you want to supply the name from a string. You
                   2169: can do this with
                   2171: doc-nextname
                   2173: E.g.,
                   2175: @example
                   2176: s" foo" nextname create
                   2177: @end example
                   2178: is equivalent to
                   2179: @example
                   2180: create foo
                   2181: @end example
                   2183: Sometimes you want to define a word without a name. You can do this with
                   2185: doc-noname
                   2187: To make any use of the newly defined word, you need its execution
                   2188: token. You can get it with
                   2190: doc-lastxt
                   2192: E.g., you can initialize a deferred word with an anonymous colon
                   2193: definition:
                   2194: @example
                   2195: Defer deferred
                   2196: noname : ( ... -- ... )
                   2197:   ... ;
                   2198: lastxt IS deferred
                   2199: @end example
                   2201: @code{lastxt} also works when the last word was not defined as
                   2202: @code{noname}. 
                   2204: The standard has also recognized the need for anonymous words and
                   2205: provides
                   2207: doc-:noname
                   2209: This leaves the execution token for the word on the stack after the
                   2210: closing @code{;}. You can rewrite the last example with @code{:noname}:
                   2211: @example
                   2212: Defer deferred
                   2213: :noname ( ... -- ... )
                   2214:   ... ;
                   2215: IS deferred
                   2216: @end example
                   2218: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
                   2219: @subsection Interpretation and Compilation Semantics
1.36      anton    2221: The @dfn{interpretation semantics} of a word are what the text
                   2222: interpreter does when it encounters the word in interpret state. It also
                   2223: appears in some other contexts, e.g., the execution token returned by
                   2224: @code{' @var{word}} identifies the interpretation semantics of
                   2225: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
                   2226: interpret-state text interpretation of @code{@var{word}}).
                   2228: The @dfn{compilation semantics} of a word are what the text interpreter
                   2229: does when it encounters the word in compile state. It also appears in
                   2230: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
                   2231: standard terminology, ``appends to the current definition''.} the
                   2232: compilation semantics of @var{word}.
                   2234: The standard also talks about @dfn{execution semantics}. They are used
                   2235: only for defining the interpretation and compilation semantics of many
                   2236: words. By default, the interpretation semantics of a word are to
                   2237: @code{execute} its execution semantics, and the compilation semantics of
                   2238: a word are to @code{compile,} its execution semantics.@footnote{In
                   2239: standard terminology: The default interpretation semantics are its
                   2240: execution semantics; the default compilation semantics are to append its
                   2241: execution semantics to the execution semantics of the current
                   2242: definition.}
                   2244: You can change the compilation semantics into @code{execute}ing the
                   2245: execution semantics with
1.35      anton    2247: doc-immediate
1.36      anton    2248: 
                   2249: You can remove the interpretation semantics of a word with
                   2251: doc-compile-only
                   2252: doc-restrict
                   2254: Note that ticking (@code{'}) compile-only words gives an error
                   2255: (``Interpreting a compile-only word'').
                   2257: Gforth also allows you to define words with arbitrary combinations of
                   2258: interpretation and compilation semantics.
1.35      anton    2260: doc-interpret/compile:
1.36      anton    2262: This feature was introduced for implementing @code{TO} and @code{S"}. I
                   2263: recommend that you do not define such words, as cute as they may be:
                   2264: they make it hard to get at both parts of the word in some contexts.
                   2265: E.g., assume you want to get an execution token for the compilation
                   2266: part. Instead, define two words, one that embodies the interpretation
                   2267: part, and one that embodies the compilation part.
                   2269: There is, however, a potentially useful application of this feature:
                   2270: Providing differing implementations for the default semantics. While
                   2271: this introduces redundancy and is therefore usually a bad idea, a
                   2272: performance improvement may be worth the trouble. E.g., consider the
                   2273: word @code{foobar}:
                   2275: @example
                   2276: : foobar
                   2277:     foo bar ;
                   2278: @end example
                   2280: Let us assume that @code{foobar} is called so frequently that the
                   2281: calling overhead would take a significant amount of the run-time. We can
                   2282: optimize it with @code{interpret/compile:}:
1.35      anton    2283: 
1.36      anton    2284: @example
                   2285: :noname
                   2286:    foo bar ;
                   2287: :noname
                   2288:    POSTPONE foo POSTPONE bar ;
                   2289: interpret/compile: foobar
                   2290: @end example
                   2292: This definition has the same interpretation semantics and essentially
                   2293: the same compilation semantics as the simple definition of
                   2294: @code{foobar}, but the implementation of the compilation semantics is
                   2295: more efficient with respect to run-time.
                   2297: Some people try to use state-smart words to emulate the feature provided
                   2298: by @code{interpret/compile:} (words are state-smart if they check
                   2299: @code{STATE} during execution). E.g., they would try to code
                   2300: @code{foobar} like this:
                   2302: @example
                   2303: : foobar
                   2304:   STATE @@
                   2305:   IF ( compilation state )
                   2306:     POSTPONE foo POSTPONE bar
                   2307:   ELSE
                   2308:     foo bar
                   2309:   ENDIF ; immediate
                   2310: @end example
                   2312: While this works if @code{foobar} is processed only by the text
                   2313: interpreter, it does not work in other contexts (like @code{'} or
                   2314: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
                   2315: for a state-smart word, not for the interpretation semantics of the
                   2316: original @code{foobar}; when you execute this execution token (directly
                   2317: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
                   2318: state, the result will not be what you expected (i.e., it will not
                   2319: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
                   2320: write them!
                   2322: It is also possible to write defining words that define words with
                   2323: arbitrary combinations of interpretation and compilation semantics (or,
                   2324: preferably, arbitrary combinations of implementations of the default
                   2325: semantics). In general, this looks like:
                   2327: @example
                   2328: : def-word
                   2329:     create-interpret/compile
                   2330:     @var{code1}
                   2331: interpretation>
                   2332:     @var{code2}
                   2333: <interpretation
                   2334: compilation>
                   2335:     @var{code3}
                   2336: <compilation ;
                   2337: @end example
                   2339: For a @var{word} defined with @code{def-word}, the interpretation
                   2340: semantics are to push the address of the body of @var{word} and perform
                   2341: @var{code2}, and the compilation semantics are to push the address of
                   2342: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
                   2343: can also be defined like this:
                   2345: @example
                   2346: : constant ( n "name" -- )
                   2347:     create-interpret/compile
                   2348:     ,
                   2349: interpretation> ( -- n )
                   2350:     @@
                   2351: <interpretation
                   2352: compilation> ( compilation. -- ; run-time. -- n )
                   2353:     @@ postpone literal
                   2354: <compilation ;
                   2355: @end example
                   2357: doc-create-interpret/compile
                   2358: doc-interpretation>
                   2359: doc-<interpretation
                   2360: doc-compilation>
                   2361: doc-<compilation
                   2363: Note that words defined with @code{interpret/compile:} and
                   2364: @code{create-interpret/compile} have an extended header structure that
                   2365: differs from other words; however, unless you try to access them with
                   2366: plain address arithmetic, you should not notice this. Words for
                   2367: accessing the header structure usually know how to deal with this; e.g.,
                   2368: @code{' word >body} also gives you the body of a word created with
                   2369: @code{create-interpret/compile}.
1.4       anton    2370: 
1.37      anton    2371: @node Tokens for Words, Wordlists, Defining Words, Words
                   2372: @section Tokens for Words
                   2374: This chapter describes the creation and use of tokens that represent
                   2375: words on the stack (and in data space).
                   2377: Named words have interpretation and compilation semantics. Unnamed words
                   2378: just have execution semantics.
                   2380: An @dfn{execution token} represents the execution semantics of an
                   2381: unnamed word. An execution token occupies one cell. As explained in
                   2382: section @ref{Supplying names}, the execution token of the last words
                   2383: defined can be produced with
                   2385: short-lastxt
                   2387: You can perform the semantics represented by an execution token with
                   2388: doc-execute
                   2389: You can compile the word with
                   2390: doc-compile,
                   2392: In Gforth, the abstract data type @emph{execution token} is implemented
                   2393: as CFA (code field address).
                   2395: The interpretation semantics of a named word are also represented by an
                   2396: execution token. You can get it with
                   2398: doc-[']
                   2399: doc-'
                   2401: For literals, you use @code{'} in interpreted code and @code{[']} in
                   2402: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
                   2403: by complaining about compile-only words. To get an execution token for a
                   2404: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
                   2405: @var{X} drop}.
                   2407: The compilation semantics are represented by a @dfn{compilation token}
                   2408: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
                   2409: execution token. The compilation semantics represented by the
                   2410: compilation token can be performed with @code{execute}, which consumes
                   2411: the whole compilation token, with an additional stack effect determined
                   2412: by the represented compilation semantics.
                   2414: doc-[comp']
                   2415: doc-comp'
1.38    ! anton    2417: You can compile the compilation semantics with @code{postpone,}. I.e.,
        !          2418: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
        !          2419: @var{word}}.
        !          2420: 
        !          2421: doc-postpone,
        !          2422: 
1.37      anton    2423: At present, the @var{w} part of a compilation token is an execution
                   2424: token, and the @var{xt} part represents either @code{execute} or
                   2425: @code{compile,}. However, don't rely on that kowledge, unless necessary;
                   2426: we may introduce unusual compilation tokens in the future (e.g.,
                   2427: compilation tokens representing the compilation semantics of literals).
                   2429: Named words are also represented by the @dfn{name token}. The abstract
                   2430: data type @emph{name token} is implemented as NFA (name field address).
                   2432: doc-find-name
                   2433: doc-name>int
                   2434: doc-name?int
                   2435: doc-name>comp
                   2436: doc-name>string
                   2438: @node Wordlists, Files, Tokens for Words, Words
1.4       anton    2439: @section Wordlists
                   2441: @node Files, Blocks, Wordlists, Words
                   2442: @section Files
                   2444: @node Blocks, Other I/O, Files, Words
                   2445: @section Blocks
                   2447: @node Other I/O, Programming Tools, Blocks, Words
                   2448: @section Other I/O
1.18      anton    2450: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4       anton    2451: @section Programming Tools
1.5       anton    2453: @menu
                   2454: * Debugging::                   Simple and quick.
                   2455: * Assertions::                  Making your programs self-checking.
                   2456: @end menu
                   2458: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4       anton    2459: @subsection Debugging
                   2461: The simple debugging aids provided in @file{debugging.fs}
                   2462: are meant to support a different style of debugging than the
                   2463: tracing/stepping debuggers used in languages with long turn-around
                   2464: times.
                   2466: A much better (faster) way in fast-compilig languages is to add
                   2467: printing code at well-selected places, let the program run, look at
                   2468: the output, see where things went wrong, add more printing code, etc.,
                   2469: until the bug is found.
                   2471: The word @code{~~} is easy to insert. It just prints debugging
                   2472: information (by default the source location and the stack contents). It
                   2473: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
                   2474: query-replace them with nothing). The deferred words
                   2475: @code{printdebugdata} and @code{printdebugline} control the output of
                   2476: @code{~~}. The default source location output format works well with
                   2477: Emacs' compilation mode, so you can step through the program at the
1.5       anton    2478: source level using @kbd{C-x `} (the advantage over a stepping debugger
                   2479: is that you can step in any direction and you know where the crash has
                   2480: happened or where the strange data has occurred).
1.4       anton    2481: 
                   2482: Note that the default actions clobber the contents of the pictured
                   2483: numeric output string, so you should not use @code{~~}, e.g., between
                   2484: @code{<#} and @code{#>}.
                   2486: doc-~~
                   2487: doc-printdebugdata
                   2488: doc-printdebugline
1.5       anton    2490: @node Assertions,  , Debugging, Programming Tools
1.4       anton    2491: @subsection Assertions
1.5       anton    2493: It is a good idea to make your programs self-checking, in particular, if
                   2494: you use an assumption (e.g., that a certain field of a data structure is
1.17      anton    2495: never zero) that may become wrong during maintenance. Gforth supports
1.5       anton    2496: assertions for this purpose. They are used like this:
                   2498: @example
                   2499: assert( @var{flag} )
                   2500: @end example
                   2502: The code between @code{assert(} and @code{)} should compute a flag, that
                   2503: should be true if everything is alright and false otherwise. It should
                   2504: not change anything else on the stack. The overall stack effect of the
                   2505: assertion is @code{( -- )}. E.g.
                   2507: @example
                   2508: assert( 1 1 + 2 = ) \ what we learn in school
                   2509: assert( dup 0<> ) \ assert that the top of stack is not zero
                   2510: assert( false ) \ this code should not be reached
                   2511: @end example
                   2513: The need for assertions is different at different times. During
                   2514: debugging, we want more checking, in production we sometimes care more
                   2515: for speed. Therefore, assertions can be turned off, i.e., the assertion
                   2516: becomes a comment. Depending on the importance of an assertion and the
                   2517: time it takes to check it, you may want to turn off some assertions and
1.17      anton    2518: keep others turned on. Gforth provides several levels of assertions for
1.5       anton    2519: this purpose:
                   2521: doc-assert0(
                   2522: doc-assert1(
                   2523: doc-assert2(
                   2524: doc-assert3(
                   2525: doc-assert(
                   2526: doc-)
                   2528: @code{Assert(} is the same as @code{assert1(}. The variable
                   2529: @code{assert-level} specifies the highest assertions that are turned
                   2530: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
                   2531: @code{assert1(} assertions perform checking, while @code{assert2(} and
                   2532: @code{assert3(} assertions are treated as comments.
                   2534: Note that the @code{assert-level} is evaluated at compile-time, not at
                   2535: run-time. I.e., you cannot turn assertions on or off at run-time, you
                   2536: have to set the @code{assert-level} appropriately before compiling a
                   2537: piece of code. You can compile several pieces of code at several
                   2538: @code{assert-level}s (e.g., a trusted library at level 1 and newly
                   2539: written code at level 3).
                   2541: doc-assert-level
                   2543: If an assertion fails, a message compatible with Emacs' compilation mode
                   2544: is produced and the execution is aborted (currently with @code{ABORT"}.
                   2545: If there is interest, we will introduce a special throw code. But if you
                   2546: intend to @code{catch} a specific condition, using @code{throw} is
                   2547: probably more appropriate than an assertion).
1.18      anton    2549: @node Assembler and Code words, Threading Words, Programming Tools, Words
                   2550: @section Assembler and Code words
                   2552: Gforth provides some words for defining primitives (words written in
                   2553: machine code), and for defining the the machine-code equivalent of
                   2554: @code{DOES>}-based defining words. However, the machine-independent
                   2555: nature of Gforth poses a few problems: First of all. Gforth runs on
                   2556: several architectures, so it can provide no standard assembler. What's
                   2557: worse is that the register allocation not only depends on the processor,
1.25      anton    2558: but also on the @code{gcc} version and options used.
1.18      anton    2559: 
1.25      anton    2560: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18      anton    2561: header structure), so a system-independent assembler may be used in
                   2562: Gforth. If you do not have an assembler, you can compile machine code
                   2563: directly with @code{,} and @code{c,}.
                   2565: doc-assembler
                   2566: doc-code
                   2567: doc-end-code
                   2568: doc-;code
                   2569: doc-flush-icache
                   2571: If @code{flush-icache} does not work correctly, @code{code} words
                   2572: etc. will not work (reliably), either.
                   2574: These words are rarely used. Therefore they reside in @code{code.fs},
                   2575: which is usually not loaded (except @code{flush-icache}, which is always
1.19      anton    2576: present). You can load them with @code{require code.fs}.
1.18      anton    2577: 
1.25      anton    2578: In the assembly code you will want to refer to the inner interpreter's
                   2579: registers (e.g., the data stack pointer) and you may want to use other
                   2580: registers for temporary storage. Unfortunately, the register allocation
                   2581: is installation-dependent.
                   2583: The easiest solution is to use explicit register declarations
                   2584: (@pxref{Explicit Reg Vars, , Variables in Specified Registers,,
                   2585: GNU C Manual}) for all of the inner interpreter's registers: You have to
                   2586: compile Gforth with @code{-DFORCE_REG} (configure option
                   2587: @code{--enable-force-reg}) and the appropriate declarations must be
                   2588: present in the @code{machine.h} file (see @code{mips.h} for an example;
                   2589: you can find a full list of all declarable register symbols with
                   2590: @code{grep register engine.c}). If you give explicit registers to all
                   2591: variables that are declared at the beginning of @code{engine()}, you
                   2592: should be able to use the other caller-saved registers for temporary
                   2593: storage. Alternatively, you can use the @code{gcc} option
                   2594: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
                   2595: Generation Conventions,, GNU C Manual}) to reserve a register
                   2596: (however, this restriction on register allocation may slow Gforth
                   2597: significantly).
                   2599: If this solution is not viable (e.g., because @code{gcc} does not allow
                   2600: you to explicitly declare all the registers you need), you have to find
                   2601: out by looking at the code where the inner interpreter's registers
                   2602: reside and which registers can be used for temporary storage. You can
                   2603: get an assembly listing of the engine's code with @code{make engine.s}.
                   2605: In any case, it is good practice to abstract your assembly code from the
                   2606: actual register allocation. E.g., if the data stack pointer resides in
                   2607: register @code{$17}, create an alias for this register called @code{sp},
                   2608: and use that in your assembly code.
1.18      anton    2610: Another option for implementing normal and defining words efficiently
                   2611: is: adding the wanted functionality to the source of Gforth. For normal
1.35      anton    2612: words you just have to edit @file{primitives} (@pxref{Automatic
                   2613: Generation}), defining words (equivalent to @code{;CODE} words, for fast
                   2614: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
                   2615: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.18      anton    2616: 
                   2618: @node Threading Words,  , Assembler and Code words, Words
1.4       anton    2619: @section Threading Words
                   2621: These words provide access to code addresses and other threading stuff
1.17      anton    2622: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4       anton    2623: abstracts away the differences between direct and indirect threading
                   2624: (and, for direct threading, the machine dependences). However, at
                   2625: present this wordset is still inclomplete. It is also pretty low-level;
                   2626: some day it will hopefully be made unnecessary by an internals words set
                   2627: that abstracts implementation details away completely.
                   2629: doc->code-address
                   2630: doc->does-code
                   2631: doc-code-address!
                   2632: doc-does-code!
                   2633: doc-does-handler!
                   2634: doc-/does-handler
1.18      anton    2636: The code addresses produced by various defining words are produced by
                   2637: the following words:
1.14      anton    2638: 
1.18      anton    2639: doc-docol:
                   2640: doc-docon:
                   2641: doc-dovar:
                   2642: doc-douser:
                   2643: doc-dodefer:
                   2644: doc-dofield:
1.35      anton    2646: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
                   2647: with @code{>DOES-CODE}. If the word was defined in that way, the value
                   2648: returned is different from 0 and identifies the @code{DOES>} used by the
                   2649: defining word.
1.14      anton    2650: 
1.4       anton    2651: @node ANS conformance, Model, Words, Top
                   2652: @chapter ANS conformance
1.17      anton    2654: To the best of our knowledge, Gforth is an
1.14      anton    2655: 
1.15      anton    2656: ANS Forth System
1.34      anton    2657: @itemize @bullet
1.15      anton    2658: @item providing the Core Extensions word set
                   2659: @item providing the Block word set
                   2660: @item providing the Block Extensions word set
                   2661: @item providing the Double-Number word set
                   2662: @item providing the Double-Number Extensions word set
                   2663: @item providing the Exception word set
                   2664: @item providing the Exception Extensions word set
                   2665: @item providing the Facility word set
                   2666: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
                   2667: @item providing the File Access word set
                   2668: @item providing the File Access Extensions word set
                   2669: @item providing the Floating-Point word set
                   2670: @item providing the Floating-Point Extensions word set
                   2671: @item providing the Locals word set
                   2672: @item providing the Locals Extensions word set
                   2673: @item providing the Memory-Allocation word set
                   2674: @item providing the Memory-Allocation Extensions word set (that one's easy)
                   2675: @item providing the Programming-Tools word set
1.34      anton    2676: @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    2677: @item providing the Search-Order word set
                   2678: @item providing the Search-Order Extensions word set
                   2679: @item providing the String word set
                   2680: @item providing the String Extensions word set (another easy one)
                   2681: @end itemize
                   2683: In addition, ANS Forth systems are required to document certain
                   2684: implementation choices. This chapter tries to meet these
                   2685: requirements. In many cases it gives a way to ask the system for the
                   2686: information instead of providing the information directly, in
                   2687: particular, if the information depends on the processor, the operating
                   2688: system or the installation options chosen, or if they are likely to
1.17      anton    2689: change during the maintenance of Gforth.
1.15      anton    2690: 
1.14      anton    2691: @comment The framework for the rest has been taken from pfe.
                   2693: @menu
                   2694: * The Core Words::              
                   2695: * The optional Block word set::  
                   2696: * The optional Double Number word set::  
                   2697: * The optional Exception word set::  
                   2698: * The optional Facility word set::  
                   2699: * The optional File-Access word set::  
                   2700: * The optional Floating-Point word set::  
                   2701: * The optional Locals word set::  
                   2702: * The optional Memory-Allocation word set::  
                   2703: * The optional Programming-Tools word set::  
                   2704: * The optional Search-Order word set::  
                   2705: @end menu
                   2708: @c =====================================================================
                   2709: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
                   2710: @comment  node-name,  next,  previous,  up
                   2711: @section The Core Words
                   2712: @c =====================================================================
                   2714: @menu
1.15      anton    2715: * core-idef::                   Implementation Defined Options                   
                   2716: * core-ambcond::                Ambiguous Conditions                
                   2717: * core-other::                  Other System Documentation                  
1.14      anton    2718: @end menu
                   2720: @c ---------------------------------------------------------------------
                   2721: @node core-idef, core-ambcond, The Core Words, The Core Words
                   2722: @subsection Implementation Defined Options
                   2723: @c ---------------------------------------------------------------------
                   2725: @table @i
                   2727: @item (Cell) aligned addresses:
1.17      anton    2728: processor-dependent. Gforth's alignment words perform natural alignment
1.14      anton    2729: (e.g., an address aligned for a datum of size 8 is divisible by
                   2730: 8). Unaligned accesses usually result in a @code{-23 THROW}.
                   2732: @item @code{EMIT} and non-graphic characters:
                   2733: The character is output using the C library function (actually, macro)
1.36      anton    2734: @code{putc}.
1.14      anton    2735: 
                   2736: @item character editing of @code{ACCEPT} and @code{EXPECT}:
                   2737: This is modeled on the GNU readline library (@pxref{Readline
                   2738: Interaction, , Command Line Editing, readline, The GNU Readline
                   2739: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
                   2740: producing a full word completion every time you type it (instead of
                   2741: producing the common prefix of all completions).
                   2743: @item character set:
                   2744: The character set of your computer and display device. Gforth is
                   2745: 8-bit-clean (but some other component in your system may make trouble).
                   2747: @item Character-aligned address requirements:
                   2748: installation-dependent. Currently a character is represented by a C
                   2749: @code{unsigned char}; in the future we might switch to @code{wchar_t}
                   2750: (Comments on that requested).
                   2752: @item character-set extensions and matching of names:
1.17      anton    2753: Any character except the ASCII NUL charcter can be used in a
1.36      anton    2754: name. Matching is case-insensitive (except in @code{TABLE}s. The
                   2755: matching is performed using the C function @code{strncasecmp}, whose
                   2756: function is probably influenced by the locale. E.g., the @code{C} locale
                   2757: does not know about accents and umlauts, so they are matched
                   2758: case-sensitively in that locale. For portability reasons it is best to
                   2759: write programs such that they work in the @code{C} locale. Then one can
                   2760: use libraries written by a Polish programmer (who might use words
                   2761: containing ISO Latin-2 encoded characters) and by a French programmer
                   2762: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
                   2763: funny results for some of the words (which ones, depends on the font you
                   2764: are using)). Also, the locale you prefer may not be available in other
                   2765: operating systems. Hopefully, Unicode will solve these problems one day.
1.14      anton    2766: 
                   2767: @item conditions under which control characters match a space delimiter:
                   2768: If @code{WORD} is called with the space character as a delimiter, all
                   2769: white-space characters (as identified by the C macro @code{isspace()})
                   2770: are delimiters. @code{PARSE}, on the other hand, treats space like other
                   2771: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
                   2772: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
                   2773: interpreter (aka text interpreter) by default, treats all white-space
                   2774: characters as delimiters.
                   2776: @item format of the control flow stack:
                   2777: The data stack is used as control flow stack. The size of a control flow
                   2778: stack item in cells is given by the constant @code{cs-item-size}. At the
                   2779: time of this writing, an item consists of a (pointer to a) locals list
                   2780: (third), an address in the code (second), and a tag for identifying the
                   2781: item (TOS). The following tags are used: @code{defstart},
                   2782: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
                   2783: @code{scopestart}.
                   2785: @item conversion of digits > 35
                   2786: The characters @code{[\]^_'} are the digits with the decimal value
                   2787: 36@minus{}41. There is no way to input many of the larger digits.
                   2789: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
                   2790: The cursor is moved to the end of the entered string. If the input is
                   2791: terminated using the @kbd{Return} key, a space is typed.
                   2793: @item exception abort sequence of @code{ABORT"}:
                   2794: The error string is stored into the variable @code{"error} and a
                   2795: @code{-2 throw} is performed.
                   2797: @item input line terminator:
1.36      anton    2798: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
                   2799: lines. One of these characters is typically produced when you type the
                   2800: @kbd{Enter} or @kbd{Return} key.
1.14      anton    2801: 
                   2802: @item maximum size of a counted string:
                   2803: @code{s" /counted-string" environment? drop .}. Currently 255 characters
                   2804: on all ports, but this may change.
                   2806: @item maximum size of a parsed string:
                   2807: Given by the constant @code{/line}. Currently 255 characters.
                   2809: @item maximum size of a definition name, in characters:
                   2810: 31
                   2812: @item maximum string length for @code{ENVIRONMENT?}, in characters:
                   2813: 31
                   2815: @item method of selecting the user input device:
1.17      anton    2816: The user input device is the standard input. There is currently no way to
                   2817: change it from within Gforth. However, the input can typically be
                   2818: redirected in the command line that starts Gforth.
1.14      anton    2819: 
                   2820: @item method of selecting the user output device:
1.36      anton    2821: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
                   2822: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
                   2823: output, so output on a terminal does not become visible before the next
                   2824: newline or buffer overflow. Output on non-terminals is invisible until
                   2825: the buffer overflows.
1.14      anton    2826: 
                   2827: @item methods of dictionary compilation:
1.17      anton    2828: What are we expected to document here?
1.14      anton    2829: 
                   2830: @item number of bits in one address unit:
                   2831: @code{s" address-units-bits" environment? drop .}. 8 in all current
                   2832: ports.
                   2834: @item number representation and arithmetic:
                   2835: Processor-dependent. Binary two's complement on all current ports.
                   2837: @item ranges for integer types:
                   2838: Installation-dependent. Make environmental queries for @code{MAX-N},
                   2839: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
                   2840: unsigned (and positive) types is 0. The lower bound for signed types on
                   2841: two's complement and one's complement machines machines can be computed
                   2842: by adding 1 to the upper bound.
                   2844: @item read-only data space regions:
                   2845: The whole Forth data space is writable.
                   2847: @item size of buffer at @code{WORD}:
                   2848: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   2849: shared with the pictured numeric output string. If overwriting
                   2850: @code{PAD} is acceptable, it is as large as the remaining dictionary
                   2851: space, although only as much can be sensibly used as fits in a counted
                   2852: string.
                   2854: @item size of one cell in address units:
                   2855: @code{1 cells .}.
                   2857: @item size of one character in address units:
                   2858: @code{1 chars .}. 1 on all current ports.
                   2860: @item size of the keyboard terminal buffer:
1.36      anton    2861: Varies. You can determine the size at a specific time using @code{lp@@
1.14      anton    2862: tib - .}. It is shared with the locals stack and TIBs of files that
                   2863: include the current file. You can change the amount of space for TIBs
1.17      anton    2864: and locals stack at Gforth startup with the command line option
1.14      anton    2865: @code{-l}.
                   2867: @item size of the pictured numeric output buffer:
                   2868: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   2869: shared with @code{WORD}.
                   2871: @item size of the scratch area returned by @code{PAD}:
                   2872: The remainder of dictionary space. You can even use the unused part of
1.36      anton    2873: the data stack space. The current size can be computed with @code{sp@@
1.14      anton    2874: pad - .}.
                   2876: @item system case-sensitivity characteristics:
1.36      anton    2877: Dictionary searches are case insensitive (except in
                   2878: @code{TABLE}s). However, as explained above under @i{character-set
                   2879: extensions}, the matching for non-ASCII characters is determined by the
                   2880: locale you are using. In the default @code{C} locale all non-ASCII
                   2881: characters are matched case-sensitively.
1.14      anton    2882: 
                   2883: @item system prompt:
                   2884: @code{ ok} in interpret state, @code{ compiled} in compile state.
                   2886: @item division rounding:
                   2887: installation dependent. @code{s" floored" environment? drop .}. We leave
1.25      anton    2888: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
1.14      anton    2889: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
                   2891: @item values of @code{STATE} when true:
                   2892: -1.
                   2894: @item values returned after arithmetic overflow:
                   2895: On two's complement machines, arithmetic is performed modulo
                   2896: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   2897: arithmetic (with appropriate mapping for signed types). Division by zero
1.36      anton    2898: typically results in a @code{-55 throw} (Floating-point unidentified
1.14      anton    2899: fault), although a @code{-10 throw} (divide by zero) would be more
                   2900: appropriate.
                   2902: @item whether the current definition can be found after @t{DOES>}:
                   2903: No.
                   2905: @end table
                   2907: @c ---------------------------------------------------------------------
                   2908: @node core-ambcond, core-other, core-idef, The Core Words
                   2909: @subsection Ambiguous conditions
                   2910: @c ---------------------------------------------------------------------
                   2912: @table @i
                   2914: @item a name is neither a word nor a number:
1.36      anton    2915: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
                   2916: preserves the data and FP stack, so you don't lose more work than
                   2917: necessary.
1.14      anton    2918: 
                   2919: @item a definition name exceeds the maximum length allowed:
                   2920: @code{-19 throw} (Word name too long)
                   2922: @item addressing a region not inside the various data spaces of the forth system:
                   2923: The stacks, code space and name space are accessible. Machine code space is
                   2924: typically readable. Accessing other addresses gives results dependent on
                   2925: the operating system. On decent systems: @code{-9 throw} (Invalid memory
                   2926: address).
                   2928: @item argument type incompatible with parameter:
                   2929: This is usually not caught. Some words perform checks, e.g., the control
                   2930: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
                   2931: mismatch).
                   2933: @item attempting to obtain the execution token of a word with undefined execution semantics:
1.36      anton    2934: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
                   2935: get an execution token for @code{compile-only-error} (which performs a
                   2936: @code{-14 throw} when executed).
1.14      anton    2937: 
                   2938: @item dividing by zero:
                   2939: typically results in a @code{-55 throw} (floating point unidentified
                   2940: fault), although a @code{-10 throw} (divide by zero) would be more
                   2941: appropriate.
                   2943: @item insufficient data stack or return stack space:
                   2944: Not checked. This typically results in mysterious illegal memory
                   2945: accesses, producing @code{-9 throw} (Invalid memory address) or
                   2946: @code{-23 throw} (Address alignment exception).
                   2948: @item insufficient space for loop control parameters:
                   2949: like other return stack overflows.
                   2951: @item insufficient space in the dictionary:
                   2952: Not checked. Similar results as stack overflows. However, typically the
                   2953: error appears at a different place when one inserts or removes code.
                   2955: @item interpreting a word with undefined interpretation semantics:
                   2956: For some words, we defined interpretation semantics. For the others:
1.36      anton    2957: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    2958: 
                   2959: @item modifying the contents of the input buffer or a string literal:
                   2960: These are located in writable memory and can be modified.
                   2962: @item overflow of the pictured numeric output string:
                   2963: Not checked.
                   2965: @item parsed string overflow:
                   2966: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
                   2968: @item producing a result out of range:
                   2969: On two's complement machines, arithmetic is performed modulo
                   2970: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   2971: arithmetic (with appropriate mapping for signed types). Division by zero
                   2972: typically results in a @code{-55 throw} (floatingpoint unidentified
                   2973: fault), although a @code{-10 throw} (divide by zero) would be more
                   2974: appropriate. @code{convert} and @code{>number} currently overflow
                   2975: silently.
                   2977: @item reading from an empty data or return stack:
                   2978: The data stack is checked by the outer (aka text) interpreter after
                   2979: every word executed. If it has underflowed, a @code{-4 throw} (Stack
                   2980: underflow) is performed. Apart from that, the stacks are not checked and
                   2981: underflows can result in similar behaviour as overflows (of adjacent
                   2982: stacks).
1.36      anton    2984: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
1.14      anton    2985: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
                   2986: use zero-length string as a name). Words like @code{'} probably will not
                   2987: find what they search. Note that it is possible to create zero-length
                   2988: names with @code{nextname} (should it not?).
                   2990: @item @code{>IN} greater than input buffer:
                   2991: The next invocation of a parsing word returns a string wih length 0.
                   2993: @item @code{RECURSE} appears after @code{DOES>}:
1.36      anton    2994: Compiles a recursive call to the defining word, not to the defined word.
1.14      anton    2995: 
                   2996: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.27      anton    2997: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
                   2998: the end of the file was reached), its source-id may be
                   2999: reused. Therefore, restoring an input source specification referencing a
                   3000: closed file may lead to unpredictable results instead of a @code{-12
                   3001: THROW}.
1.36      anton    3003: In the future, Gforth may be able to restore input source specifications
1.27      anton    3004: from other than the current input soruce.
1.14      anton    3005: 
                   3006: @item data space containing definitions gets de-allocated:
                   3007: Deallocation with @code{allot} is not checked. This typically resuls in
                   3008: memory access faults or execution of illegal instructions.
                   3010: @item data space read/write with incorrect alignment:
                   3011: Processor-dependent. Typically results in a @code{-23 throw} (Address
                   3012: alignment exception). Under Linux on a 486 or later processor with
                   3013: alignment turned on, incorrect alignment results in a @code{-9 throw}
                   3014: (Invalid memory address). There are reportedly some processors with
                   3015: alignment restrictions that do not report them.
                   3017: @item data space pointer not properly aligned, @code{,}, @code{C,}:
                   3018: Like other alignment errors.
                   3020: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
                   3021: Not checked. May cause an illegal memory access.
                   3023: @item loop control parameters not available:
                   3024: Not checked. The counted loop words simply assume that the top of return
                   3025: stack items are loop control parameters and behave accordingly.
                   3027: @item most recent definition does not have a name (@code{IMMEDIATE}):
                   3028: @code{abort" last word was headerless"}.
                   3030: @item name not defined by @code{VALUE} used by @code{TO}:
1.36      anton    3031: @code{-32 throw} (Invalid name argument) (unless name was defined by
                   3032: @code{CONSTANT}; then it just changes the constant).
1.14      anton    3033: 
1.15      anton    3034: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.14      anton    3035: @code{-13 throw} (Undefined word)
                   3037: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
                   3038: Gforth behaves as if they were of the same type. I.e., you can predict
                   3039: the behaviour by interpreting all parameters as, e.g., signed.
                   3041: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
1.36      anton    3042: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
                   3043: compilation semantics of @code{TO}.
1.14      anton    3044: 
                   3045: @item String longer than a counted string returned by @code{WORD}:
                   3046: Not checked. The string will be ok, but the count will, of course,
                   3047: contain only the least significant bits of the length.
1.15      anton    3049: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.14      anton    3050: Processor-dependent. Typical behaviours are returning 0 and using only
                   3051: the low bits of the shift count.
                   3053: @item word not defined via @code{CREATE}:
                   3054: @code{>BODY} produces the PFA of the word no matter how it was defined.
                   3056: @code{DOES>} changes the execution semantics of the last defined word no
                   3057: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
                   3058: @code{CREATE , DOES>}.
                   3060: @item words improperly used outside @code{<#} and @code{#>}:
                   3061: Not checked. As usual, you can expect memory faults.
                   3063: @end table
                   3066: @c ---------------------------------------------------------------------
                   3067: @node core-other,  , core-ambcond, The Core Words
                   3068: @subsection Other system documentation
                   3069: @c ---------------------------------------------------------------------
                   3071: @table @i
                   3073: @item nonstandard words using @code{PAD}:
                   3074: None.
                   3076: @item operator's terminal facilities available:
1.26      anton    3077: After processing the command line, Gforth goes into interactive mode,
                   3078: and you can give commands to Gforth interactively. The actual facilities
                   3079: available depend on how you invoke Gforth.
1.14      anton    3080: 
                   3081: @item program data space available:
1.36      anton    3082: @code{sp@@ here - .} gives the space remaining for dictionary and data
1.14      anton    3083: stack together.
                   3085: @item return stack space available:
1.26      anton    3086: By default 16 KBytes. The default can be overridden with the @code{-r}
                   3087: switch (@pxref{Invocation}) when Gforth starts up.
1.14      anton    3088: 
                   3089: @item stack space available:
1.36      anton    3090: @code{sp@@ here - .} gives the space remaining for dictionary and data
1.14      anton    3091: stack together.
                   3093: @item system dictionary space required, in address units:
                   3094: Type @code{here forthstart - .} after startup. At the time of this
                   3095: writing, this gives 70108 (bytes) on a 32-bit system.
                   3096: @end table
                   3099: @c =====================================================================
                   3100: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
                   3101: @section The optional Block word set
                   3102: @c =====================================================================
                   3104: @menu
1.15      anton    3105: * block-idef::                  Implementation Defined Options                  
                   3106: * block-ambcond::               Ambiguous Conditions               
                   3107: * block-other::                 Other System Documentation                 
1.14      anton    3108: @end menu
                   3111: @c ---------------------------------------------------------------------
                   3112: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
                   3113: @subsection Implementation Defined Options
                   3114: @c ---------------------------------------------------------------------
                   3116: @table @i
                   3118: @item the format for display by @code{LIST}:
                   3119: First the screen number is displayed, then 16 lines of 64 characters,
                   3120: each line preceded by the line number.
                   3122: @item the length of a line affected by @code{\}:
                   3123: 64 characters.
                   3124: @end table
                   3127: @c ---------------------------------------------------------------------
                   3128: @node block-ambcond, block-other, block-idef, The optional Block word set
                   3129: @subsection Ambiguous conditions
                   3130: @c ---------------------------------------------------------------------
                   3132: @table @i
                   3134: @item correct block read was not possible:
                   3135: Typically results in a @code{throw} of some OS-derived value (between
                   3136: -512 and -2048). If the blocks file was just not long enough, blanks are
                   3137: supplied for the missing portion.
                   3139: @item I/O exception in block transfer:
                   3140: Typically results in a @code{throw} of some OS-derived value (between
                   3141: -512 and -2048).
                   3143: @item invalid block number:
                   3144: @code{-35 throw} (Invalid block number)
                   3146: @item a program directly alters the contents of @code{BLK}:
                   3147: The input stream is switched to that other block, at the same
                   3148: position. If the storing to @code{BLK} happens when interpreting
                   3149: non-block input, the system will get quite confused when the block ends.
                   3151: @item no current block buffer for @code{UPDATE}:
                   3152: @code{UPDATE} has no effect.
                   3154: @end table
                   3157: @c ---------------------------------------------------------------------
                   3158: @node block-other,  , block-ambcond, The optional Block word set
                   3159: @subsection Other system documentation
                   3160: @c ---------------------------------------------------------------------
                   3162: @table @i
                   3164: @item any restrictions a multiprogramming system places on the use of buffer addresses:
                   3165: No restrictions (yet).
                   3167: @item the number of blocks available for source and data:
                   3168: depends on your disk space.
                   3170: @end table
                   3173: @c =====================================================================
                   3174: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
                   3175: @section The optional Double Number word set
                   3176: @c =====================================================================
                   3178: @menu
1.15      anton    3179: * double-ambcond::              Ambiguous Conditions              
1.14      anton    3180: @end menu
                   3183: @c ---------------------------------------------------------------------
1.15      anton    3184: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
1.14      anton    3185: @subsection Ambiguous conditions
                   3186: @c ---------------------------------------------------------------------
                   3188: @table @i
1.15      anton    3190: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.14      anton    3191: The least significant cell of @var{d} is produced.
                   3193: @end table
                   3196: @c =====================================================================
                   3197: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
                   3198: @section The optional Exception word set
                   3199: @c =====================================================================
                   3201: @menu
1.15      anton    3202: * exception-idef::              Implementation Defined Options              
1.14      anton    3203: @end menu
                   3206: @c ---------------------------------------------------------------------
1.15      anton    3207: @node exception-idef,  , The optional Exception word set, The optional Exception word set
1.14      anton    3208: @subsection Implementation Defined Options
                   3209: @c ---------------------------------------------------------------------
                   3211: @table @i
                   3212: @item @code{THROW}-codes used in the system:
                   3213: The codes -256@minus{}-511 are used for reporting signals (see
                   3214: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
                   3215: (for file and memory allocation operations). The mapping from OS error
1.37      anton    3216: numbers to throw code is -512@minus{}@code{errno}. One side effect of
1.14      anton    3217: this mapping is that undefined OS errors produce a message with a
                   3218: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
                   3219: 488} on my system.
                   3220: @end table
                   3222: @c =====================================================================
                   3223: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
                   3224: @section The optional Facility word set
                   3225: @c =====================================================================
                   3227: @menu
1.15      anton    3228: * facility-idef::               Implementation Defined Options               
                   3229: * facility-ambcond::            Ambiguous Conditions            
1.14      anton    3230: @end menu
                   3233: @c ---------------------------------------------------------------------
                   3234: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
                   3235: @subsection Implementation Defined Options
                   3236: @c ---------------------------------------------------------------------
                   3238: @table @i
                   3240: @item encoding of keyboard events (@code{EKEY}):
                   3241: Not yet implemeted.
                   3243: @item duration of a system clock tick
                   3244: System dependent. With respect to @code{MS}, the time is specified in
                   3245: microseconds. How well the OS and the hardware implement this, is
                   3246: another question.
                   3248: @item repeatability to be expected from the execution of @code{MS}:
                   3249: System dependent. On Unix, a lot depends on load. If the system is
1.17      anton    3250: lightly loaded, and the delay is short enough that Gforth does not get
1.14      anton    3251: swapped out, the performance should be acceptable. Under MS-DOS and
                   3252: other single-tasking systems, it should be good.
                   3254: @end table
                   3257: @c ---------------------------------------------------------------------
1.15      anton    3258: @node facility-ambcond,  , facility-idef, The optional Facility word set
1.14      anton    3259: @subsection Ambiguous conditions
                   3260: @c ---------------------------------------------------------------------
                   3262: @table @i
                   3264: @item @code{AT-XY} can't be performed on user output device:
                   3265: Largely terminal dependant. No range checks are done on the arguments.
                   3266: No errors are reported. You may see some garbage appearing, you may see
                   3267: simply nothing happen.
                   3269: @end table
                   3272: @c =====================================================================
                   3273: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
                   3274: @section The optional File-Access word set
                   3275: @c =====================================================================
                   3277: @menu
1.15      anton    3278: * file-idef::                   Implementation Defined Options                   
                   3279: * file-ambcond::                Ambiguous Conditions                
1.14      anton    3280: @end menu
                   3283: @c ---------------------------------------------------------------------
                   3284: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
                   3285: @subsection Implementation Defined Options
                   3286: @c ---------------------------------------------------------------------
                   3288: @table @i
                   3290: @item File access methods used:
                   3291: @code{R/O}, @code{R/W} and @code{BIN} work as you would
                   3292: expect. @code{W/O} translates into the C file opening mode @code{w} (or
                   3293: @code{wb}): The file is cleared, if it exists, and created, if it does
1.15      anton    3294: not (both with @code{open-file} and @code{create-file}).  Under Unix
1.14      anton    3295: @code{create-file} creates a file with 666 permissions modified by your
                   3296: umask.
                   3298: @item file exceptions:
                   3299: The file words do not raise exceptions (except, perhaps, memory access
                   3300: faults when you pass illegal addresses or file-ids).
                   3302: @item file line terminator:
                   3303: System-dependent. Gforth uses C's newline character as line
                   3304: terminator. What the actual character code(s) of this are is
                   3305: system-dependent.
                   3307: @item file name format
                   3308: System dependent. Gforth just uses the file name format of your OS.
                   3310: @item information returned by @code{FILE-STATUS}:
                   3311: @code{FILE-STATUS} returns the most powerful file access mode allowed
                   3312: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
                   3313: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
                   3314: along with the retured mode.
                   3316: @item input file state after an exception when including source:
                   3317: All files that are left via the exception are closed.
                   3319: @item @var{ior} values and meaning:
1.15      anton    3320: The @var{ior}s returned by the file and memory allocation words are
                   3321: intended as throw codes. They typically are in the range
                   3322: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3323: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3324: 
                   3325: @item maximum depth of file input nesting:
                   3326: limited by the amount of return stack, locals/TIB stack, and the number
                   3327: of open files available. This should not give you troubles.
                   3329: @item maximum size of input line:
                   3330: @code{/line}. Currently 255.
                   3332: @item methods of mapping block ranges to files:
1.37      anton    3333: By default, blocks are accessed in the file @file{blocks.fb} in the
                   3334: current working directory. The file can be switched with @code{USE}.
1.14      anton    3335: 
                   3336: @item number of string buffers provided by @code{S"}:
                   3337: 1
                   3339: @item size of string buffer used by @code{S"}:
                   3340: @code{/line}. currently 255.
                   3342: @end table
                   3344: @c ---------------------------------------------------------------------
1.15      anton    3345: @node file-ambcond,  , file-idef, The optional File-Access word set
1.14      anton    3346: @subsection Ambiguous conditions
                   3347: @c ---------------------------------------------------------------------
                   3349: @table @i
                   3351: @item attempting to position a file outside it's boundaries:
                   3352: @code{REPOSITION-FILE} is performed as usual: Afterwards,
                   3353: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
                   3355: @item attempting to read from file positions not yet written:
                   3356: End-of-file, i.e., zero characters are read and no error is reported.
                   3358: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
                   3359: An appropriate exception may be thrown, but a memory fault or other
                   3360: problem is more probable.
                   3362: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
                   3363: The @var{ior} produced by the operation, that discovered the problem, is
                   3364: thrown.
                   3366: @item named file cannot be opened (@code{included}):
                   3367: The @var{ior} produced by @code{open-file} is thrown.
                   3369: @item requesting an unmapped block number:
                   3370: There are no unmapped legal block numbers. On some operating systems,
                   3371: writing a block with a large number may overflow the file system and
                   3372: have an error message as consequence.
                   3374: @item using @code{source-id} when @code{blk} is non-zero:
                   3375: @code{source-id} performs its function. Typically it will give the id of
                   3376: the source which loaded the block. (Better ideas?)
                   3378: @end table
                   3381: @c =====================================================================
                   3382: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15      anton    3383: @section The optional Floating-Point word set
1.14      anton    3384: @c =====================================================================
                   3386: @menu
1.15      anton    3387: * floating-idef::               Implementation Defined Options
                   3388: * floating-ambcond::            Ambiguous Conditions            
1.14      anton    3389: @end menu
                   3392: @c ---------------------------------------------------------------------
                   3393: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
                   3394: @subsection Implementation Defined Options
                   3395: @c ---------------------------------------------------------------------
                   3397: @table @i
1.15      anton    3399: @item format and range of floating point numbers:
                   3400: System-dependent; the @code{double} type of C.
1.14      anton    3401: 
1.15      anton    3402: @item results of @code{REPRESENT} when @var{float} is out of range:
                   3403: System dependent; @code{REPRESENT} is implemented using the C library
                   3404: function @code{ecvt()} and inherits its behaviour in this respect.
1.14      anton    3405: 
1.15      anton    3406: @item rounding or truncation of floating-point numbers:
1.26      anton    3407: System dependent; the rounding behaviour is inherited from the hosting C
                   3408: compiler. IEEE-FP-based (i.e., most) systems by default round to
                   3409: nearest, and break ties by rounding to even (i.e., such that the last
                   3410: bit of the mantissa is 0).
1.14      anton    3411: 
1.15      anton    3412: @item size of floating-point stack:
                   3413: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
                   3414: with the command-line option @code{-f}.
1.14      anton    3415: 
1.15      anton    3416: @item width of floating-point stack:
                   3417: @code{1 floats}.
1.14      anton    3418: 
                   3419: @end table
                   3422: @c ---------------------------------------------------------------------
1.15      anton    3423: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
                   3424: @subsection Ambiguous conditions
1.14      anton    3425: @c ---------------------------------------------------------------------
                   3427: @table @i
1.15      anton    3429: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
1.37      anton    3430: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3431: alignment violations.
1.14      anton    3432: 
1.15      anton    3433: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
1.37      anton    3434: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3435: alignment violations.
1.14      anton    3436: 
1.15      anton    3437: @item Floating-point result out of range:
                   3438: System-dependent. Can result in a @code{-55 THROW} (Floating-point
                   3439: unidentified fault), or can produce a special value representing, e.g.,
                   3440: Infinity.
1.14      anton    3441: 
1.15      anton    3442: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
                   3443: System-dependent. Typically results in an alignment fault like other
                   3444: alignment violations.
1.14      anton    3445: 
1.15      anton    3446: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
                   3447: The floating-point number is converted into decimal nonetheless.
1.14      anton    3448: 
1.15      anton    3449: @item Both arguments are equal to zero (@code{FATAN2}):
                   3450: System-dependent. @code{FATAN2} is implemented using the C library
                   3451: function @code{atan2()}.
1.14      anton    3452: 
1.15      anton    3453: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
                   3454: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
                   3455: because of small errors and the tan will be a very large (or very small)
                   3456: but finite number.
1.14      anton    3457: 
1.15      anton    3458: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
                   3459: The result is rounded to the nearest float.
1.14      anton    3460: 
1.15      anton    3461: @item dividing by zero:
                   3462: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3463: 
1.15      anton    3464: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
                   3465: System dependent. On IEEE-FP based systems the number is converted into
                   3466: an infinity.
1.14      anton    3467: 
1.15      anton    3468: @item @var{float}<1 (@code{facosh}):
                   3469: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3470: 
1.15      anton    3471: @item @var{float}=<-1 (@code{flnp1}):
                   3472: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3473: negative infinity is typically produced for @var{float}=-1.
1.14      anton    3474: 
1.15      anton    3475: @item @var{float}=<0 (@code{fln}, @code{flog}):
                   3476: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3477: negative infinity is typically produced for @var{float}=0.
1.14      anton    3478: 
1.15      anton    3479: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
                   3480: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
                   3481: produces values for these inputs on my Linux box (Bug in the C library?)
1.14      anton    3482: 
1.15      anton    3483: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
                   3484: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    3485: 
1.15      anton    3486: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
                   3487: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    3488: 
1.15      anton    3489: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
                   3490: This does not happen.
                   3491: @end table
1.14      anton    3492: 
                   3495: @c =====================================================================
1.15      anton    3496: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
                   3497: @section The optional Locals word set
1.14      anton    3498: @c =====================================================================
                   3500: @menu
1.15      anton    3501: * locals-idef::                 Implementation Defined Options                 
                   3502: * locals-ambcond::              Ambiguous Conditions              
1.14      anton    3503: @end menu
                   3506: @c ---------------------------------------------------------------------
1.15      anton    3507: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14      anton    3508: @subsection Implementation Defined Options
                   3509: @c ---------------------------------------------------------------------
                   3511: @table @i
1.15      anton    3513: @item maximum number of locals in a definition:
                   3514: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
                   3515: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
                   3516: characters. The number of locals in a definition is bounded by the size
                   3517: of locals-buffer, which contains the names of the locals.
1.14      anton    3518: 
                   3519: @end table
                   3522: @c ---------------------------------------------------------------------
1.15      anton    3523: @node locals-ambcond,  , locals-idef, The optional Locals word set
1.14      anton    3524: @subsection Ambiguous conditions
                   3525: @c ---------------------------------------------------------------------
                   3527: @table @i
1.15      anton    3529: @item executing a named local in interpretation state:
                   3530: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    3531: 
1.15      anton    3532: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
                   3533: @code{-32 throw} (Invalid name argument)
1.14      anton    3534: 
                   3535: @end table
                   3538: @c =====================================================================
1.15      anton    3539: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
                   3540: @section The optional Memory-Allocation word set
1.14      anton    3541: @c =====================================================================
                   3543: @menu
1.15      anton    3544: * memory-idef::                 Implementation Defined Options                 
1.14      anton    3545: @end menu
                   3548: @c ---------------------------------------------------------------------
1.15      anton    3549: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14      anton    3550: @subsection Implementation Defined Options
                   3551: @c ---------------------------------------------------------------------
                   3553: @table @i
1.15      anton    3555: @item values and meaning of @var{ior}:
                   3556: The @var{ior}s returned by the file and memory allocation words are
                   3557: intended as throw codes. They typically are in the range
                   3558: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3559: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3560: 
                   3561: @end table
                   3563: @c =====================================================================
1.15      anton    3564: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
                   3565: @section The optional Programming-Tools word set
1.14      anton    3566: @c =====================================================================
                   3568: @menu
1.15      anton    3569: * programming-idef::            Implementation Defined Options            
                   3570: * programming-ambcond::         Ambiguous Conditions         
1.14      anton    3571: @end menu
                   3574: @c ---------------------------------------------------------------------
1.15      anton    3575: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14      anton    3576: @subsection Implementation Defined Options
                   3577: @c ---------------------------------------------------------------------
                   3579: @table @i
1.15      anton    3581: @item ending sequence for input following @code{;code} and @code{code}:
1.37      anton    3582: @code{end-code}
1.14      anton    3583: 
1.15      anton    3584: @item manner of processing input following @code{;code} and @code{code}:
1.37      anton    3585: The @code{assembler} vocabulary is pushed on the search order stack, and
                   3586: the input is processed by the text interpreter, (starting) in interpret
                   3587: state.
1.15      anton    3588: 
                   3589: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
1.37      anton    3590: The ANS Forth search order word set.
1.15      anton    3591: 
                   3592: @item source and format of display by @code{SEE}:
                   3593: The source for @code{see} is the intermediate code used by the inner
                   3594: interpreter.  The current @code{see} tries to output Forth source code
                   3595: as well as possible.
1.14      anton    3597: @end table
                   3599: @c ---------------------------------------------------------------------
1.15      anton    3600: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
1.14      anton    3601: @subsection Ambiguous conditions
                   3602: @c ---------------------------------------------------------------------
                   3604: @table @i
1.15      anton    3606: @item deleting the compilation wordlist (@code{FORGET}):
                   3607: Not implemented (yet).
1.14      anton    3608: 
1.15      anton    3609: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
                   3610: This typically results in an @code{abort"} with a descriptive error
                   3611: message (may change into a @code{-22 throw} (Control structure mismatch)
                   3612: in the future). You may also get a memory access error. If you are
                   3613: unlucky, this ambiguous condition is not caught.
                   3615: @item @var{name} can't be found (@code{forget}):
                   3616: Not implemented (yet).
1.14      anton    3617: 
1.15      anton    3618: @item @var{name} not defined via @code{CREATE}:
1.37      anton    3619: @code{;code} behaves like @code{DOES>} in this respect, i.e., it changes
                   3620: the execution semantics of the last defined word no matter how it was
                   3621: defined.
1.14      anton    3622: 
1.15      anton    3623: @item @code{POSTPONE} applied to @code{[IF]}:
                   3624: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
                   3625: equivalent to @code{[IF]}.
1.14      anton    3626: 
1.15      anton    3627: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
                   3628: Continue in the same state of conditional compilation in the next outer
                   3629: input source. Currently there is no warning to the user about this.
1.14      anton    3630: 
1.15      anton    3631: @item removing a needed definition (@code{FORGET}):
                   3632: Not implemented (yet).
1.14      anton    3633: 
                   3634: @end table
                   3637: @c =====================================================================
1.15      anton    3638: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
                   3639: @section The optional Search-Order word set
1.14      anton    3640: @c =====================================================================
                   3642: @menu
1.15      anton    3643: * search-idef::                 Implementation Defined Options                 
                   3644: * search-ambcond::              Ambiguous Conditions              
1.14      anton    3645: @end menu
                   3648: @c ---------------------------------------------------------------------
1.15      anton    3649: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14      anton    3650: @subsection Implementation Defined Options
                   3651: @c ---------------------------------------------------------------------
                   3653: @table @i
1.15      anton    3655: @item maximum number of word lists in search order:
                   3656: @code{s" wordlists" environment? drop .}. Currently 16.
                   3658: @item minimum search order:
                   3659: @code{root root}.
1.14      anton    3660: 
                   3661: @end table
                   3663: @c ---------------------------------------------------------------------
1.15      anton    3664: @node search-ambcond,  , search-idef, The optional Search-Order word set
1.14      anton    3665: @subsection Ambiguous conditions
                   3666: @c ---------------------------------------------------------------------
                   3668: @table @i
1.15      anton    3670: @item changing the compilation wordlist (during compilation):
1.33      anton    3671: The word is entered into the wordlist that was the compilation wordlist
                   3672: at the start of the definition. Any changes to the name field (e.g.,
                   3673: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
                   3674: are applied to the latest defined word (as reported by @code{last} or
                   3675: @code{lastxt}), if possible, irrespective of the compilation wordlist.
1.14      anton    3676: 
1.15      anton    3677: @item search order empty (@code{previous}):
                   3678: @code{abort" Vocstack empty"}.
1.14      anton    3679: 
1.15      anton    3680: @item too many word lists in search order (@code{also}):
                   3681: @code{abort" Vocstack full"}.
1.14      anton    3682: 
                   3683: @end table
1.13      anton    3684: 
1.34      anton    3685: @node Model, Integrating Gforth, ANS conformance, Top
                   3686: @chapter Model
                   3688: This chapter has yet to be written. It will contain information, on
                   3689: which internal structures you can rely.
                   3691: @node Integrating Gforth, Emacs and Gforth, Model, Top
                   3692: @chapter Integrating Gforth into C programs
                   3694: This is not yet implemented.
                   3696: Several people like to use Forth as scripting language for applications
                   3697: that are otherwise written in C, C++, or some other language.
                   3699: The Forth system ATLAST provides facilities for embedding it into
                   3700: applications; unfortunately it has several disadvantages: most
1.36      anton    3701: importantly, it is not based on ANS Forth, and it is apparently dead
1.34      anton    3702: (i.e., not developed further and not supported). The facilities
                   3703: provided by Gforth in this area are inspired by ATLASTs facilities, so
                   3704: making the switch should not be hard.
                   3706: We also tried to design the interface such that it can easily be
                   3707: implemented by other Forth systems, so that we may one day arrive at a
                   3708: standardized interface. Such a standard interface would allow you to
                   3709: replace the Forth system without having to rewrite C code.
                   3711: You embed the Gforth interpreter by linking with the library
                   3712: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
                   3713: global symbols in this library that belong to the interface, have the
                   3714: prefix @code{forth_}. (Global symbols that are used internally have the
                   3715: prefix @code{gforth_}).
                   3717: You can include the declarations of Forth types and the functions and
1.36      anton    3718: variables of the interface with @code{#include <forth.h>}.
1.34      anton    3719: 
                   3720: Types.
1.13      anton    3721: 
1.34      anton    3722: Variables.
                   3724: Data and FP Stack pointer. Area sizes.
                   3726: functions.
                   3728: forth_init(imagefile)
                   3729: forth_evaluate(string) exceptions?
                   3730: forth_goto(address) (or forth_execute(xt)?)
                   3731: forth_continue() (a corountining mechanism)
                   3733: Adding primitives.
                   3735: No checking.
                   3737: Signals?
                   3739: Accessing the Stacks
1.4       anton    3740: 
1.34      anton    3741: @node Emacs and Gforth, Internals, Integrating Gforth, Top
1.17      anton    3742: @chapter Emacs and Gforth
1.4       anton    3743: 
1.17      anton    3744: Gforth comes with @file{gforth.el}, an improved version of
1.33      anton    3745: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.4       anton    3746: improvements are a better (but still not perfect) handling of
                   3747: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8       anton    3748: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
                   3749: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
                   3750: stuff I do not use alone, even though some of it only makes sense for
                   3751: TILE. To get a description of these features, enter Forth mode and type
                   3752: @kbd{C-h m}.
1.4       anton    3753: 
1.17      anton    3754: In addition, Gforth supports Emacs quite well: The source code locations
1.4       anton    3755: given in error messages, debugging output (from @code{~~}) and failed
                   3756: assertion messages are in the right format for Emacs' compilation mode
                   3757: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
                   3758: Manual}) so the source location corresponding to an error or other
                   3759: message is only a few keystrokes away (@kbd{C-x `} for the next error,
                   3760: @kbd{C-c C-c} for the error under the cursor).
                   3762: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
                   3763: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
                   3764: contains the definitions of all words defined afterwards. You can then
                   3765: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17      anton    3766: several tags files at the same time (e.g., one for the Gforth sources
1.28      anton    3767: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
                   3768: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
                   3769: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
1.33      anton    3770: @file{/usr/local/share/gforth/0.2.0/TAGS}).
1.4       anton    3771: 
                   3772: To get all these benefits, add the following lines to your @file{.emacs}
                   3773: file:
                   3775: @example
                   3776: (autoload 'forth-mode "gforth.el")
                   3777: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
                   3778: @end example
1.17      anton    3780: @node Internals, Bugs, Emacs and Gforth, Top
1.3       anton    3781: @chapter Internals
1.17      anton    3783: Reading this section is not necessary for programming with Gforth. It
                   3784: should be helpful for finding your way in the Gforth sources.
1.3       anton    3785: 
1.24      anton    3786: The ideas in this section have also been published in the papers
                   3787: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
                   3788: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
                   3789: Ertl, presented at EuroForth '93; the latter is available at
                   3790: @*@file{}.
1.4       anton    3792: @menu
                   3793: * Portability::                 
                   3794: * Threading::                   
                   3795: * Primitives::                  
                   3796: * System Architecture::         
1.17      anton    3797: * Performance::                 
1.4       anton    3798: @end menu
                   3800: @node Portability, Threading, Internals, Internals
1.3       anton    3801: @section Portability
                   3803: One of the main goals of the effort is availability across a wide range
                   3804: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
                   3805: this goal by manually coding the engine in assembly language for several
                   3806: then-popular processors. This approach is very labor-intensive and the
                   3807: results are short-lived due to progress in computer architecture.
                   3809: Others have avoided this problem by coding in C, e.g., Mitch Bradley
                   3810: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
                   3811: particularly popular for UNIX-based Forths due to the large variety of
                   3812: architectures of UNIX machines. Unfortunately an implementation in C
                   3813: does not mix well with the goals of efficiency and with using
                   3814: traditional techniques: Indirect or direct threading cannot be expressed
                   3815: in C, and switch threading, the fastest technique available in C, is
                   3816: significantly slower. Another problem with C is that it's very
                   3817: cumbersome to express double integer arithmetic.
                   3819: Fortunately, there is a portable language that does not have these
                   3820: limitations: GNU C, the version of C processed by the GNU C compiler
                   3821: (@pxref{C Extensions, , Extensions to the C Language Family,,
                   3822: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
                   3823: Labels as Values,, GNU C Manual}) makes direct and indirect
                   3824: threading possible, its @code{long long} type (@pxref{Long Long, ,
1.33      anton    3825: Double-Word Integers,, GNU C Manual}) corresponds to Forth's
1.32      anton    3826: double numbers@footnote{Unfortunately, long longs are not implemented
                   3827: properly on all machines (e.g., on alpha-osf1, long longs are only 64
                   3828: bits, the same size as longs (and pointers), but they should be twice as
                   3829: long according to @ref{Long Long, , Double-Word Integers,, GNU
                   3830: C Manual}). So, we had to implement doubles in C after all. Still, on
                   3831: most machines we can use long longs and achieve better performance than
                   3832: with the emulation package.}. GNU C is available for free on all
                   3833: important (and many unimportant) UNIX machines, VMS, 80386s running
                   3834: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
                   3835: on all these machines.
1.3       anton    3836: 
                   3837: Writing in a portable language has the reputation of producing code that
                   3838: is slower than assembly. For our Forth engine we repeatedly looked at
                   3839: the code produced by the compiler and eliminated most compiler-induced
                   3840: inefficiencies by appropriate changes in the source-code.
                   3842: However, register allocation cannot be portably influenced by the
                   3843: programmer, leading to some inefficiencies on register-starved
                   3844: machines. We use explicit register declarations (@pxref{Explicit Reg
                   3845: Vars, , Variables in Specified Registers,, GNU C Manual}) to
                   3846: improve the speed on some machines. They are turned on by using the
                   3847: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
                   3848: only depends on the machine, but also on the compiler version: On some
                   3849: machines some compiler versions produce incorrect code when certain
                   3850: explicit register declarations are used. So by default
                   3851: @code{-DFORCE_REG} is not used.
1.4       anton    3853: @node Threading, Primitives, Portability, Internals
1.3       anton    3854: @section Threading
                   3856: GNU C's labels as values extension (available since @code{gcc-2.0},
                   3857: @pxref{Labels as Values, , Labels as Values,, GNU C Manual})
                   3858: makes it possible to take the address of @var{label} by writing
                   3859: @code{&&@var{label}}.  This address can then be used in a statement like
                   3860: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
                   3861: @code{goto x}.
                   3863: With this feature an indirect threaded NEXT looks like:
                   3864: @example
                   3865: cfa = *ip++;
                   3866: ca = *cfa;
                   3867: goto *ca;
                   3868: @end example
                   3869: For those unfamiliar with the names: @code{ip} is the Forth instruction
                   3870: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
                   3871: execution token and points to the code field of the next word to be
                   3872: executed; The @code{ca} (code address) fetched from there points to some
                   3873: executable code, e.g., a primitive or the colon definition handler
                   3874: @code{docol}.
                   3876: Direct threading is even simpler:
                   3877: @example
                   3878: ca = *ip++;
                   3879: goto *ca;
                   3880: @end example
                   3882: Of course we have packaged the whole thing neatly in macros called
                   3883: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
1.4       anton    3885: @menu
                   3886: * Scheduling::                  
                   3887: * Direct or Indirect Threaded?::  
                   3888: * DOES>::                       
                   3889: @end menu
                   3891: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3       anton    3892: @subsection Scheduling
                   3894: There is a little complication: Pipelined and superscalar processors,
                   3895: i.e., RISC and some modern CISC machines can process independent
                   3896: instructions while waiting for the results of an instruction. The
                   3897: compiler usually reorders (schedules) the instructions in a way that
                   3898: achieves good usage of these delay slots. However, on our first tries
                   3899: the compiler did not do well on scheduling primitives. E.g., for
                   3900: @code{+} implemented as
                   3901: @example
                   3902: n=sp[0]+sp[1];
                   3903: sp++;
                   3904: sp[0]=n;
                   3905: NEXT;
                   3906: @end example
                   3907: the NEXT comes strictly after the other code, i.e., there is nearly no
                   3908: scheduling. After a little thought the problem becomes clear: The
                   3909: compiler cannot know that sp and ip point to different addresses (and
1.4       anton    3910: the version of @code{gcc} we used would not know it even if it was
                   3911: possible), so it could not move the load of the cfa above the store to
                   3912: the TOS. Indeed the pointers could be the same, if code on or very near
                   3913: the top of stack were executed. In the interest of speed we chose to
                   3914: forbid this probably unused ``feature'' and helped the compiler in
                   3915: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
                   3916: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3       anton    3917: @example
                   3918: n=sp[0]+sp[1];
                   3919: sp++;
                   3920: NEXT_P1;
                   3921: sp[0]=n;
                   3922: NEXT_P2;
                   3923: @end example
1.4       anton    3924: This can be scheduled optimally by the compiler.
1.3       anton    3925: 
                   3926: This division can be turned off with the switch @code{-DCISC_NEXT}. This
                   3927: switch is on by default on machines that do not profit from scheduling
                   3928: (e.g., the 80386), in order to preserve registers.
1.4       anton    3930: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3       anton    3931: @subsection Direct or Indirect Threaded?
                   3933: Both! After packaging the nasty details in macro definitions we
                   3934: realized that we could switch between direct and indirect threading by
                   3935: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
                   3936: defining a few machine-specific macros for the direct-threading case.
                   3937: On the Forth level we also offer access words that hide the
                   3938: differences between the threading methods (@pxref{Threading Words}).
                   3940: Indirect threading is implemented completely
                   3941: machine-independently. Direct threading needs routines for creating
                   3942: jumps to the executable code (e.g. to docol or dodoes). These routines
                   3943: are inherently machine-dependent, but they do not amount to many source
                   3944: lines. I.e., even porting direct threading to a new machine is a small
                   3945: effort.
1.4       anton    3947: @node DOES>,  , Direct or Indirect Threaded?, Threading
1.3       anton    3948: @subsection DOES>
                   3949: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
                   3950: the chunk of code executed by every word defined by a
                   3951: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
                   3952: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
                   3953: DOES-code)? There are two solutions:
                   3955: In fig-Forth the code field points directly to the dodoes and the
                   3956: DOES-code address is stored in the cell after the code address
                   3957: (i.e. at cfa cell+). It may seem that this solution is illegal in the
                   3958: Forth-79 and all later standards, because in fig-Forth this address
                   3959: lies in the body (which is illegal in these standards). However, by
                   3960: making the code field larger for all words this solution becomes legal
                   3961: again. We use this approach for the indirect threaded version. Leaving
                   3962: a cell unused in most words is a bit wasteful, but on the machines we
                   3963: are targetting this is hardly a problem. The other reason for having a
                   3964: code field size of two cells is to avoid having different image files
1.4       anton    3965: for direct and indirect threaded systems (@pxref{System Architecture}).
1.3       anton    3966: 
                   3967: The other approach is that the code field points or jumps to the cell
                   3968: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
                   3969: this address. @code{dodoes} can then get the DOES-code address by
                   3970: computing the code address, i.e., the address of the jump to dodoes,
                   3971: and add the length of that jump field. A variant of this is to have a
                   3972: call to @code{dodoes} after the @code{DOES>}; then the return address
                   3973: (which can be found in the return register on RISCs) is the DOES-code
                   3974: address. Since the two cells available in the code field are usually
                   3975: used up by the jump to the code address in direct threading, we use
                   3976: this approach for direct threading. We did not want to add another
                   3977: cell to the code field.
1.4       anton    3979: @node Primitives, System Architecture, Threading, Internals
1.3       anton    3980: @section Primitives
1.4       anton    3982: @menu
                   3983: * Automatic Generation::        
                   3984: * TOS Optimization::            
                   3985: * Produced code::               
                   3986: @end menu
                   3988: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3       anton    3989: @subsection Automatic Generation
                   3991: Since the primitives are implemented in a portable language, there is no
                   3992: longer any need to minimize the number of primitives. On the contrary,
                   3993: having many primitives is an advantage: speed. In order to reduce the
                   3994: number of errors in primitives and to make programming them easier, we
                   3995: provide a tool, the primitive generator (@file{prims2x.fs}), that
                   3996: automatically generates most (and sometimes all) of the C code for a
                   3997: primitive from the stack effect notation.  The source for a primitive
                   3998: has the following form:
                   4000: @format
                   4001: @var{Forth-name}       @var{stack-effect}      @var{category}  [@var{pronounc.}]
                   4002: [@code{""}@var{glossary entry}@code{""}]
                   4003: @var{C code}
                   4004: [@code{:}
                   4005: @var{Forth code}]
                   4006: @end format
                   4008: The items in brackets are optional. The category and glossary fields
                   4009: are there for generating the documentation, the Forth code is there
                   4010: for manual implementations on machines without GNU C. E.g., the source
                   4011: for the primitive @code{+} is:
                   4012: @example
                   4013: +    n1 n2 -- n    core    plus
                   4014: n = n1+n2;
                   4015: @end example
                   4017: This looks like a specification, but in fact @code{n = n1+n2} is C
                   4018: code. Our primitive generation tool extracts a lot of information from
                   4019: the stack effect notations@footnote{We use a one-stack notation, even
                   4020: though we have separate data and floating-point stacks; The separate
                   4021: notation can be generated easily from the unified notation.}: The number
                   4022: of items popped from and pushed on the stack, their type, and by what
                   4023: name they are referred to in the C code. It then generates a C code
                   4024: prelude and postlude for each primitive. The final C code for @code{+}
                   4025: looks like this:
                   4027: @example
                   4028: I_plus:        /* + ( n1 n2 -- n ) */  /* label, stack effect */
                   4029: /*  */                          /* documentation */
1.4       anton    4030: @{
1.3       anton    4031: DEF_CA                          /* definition of variable ca (indirect threading) */
                   4032: Cell n1;                        /* definitions of variables */
                   4033: Cell n2;
                   4034: Cell n;
                   4035: n1 = (Cell) sp[1];              /* input */
                   4036: n2 = (Cell) TOS;
                   4037: sp += 1;                        /* stack adjustment */
                   4038: NAME("+")                       /* debugging output (with -DDEBUG) */
1.4       anton    4039: @{
1.3       anton    4040: n = n1+n2;                      /* C code taken from the source */
1.4       anton    4041: @}
1.3       anton    4042: NEXT_P1;                        /* NEXT part 1 */
                   4043: TOS = (Cell)n;                  /* output */
                   4044: NEXT_P2;                        /* NEXT part 2 */
1.4       anton    4045: @}
1.3       anton    4046: @end example
                   4048: This looks long and inefficient, but the GNU C compiler optimizes quite
                   4049: well and produces optimal code for @code{+} on, e.g., the R3000 and the
                   4050: HP RISC machines: Defining the @code{n}s does not produce any code, and
                   4051: using them as intermediate storage also adds no cost.
                   4053: There are also other optimizations, that are not illustrated by this
                   4054: example: Assignments between simple variables are usually for free (copy
                   4055: propagation). If one of the stack items is not used by the primitive
                   4056: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
                   4057: (dead code elimination). On the other hand, there are some things that
                   4058: the compiler does not do, therefore they are performed by
                   4059: @file{prims2x.fs}: The compiler does not optimize code away that stores
                   4060: a stack item to the place where it just came from (e.g., @code{over}).
                   4062: While programming a primitive is usually easy, there are a few cases
                   4063: where the programmer has to take the actions of the generator into
                   4064: account, most notably @code{?dup}, but also words that do not (always)
                   4065: fall through to NEXT.
1.4       anton    4067: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3       anton    4068: @subsection TOS Optimization
                   4070: An important optimization for stack machine emulators, e.g., Forth
                   4071: engines, is keeping  one or more of the top stack items in
1.4       anton    4072: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
                   4073: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.34      anton    4074: @itemize @bullet
1.3       anton    4075: @item
                   4076: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
                   4077: due to fewer loads from and stores to the stack.
                   4078: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
                   4079: @var{y<n}, due to additional moves between registers.
                   4080: @end itemize
                   4082: In particular, keeping one item in a register is never a disadvantage,
                   4083: if there are enough registers. Keeping two items in registers is a
                   4084: disadvantage for frequent words like @code{?branch}, constants,
                   4085: variables, literals and @code{i}. Therefore our generator only produces
                   4086: code that keeps zero or one items in registers. The generated C code
                   4087: covers both cases; the selection between these alternatives is made at
                   4088: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
                   4089: code for @code{+} is just a simple variable name in the one-item case,
                   4090: otherwise it is a macro that expands into @code{sp[0]}. Note that the
                   4091: GNU C compiler tries to keep simple variables like @code{TOS} in
                   4092: registers, and it usually succeeds, if there are enough registers.
                   4094: The primitive generator performs the TOS optimization for the
                   4095: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
                   4096: operations the benefit of this optimization is even larger:
                   4097: floating-point operations take quite long on most processors, but can be
                   4098: performed in parallel with other operations as long as their results are
                   4099: not used. If the FP-TOS is kept in a register, this works. If
                   4100: it is kept on the stack, i.e., in memory, the store into memory has to
                   4101: wait for the result of the floating-point operation, lengthening the
                   4102: execution time of the primitive considerably.
                   4104: The TOS optimization makes the automatic generation of primitives a
                   4105: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
                   4106: @code{TOS} is not sufficient. There are some special cases to
                   4107: consider:
1.34      anton    4108: @itemize @bullet
1.3       anton    4109: @item In the case of @code{dup ( w -- w w )} the generator must not
                   4110: eliminate the store to the original location of the item on the stack,
                   4111: if the TOS optimization is turned on.
1.4       anton    4112: @item Primitives with stack effects of the form @code{--}
                   4113: @var{out1}...@var{outy} must store the TOS to the stack at the start.
                   4114: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3       anton    4115: must load the TOS from the stack at the end. But for the null stack
                   4116: effect @code{--} no stores or loads should be generated.
                   4117: @end itemize
1.4       anton    4119: @node Produced code,  , TOS Optimization, Primitives
1.3       anton    4120: @subsection Produced code
                   4122: To see what assembly code is produced for the primitives on your machine
                   4123: with your compiler and your flag settings, type @code{make engine.s} and
1.4       anton    4124: look at the resulting file @file{engine.s}.
1.3       anton    4125: 
1.17      anton    4126: @node System Architecture, Performance, Primitives, Internals
1.3       anton    4127: @section System Architecture
                   4129: Our Forth system consists not only of primitives, but also of
                   4130: definitions written in Forth. Since the Forth compiler itself belongs
                   4131: to those definitions, it is not possible to start the system with the
                   4132: primitives and the Forth source alone. Therefore we provide the Forth
                   4133: code as an image file in nearly executable form. At the start of the
                   4134: system a C routine loads the image file into memory, sets up the
                   4135: memory (stacks etc.) according to information in the image file, and
                   4136: starts executing Forth code.
                   4138: The image file format is a compromise between the goals of making it
                   4139: easy to generate image files and making them portable. The easiest way
                   4140: to generate an image file is to just generate a memory dump. However,
                   4141: this kind of image file cannot be used on a different machine, or on
                   4142: the next version of the engine on the same machine, it even might not
                   4143: work with the same engine compiled by a different version of the C
                   4144: compiler. We would like to have as few versions of the image file as
                   4145: possible, because we do not want to distribute many versions of the
                   4146: same image file, and to make it easy for the users to use their image
                   4147: files on many machines. We currently need to create a different image
                   4148: file for machines with different cell sizes and different byte order
1.17      anton    4149: (little- or big-endian)@footnote{We are considering adding information to the
1.3       anton    4150: image file that enables the loader to change the byte order.}.
                   4152: Forth code that is going to end up in a portable image file has to
1.4       anton    4153: comply to some restrictions: addresses have to be stored in memory with
                   4154: special words (@code{A!}, @code{A,}, etc.) in order to make the code
                   4155: relocatable. Cells, floats, etc., have to be stored at the natural
                   4156: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
                   4157: dividable by~8. This happens automatically in our system when you use
                   4158: the ANS Forth alignment words.}, in order to avoid alignment faults on
                   4159: machines with stricter alignment. The image file is produced by a
                   4160: metacompiler (@file{cross.fs}).
1.3       anton    4161: 
                   4162: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
                   4163: file is not directly executable, but has to undergo some manipulations
                   4164: during loading. Address relocation is performed at image load-time, not
                   4165: at run-time. The loader also has to replace tokens standing for
                   4166: primitive calls with the appropriate code-field addresses (or code
                   4167: addresses in the case of direct threading).
1.4       anton    4168: 
1.17      anton    4169: @node  Performance,  , System Architecture, Internals
                   4170: @section Performance
                   4172: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
                   4173: impossible to write a significantly faster engine.
                   4175: On register-starved machines like the 386 architecture processors
                   4176: improvements are possible, because @code{gcc} does not utilize the
                   4177: registers as well as a human, even with explicit register declarations;
                   4178: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
                   4179: and hand-tuned it for the 486; this system is 1.19 times faster on the
                   4180: Sieve benchmark on a 486DX2/66 than Gforth compiled with
                   4181: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
                   4183: However, this potential advantage of assembly language implementations
                   4184: is not necessarily realized in complete Forth systems: We compared
1.26      anton    4185: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
                   4186: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
                   4187: 1994) and Eforth (with and without peephole (aka pinhole) optimization
                   4188: of the threaded code); all these systems were written in assembly
1.30      anton    4189: language. We also compared Gforth with three systems written in C:
1.32      anton    4190: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
                   4191: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
                   4192: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
                   4193: -fomit-frame-pointer; ThisForth employs peephole optimization of the
                   4194: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
                   4195: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
                   4196: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
                   4197: 486DX2/66 with similar memory performance under Windows NT. Marcel
                   4198: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
                   4199: added the peephole optimizer, ran the benchmarks and reported the
                   4200: results.
1.17      anton    4201:  
                   4202: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
                   4203: matrix multiplication come from the Stanford integer benchmarks and have
                   4204: been translated into Forth by Martin Fraeman; we used the versions
1.30      anton    4205: included in the TILE Forth package, but with bigger data set sizes; and
                   4206: a recursive Fibonacci number computation for benchmarking calling
                   4207: performance. The following table shows the time taken for the benchmarks
                   4208: scaled by the time taken by Gforth (in other words, it shows the speedup
                   4209: factor that Gforth achieved over the other systems).
1.17      anton    4210: 
                   4211: @example
1.30      anton    4212: relative      Win32-    NT       eforth       This-
                   4213:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
1.32      anton    4214: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
                   4215: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
1.38    ! anton    4216: matmul    1.00  1.47  1.35   1.46  0.74  1.58        4.09
        !          4217: fib       1.00  1.52  1.34   1.22  0.86  1.74  2.99  4.30
1.17      anton    4218: @end example
                   4220: You may find the good performance of Gforth compared with the systems
                   4221: written in assembly language quite surprising. One important reason for
                   4222: the disappointing performance of these systems is probably that they are
                   4223: not written optimally for the 486 (e.g., they use the @code{lods}
                   4224: instruction). In addition, Win32Forth uses a comfortable, but costly
                   4225: method for relocating the Forth image: like @code{cforth}, it computes
                   4226: the actual addresses at run time, resulting in two address computations
                   4227: per NEXT (@pxref{System Architecture}).
1.26      anton    4229: Only Eforth with the peephole optimizer performs comparable to
                   4230: Gforth. The speedups achieved with peephole optimization of threaded
                   4231: code are quite remarkable. Adding a peephole optimizer to Gforth should
                   4232: cause similar speedups.
1.30      anton    4234: The speedup of Gforth over PFE, ThisForth and TILE can be easily
                   4235: explained with the self-imposed restriction to standard C, which makes
                   4236: efficient threading impossible (however, the measured implementation of
                   4237: PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
                   4238: Register Variables,, GNU C Manual}).  Moreover, current C
                   4239: compilers have a hard time optimizing other aspects of the ThisForth
                   4240: and the TILE source.
1.17      anton    4241: 
                   4242: Note that the performance of Gforth on 386 architecture processors
                   4243: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
                   4244: failed to allocate any of the virtual machine registers into real
                   4245: machine registers by itself and would not work correctly with explicit
                   4246: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
                   4247: running the Sieve) than the one measured above.
1.26      anton    4249: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
                   4250: Maierhofer (presented at EuroForth '95), an indirect threaded version of
                   4251: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
                   4252: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
                   4253: used here. The paper available at
1.24      anton    4254: @*@file{};
                   4255: it also contains numbers for some native code systems. You can find
                   4256: numbers for Gforth on various machines in @file{Benchres}.
1.29      anton    4258: @node Bugs, Origin, Internals, Top
1.4       anton    4259: @chapter Bugs
1.17      anton    4261: Known bugs are described in the file BUGS in the Gforth distribution.
1.24      anton    4263: If you find a bug, please send a bug report to
1.32      anton    4264: @code{}. A bug report should
1.17      anton    4265: describe the Gforth version used (it is announced at the start of an
                   4266: interactive Gforth session), the machine and operating system (on Unix
                   4267: systems you can use @code{uname -a} to produce this information), the
1.24      anton    4268: installation options (send the @code{config.status} file), and a
                   4269: complete list of changes you (or your installer) have made to the Gforth
                   4270: sources (if any); it should contain a program (or a sequence of keyboard
                   4271: commands) that reproduces the bug and a description of what you think
                   4272: constitutes the buggy behaviour.
1.17      anton    4273: 
                   4274: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
                   4275: to Report Bugs,, GNU C Manual}.
1.29      anton    4278: @node Origin, Word Index, Bugs, Top
                   4279: @chapter Authors and Ancestors of Gforth
                   4281: @section Authors and Contributors
                   4283: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
1.30      anton    4284: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
                   4285: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
                   4286: with their continuous feedback. Lennart Benshop contributed
1.29      anton    4287: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
                   4288: support for calling C libraries. Helpful comments also came from Paul
1.37      anton    4289: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
                   4290: Wavrik, Stott Bolton and Marc de Groot.
1.29      anton    4291: 
1.30      anton    4292: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
                   4293: and autoconf, among others), and to the creators of the Internet: Gforth
                   4294: was developed across the Internet, and its authors have not met
                   4295: physically yet.
1.29      anton    4297: @section Pedigree
1.4       anton    4298: 
1.17      anton    4299: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24      anton    4300: Dirk Zoller) will cross-fertilize each other. Of course, a significant
                   4301: part of the design of Gforth was prescribed by ANS Forth.
1.17      anton    4302: 
1.23      pazsan   4303: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
                   4304: 32 bit native code version of VolksForth for the Atari ST, written
                   4305: mostly by Dietrich Weineck.
                   4307: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
                   4308: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24      anton    4309: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17      anton    4310: 
1.34      anton    4311: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17      anton    4312: Forth-83 standard. !! Pedigree? When?
                   4314: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24      anton    4315: 1979. Robert Selzer and Bill Ragsdale developed the original
                   4316: implementation of fig-Forth for the 6502 based on microForth.
                   4318: The principal architect of microForth was Dean Sanderson. microForth was
                   4319: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
                   4320: the 1802, and subsequently implemented on the 8080, the 6800 and the
                   4321: Z80.
1.17      anton    4322: 
1.24      anton    4323: All earlier Forth systems were custom-made, usually by Charles Moore,
1.30      anton    4324: who discovered (as he puts it) Forth during the late 60s. The first full
                   4325: Forth existed in 1971.
1.17      anton    4326: 
                   4327: A part of the information in this section comes from @cite{The Evolution
                   4328: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
                   4329: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
                   4330: Notices 28(3), 1993.  You can find more historical and genealogical
                   4331: information about Forth there.
1.29      anton    4333: @node Word Index, Node Index, Origin, Top
1.4       anton    4334: @chapter Word Index
1.18      anton    4336: This index is as incomplete as the manual. Each word is listed with
                   4337: stack effect and wordset.
1.17      anton    4338: 
                   4339: @printindex fn
1.4       anton    4341: @node Node Index,  , Word Index, Top
                   4342: @chapter Node Index
1.17      anton    4343: 
                   4344: This index is even less complete than the manual.
1.1       anton    4345: 
                   4346: @contents
                   4347: @bye

FreeBSD-CVSweb <>