Annotation of gforth/gforth.ds, revision 1.36

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::              
                    676: * Wordlists::                   
                    677: * Files::                       
                    678: * Blocks::                      
                    679: * Other I/O::                   
                    680: * Programming Tools::           
1.18      anton     681: * Assembler and Code words::    
1.4       anton     682: * Threading Words::             
1.1       anton     683: @end menu
                    685: @node Notation, Arithmetic, Words, Words
                    686: @section Notation
                    688: The Forth words are described in this section in the glossary notation
                    689: that has become a de-facto standard for Forth texts, i.e.
1.4       anton     691: @format
1.1       anton     692: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
1.4       anton     693: @end format
1.1       anton     694: @var{Description}
                    696: @table @var
                    697: @item word
1.17      anton     698: The name of the word. BTW, Gforth is case insensitive, so you can
1.14      anton     699: type the words in in lower case (However, @pxref{core-idef}).
1.1       anton     700: 
                    701: @item Stack effect
                    702: The stack effect is written in the notation @code{@var{before} --
                    703: @var{after}}, where @var{before} and @var{after} describe the top of
                    704: stack entries before and after the execution of the word. The rest of
                    705: the stack is not touched by the word. The top of stack is rightmost,
1.17      anton     706: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1       anton     707: uses a separate floating point stack, but a unified stack
                    708: notation. Also, return stack effects are not shown in @var{stack
                    709: effect}, but in @var{Description}. The name of a stack item describes
                    710: the type and/or the function of the item. See below for a discussion of
                    711: the types.
1.19      anton     713: All words have two stack effects: A compile-time stack effect and a
                    714: run-time stack effect. The compile-time stack-effect of most words is
                    715: @var{ -- }. If the compile-time stack-effect of a word deviates from
                    716: this standard behaviour, or the word does other unusual things at
                    717: compile time, both stack effects are shown; otherwise only the run-time
                    718: stack effect is shown.
1.1       anton     720: @item pronunciation
                    721: How the word is pronounced
                    723: @item wordset
                    724: The ANS Forth standard is divided into several wordsets. A standard
                    725: system need not support all of them. So, the fewer wordsets your program
                    726: uses the more portable it will be in theory. However, we suspect that
                    727: most ANS Forth systems on personal machines will feature all
                    728: wordsets. Words that are not defined in the ANS standard have
1.19      anton     729: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
                    730: describes words that will work in future releases of Gforth;
                    731: @code{gforth-internal} words are more volatile. Environmental query
                    732: strings are also displayed like words; you can recognize them by the
                    733: @code{environment} in the wordset field.
1.1       anton     734: 
                    735: @item Description
                    736: A description of the behaviour of the word.
                    737: @end table
1.4       anton     739: The type of a stack item is specified by the character(s) the name
                    740: starts with:
1.1       anton     741: 
                    742: @table @code
                    743: @item f
                    744: Bool, i.e. @code{false} or @code{true}.
                    745: @item c
                    746: Char
                    747: @item w
                    748: Cell, can contain an integer or an address
                    749: @item n
                    750: signed integer
                    751: @item u
                    752: unsigned integer
                    753: @item d
                    754: double sized signed integer
                    755: @item ud
                    756: double sized unsigned integer
                    757: @item r
1.36    ! anton     758: Float (on the FP stack)
1.1       anton     759: @item a_
                    760: Cell-aligned address
                    761: @item c_
1.36    ! anton     762: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.1       anton     763: @item f_
                    764: Float-aligned address
                    765: @item df_
                    766: Address aligned for IEEE double precision float
                    767: @item sf_
                    768: Address aligned for IEEE single precision float
                    769: @item xt
                    770: Execution token, same size as Cell
                    771: @item wid
                    772: Wordlist ID, same size as Cell
                    773: @item f83name
                    774: Pointer to a name structure
1.36    ! anton     775: @item "
        !           776: string in the input stream (not the stack). The terminating character is
        !           777: a blank by default. If it is not a blank, it is shown in @code{<>}
        !           778: quotes.
        !           779: 
1.1       anton     780: @end table
1.4       anton     782: @node Arithmetic, Stack Manipulation, Notation, Words
1.1       anton     783: @section Arithmetic
                    784: Forth arithmetic is not checked, i.e., you will not hear about integer
                    785: overflow on addition or multiplication, you may hear about division by
                    786: zero if you are lucky. The operator is written after the operands, but
                    787: the operands are still in the original order. I.e., the infix @code{2-1}
                    788: corresponds to @code{2 1 -}. Forth offers a variety of division
                    789: operators. If you perform division with potentially negative operands,
                    790: you do not want to use @code{/} or @code{/mod} with its undefined
                    791: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4       anton     792: former, @pxref{Mixed precision}).
                    794: @menu
                    795: * Single precision::            
                    796: * Bitwise operations::          
                    797: * Mixed precision::             operations with single and double-cell integers
                    798: * Double precision::            Double-cell integer arithmetic
                    799: * Floating Point::              
                    800: @end menu
1.1       anton     801: 
1.4       anton     802: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1       anton     803: @subsection Single precision
                    804: doc-+
                    805: doc--
                    806: doc-*
                    807: doc-/
                    808: doc-mod
                    809: doc-/mod
                    810: doc-negate
                    811: doc-abs
                    812: doc-min
                    813: doc-max
1.4       anton     815: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1       anton     816: @subsection Bitwise operations
                    817: doc-and
                    818: doc-or
                    819: doc-xor
                    820: doc-invert
                    821: doc-2*
                    822: doc-2/
1.4       anton     824: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1       anton     825: @subsection Mixed precision
                    826: doc-m+
                    827: doc-*/
                    828: doc-*/mod
                    829: doc-m*
                    830: doc-um*
                    831: doc-m*/
                    832: doc-um/mod
                    833: doc-fm/mod
                    834: doc-sm/rem
1.4       anton     836: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1       anton     837: @subsection Double precision
1.16      anton     838: 
                    839: The outer (aka text) interpreter converts numbers containing a dot into
                    840: a double precision number. Note that only numbers with the dot as last
                    841: character are standard-conforming.
1.1       anton     843: doc-d+
                    844: doc-d-
                    845: doc-dnegate
                    846: doc-dabs
                    847: doc-dmin
                    848: doc-dmax
1.4       anton     850: @node Floating Point,  , Double precision, Arithmetic
                    851: @subsection Floating Point
1.16      anton     852: 
                    853: The format of floating point numbers recognized by the outer (aka text)
                    854: interpreter is: a signed decimal number, possibly containing a decimal
                    855: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
                    856: by a signed integer (the exponent). E.g., @code{1e} ist the same as
1.35      anton     857: @code{+1.0e+0}. Note that a number without @code{e}
1.16      anton     858: is not interpreted as floating-point number, but as double (if the
                    859: number contains a @code{.}) or single precision integer. Also,
                    860: conversions between string and floating point numbers always use base
                    861: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
                    862: value greater then 14, the @code{E} may be interpreted as digit and the
                    863: number will be interpreted as integer, unless it has a signed exponent
                    864: (both @code{+} and @code{-} are allowed as signs).
1.4       anton     865: 
                    866: Angles in floating point operations are given in radians (a full circle
1.17      anton     867: has 2 pi radians). Note, that Gforth has a separate floating point
1.4       anton     868: stack, but we use the unified notation.
                    870: Floating point numbers have a number of unpleasant surprises for the
                    871: unwary (e.g., floating point addition is not associative) and even a few
                    872: for the wary. You should not use them unless you know what you are doing
                    873: or you don't care that the results you get are totally bogus. If you
                    874: want to learn about the problems of floating point numbers (and how to
1.11      anton     875: avoid them), you might start with @cite{David Goldberg, What Every
1.6       anton     876: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
                    877: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4       anton     878: 
                    879: doc-f+
                    880: doc-f-
                    881: doc-f*
                    882: doc-f/
                    883: doc-fnegate
                    884: doc-fabs
                    885: doc-fmax
                    886: doc-fmin
                    887: doc-floor
                    888: doc-fround
                    889: doc-f**
                    890: doc-fsqrt
                    891: doc-fexp
                    892: doc-fexpm1
                    893: doc-fln
                    894: doc-flnp1
                    895: doc-flog
1.6       anton     896: doc-falog
1.4       anton     897: doc-fsin
                    898: doc-fcos
                    899: doc-fsincos
                    900: doc-ftan
                    901: doc-fasin
                    902: doc-facos
                    903: doc-fatan
                    904: doc-fatan2
                    905: doc-fsinh
                    906: doc-fcosh
                    907: doc-ftanh
                    908: doc-fasinh
                    909: doc-facosh
                    910: doc-fatanh
                    912: @node Stack Manipulation, Memory access, Arithmetic, Words
1.1       anton     913: @section Stack Manipulation
1.17      anton     915: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1       anton     916: addresses, and double cells, a floating point stack for floating point
                    917: numbers, a return stack for storing the return addresses of colon
                    918: definitions and other data, and a locals stack for storing local
                    919: variables. Note that while every sane Forth has a separate floating
                    920: point stack, this is not strictly required; an ANS Forth system could
                    921: theoretically keep floating point numbers on the data stack. As an
                    922: additional difficulty, you don't know how many cells a floating point
                    923: number takes. It is reportedly possible to write words in a way that
                    924: they work also for a unified stack model, but we do not recommend trying
1.4       anton     925: it. Instead, just say that your program has an environmental dependency
                    926: on a separate FP stack.
                    928: Also, a Forth system is allowed to keep the local variables on the
1.1       anton     929: return stack. This is reasonable, as local variables usually eliminate
                    930: the need to use the return stack explicitly. So, if you want to produce
                    931: a standard complying program and if you are using local variables in a
                    932: word, forget about return stack manipulations in that word (see the
                    933: standard document for the exact rules).
1.4       anton     935: @menu
                    936: * Data stack::                  
                    937: * Floating point stack::        
                    938: * Return stack::                
                    939: * Locals stack::                
                    940: * Stack pointer manipulation::  
                    941: @end menu
                    943: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1       anton     944: @subsection Data stack
                    945: doc-drop
                    946: doc-nip
                    947: doc-dup
                    948: doc-over
                    949: doc-tuck
                    950: doc-swap
                    951: doc-rot
                    952: doc--rot
                    953: doc-?dup
                    954: doc-pick
                    955: doc-roll
                    956: doc-2drop
                    957: doc-2nip
                    958: doc-2dup
                    959: doc-2over
                    960: doc-2tuck
                    961: doc-2swap
                    962: doc-2rot
1.4       anton     964: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1       anton     965: @subsection Floating point stack
                    966: doc-fdrop
                    967: doc-fnip
                    968: doc-fdup
                    969: doc-fover
                    970: doc-ftuck
                    971: doc-fswap
                    972: doc-frot
1.4       anton     974: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1       anton     975: @subsection Return stack
                    976: doc->r
                    977: doc-r>
                    978: doc-r@
                    979: doc-rdrop
                    980: doc-2>r
                    981: doc-2r>
                    982: doc-2r@
                    983: doc-2rdrop
1.4       anton     985: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1       anton     986: @subsection Locals stack
1.4       anton     988: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
1.1       anton     989: @subsection Stack pointer manipulation
                    990: doc-sp@
                    991: doc-sp!
                    992: doc-fp@
                    993: doc-fp!
                    994: doc-rp@
                    995: doc-rp!
                    996: doc-lp@
                    997: doc-lp!
1.4       anton     999: @node Memory access, Control Structures, Stack Manipulation, Words
1.1       anton    1000: @section Memory access
1.4       anton    1002: @menu
                   1003: * Stack-Memory transfers::      
                   1004: * Address arithmetic::          
                   1005: * Memory block access::         
                   1006: @end menu
                   1008: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1.1       anton    1009: @subsection Stack-Memory transfers
                   1011: doc-@
                   1012: doc-!
                   1013: doc-+!
                   1014: doc-c@
                   1015: doc-c!
                   1016: doc-2@
                   1017: doc-2!
                   1018: doc-f@
                   1019: doc-f!
                   1020: doc-sf@
                   1021: doc-sf!
                   1022: doc-df@
                   1023: doc-df!
1.4       anton    1025: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1.1       anton    1026: @subsection Address arithmetic
                   1028: ANS Forth does not specify the sizes of the data types. Instead, it
                   1029: offers a number of words for computing sizes and doing address
                   1030: arithmetic. Basically, address arithmetic is performed in terms of
                   1031: address units (aus); on most systems the address unit is one byte. Note
                   1032: that a character may have more than one au, so @code{chars} is no noop
                   1033: (on systems where it is a noop, it compiles to nothing).
                   1035: ANS Forth also defines words for aligning addresses for specific
                   1036: addresses. Many computers require that accesses to specific data types
                   1037: must only occur at specific addresses; e.g., that cells may only be
                   1038: accessed at addresses divisible by 4. Even if a machine allows unaligned
                   1039: accesses, it can usually perform aligned accesses faster. 
1.17      anton    1041: For the performance-conscious: alignment operations are usually only
1.1       anton    1042: necessary during the definition of a data structure, not during the
                   1043: (more frequent) accesses to it.
                   1045: ANS Forth defines no words for character-aligning addresses. This is not
                   1046: an oversight, but reflects the fact that addresses that are not
                   1047: char-aligned have no use in the standard and therefore will not be
                   1048: created.
                   1050: The standard guarantees that addresses returned by @code{CREATE}d words
1.17      anton    1051: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1       anton    1052: are aligned for all purposes.
1.9       anton    1054: Note that the standard defines a word @code{char}, which has nothing to
                   1055: do with address arithmetic.
1.1       anton    1057: doc-chars
                   1058: doc-char+
                   1059: doc-cells
                   1060: doc-cell+
                   1061: doc-align
                   1062: doc-aligned
                   1063: doc-floats
                   1064: doc-float+
                   1065: doc-falign
                   1066: doc-faligned
                   1067: doc-sfloats
                   1068: doc-sfloat+
                   1069: doc-sfalign
                   1070: doc-sfaligned
                   1071: doc-dfloats
                   1072: doc-dfloat+
                   1073: doc-dfalign
                   1074: doc-dfaligned
1.10      anton    1075: doc-maxalign
                   1076: doc-maxaligned
                   1077: doc-cfalign
                   1078: doc-cfaligned
1.1       anton    1079: doc-address-unit-bits
1.4       anton    1081: @node Memory block access,  , Address arithmetic, Memory access
1.1       anton    1082: @subsection Memory block access
                   1084: doc-move
                   1085: doc-erase
                   1087: While the previous words work on address units, the rest works on
                   1088: characters.
                   1090: doc-cmove
                   1091: doc-cmove>
                   1092: doc-fill
                   1093: doc-blank
1.4       anton    1095: @node Control Structures, Locals, Memory access, Words
1.1       anton    1096: @section Control Structures
                   1098: Control structures in Forth cannot be used in interpret state, only in
                   1099: compile state, i.e., in a colon definition. We do not like this
                   1100: limitation, but have not seen a satisfying way around it yet, although
                   1101: many schemes have been proposed.
1.4       anton    1103: @menu
                   1104: * Selection::                   
                   1105: * Simple Loops::                
                   1106: * Counted Loops::               
                   1107: * Arbitrary control structures::  
                   1108: * Calls and returns::           
                   1109: * Exception Handling::          
                   1110: @end menu
                   1112: @node Selection, Simple Loops, Control Structures, Control Structures
1.1       anton    1113: @subsection Selection
                   1115: @example
                   1116: @var{flag}
                   1117: IF
                   1118:   @var{code}
                   1119: ENDIF
                   1120: @end example
                   1121: or
                   1122: @example
                   1123: @var{flag}
                   1124: IF
                   1125:   @var{code1}
                   1126: ELSE
                   1127:   @var{code2}
                   1128: ENDIF
                   1129: @end example
1.4       anton    1131: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1       anton    1132: standard, and @code{ENDIF} is not, although it is quite popular. We
                   1133: recommend using @code{ENDIF}, because it is less confusing for people
                   1134: who also know other languages (and is not prone to reinforcing negative
                   1135: prejudices against Forth in these people). Adding @code{ENDIF} to a
                   1136: system that only supplies @code{THEN} is simple:
                   1137: @example
                   1138: : endif   POSTPONE then ; immediate
                   1139: @end example
                   1141: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
                   1142: (adv.)}  has the following meanings:
                   1143: @quotation
                   1144: ... 2b: following next after in order ... 3d: as a necessary consequence
                   1145: (if you were there, then you saw them).
                   1146: @end quotation
                   1147: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
                   1148: and many other programming languages has the meaning 3d.]
1.31      anton    1150: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
                   1151: you can avoid using @code{?dup}. Using these alternatives is also more
                   1152: efficient than using @code{?dup}. Definitions in plain standard Forth
                   1153: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
                   1154: @file{compat/control.fs}.
1.1       anton    1155: 
                   1156: @example
                   1157: @var{n}
                   1158: CASE
                   1159:   @var{n1} OF @var{code1} ENDOF
                   1160:   @var{n2} OF @var{code2} ENDOF
1.4       anton    1161:   @dots{}
1.1       anton    1162: ENDCASE
                   1163: @end example
                   1165: Executes the first @var{codei}, where the @var{ni} is equal to
                   1166: @var{n}. A default case can be added by simply writing the code after
                   1167: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
                   1168: but must not consume it.
1.4       anton    1170: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1       anton    1171: @subsection Simple Loops
                   1173: @example
                   1174: BEGIN
                   1175:   @var{code1}
                   1176:   @var{flag}
                   1177: WHILE
                   1178:   @var{code2}
                   1179: REPEAT
                   1180: @end example
                   1182: @var{code1} is executed and @var{flag} is computed. If it is true,
                   1183: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
                   1185: @example
                   1186: BEGIN
                   1187:   @var{code}
                   1188:   @var{flag}
                   1189: UNTIL
                   1190: @end example
                   1192: @var{code} is executed. The loop is restarted if @code{flag} is false.
                   1194: @example
                   1195: BEGIN
                   1196:   @var{code}
                   1197: AGAIN
                   1198: @end example
                   1200: This is an endless loop.
1.4       anton    1202: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1       anton    1203: @subsection Counted Loops
                   1205: The basic counted loop is:
                   1206: @example
                   1207: @var{limit} @var{start}
                   1208: ?DO
                   1209:   @var{body}
                   1210: LOOP
                   1211: @end example
                   1213: This performs one iteration for every integer, starting from @var{start}
                   1214: and up to, but excluding @var{limit}. The counter, aka index, can be
                   1215: accessed with @code{i}. E.g., the loop
                   1216: @example
                   1217: 10 0 ?DO
                   1218:   i .
                   1219: LOOP
                   1220: @end example
                   1221: prints
                   1222: @example
                   1223: 0 1 2 3 4 5 6 7 8 9
                   1224: @end example
                   1225: The index of the innermost loop can be accessed with @code{i}, the index
                   1226: of the next loop with @code{j}, and the index of the third loop with
                   1227: @code{k}.
                   1229: The loop control data are kept on the return stack, so there are some
                   1230: restrictions on mixing return stack accesses and counted loop
                   1231: words. E.g., if you put values on the return stack outside the loop, you
                   1232: cannot read them inside the loop. If you put values on the return stack
                   1233: within a loop, you have to remove them before the end of the loop and
                   1234: before accessing the index of the loop.
                   1236: There are several variations on the counted loop:
                   1238: @code{LEAVE} leaves the innermost counted loop immediately.
1.18      anton    1240: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
                   1241: (and @code{LOOP} iterates until they become equal by wrap-around
                   1242: arithmetic). This behaviour is usually not what you want. Therefore,
                   1243: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
                   1244: @code{?DO}), which do not enter the loop if @var{start} is greater than
                   1245: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.30      anton    1246: unsigned loop parameters.
1.18      anton    1247: 
1.1       anton    1248: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
                   1249: index by @var{n} instead of by 1. The loop is terminated when the border
                   1250: between @var{limit-1} and @var{limit} is crossed. E.g.:
1.18      anton    1252: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
1.1       anton    1253: 
1.18      anton    1254: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
1.1       anton    1255: 
                   1256: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1.2       anton    1258: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
1.1       anton    1259: 
1.2       anton    1260: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
1.1       anton    1261: 
1.18      anton    1262: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
                   1263: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
                   1264: index by @var{u} each iteration. The loop is terminated when the border
                   1265: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
                   1266: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1       anton    1267: 
1.18      anton    1268: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
1.1       anton    1269: 
1.18      anton    1270: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
1.1       anton    1271: 
1.18      anton    1272: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
1.1       anton    1273: 
1.30      anton    1274: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
                   1275: @code{-LOOP} are not in the ANS Forth standard. However, an
                   1276: implementation for these words that uses only standard words is provided
                   1277: in @file{compat/loops.fs}.
1.18      anton    1278: 
                   1279: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
                   1280: the loop, independent of the loop parameters. Do not use @code{DO}, even
                   1281: if you know that the loop is entered in any case. Such knowledge tends
                   1282: to become invalid during maintenance of a program, and then the
                   1283: @code{DO} will make trouble.
1.1       anton    1284: 
                   1285: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
                   1286: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
                   1287: return stack so @code{EXIT} can get to its return address.
                   1289: Another counted loop is
                   1290: @example
                   1291: @var{n}
                   1292: FOR
                   1293:   @var{body}
                   1294: NEXT
                   1295: @end example
                   1296: This is the preferred loop of native code compiler writers who are too
1.17      anton    1297: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1       anton    1298: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
                   1299: and ending with 0. Other Forth systems may behave differently, even if
1.30      anton    1300: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
                   1301: loops.
1.1       anton    1302: 
1.4       anton    1303: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2       anton    1304: @subsection Arbitrary control structures
                   1306: ANS Forth permits and supports using control structures in a non-nested
                   1307: way. Information about incomplete control structures is stored on the
                   1308: control-flow stack. This stack may be implemented on the Forth data
1.17      anton    1309: stack, and this is what we have done in Gforth.
1.2       anton    1310: 
                   1311: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
                   1312: entry represents a backward branch target. A few words are the basis for
                   1313: building any control structure possible (except control structures that
                   1314: need storage, like calls, coroutines, and backtracking).
1.3       anton    1316: doc-if
                   1317: doc-ahead
                   1318: doc-then
                   1319: doc-begin
                   1320: doc-until
                   1321: doc-again
                   1322: doc-cs-pick
                   1323: doc-cs-roll
1.2       anton    1324: 
1.17      anton    1325: On many systems control-flow stack items take one word, in Gforth they
1.2       anton    1326: currently take three (this may change in the future). Therefore it is a
                   1327: really good idea to manipulate the control flow stack with
                   1328: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
                   1329: words.
                   1331: Some standard control structure words are built from these words:
1.3       anton    1333: doc-else
                   1334: doc-while
                   1335: doc-repeat
1.2       anton    1336: 
1.31      anton    1337: Gforth adds some more control-structure words:
                   1339: doc-endif
                   1340: doc-?dup-if
                   1341: doc-?dup-0=-if
1.2       anton    1343: Counted loop words constitute a separate group of words:
1.3       anton    1345: doc-?do
1.18      anton    1346: doc-+do
                   1347: doc-u+do
                   1348: doc--do
                   1349: doc-u-do
1.3       anton    1350: doc-do
                   1351: doc-for
                   1352: doc-loop
                   1353: doc-+loop
1.18      anton    1354: doc--loop
1.3       anton    1355: doc-next
                   1356: doc-leave
                   1357: doc-?leave
                   1358: doc-unloop
1.10      anton    1359: doc-done
1.2       anton    1360: 
                   1361: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
                   1362: @i{do-sys}. Our system allows it, but it's your job to ensure that for
                   1363: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3       anton    1364: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
                   1365: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7       pazsan   1366: resolved (by using one of the loop-ending words or @code{DONE}).
1.2       anton    1367: 
                   1368: Another group of control structure words are
1.3       anton    1370: doc-case
                   1371: doc-endcase
                   1372: doc-of
                   1373: doc-endof
1.2       anton    1374: 
                   1375: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
                   1376: @code{cs-roll}.
1.3       anton    1378: @subsubsection Programming Style
                   1380: In order to ensure readability we recommend that you do not create
                   1381: arbitrary control structures directly, but define new control structure
                   1382: words for the control structure you want and use these words in your
                   1383: program.
                   1385: E.g., instead of writing
                   1387: @example
                   1388: begin
                   1389:   ...
                   1390: if [ 1 cs-roll ]
                   1391:   ...
                   1392: again then
                   1393: @end example
                   1395: we recommend defining control structure words, e.g.,
                   1397: @example
                   1398: : while ( dest -- orig dest )
                   1399:  POSTPONE if
                   1400:  1 cs-roll ; immediate
                   1402: : repeat ( orig dest -- )
                   1403:  POSTPONE again
                   1404:  POSTPONE then ; immediate
                   1405: @end example
                   1407: and then using these to create the control structure:
                   1409: @example
                   1410: begin
                   1411:   ...
                   1412: while
                   1413:   ...
                   1414: repeat
                   1415: @end example
1.30      anton    1417: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1.3       anton    1418: @code{WHILE} are predefined, so in this example it would not be
                   1419: necessary to define them.
1.4       anton    1421: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3       anton    1422: @subsection Calls and returns
                   1424: A definition can be called simply be writing the name of the
1.17      anton    1425: definition. When the end of the definition is reached, it returns. An
                   1426: earlier return can be forced using
1.3       anton    1427: 
                   1428: doc-exit
                   1430: Don't forget to clean up the return stack and @code{UNLOOP} any
                   1431: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
                   1432: primitive compiled by @code{EXIT} is
                   1434: doc-;s
1.4       anton    1436: @node Exception Handling,  , Calls and returns, Control Structures
1.3       anton    1437: @subsection Exception Handling
                   1439: doc-catch
                   1440: doc-throw
1.4       anton    1442: @node Locals, Defining Words, Control Structures, Words
1.1       anton    1443: @section Locals
1.2       anton    1445: Local variables can make Forth programming more enjoyable and Forth
                   1446: programs easier to read. Unfortunately, the locals of ANS Forth are
                   1447: laden with restrictions. Therefore, we provide not only the ANS Forth
                   1448: locals wordset, but also our own, more powerful locals wordset (we
                   1449: implemented the ANS Forth locals wordset through our locals wordset).
1.24      anton    1451: The ideas in this section have also been published in the paper
                   1452: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
                   1453: at EuroForth '94; it is available at
                   1454: @*@file{}.
1.2       anton    1456: @menu
1.17      anton    1457: * Gforth locals::               
1.4       anton    1458: * ANS Forth locals::            
1.2       anton    1459: @end menu
1.17      anton    1461: @node Gforth locals, ANS Forth locals, Locals, Locals
                   1462: @subsection Gforth locals
1.2       anton    1463: 
                   1464: Locals can be defined with
                   1466: @example
                   1467: @{ local1 local2 ... -- comment @}
                   1468: @end example
                   1469: or
                   1470: @example
                   1471: @{ local1 local2 ... @}
                   1472: @end example
                   1474: E.g.,
                   1475: @example
                   1476: : max @{ n1 n2 -- n3 @}
                   1477:  n1 n2 > if
                   1478:    n1
                   1479:  else
                   1480:    n2
                   1481:  endif ;
                   1482: @end example
                   1484: The similarity of locals definitions with stack comments is intended. A
                   1485: locals definition often replaces the stack comment of a word. The order
                   1486: of the locals corresponds to the order in a stack comment and everything
                   1487: after the @code{--} is really a comment.
                   1489: This similarity has one disadvantage: It is too easy to confuse locals
                   1490: declarations with stack comments, causing bugs and making them hard to
                   1491: find. However, this problem can be avoided by appropriate coding
                   1492: conventions: Do not use both notations in the same program. If you do,
                   1493: they should be distinguished using additional means, e.g. by position.
                   1495: The name of the local may be preceded by a type specifier, e.g.,
                   1496: @code{F:} for a floating point value:
                   1498: @example
                   1499: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
                   1500: \ complex multiplication
                   1501:  Ar Br f* Ai Bi f* f-
                   1502:  Ar Bi f* Ai Br f* f+ ;
                   1503: @end example
1.17      anton    1505: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2       anton    1506: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
                   1507: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
                   1508: with @code{W:}, @code{D:} etc.) produces its value and can be changed
                   1509: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
                   1510: produces its address (which becomes invalid when the variable's scope is
                   1511: left). E.g., the standard word @code{emit} can be defined in therms of
                   1512: @code{type} like this:
                   1514: @example
                   1515: : emit @{ C^ char* -- @}
                   1516:     char* 1 type ;
                   1517: @end example
                   1519: A local without type specifier is a @code{W:} local. Both flavours of
                   1520: locals are initialized with values from the data or FP stack.
                   1522: Currently there is no way to define locals with user-defined data
                   1523: structures, but we are working on it.
1.17      anton    1525: Gforth allows defining locals everywhere in a colon definition. This
1.7       pazsan   1526: poses the following questions:
1.2       anton    1527: 
1.4       anton    1528: @menu
                   1529: * Where are locals visible by name?::  
1.14      anton    1530: * How long do locals live?::    
1.4       anton    1531: * Programming Style::           
                   1532: * Implementation::              
                   1533: @end menu
1.17      anton    1535: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2       anton    1536: @subsubsection Where are locals visible by name?
                   1538: Basically, the answer is that locals are visible where you would expect
                   1539: it in block-structured languages, and sometimes a little longer. If you
                   1540: want to restrict the scope of a local, enclose its definition in
                   1541: @code{SCOPE}...@code{ENDSCOPE}.
                   1543: doc-scope
                   1544: doc-endscope
                   1546: These words behave like control structure words, so you can use them
                   1547: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
                   1548: arbitrary ways.
                   1550: If you want a more exact answer to the visibility question, here's the
                   1551: basic principle: A local is visible in all places that can only be
                   1552: reached through the definition of the local@footnote{In compiler
                   1553: construction terminology, all places dominated by the definition of the
                   1554: local.}. In other words, it is not visible in places that can be reached
                   1555: without going through the definition of the local. E.g., locals defined
                   1556: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
                   1557: defined in @code{BEGIN}...@code{UNTIL} are visible after the
                   1558: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
                   1560: The reasoning behind this solution is: We want to have the locals
                   1561: visible as long as it is meaningful. The user can always make the
                   1562: visibility shorter by using explicit scoping. In a place that can
                   1563: only be reached through the definition of a local, the meaning of a
                   1564: local name is clear. In other places it is not: How is the local
                   1565: initialized at the control flow path that does not contain the
                   1566: definition? Which local is meant, if the same name is defined twice in
                   1567: two independent control flow paths?
                   1569: This should be enough detail for nearly all users, so you can skip the
                   1570: rest of this section. If you relly must know all the gory details and
                   1571: options, read on.
                   1573: In order to implement this rule, the compiler has to know which places
                   1574: are unreachable. It knows this automatically after @code{AHEAD},
                   1575: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
                   1576: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
                   1577: compiler that the control flow never reaches that place. If
                   1578: @code{UNREACHABLE} is not used where it could, the only consequence is
                   1579: that the visibility of some locals is more limited than the rule above
                   1580: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
                   1581: lie to the compiler), buggy code will be produced.
                   1583: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3       anton    1584: does not know which locals will be visible on the incoming
                   1585: back-edge. All problems discussed in the following are due to this
                   1586: ignorance of the compiler (we discuss the problems using @code{BEGIN}
                   1587: loops as examples; the discussion also applies to @code{?DO} and other
1.2       anton    1588: loops). Perhaps the most insidious example is:
                   1589: @example
                   1590: AHEAD
                   1591: BEGIN
                   1592:   x
                   1593: [ 1 CS-ROLL ] THEN
1.4       anton    1594:   @{ x @}
1.2       anton    1595:   ...
                   1596: UNTIL
                   1597: @end example
                   1599: This should be legal according to the visibility rule. The use of
                   1600: @code{x} can only be reached through the definition; but that appears
                   1601: textually below the use.
                   1603: From this example it is clear that the visibility rules cannot be fully
                   1604: implemented without major headaches. Our implementation treats common
                   1605: cases as advertised and the exceptions are treated in a safe way: The
                   1606: compiler makes a reasonable guess about the locals visible after a
                   1607: @code{BEGIN}; if it is too pessimistic, the
                   1608: user will get a spurious error about the local not being defined; if the
                   1609: compiler is too optimistic, it will notice this later and issue a
                   1610: warning. In the case above the compiler would complain about @code{x}
                   1611: being undefined at its use. You can see from the obscure examples in
                   1612: this section that it takes quite unusual control structures to get the
                   1613: compiler into trouble, and even then it will often do fine.
                   1615: If the @code{BEGIN} is reachable from above, the most optimistic guess
                   1616: is that all locals visible before the @code{BEGIN} will also be
                   1617: visible after the @code{BEGIN}. This guess is valid for all loops that
                   1618: are entered only through the @code{BEGIN}, in particular, for normal
                   1619: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
                   1620: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
                   1621: compiler. When the branch to the @code{BEGIN} is finally generated by
                   1622: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
                   1623: warns the user if it was too optimisitic:
                   1624: @example
                   1625: IF
1.4       anton    1626:   @{ x @}
1.2       anton    1627: BEGIN
                   1628:   \ x ? 
                   1629: [ 1 cs-roll ] THEN
                   1630:   ...
                   1631: UNTIL
                   1632: @end example
                   1634: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
                   1635: optimistically assumes that it lives until the @code{THEN}. It notices
                   1636: this difference when it compiles the @code{UNTIL} and issues a
                   1637: warning. The user can avoid the warning, and make sure that @code{x}
                   1638: is not used in the wrong area by using explicit scoping:
                   1639: @example
                   1640: IF
                   1641:   SCOPE
1.4       anton    1642:   @{ x @}
1.2       anton    1643:   ENDSCOPE
                   1644: BEGIN
                   1645: [ 1 cs-roll ] THEN
                   1646:   ...
                   1647: UNTIL
                   1648: @end example
                   1650: Since the guess is optimistic, there will be no spurious error messages
                   1651: about undefined locals.
                   1653: If the @code{BEGIN} is not reachable from above (e.g., after
                   1654: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
                   1655: optimistic guess, as the locals visible after the @code{BEGIN} may be
                   1656: defined later. Therefore, the compiler assumes that no locals are
1.17      anton    1657: visible after the @code{BEGIN}. However, the user can use
1.2       anton    1658: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17      anton    1659: visible at the BEGIN as at the point where the top control-flow stack
                   1660: item was created.
1.2       anton    1661: 
                   1662: doc-assume-live
                   1664: E.g.,
                   1665: @example
1.4       anton    1666: @{ x @}
1.2       anton    1667: AHEAD
                   1668: ASSUME-LIVE
                   1669: BEGIN
                   1670:   x
                   1671: [ 1 CS-ROLL ] THEN
                   1672:   ...
                   1673: UNTIL
                   1674: @end example
                   1676: Other cases where the locals are defined before the @code{BEGIN} can be
                   1677: handled by inserting an appropriate @code{CS-ROLL} before the
                   1678: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
                   1679: behind the @code{ASSUME-LIVE}).
                   1681: Cases where locals are defined after the @code{BEGIN} (but should be
                   1682: visible immediately after the @code{BEGIN}) can only be handled by
                   1683: rearranging the loop. E.g., the ``most insidious'' example above can be
                   1684: arranged into:
                   1685: @example
                   1686: BEGIN
1.4       anton    1687:   @{ x @}
1.2       anton    1688:   ... 0=
                   1689: WHILE
                   1690:   x
                   1691: REPEAT
                   1692: @end example
1.17      anton    1694: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2       anton    1695: @subsubsection How long do locals live?
                   1697: The right answer for the lifetime question would be: A local lives at
                   1698: least as long as it can be accessed. For a value-flavoured local this
                   1699: means: until the end of its visibility. However, a variable-flavoured
                   1700: local could be accessed through its address far beyond its visibility
                   1701: scope. Ultimately, this would mean that such locals would have to be
                   1702: garbage collected. Since this entails un-Forth-like implementation
                   1703: complexities, I adopted the same cowardly solution as some other
                   1704: languages (e.g., C): The local lives only as long as it is visible;
                   1705: afterwards its address is invalid (and programs that access it
                   1706: afterwards are erroneous).
1.17      anton    1708: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2       anton    1709: @subsubsection Programming Style
                   1711: The freedom to define locals anywhere has the potential to change
                   1712: programming styles dramatically. In particular, the need to use the
                   1713: return stack for intermediate storage vanishes. Moreover, all stack
                   1714: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
                   1715: determined arguments) can be eliminated: If the stack items are in the
                   1716: wrong order, just write a locals definition for all of them; then
                   1717: write the items in the order you want.
                   1719: This seems a little far-fetched and eliminating stack manipulations is
1.4       anton    1720: unlikely to become a conscious programming objective. Still, the number
                   1721: of stack manipulations will be reduced dramatically if local variables
1.17      anton    1722: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4       anton    1723: a traditional implementation of @code{max}).
1.2       anton    1724: 
                   1725: This shows one potential benefit of locals: making Forth programs more
                   1726: readable. Of course, this benefit will only be realized if the
                   1727: programmers continue to honour the principle of factoring instead of
                   1728: using the added latitude to make the words longer.
                   1730: Using @code{TO} can and should be avoided.  Without @code{TO},
                   1731: every value-flavoured local has only a single assignment and many
                   1732: advantages of functional languages apply to Forth. I.e., programs are
                   1733: easier to analyse, to optimize and to read: It is clear from the
                   1734: definition what the local stands for, it does not turn into something
                   1735: different later.
                   1737: E.g., a definition using @code{TO} might look like this:
                   1738: @example
                   1739: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1740:  u1 u2 min 0
                   1741:  ?do
1.36    ! anton    1742:    addr1 c@@ addr2 c@@ -
1.31      anton    1743:    ?dup-if
1.2       anton    1744:      unloop exit
                   1745:    then
                   1746:    addr1 char+ TO addr1
                   1747:    addr2 char+ TO addr2
                   1748:  loop
                   1749:  u1 u2 - ;
                   1750: @end example
                   1751: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
                   1752: every loop iteration. @code{strcmp} is a typical example of the
                   1753: readability problems of using @code{TO}. When you start reading
                   1754: @code{strcmp}, you think that @code{addr1} refers to the start of the
                   1755: string. Only near the end of the loop you realize that it is something
                   1756: else.
                   1758: This can be avoided by defining two locals at the start of the loop that
                   1759: are initialized with the right value for the current iteration.
                   1760: @example
                   1761: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1762:  addr1 addr2
                   1763:  u1 u2 min 0 
                   1764:  ?do @{ s1 s2 @}
1.36    ! anton    1765:    s1 c@@ s2 c@@ -
1.31      anton    1766:    ?dup-if
1.2       anton    1767:      unloop exit
                   1768:    then
                   1769:    s1 char+ s2 char+
                   1770:  loop
                   1771:  2drop
                   1772:  u1 u2 - ;
                   1773: @end example
                   1774: Here it is clear from the start that @code{s1} has a different value
                   1775: in every loop iteration.
1.17      anton    1777: @node Implementation,  , Programming Style, Gforth locals
1.2       anton    1778: @subsubsection Implementation
1.17      anton    1780: Gforth uses an extra locals stack. The most compelling reason for
1.2       anton    1781: this is that the return stack is not float-aligned; using an extra stack
                   1782: also eliminates the problems and restrictions of using the return stack
                   1783: as locals stack. Like the other stacks, the locals stack grows toward
                   1784: lower addresses. A few primitives allow an efficient implementation:
                   1786: doc-@local#
                   1787: doc-f@local#
                   1788: doc-laddr#
                   1789: doc-lp+!#
                   1790: doc-lp!
                   1791: doc->l
                   1792: doc-f>l
                   1794: In addition to these primitives, some specializations of these
                   1795: primitives for commonly occurring inline arguments are provided for
                   1796: efficiency reasons, e.g., @code{@@local0} as specialization of
                   1797: @code{@@local#} for the inline argument 0. The following compiling words
                   1798: compile the right specialized version, or the general version, as
                   1799: appropriate:
1.12      anton    1801: doc-compile-@local
                   1802: doc-compile-f@local
1.2       anton    1803: doc-compile-lp+!
                   1805: Combinations of conditional branches and @code{lp+!#} like
                   1806: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
                   1807: is taken) are provided for efficiency and correctness in loops.
                   1809: A special area in the dictionary space is reserved for keeping the
                   1810: local variable names. @code{@{} switches the dictionary pointer to this
                   1811: area and @code{@}} switches it back and generates the locals
                   1812: initializing code. @code{W:} etc.@ are normal defining words. This
                   1813: special area is cleared at the start of every colon definition.
1.17      anton    1815: A special feature of Gforth's dictionary is used to implement the
1.2       anton    1816: definition of locals without type specifiers: every wordlist (aka
                   1817: vocabulary) has its own methods for searching
1.4       anton    1818: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2       anton    1819: with a special search method: When it is searched for a word, it
                   1820: actually creates that word using @code{W:}. @code{@{} changes the search
                   1821: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
                   1822: and then the wordlist for defining locals without type specifiers.
                   1824: The lifetime rules support a stack discipline within a colon
                   1825: definition: The lifetime of a local is either nested with other locals
                   1826: lifetimes or it does not overlap them.
                   1828: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
                   1829: pointer manipulation is generated. Between control structure words
                   1830: locals definitions can push locals onto the locals stack. @code{AGAIN}
                   1831: is the simplest of the other three control flow words. It has to
                   1832: restore the locals stack depth of the corresponding @code{BEGIN}
                   1833: before branching. The code looks like this:
                   1834: @format
                   1835: @code{lp+!#} current-locals-size @minus{} dest-locals-size
                   1836: @code{branch} <begin>
                   1837: @end format
                   1839: @code{UNTIL} is a little more complicated: If it branches back, it
                   1840: must adjust the stack just like @code{AGAIN}. But if it falls through,
                   1841: the locals stack must not be changed. The compiler generates the
                   1842: following code:
                   1843: @format
                   1844: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
                   1845: @end format
                   1846: The locals stack pointer is only adjusted if the branch is taken.
                   1848: @code{THEN} can produce somewhat inefficient code:
                   1849: @format
                   1850: @code{lp+!#} current-locals-size @minus{} orig-locals-size
                   1851: <orig target>:
                   1852: @code{lp+!#} orig-locals-size @minus{} new-locals-size
                   1853: @end format
                   1854: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4       anton    1855: level at the @var{orig} point to the level after the @code{THEN}. The
1.2       anton    1856: first @code{lp+!#} adjusts the locals stack pointer from the current
                   1857: level to the level at the orig point, so the complete effect is an
                   1858: adjustment from the current level to the right level after the
                   1859: @code{THEN}.
                   1861: In a conventional Forth implementation a dest control-flow stack entry
                   1862: is just the target address and an orig entry is just the address to be
                   1863: patched. Our locals implementation adds a wordlist to every orig or dest
                   1864: item. It is the list of locals visible (or assumed visible) at the point
                   1865: described by the entry. Our implementation also adds a tag to identify
                   1866: the kind of entry, in particular to differentiate between live and dead
                   1867: (reachable and unreachable) orig entries.
                   1869: A few unusual operations have to be performed on locals wordlists:
                   1871: doc-common-list
                   1872: doc-sub-list?
                   1873: doc-list-size
                   1875: Several features of our locals wordlist implementation make these
                   1876: operations easy to implement: The locals wordlists are organised as
                   1877: linked lists; the tails of these lists are shared, if the lists
                   1878: contain some of the same locals; and the address of a name is greater
                   1879: than the address of the names behind it in the list.
                   1881: Another important implementation detail is the variable
                   1882: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
                   1883: determine if they can be reached directly or only through the branch
                   1884: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
                   1885: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
                   1886: definition, by @code{BEGIN} and usually by @code{THEN}.
                   1888: Counted loops are similar to other loops in most respects, but
                   1889: @code{LEAVE} requires special attention: It performs basically the same
                   1890: service as @code{AHEAD}, but it does not create a control-flow stack
                   1891: entry. Therefore the information has to be stored elsewhere;
                   1892: traditionally, the information was stored in the target fields of the
                   1893: branches created by the @code{LEAVE}s, by organizing these fields into a
                   1894: linked list. Unfortunately, this clever trick does not provide enough
                   1895: space for storing our extended control flow information. Therefore, we
                   1896: introduce another stack, the leave stack. It contains the control-flow
                   1897: stack entries for all unresolved @code{LEAVE}s.
                   1899: Local names are kept until the end of the colon definition, even if
                   1900: they are no longer visible in any control-flow path. In a few cases
                   1901: this may lead to increased space needs for the locals name area, but
                   1902: usually less than reclaiming this space would cost in code size.
1.17      anton    1905: @node ANS Forth locals,  , Gforth locals, Locals
1.2       anton    1906: @subsection ANS Forth locals
                   1908: The ANS Forth locals wordset does not define a syntax for locals, but
                   1909: words that make it possible to define various syntaxes. One of the
1.17      anton    1910: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2       anton    1911: wordset, i.e.:
                   1913: @example
                   1914: @{ local1 local2 ... -- comment @}
                   1915: @end example
                   1916: or
                   1917: @example
                   1918: @{ local1 local2 ... @}
                   1919: @end example
                   1921: The order of the locals corresponds to the order in a stack comment. The
                   1922: restrictions are:
1.1       anton    1923: 
1.2       anton    1924: @itemize @bullet
                   1925: @item
1.17      anton    1926: Locals can only be cell-sized values (no type specifiers are allowed).
1.2       anton    1927: @item
                   1928: Locals can be defined only outside control structures.
                   1929: @item
                   1930: Locals can interfere with explicit usage of the return stack. For the
                   1931: exact (and long) rules, see the standard. If you don't use return stack
1.17      anton    1932: accessing words in a definition using locals, you will be all right. The
1.2       anton    1933: purpose of this rule is to make locals implementation on the return
                   1934: stack easier.
                   1935: @item
                   1936: The whole definition must be in one line.
                   1937: @end itemize
1.35      anton    1939: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
                   1940: Defining Words}). I.e., they are initialized from the stack. Using their
1.2       anton    1941: name produces their value. Their value can be changed using @code{TO}.
1.17      anton    1943: Since this syntax is supported by Gforth directly, you need not do
1.2       anton    1944: anything to use it. If you want to port a program using this syntax to
1.30      anton    1945: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
                   1946: syntax on the other system.
1.2       anton    1947: 
                   1948: Note that a syntax shown in the standard, section A.13 looks
                   1949: similar, but is quite different in having the order of locals
                   1950: reversed. Beware!
                   1952: The ANS Forth locals wordset itself consists of the following word
                   1954: doc-(local)
                   1956: The ANS Forth locals extension wordset defines a syntax, but it is so
                   1957: awful that we strongly recommend not to use it. We have implemented this
1.17      anton    1958: syntax to make porting to Gforth easy, but do not document it here. The
1.2       anton    1959: problem with this syntax is that the locals are defined in an order
                   1960: reversed with respect to the standard stack comment notation, making
                   1961: programs harder to read, and easier to misread and miswrite. The only
                   1962: merit of this syntax is that it is easy to implement using the ANS Forth
                   1963: locals wordset.
1.3       anton    1964: 
1.4       anton    1965: @node Defining Words, Wordlists, Locals, Words
                   1966: @section Defining Words
1.14      anton    1968: @menu
1.35      anton    1969: * Simple Defining Words::       
                   1970: * Colon Definitions::           
                   1971: * User-defined Defining Words::  
                   1972: * Supplying names::             
                   1973: * Interpretation and Compilation Semantics::  
1.14      anton    1974: @end menu
1.35      anton    1976: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
                   1977: @subsection Simple Defining Words
                   1979: doc-constant
                   1980: doc-2constant
                   1981: doc-fconstant
                   1982: doc-variable
                   1983: doc-2variable
                   1984: doc-fvariable
                   1985: doc-create
                   1986: doc-user
                   1987: doc-value
                   1988: doc-to
                   1989: doc-defer
                   1990: doc-is
                   1992: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
                   1993: @subsection Colon Definitions
                   1995: @example
                   1996: : name ( ... -- ... )
                   1997:     word1 word2 word3 ;
                   1998: @end example
                   2000: creates a word called @code{name}, that, upon execution, executes
                   2001: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
                   2003: The explanation above is somewhat superficial. @xref{Interpretation and
                   2004: Compilation Semantics} for an in-depth discussion of some of the issues
                   2005: involved.
                   2007: doc-:
                   2008: doc-;
                   2010: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
                   2011: @subsection User-defined Defining Words
                   2013: You can create new defining words simply by wrapping defining-time code
                   2014: around existing defining words and putting the sequence in a colon
                   2015: definition.
1.36    ! anton    2017: If you want the words defined with your defining words to behave
        !          2018: differently from words defined with standard defining words, you can
1.35      anton    2019: write your defining word like this:
                   2021: @example
                   2022: : def-word ( "name" -- )
                   2023:     Create @var{code1}
                   2024: DOES> ( ... -- ... )
                   2025:     @var{code2} ;
                   2027: def-word name
                   2028: @end example
                   2030: Technically, this fragment defines a defining word @code{def-word}, and
                   2031: a word @code{name}; when you execute @code{name}, the address of the
                   2032: body of @code{name} is put on the data stack and @var{code2} is executed
                   2033: (the address of the body of @code{name} is the address @code{HERE}
1.36    ! anton    2034: returns immediately after the @code{CREATE}).
        !          2035: 
        !          2036: In other words, if you make the following definitions:
        !          2037: 
        !          2038: @example
        !          2039: : def-word1 ( "name" -- )
        !          2040:     Create @var{code1} ;
        !          2041: 
        !          2042: : action1 ( ... -- ... )
        !          2043:     @var{code2} ;
        !          2044: 
        !          2045: def-word name1
        !          2046: @end example
        !          2047: 
        !          2048: Using @code{name1 action1} is equivalent to using @code{name}.
        !          2049: 
        !          2050: E.g., you can implement @code{Constant} in this way:
1.35      anton    2051: 
                   2052: @example
                   2053: : constant ( w "name" -- )
                   2054:     create ,
                   2055: DOES> ( -- w )
1.36    ! anton    2056:     @@ ;
1.35      anton    2057: @end example
                   2059: When you create a constant with @code{5 constant five}, first a new word
                   2060: @code{five} is created, then the value 5 is laid down in the body of
                   2061: @code{five} with @code{,}. When @code{five} is invoked, the address of
                   2062: the body is put on the stack, and @code{@@} retrieves the value 5.
                   2064: In the example above the stack comment after the @code{DOES>} specifies
                   2065: the stack effect of the defined words, not the stack effect of the
                   2066: following code (the following code expects the address of the body on
                   2067: the top of stack, which is not reflected in the stack comment). This is
                   2068: the convention that I use and recommend (it clashes a bit with using
                   2069: locals declarations for stack effect specification, though).
                   2071: @subsubsection Applications of @code{CREATE..DOES>}
1.36    ! anton    2073: You may wonder how to use this feature. Here are some usage patterns:
1.35      anton    2074: 
                   2075: When you see a sequence of code occurring several times, and you can
                   2076: identify a meaning, you will factor it out as a colon definition. When
                   2077: you see similar colon definitions, you can factor them using
                   2078: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
                   2079: that look very similar:
                   2080: @example
                   2081: : ori, ( reg-taget reg-source n -- )
                   2082:     0 asm-reg-reg-imm ;
                   2083: : andi, ( reg-taget reg-source n -- )
                   2084:     1 asm-reg-reg-imm ;
                   2085: @end example
                   2087: This could be factored with:
                   2088: @example
                   2089: : reg-reg-imm ( op-code -- )
                   2090:     create ,
                   2091: DOES> ( reg-taget reg-source n -- )
1.36    ! anton    2092:     @@ asm-reg-reg-imm ;
1.35      anton    2093: 
                   2094: 0 reg-reg-imm ori,
                   2095: 1 reg-reg-imm andi,
                   2096: @end example
                   2098: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
                   2099: supply a part of the parameters for a word (known as @dfn{currying} in
                   2100: the functional language community). E.g., @code{+} needs two
                   2101: parameters. Creating versions of @code{+} with one parameter fixed can
                   2102: be done like this:
                   2103: @example
                   2104: : curry+ ( n1 -- )
                   2105:     create ,
                   2106: DOES> ( n2 -- n1+n2 )
1.36    ! anton    2107:     @@ + ;
1.35      anton    2108: 
                   2109:  3 curry+ 3+
                   2110: -2 curry+ 2-
                   2111: @end example
                   2113: @subsubsection The gory details of @code{CREATE..DOES>}
                   2115: doc-does>
                   2117: This means that you need not use @code{CREATE} and @code{DOES>} in the
                   2118: same definition; E.g., you can put the @code{DOES>}-part in a separate
                   2119: definition. This allows us to, e.g., select among different DOES>-parts:
                   2120: @example
                   2121: : does1 
                   2122: DOES> ( ... -- ... )
                   2123:     ... ;
                   2125: : does2
                   2126: DOES> ( ... -- ... )
                   2127:     ... ;
                   2129: : def-word ( ... -- ... )
                   2130:     create ...
                   2131:     IF
                   2132:        does1
                   2133:     ELSE
                   2134:        does2
                   2135:     ENDIF ;
                   2136: @end example
                   2138: In a standard program you can apply a @code{DOES>}-part only if the last
                   2139: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
                   2140: will override the behaviour of the last word defined in any case. In a
                   2141: standard program, you can use @code{DOES>} only in a colon
                   2142: definition. In Gforth, you can also use it in interpretation state, in a
                   2143: kind of one-shot mode:
                   2144: @example
                   2145: CREATE name ( ... -- ... )
                   2146:   @var{initialization}
                   2147: DOES>
                   2148:   @var{code} ;
                   2149: @end example
                   2150: This is equivalwent to the standard
                   2151: @example
                   2152: :noname
                   2153: DOES>
                   2154:     @var{code} ;
                   2155: CREATE name EXECUTE ( ... -- ... )
                   2156:     @var{initialization}
                   2157: @end example
                   2159: You can get the address of the body of a word with
                   2161: doc->body
                   2163: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
                   2164: @subsection Supplying names for the defined words
                   2166: By default, defining words take the names for the defined words from the
                   2167: input stream. Sometimes you want to supply the name from a string. You
                   2168: can do this with
                   2170: doc-nextname
                   2172: E.g.,
                   2174: @example
                   2175: s" foo" nextname create
                   2176: @end example
                   2177: is equivalent to
                   2178: @example
                   2179: create foo
                   2180: @end example
                   2182: Sometimes you want to define a word without a name. You can do this with
                   2184: doc-noname
                   2186: To make any use of the newly defined word, you need its execution
                   2187: token. You can get it with
                   2189: doc-lastxt
                   2191: E.g., you can initialize a deferred word with an anonymous colon
                   2192: definition:
                   2193: @example
                   2194: Defer deferred
                   2195: noname : ( ... -- ... )
                   2196:   ... ;
                   2197: lastxt IS deferred
                   2198: @end example
                   2200: @code{lastxt} also works when the last word was not defined as
                   2201: @code{noname}. 
                   2203: The standard has also recognized the need for anonymous words and
                   2204: provides
                   2206: doc-:noname
                   2208: This leaves the execution token for the word on the stack after the
                   2209: closing @code{;}. You can rewrite the last example with @code{:noname}:
                   2210: @example
                   2211: Defer deferred
                   2212: :noname ( ... -- ... )
                   2213:   ... ;
                   2214: IS deferred
                   2215: @end example
                   2217: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
                   2218: @subsection Interpretation and Compilation Semantics
1.36    ! anton    2220: The @dfn{interpretation semantics} of a word are what the text
        !          2221: interpreter does when it encounters the word in interpret state. It also
        !          2222: appears in some other contexts, e.g., the execution token returned by
        !          2223: @code{' @var{word}} identifies the interpretation semantics of
        !          2224: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
        !          2225: interpret-state text interpretation of @code{@var{word}}).
        !          2226: 
        !          2227: The @dfn{compilation semantics} of a word are what the text interpreter
        !          2228: does when it encounters the word in compile state. It also appears in
        !          2229: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
        !          2230: standard terminology, ``appends to the current definition''.} the
        !          2231: compilation semantics of @var{word}.
        !          2232: 
        !          2233: The standard also talks about @dfn{execution semantics}. They are used
        !          2234: only for defining the interpretation and compilation semantics of many
        !          2235: words. By default, the interpretation semantics of a word are to
        !          2236: @code{execute} its execution semantics, and the compilation semantics of
        !          2237: a word are to @code{compile,} its execution semantics.@footnote{In
        !          2238: standard terminology: The default interpretation semantics are its
        !          2239: execution semantics; the default compilation semantics are to append its
        !          2240: execution semantics to the execution semantics of the current
        !          2241: definition.}
        !          2242: 
        !          2243: You can change the compilation semantics into @code{execute}ing the
        !          2244: execution semantics with
        !          2245: 
1.35      anton    2246: doc-immediate
1.36    ! anton    2247: 
        !          2248: You can remove the interpretation semantics of a word with
        !          2249: 
        !          2250: doc-compile-only
        !          2251: doc-restrict
        !          2252: 
        !          2253: Note that ticking (@code{'}) compile-only words gives an error
        !          2254: (``Interpreting a compile-only word'').
        !          2255: 
        !          2256: Gforth also allows you to define words with arbitrary combinations of
        !          2257: interpretation and compilation semantics.
        !          2258: 
1.35      anton    2259: doc-interpret/compile:
1.36    ! anton    2261: This feature was introduced for implementing @code{TO} and @code{S"}. I
        !          2262: recommend that you do not define such words, as cute as they may be:
        !          2263: they make it hard to get at both parts of the word in some contexts.
        !          2264: E.g., assume you want to get an execution token for the compilation
        !          2265: part. Instead, define two words, one that embodies the interpretation
        !          2266: part, and one that embodies the compilation part.
        !          2267: 
        !          2268: There is, however, a potentially useful application of this feature:
        !          2269: Providing differing implementations for the default semantics. While
        !          2270: this introduces redundancy and is therefore usually a bad idea, a
        !          2271: performance improvement may be worth the trouble. E.g., consider the
        !          2272: word @code{foobar}:
        !          2273: 
        !          2274: @example
        !          2275: : foobar
        !          2276:     foo bar ;
        !          2277: @end example
        !          2278: 
        !          2279: Let us assume that @code{foobar} is called so frequently that the
        !          2280: calling overhead would take a significant amount of the run-time. We can
        !          2281: optimize it with @code{interpret/compile:}:
1.35      anton    2282: 
1.36    ! anton    2283: @example
        !          2284: :noname
        !          2285:    foo bar ;
        !          2286: :noname
        !          2287:    POSTPONE foo POSTPONE bar ;
        !          2288: interpret/compile: foobar
        !          2289: @end example
        !          2290: 
        !          2291: This definition has the same interpretation semantics and essentially
        !          2292: the same compilation semantics as the simple definition of
        !          2293: @code{foobar}, but the implementation of the compilation semantics is
        !          2294: more efficient with respect to run-time.
        !          2295: 
        !          2296: Some people try to use state-smart words to emulate the feature provided
        !          2297: by @code{interpret/compile:} (words are state-smart if they check
        !          2298: @code{STATE} during execution). E.g., they would try to code
        !          2299: @code{foobar} like this:
        !          2300: 
        !          2301: @example
        !          2302: : foobar
        !          2303:   STATE @@
        !          2304:   IF ( compilation state )
        !          2305:     POSTPONE foo POSTPONE bar
        !          2306:   ELSE
        !          2307:     foo bar
        !          2308:   ENDIF ; immediate
        !          2309: @end example
        !          2310: 
        !          2311: While this works if @code{foobar} is processed only by the text
        !          2312: interpreter, it does not work in other contexts (like @code{'} or
        !          2313: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
        !          2314: for a state-smart word, not for the interpretation semantics of the
        !          2315: original @code{foobar}; when you execute this execution token (directly
        !          2316: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
        !          2317: state, the result will not be what you expected (i.e., it will not
        !          2318: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
        !          2319: write them!
        !          2320: 
        !          2321: It is also possible to write defining words that define words with
        !          2322: arbitrary combinations of interpretation and compilation semantics (or,
        !          2323: preferably, arbitrary combinations of implementations of the default
        !          2324: semantics). In general, this looks like:
        !          2325: 
        !          2326: @example
        !          2327: : def-word
        !          2328:     create-interpret/compile
        !          2329:     @var{code1}
        !          2330: interpretation>
        !          2331:     @var{code2}
        !          2332: <interpretation
        !          2333: compilation>
        !          2334:     @var{code3}
        !          2335: <compilation ;
        !          2336: @end example
        !          2337: 
        !          2338: For a @var{word} defined with @code{def-word}, the interpretation
        !          2339: semantics are to push the address of the body of @var{word} and perform
        !          2340: @var{code2}, and the compilation semantics are to push the address of
        !          2341: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
        !          2342: can also be defined like this:
        !          2343: 
        !          2344: @example
        !          2345: : constant ( n "name" -- )
        !          2346:     create-interpret/compile
        !          2347:     ,
        !          2348: interpretation> ( -- n )
        !          2349:     @@
        !          2350: <interpretation
        !          2351: compilation> ( compilation. -- ; run-time. -- n )
        !          2352:     @@ postpone literal
        !          2353: <compilation ;
        !          2354: @end example
        !          2355: 
        !          2356: doc-create-interpret/compile
        !          2357: doc-interpretation>
        !          2358: doc-<interpretation
        !          2359: doc-compilation>
        !          2360: doc-<compilation
        !          2361: 
        !          2362: Note that words defined with @code{interpret/compile:} and
        !          2363: @code{create-interpret/compile} have an extended header structure that
        !          2364: differs from other words; however, unless you try to access them with
        !          2365: plain address arithmetic, you should not notice this. Words for
        !          2366: accessing the header structure usually know how to deal with this; e.g.,
        !          2367: @code{' word >body} also gives you the body of a word created with
        !          2368: @code{create-interpret/compile}.
1.4       anton    2369: 
                   2370: @node Wordlists, Files, Defining Words, Words
                   2371: @section Wordlists
                   2373: @node Files, Blocks, Wordlists, Words
                   2374: @section Files
                   2376: @node Blocks, Other I/O, Files, Words
                   2377: @section Blocks
                   2379: @node Other I/O, Programming Tools, Blocks, Words
                   2380: @section Other I/O
1.18      anton    2382: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4       anton    2383: @section Programming Tools
1.5       anton    2385: @menu
                   2386: * Debugging::                   Simple and quick.
                   2387: * Assertions::                  Making your programs self-checking.
                   2388: @end menu
                   2390: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4       anton    2391: @subsection Debugging
                   2393: The simple debugging aids provided in @file{debugging.fs}
                   2394: are meant to support a different style of debugging than the
                   2395: tracing/stepping debuggers used in languages with long turn-around
                   2396: times.
                   2398: A much better (faster) way in fast-compilig languages is to add
                   2399: printing code at well-selected places, let the program run, look at
                   2400: the output, see where things went wrong, add more printing code, etc.,
                   2401: until the bug is found.
                   2403: The word @code{~~} is easy to insert. It just prints debugging
                   2404: information (by default the source location and the stack contents). It
                   2405: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
                   2406: query-replace them with nothing). The deferred words
                   2407: @code{printdebugdata} and @code{printdebugline} control the output of
                   2408: @code{~~}. The default source location output format works well with
                   2409: Emacs' compilation mode, so you can step through the program at the
1.5       anton    2410: source level using @kbd{C-x `} (the advantage over a stepping debugger
                   2411: is that you can step in any direction and you know where the crash has
                   2412: happened or where the strange data has occurred).
1.4       anton    2413: 
                   2414: Note that the default actions clobber the contents of the pictured
                   2415: numeric output string, so you should not use @code{~~}, e.g., between
                   2416: @code{<#} and @code{#>}.
                   2418: doc-~~
                   2419: doc-printdebugdata
                   2420: doc-printdebugline
1.5       anton    2422: @node Assertions,  , Debugging, Programming Tools
1.4       anton    2423: @subsection Assertions
1.5       anton    2425: It is a good idea to make your programs self-checking, in particular, if
                   2426: you use an assumption (e.g., that a certain field of a data structure is
1.17      anton    2427: never zero) that may become wrong during maintenance. Gforth supports
1.5       anton    2428: assertions for this purpose. They are used like this:
                   2430: @example
                   2431: assert( @var{flag} )
                   2432: @end example
                   2434: The code between @code{assert(} and @code{)} should compute a flag, that
                   2435: should be true if everything is alright and false otherwise. It should
                   2436: not change anything else on the stack. The overall stack effect of the
                   2437: assertion is @code{( -- )}. E.g.
                   2439: @example
                   2440: assert( 1 1 + 2 = ) \ what we learn in school
                   2441: assert( dup 0<> ) \ assert that the top of stack is not zero
                   2442: assert( false ) \ this code should not be reached
                   2443: @end example
                   2445: The need for assertions is different at different times. During
                   2446: debugging, we want more checking, in production we sometimes care more
                   2447: for speed. Therefore, assertions can be turned off, i.e., the assertion
                   2448: becomes a comment. Depending on the importance of an assertion and the
                   2449: time it takes to check it, you may want to turn off some assertions and
1.17      anton    2450: keep others turned on. Gforth provides several levels of assertions for
1.5       anton    2451: this purpose:
                   2453: doc-assert0(
                   2454: doc-assert1(
                   2455: doc-assert2(
                   2456: doc-assert3(
                   2457: doc-assert(
                   2458: doc-)
                   2460: @code{Assert(} is the same as @code{assert1(}. The variable
                   2461: @code{assert-level} specifies the highest assertions that are turned
                   2462: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
                   2463: @code{assert1(} assertions perform checking, while @code{assert2(} and
                   2464: @code{assert3(} assertions are treated as comments.
                   2466: Note that the @code{assert-level} is evaluated at compile-time, not at
                   2467: run-time. I.e., you cannot turn assertions on or off at run-time, you
                   2468: have to set the @code{assert-level} appropriately before compiling a
                   2469: piece of code. You can compile several pieces of code at several
                   2470: @code{assert-level}s (e.g., a trusted library at level 1 and newly
                   2471: written code at level 3).
                   2473: doc-assert-level
                   2475: If an assertion fails, a message compatible with Emacs' compilation mode
                   2476: is produced and the execution is aborted (currently with @code{ABORT"}.
                   2477: If there is interest, we will introduce a special throw code. But if you
                   2478: intend to @code{catch} a specific condition, using @code{throw} is
                   2479: probably more appropriate than an assertion).
1.18      anton    2481: @node Assembler and Code words, Threading Words, Programming Tools, Words
                   2482: @section Assembler and Code words
                   2484: Gforth provides some words for defining primitives (words written in
                   2485: machine code), and for defining the the machine-code equivalent of
                   2486: @code{DOES>}-based defining words. However, the machine-independent
                   2487: nature of Gforth poses a few problems: First of all. Gforth runs on
                   2488: several architectures, so it can provide no standard assembler. What's
                   2489: worse is that the register allocation not only depends on the processor,
1.25      anton    2490: but also on the @code{gcc} version and options used.
1.18      anton    2491: 
1.25      anton    2492: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18      anton    2493: header structure), so a system-independent assembler may be used in
                   2494: Gforth. If you do not have an assembler, you can compile machine code
                   2495: directly with @code{,} and @code{c,}.
                   2497: doc-assembler
                   2498: doc-code
                   2499: doc-end-code
                   2500: doc-;code
                   2501: doc-flush-icache
                   2503: If @code{flush-icache} does not work correctly, @code{code} words
                   2504: etc. will not work (reliably), either.
                   2506: These words are rarely used. Therefore they reside in @code{code.fs},
                   2507: which is usually not loaded (except @code{flush-icache}, which is always
1.19      anton    2508: present). You can load them with @code{require code.fs}.
1.18      anton    2509: 
1.25      anton    2510: In the assembly code you will want to refer to the inner interpreter's
                   2511: registers (e.g., the data stack pointer) and you may want to use other
                   2512: registers for temporary storage. Unfortunately, the register allocation
                   2513: is installation-dependent.
                   2515: The easiest solution is to use explicit register declarations
                   2516: (@pxref{Explicit Reg Vars, , Variables in Specified Registers,,
                   2517: GNU C Manual}) for all of the inner interpreter's registers: You have to
                   2518: compile Gforth with @code{-DFORCE_REG} (configure option
                   2519: @code{--enable-force-reg}) and the appropriate declarations must be
                   2520: present in the @code{machine.h} file (see @code{mips.h} for an example;
                   2521: you can find a full list of all declarable register symbols with
                   2522: @code{grep register engine.c}). If you give explicit registers to all
                   2523: variables that are declared at the beginning of @code{engine()}, you
                   2524: should be able to use the other caller-saved registers for temporary
                   2525: storage. Alternatively, you can use the @code{gcc} option
                   2526: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
                   2527: Generation Conventions,, GNU C Manual}) to reserve a register
                   2528: (however, this restriction on register allocation may slow Gforth
                   2529: significantly).
                   2531: If this solution is not viable (e.g., because @code{gcc} does not allow
                   2532: you to explicitly declare all the registers you need), you have to find
                   2533: out by looking at the code where the inner interpreter's registers
                   2534: reside and which registers can be used for temporary storage. You can
                   2535: get an assembly listing of the engine's code with @code{make engine.s}.
                   2537: In any case, it is good practice to abstract your assembly code from the
                   2538: actual register allocation. E.g., if the data stack pointer resides in
                   2539: register @code{$17}, create an alias for this register called @code{sp},
                   2540: and use that in your assembly code.
1.18      anton    2542: Another option for implementing normal and defining words efficiently
                   2543: is: adding the wanted functionality to the source of Gforth. For normal
1.35      anton    2544: words you just have to edit @file{primitives} (@pxref{Automatic
                   2545: Generation}), defining words (equivalent to @code{;CODE} words, for fast
                   2546: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
                   2547: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.18      anton    2548: 
                   2550: @node Threading Words,  , Assembler and Code words, Words
1.4       anton    2551: @section Threading Words
                   2553: These words provide access to code addresses and other threading stuff
1.17      anton    2554: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4       anton    2555: abstracts away the differences between direct and indirect threading
                   2556: (and, for direct threading, the machine dependences). However, at
                   2557: present this wordset is still inclomplete. It is also pretty low-level;
                   2558: some day it will hopefully be made unnecessary by an internals words set
                   2559: that abstracts implementation details away completely.
                   2561: doc->code-address
                   2562: doc->does-code
                   2563: doc-code-address!
                   2564: doc-does-code!
                   2565: doc-does-handler!
                   2566: doc-/does-handler
1.18      anton    2568: The code addresses produced by various defining words are produced by
                   2569: the following words:
1.14      anton    2570: 
1.18      anton    2571: doc-docol:
                   2572: doc-docon:
                   2573: doc-dovar:
                   2574: doc-douser:
                   2575: doc-dodefer:
                   2576: doc-dofield:
1.35      anton    2578: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
                   2579: with @code{>DOES-CODE}. If the word was defined in that way, the value
                   2580: returned is different from 0 and identifies the @code{DOES>} used by the
                   2581: defining word.
1.14      anton    2582: 
1.4       anton    2583: @node ANS conformance, Model, Words, Top
                   2584: @chapter ANS conformance
1.17      anton    2586: To the best of our knowledge, Gforth is an
1.14      anton    2587: 
1.15      anton    2588: ANS Forth System
1.34      anton    2589: @itemize @bullet
1.15      anton    2590: @item providing the Core Extensions word set
                   2591: @item providing the Block word set
                   2592: @item providing the Block Extensions word set
                   2593: @item providing the Double-Number word set
                   2594: @item providing the Double-Number Extensions word set
                   2595: @item providing the Exception word set
                   2596: @item providing the Exception Extensions word set
                   2597: @item providing the Facility word set
                   2598: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
                   2599: @item providing the File Access word set
                   2600: @item providing the File Access Extensions word set
                   2601: @item providing the Floating-Point word set
                   2602: @item providing the Floating-Point Extensions word set
                   2603: @item providing the Locals word set
                   2604: @item providing the Locals Extensions word set
                   2605: @item providing the Memory-Allocation word set
                   2606: @item providing the Memory-Allocation Extensions word set (that one's easy)
                   2607: @item providing the Programming-Tools word set
1.34      anton    2608: @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    2609: @item providing the Search-Order word set
                   2610: @item providing the Search-Order Extensions word set
                   2611: @item providing the String word set
                   2612: @item providing the String Extensions word set (another easy one)
                   2613: @end itemize
                   2615: In addition, ANS Forth systems are required to document certain
                   2616: implementation choices. This chapter tries to meet these
                   2617: requirements. In many cases it gives a way to ask the system for the
                   2618: information instead of providing the information directly, in
                   2619: particular, if the information depends on the processor, the operating
                   2620: system or the installation options chosen, or if they are likely to
1.17      anton    2621: change during the maintenance of Gforth.
1.15      anton    2622: 
1.14      anton    2623: @comment The framework for the rest has been taken from pfe.
                   2625: @menu
                   2626: * The Core Words::              
                   2627: * The optional Block word set::  
                   2628: * The optional Double Number word set::  
                   2629: * The optional Exception word set::  
                   2630: * The optional Facility word set::  
                   2631: * The optional File-Access word set::  
                   2632: * The optional Floating-Point word set::  
                   2633: * The optional Locals word set::  
                   2634: * The optional Memory-Allocation word set::  
                   2635: * The optional Programming-Tools word set::  
                   2636: * The optional Search-Order word set::  
                   2637: @end menu
                   2640: @c =====================================================================
                   2641: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
                   2642: @comment  node-name,  next,  previous,  up
                   2643: @section The Core Words
                   2644: @c =====================================================================
                   2646: @menu
1.15      anton    2647: * core-idef::                   Implementation Defined Options                   
                   2648: * core-ambcond::                Ambiguous Conditions                
                   2649: * core-other::                  Other System Documentation                  
1.14      anton    2650: @end menu
                   2652: @c ---------------------------------------------------------------------
                   2653: @node core-idef, core-ambcond, The Core Words, The Core Words
                   2654: @subsection Implementation Defined Options
                   2655: @c ---------------------------------------------------------------------
                   2657: @table @i
                   2659: @item (Cell) aligned addresses:
1.17      anton    2660: processor-dependent. Gforth's alignment words perform natural alignment
1.14      anton    2661: (e.g., an address aligned for a datum of size 8 is divisible by
                   2662: 8). Unaligned accesses usually result in a @code{-23 THROW}.
                   2664: @item @code{EMIT} and non-graphic characters:
                   2665: The character is output using the C library function (actually, macro)
1.36    ! anton    2666: @code{putc}.
1.14      anton    2667: 
                   2668: @item character editing of @code{ACCEPT} and @code{EXPECT}:
                   2669: This is modeled on the GNU readline library (@pxref{Readline
                   2670: Interaction, , Command Line Editing, readline, The GNU Readline
                   2671: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
                   2672: producing a full word completion every time you type it (instead of
                   2673: producing the common prefix of all completions).
                   2675: @item character set:
                   2676: The character set of your computer and display device. Gforth is
                   2677: 8-bit-clean (but some other component in your system may make trouble).
                   2679: @item Character-aligned address requirements:
                   2680: installation-dependent. Currently a character is represented by a C
                   2681: @code{unsigned char}; in the future we might switch to @code{wchar_t}
                   2682: (Comments on that requested).
                   2684: @item character-set extensions and matching of names:
1.17      anton    2685: Any character except the ASCII NUL charcter can be used in a
1.36    ! anton    2686: name. Matching is case-insensitive (except in @code{TABLE}s. The
        !          2687: matching is performed using the C function @code{strncasecmp}, whose
        !          2688: function is probably influenced by the locale. E.g., the @code{C} locale
        !          2689: does not know about accents and umlauts, so they are matched
        !          2690: case-sensitively in that locale. For portability reasons it is best to
        !          2691: write programs such that they work in the @code{C} locale. Then one can
        !          2692: use libraries written by a Polish programmer (who might use words
        !          2693: containing ISO Latin-2 encoded characters) and by a French programmer
        !          2694: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
        !          2695: funny results for some of the words (which ones, depends on the font you
        !          2696: are using)). Also, the locale you prefer may not be available in other
        !          2697: operating systems. Hopefully, Unicode will solve these problems one day.
1.14      anton    2698: 
                   2699: @item conditions under which control characters match a space delimiter:
                   2700: If @code{WORD} is called with the space character as a delimiter, all
                   2701: white-space characters (as identified by the C macro @code{isspace()})
                   2702: are delimiters. @code{PARSE}, on the other hand, treats space like other
                   2703: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
                   2704: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
                   2705: interpreter (aka text interpreter) by default, treats all white-space
                   2706: characters as delimiters.
                   2708: @item format of the control flow stack:
                   2709: The data stack is used as control flow stack. The size of a control flow
                   2710: stack item in cells is given by the constant @code{cs-item-size}. At the
                   2711: time of this writing, an item consists of a (pointer to a) locals list
                   2712: (third), an address in the code (second), and a tag for identifying the
                   2713: item (TOS). The following tags are used: @code{defstart},
                   2714: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
                   2715: @code{scopestart}.
                   2717: @item conversion of digits > 35
                   2718: The characters @code{[\]^_'} are the digits with the decimal value
                   2719: 36@minus{}41. There is no way to input many of the larger digits.
                   2721: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
                   2722: The cursor is moved to the end of the entered string. If the input is
                   2723: terminated using the @kbd{Return} key, a space is typed.
                   2725: @item exception abort sequence of @code{ABORT"}:
                   2726: The error string is stored into the variable @code{"error} and a
                   2727: @code{-2 throw} is performed.
                   2729: @item input line terminator:
1.36    ! anton    2730: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
        !          2731: lines. One of these characters is typically produced when you type the
        !          2732: @kbd{Enter} or @kbd{Return} key.
1.14      anton    2733: 
                   2734: @item maximum size of a counted string:
                   2735: @code{s" /counted-string" environment? drop .}. Currently 255 characters
                   2736: on all ports, but this may change.
                   2738: @item maximum size of a parsed string:
                   2739: Given by the constant @code{/line}. Currently 255 characters.
                   2741: @item maximum size of a definition name, in characters:
                   2742: 31
                   2744: @item maximum string length for @code{ENVIRONMENT?}, in characters:
                   2745: 31
                   2747: @item method of selecting the user input device:
1.17      anton    2748: The user input device is the standard input. There is currently no way to
                   2749: change it from within Gforth. However, the input can typically be
                   2750: redirected in the command line that starts Gforth.
1.14      anton    2751: 
                   2752: @item method of selecting the user output device:
1.36    ! anton    2753: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
        !          2754: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
        !          2755: output, so output on a terminal does not become visible before the next
        !          2756: newline or buffer overflow. Output on non-terminals is invisible until
        !          2757: the buffer overflows.
1.14      anton    2758: 
                   2759: @item methods of dictionary compilation:
1.17      anton    2760: What are we expected to document here?
1.14      anton    2761: 
                   2762: @item number of bits in one address unit:
                   2763: @code{s" address-units-bits" environment? drop .}. 8 in all current
                   2764: ports.
                   2766: @item number representation and arithmetic:
                   2767: Processor-dependent. Binary two's complement on all current ports.
                   2769: @item ranges for integer types:
                   2770: Installation-dependent. Make environmental queries for @code{MAX-N},
                   2771: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
                   2772: unsigned (and positive) types is 0. The lower bound for signed types on
                   2773: two's complement and one's complement machines machines can be computed
                   2774: by adding 1 to the upper bound.
                   2776: @item read-only data space regions:
                   2777: The whole Forth data space is writable.
                   2779: @item size of buffer at @code{WORD}:
                   2780: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   2781: shared with the pictured numeric output string. If overwriting
                   2782: @code{PAD} is acceptable, it is as large as the remaining dictionary
                   2783: space, although only as much can be sensibly used as fits in a counted
                   2784: string.
                   2786: @item size of one cell in address units:
                   2787: @code{1 cells .}.
                   2789: @item size of one character in address units:
                   2790: @code{1 chars .}. 1 on all current ports.
                   2792: @item size of the keyboard terminal buffer:
1.36    ! anton    2793: Varies. You can determine the size at a specific time using @code{lp@@
1.14      anton    2794: tib - .}. It is shared with the locals stack and TIBs of files that
                   2795: include the current file. You can change the amount of space for TIBs
1.17      anton    2796: and locals stack at Gforth startup with the command line option
1.14      anton    2797: @code{-l}.
                   2799: @item size of the pictured numeric output buffer:
                   2800: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   2801: shared with @code{WORD}.
                   2803: @item size of the scratch area returned by @code{PAD}:
                   2804: The remainder of dictionary space. You can even use the unused part of
1.36    ! anton    2805: the data stack space. The current size can be computed with @code{sp@@
1.14      anton    2806: pad - .}.
                   2808: @item system case-sensitivity characteristics:
1.36    ! anton    2809: Dictionary searches are case insensitive (except in
        !          2810: @code{TABLE}s). However, as explained above under @i{character-set
        !          2811: extensions}, the matching for non-ASCII characters is determined by the
        !          2812: locale you are using. In the default @code{C} locale all non-ASCII
        !          2813: characters are matched case-sensitively.
1.14      anton    2814: 
                   2815: @item system prompt:
                   2816: @code{ ok} in interpret state, @code{ compiled} in compile state.
                   2818: @item division rounding:
                   2819: installation dependent. @code{s" floored" environment? drop .}. We leave
1.25      anton    2820: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
1.14      anton    2821: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
                   2823: @item values of @code{STATE} when true:
                   2824: -1.
                   2826: @item values returned after arithmetic overflow:
                   2827: On two's complement machines, arithmetic is performed modulo
                   2828: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   2829: arithmetic (with appropriate mapping for signed types). Division by zero
1.36    ! anton    2830: typically results in a @code{-55 throw} (Floating-point unidentified
1.14      anton    2831: fault), although a @code{-10 throw} (divide by zero) would be more
                   2832: appropriate.
                   2834: @item whether the current definition can be found after @t{DOES>}:
                   2835: No.
                   2837: @end table
                   2839: @c ---------------------------------------------------------------------
                   2840: @node core-ambcond, core-other, core-idef, The Core Words
                   2841: @subsection Ambiguous conditions
                   2842: @c ---------------------------------------------------------------------
                   2844: @table @i
                   2846: @item a name is neither a word nor a number:
1.36    ! anton    2847: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
        !          2848: preserves the data and FP stack, so you don't lose more work than
        !          2849: necessary.
1.14      anton    2850: 
                   2851: @item a definition name exceeds the maximum length allowed:
                   2852: @code{-19 throw} (Word name too long)
                   2854: @item addressing a region not inside the various data spaces of the forth system:
                   2855: The stacks, code space and name space are accessible. Machine code space is
                   2856: typically readable. Accessing other addresses gives results dependent on
                   2857: the operating system. On decent systems: @code{-9 throw} (Invalid memory
                   2858: address).
                   2860: @item argument type incompatible with parameter:
                   2861: This is usually not caught. Some words perform checks, e.g., the control
                   2862: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
                   2863: mismatch).
                   2865: @item attempting to obtain the execution token of a word with undefined execution semantics:
1.36    ! anton    2866: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
        !          2867: get an execution token for @code{compile-only-error} (which performs a
        !          2868: @code{-14 throw} when executed).
1.14      anton    2869: 
                   2870: @item dividing by zero:
                   2871: typically results in a @code{-55 throw} (floating point unidentified
                   2872: fault), although a @code{-10 throw} (divide by zero) would be more
                   2873: appropriate.
                   2875: @item insufficient data stack or return stack space:
                   2876: Not checked. This typically results in mysterious illegal memory
                   2877: accesses, producing @code{-9 throw} (Invalid memory address) or
                   2878: @code{-23 throw} (Address alignment exception).
                   2880: @item insufficient space for loop control parameters:
                   2881: like other return stack overflows.
                   2883: @item insufficient space in the dictionary:
                   2884: Not checked. Similar results as stack overflows. However, typically the
                   2885: error appears at a different place when one inserts or removes code.
                   2887: @item interpreting a word with undefined interpretation semantics:
                   2888: For some words, we defined interpretation semantics. For the others:
1.36    ! anton    2889: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    2890: 
                   2891: @item modifying the contents of the input buffer or a string literal:
                   2892: These are located in writable memory and can be modified.
                   2894: @item overflow of the pictured numeric output string:
                   2895: Not checked.
                   2897: @item parsed string overflow:
                   2898: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
                   2900: @item producing a result out of range:
                   2901: On two's complement machines, arithmetic is performed modulo
                   2902: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   2903: arithmetic (with appropriate mapping for signed types). Division by zero
                   2904: typically results in a @code{-55 throw} (floatingpoint unidentified
                   2905: fault), although a @code{-10 throw} (divide by zero) would be more
                   2906: appropriate. @code{convert} and @code{>number} currently overflow
                   2907: silently.
                   2909: @item reading from an empty data or return stack:
                   2910: The data stack is checked by the outer (aka text) interpreter after
                   2911: every word executed. If it has underflowed, a @code{-4 throw} (Stack
                   2912: underflow) is performed. Apart from that, the stacks are not checked and
                   2913: underflows can result in similar behaviour as overflows (of adjacent
                   2914: stacks).
1.36    ! anton    2916: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
1.14      anton    2917: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
                   2918: use zero-length string as a name). Words like @code{'} probably will not
                   2919: find what they search. Note that it is possible to create zero-length
                   2920: names with @code{nextname} (should it not?).
                   2922: @item @code{>IN} greater than input buffer:
                   2923: The next invocation of a parsing word returns a string wih length 0.
                   2925: @item @code{RECURSE} appears after @code{DOES>}:
1.36    ! anton    2926: Compiles a recursive call to the defining word, not to the defined word.
1.14      anton    2927: 
                   2928: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.27      anton    2929: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
                   2930: the end of the file was reached), its source-id may be
                   2931: reused. Therefore, restoring an input source specification referencing a
                   2932: closed file may lead to unpredictable results instead of a @code{-12
                   2933: THROW}.
1.36    ! anton    2935: In the future, Gforth may be able to restore input source specifications
1.27      anton    2936: from other than the current input soruce.
1.14      anton    2937: 
                   2938: @item data space containing definitions gets de-allocated:
                   2939: Deallocation with @code{allot} is not checked. This typically resuls in
                   2940: memory access faults or execution of illegal instructions.
                   2942: @item data space read/write with incorrect alignment:
                   2943: Processor-dependent. Typically results in a @code{-23 throw} (Address
                   2944: alignment exception). Under Linux on a 486 or later processor with
                   2945: alignment turned on, incorrect alignment results in a @code{-9 throw}
                   2946: (Invalid memory address). There are reportedly some processors with
                   2947: alignment restrictions that do not report them.
                   2949: @item data space pointer not properly aligned, @code{,}, @code{C,}:
                   2950: Like other alignment errors.
                   2952: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
                   2953: Not checked. May cause an illegal memory access.
                   2955: @item loop control parameters not available:
                   2956: Not checked. The counted loop words simply assume that the top of return
                   2957: stack items are loop control parameters and behave accordingly.
                   2959: @item most recent definition does not have a name (@code{IMMEDIATE}):
                   2960: @code{abort" last word was headerless"}.
                   2962: @item name not defined by @code{VALUE} used by @code{TO}:
1.36    ! anton    2963: @code{-32 throw} (Invalid name argument) (unless name was defined by
        !          2964: @code{CONSTANT}; then it just changes the constant).
1.14      anton    2965: 
1.15      anton    2966: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.14      anton    2967: @code{-13 throw} (Undefined word)
                   2969: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
                   2970: Gforth behaves as if they were of the same type. I.e., you can predict
                   2971: the behaviour by interpreting all parameters as, e.g., signed.
                   2973: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
1.36    ! anton    2974: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
        !          2975: compilation semantics of @code{TO}.
1.14      anton    2976: 
                   2977: @item String longer than a counted string returned by @code{WORD}:
                   2978: Not checked. The string will be ok, but the count will, of course,
                   2979: contain only the least significant bits of the length.
1.15      anton    2981: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.14      anton    2982: Processor-dependent. Typical behaviours are returning 0 and using only
                   2983: the low bits of the shift count.
                   2985: @item word not defined via @code{CREATE}:
                   2986: @code{>BODY} produces the PFA of the word no matter how it was defined.
                   2988: @code{DOES>} changes the execution semantics of the last defined word no
                   2989: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
                   2990: @code{CREATE , DOES>}.
                   2992: @item words improperly used outside @code{<#} and @code{#>}:
                   2993: Not checked. As usual, you can expect memory faults.
                   2995: @end table
                   2998: @c ---------------------------------------------------------------------
                   2999: @node core-other,  , core-ambcond, The Core Words
                   3000: @subsection Other system documentation
                   3001: @c ---------------------------------------------------------------------
                   3003: @table @i
                   3005: @item nonstandard words using @code{PAD}:
                   3006: None.
                   3008: @item operator's terminal facilities available:
1.26      anton    3009: After processing the command line, Gforth goes into interactive mode,
                   3010: and you can give commands to Gforth interactively. The actual facilities
                   3011: available depend on how you invoke Gforth.
1.14      anton    3012: 
                   3013: @item program data space available:
1.36    ! anton    3014: @code{sp@@ here - .} gives the space remaining for dictionary and data
1.14      anton    3015: stack together.
                   3017: @item return stack space available:
1.26      anton    3018: By default 16 KBytes. The default can be overridden with the @code{-r}
                   3019: switch (@pxref{Invocation}) when Gforth starts up.
1.14      anton    3020: 
                   3021: @item stack space available:
1.36    ! anton    3022: @code{sp@@ here - .} gives the space remaining for dictionary and data
1.14      anton    3023: stack together.
                   3025: @item system dictionary space required, in address units:
                   3026: Type @code{here forthstart - .} after startup. At the time of this
                   3027: writing, this gives 70108 (bytes) on a 32-bit system.
                   3028: @end table
                   3031: @c =====================================================================
                   3032: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
                   3033: @section The optional Block word set
                   3034: @c =====================================================================
                   3036: @menu
1.15      anton    3037: * block-idef::                  Implementation Defined Options                  
                   3038: * block-ambcond::               Ambiguous Conditions               
                   3039: * block-other::                 Other System Documentation                 
1.14      anton    3040: @end menu
                   3043: @c ---------------------------------------------------------------------
                   3044: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
                   3045: @subsection Implementation Defined Options
                   3046: @c ---------------------------------------------------------------------
                   3048: @table @i
                   3050: @item the format for display by @code{LIST}:
                   3051: First the screen number is displayed, then 16 lines of 64 characters,
                   3052: each line preceded by the line number.
                   3054: @item the length of a line affected by @code{\}:
                   3055: 64 characters.
                   3056: @end table
                   3059: @c ---------------------------------------------------------------------
                   3060: @node block-ambcond, block-other, block-idef, The optional Block word set
                   3061: @subsection Ambiguous conditions
                   3062: @c ---------------------------------------------------------------------
                   3064: @table @i
                   3066: @item correct block read was not possible:
                   3067: Typically results in a @code{throw} of some OS-derived value (between
                   3068: -512 and -2048). If the blocks file was just not long enough, blanks are
                   3069: supplied for the missing portion.
                   3071: @item I/O exception in block transfer:
                   3072: Typically results in a @code{throw} of some OS-derived value (between
                   3073: -512 and -2048).
                   3075: @item invalid block number:
                   3076: @code{-35 throw} (Invalid block number)
                   3078: @item a program directly alters the contents of @code{BLK}:
                   3079: The input stream is switched to that other block, at the same
                   3080: position. If the storing to @code{BLK} happens when interpreting
                   3081: non-block input, the system will get quite confused when the block ends.
                   3083: @item no current block buffer for @code{UPDATE}:
                   3084: @code{UPDATE} has no effect.
                   3086: @end table
                   3089: @c ---------------------------------------------------------------------
                   3090: @node block-other,  , block-ambcond, The optional Block word set
                   3091: @subsection Other system documentation
                   3092: @c ---------------------------------------------------------------------
                   3094: @table @i
                   3096: @item any restrictions a multiprogramming system places on the use of buffer addresses:
                   3097: No restrictions (yet).
                   3099: @item the number of blocks available for source and data:
                   3100: depends on your disk space.
                   3102: @end table
                   3105: @c =====================================================================
                   3106: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
                   3107: @section The optional Double Number word set
                   3108: @c =====================================================================
                   3110: @menu
1.15      anton    3111: * double-ambcond::              Ambiguous Conditions              
1.14      anton    3112: @end menu
                   3115: @c ---------------------------------------------------------------------
1.15      anton    3116: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
1.14      anton    3117: @subsection Ambiguous conditions
                   3118: @c ---------------------------------------------------------------------
                   3120: @table @i
1.15      anton    3122: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.14      anton    3123: The least significant cell of @var{d} is produced.
                   3125: @end table
                   3128: @c =====================================================================
                   3129: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
                   3130: @section The optional Exception word set
                   3131: @c =====================================================================
                   3133: @menu
1.15      anton    3134: * exception-idef::              Implementation Defined Options              
1.14      anton    3135: @end menu
                   3138: @c ---------------------------------------------------------------------
1.15      anton    3139: @node exception-idef,  , The optional Exception word set, The optional Exception word set
1.14      anton    3140: @subsection Implementation Defined Options
                   3141: @c ---------------------------------------------------------------------
                   3143: @table @i
                   3144: @item @code{THROW}-codes used in the system:
                   3145: The codes -256@minus{}-511 are used for reporting signals (see
                   3146: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
                   3147: (for file and memory allocation operations). The mapping from OS error
                   3148: numbers to throw code is -512@minus{}@var{errno}. One side effect of
                   3149: this mapping is that undefined OS errors produce a message with a
                   3150: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
                   3151: 488} on my system.
                   3152: @end table
                   3154: @c =====================================================================
                   3155: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
                   3156: @section The optional Facility word set
                   3157: @c =====================================================================
                   3159: @menu
1.15      anton    3160: * facility-idef::               Implementation Defined Options               
                   3161: * facility-ambcond::            Ambiguous Conditions            
1.14      anton    3162: @end menu
                   3165: @c ---------------------------------------------------------------------
                   3166: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
                   3167: @subsection Implementation Defined Options
                   3168: @c ---------------------------------------------------------------------
                   3170: @table @i
                   3172: @item encoding of keyboard events (@code{EKEY}):
                   3173: Not yet implemeted.
                   3175: @item duration of a system clock tick
                   3176: System dependent. With respect to @code{MS}, the time is specified in
                   3177: microseconds. How well the OS and the hardware implement this, is
                   3178: another question.
                   3180: @item repeatability to be expected from the execution of @code{MS}:
                   3181: System dependent. On Unix, a lot depends on load. If the system is
1.17      anton    3182: lightly loaded, and the delay is short enough that Gforth does not get
1.14      anton    3183: swapped out, the performance should be acceptable. Under MS-DOS and
                   3184: other single-tasking systems, it should be good.
                   3186: @end table
                   3189: @c ---------------------------------------------------------------------
1.15      anton    3190: @node facility-ambcond,  , facility-idef, The optional Facility word set
1.14      anton    3191: @subsection Ambiguous conditions
                   3192: @c ---------------------------------------------------------------------
                   3194: @table @i
                   3196: @item @code{AT-XY} can't be performed on user output device:
                   3197: Largely terminal dependant. No range checks are done on the arguments.
                   3198: No errors are reported. You may see some garbage appearing, you may see
                   3199: simply nothing happen.
                   3201: @end table
                   3204: @c =====================================================================
                   3205: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
                   3206: @section The optional File-Access word set
                   3207: @c =====================================================================
                   3209: @menu
1.15      anton    3210: * file-idef::                   Implementation Defined Options                   
                   3211: * file-ambcond::                Ambiguous Conditions                
1.14      anton    3212: @end menu
                   3215: @c ---------------------------------------------------------------------
                   3216: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
                   3217: @subsection Implementation Defined Options
                   3218: @c ---------------------------------------------------------------------
                   3220: @table @i
                   3222: @item File access methods used:
                   3223: @code{R/O}, @code{R/W} and @code{BIN} work as you would
                   3224: expect. @code{W/O} translates into the C file opening mode @code{w} (or
                   3225: @code{wb}): The file is cleared, if it exists, and created, if it does
1.15      anton    3226: not (both with @code{open-file} and @code{create-file}).  Under Unix
1.14      anton    3227: @code{create-file} creates a file with 666 permissions modified by your
                   3228: umask.
                   3230: @item file exceptions:
                   3231: The file words do not raise exceptions (except, perhaps, memory access
                   3232: faults when you pass illegal addresses or file-ids).
                   3234: @item file line terminator:
                   3235: System-dependent. Gforth uses C's newline character as line
                   3236: terminator. What the actual character code(s) of this are is
                   3237: system-dependent.
                   3239: @item file name format
                   3240: System dependent. Gforth just uses the file name format of your OS.
                   3242: @item information returned by @code{FILE-STATUS}:
                   3243: @code{FILE-STATUS} returns the most powerful file access mode allowed
                   3244: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
                   3245: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
                   3246: along with the retured mode.
                   3248: @item input file state after an exception when including source:
                   3249: All files that are left via the exception are closed.
                   3251: @item @var{ior} values and meaning:
1.15      anton    3252: The @var{ior}s returned by the file and memory allocation words are
                   3253: intended as throw codes. They typically are in the range
                   3254: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3255: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3256: 
                   3257: @item maximum depth of file input nesting:
                   3258: limited by the amount of return stack, locals/TIB stack, and the number
                   3259: of open files available. This should not give you troubles.
                   3261: @item maximum size of input line:
                   3262: @code{/line}. Currently 255.
                   3264: @item methods of mapping block ranges to files:
                   3265: Currently, the block words automatically access the file
                   3266: @file{blocks.fb} in the currend working directory. More sophisticated
                   3267: methods could be implemented if there is demand (and a volunteer).
                   3269: @item number of string buffers provided by @code{S"}:
                   3270: 1
                   3272: @item size of string buffer used by @code{S"}:
                   3273: @code{/line}. currently 255.
                   3275: @end table
                   3277: @c ---------------------------------------------------------------------
1.15      anton    3278: @node file-ambcond,  , file-idef, The optional File-Access word set
1.14      anton    3279: @subsection Ambiguous conditions
                   3280: @c ---------------------------------------------------------------------
                   3282: @table @i
                   3284: @item attempting to position a file outside it's boundaries:
                   3285: @code{REPOSITION-FILE} is performed as usual: Afterwards,
                   3286: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
                   3288: @item attempting to read from file positions not yet written:
                   3289: End-of-file, i.e., zero characters are read and no error is reported.
                   3291: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
                   3292: An appropriate exception may be thrown, but a memory fault or other
                   3293: problem is more probable.
                   3295: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
                   3296: The @var{ior} produced by the operation, that discovered the problem, is
                   3297: thrown.
                   3299: @item named file cannot be opened (@code{included}):
                   3300: The @var{ior} produced by @code{open-file} is thrown.
                   3302: @item requesting an unmapped block number:
                   3303: There are no unmapped legal block numbers. On some operating systems,
                   3304: writing a block with a large number may overflow the file system and
                   3305: have an error message as consequence.
                   3307: @item using @code{source-id} when @code{blk} is non-zero:
                   3308: @code{source-id} performs its function. Typically it will give the id of
                   3309: the source which loaded the block. (Better ideas?)
                   3311: @end table
                   3314: @c =====================================================================
                   3315: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15      anton    3316: @section The optional Floating-Point word set
1.14      anton    3317: @c =====================================================================
                   3319: @menu
1.15      anton    3320: * floating-idef::               Implementation Defined Options
                   3321: * floating-ambcond::            Ambiguous Conditions            
1.14      anton    3322: @end menu
                   3325: @c ---------------------------------------------------------------------
                   3326: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
                   3327: @subsection Implementation Defined Options
                   3328: @c ---------------------------------------------------------------------
                   3330: @table @i
1.15      anton    3332: @item format and range of floating point numbers:
                   3333: System-dependent; the @code{double} type of C.
1.14      anton    3334: 
1.15      anton    3335: @item results of @code{REPRESENT} when @var{float} is out of range:
                   3336: System dependent; @code{REPRESENT} is implemented using the C library
                   3337: function @code{ecvt()} and inherits its behaviour in this respect.
1.14      anton    3338: 
1.15      anton    3339: @item rounding or truncation of floating-point numbers:
1.26      anton    3340: System dependent; the rounding behaviour is inherited from the hosting C
                   3341: compiler. IEEE-FP-based (i.e., most) systems by default round to
                   3342: nearest, and break ties by rounding to even (i.e., such that the last
                   3343: bit of the mantissa is 0).
1.14      anton    3344: 
1.15      anton    3345: @item size of floating-point stack:
                   3346: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
                   3347: with the command-line option @code{-f}.
1.14      anton    3348: 
1.15      anton    3349: @item width of floating-point stack:
                   3350: @code{1 floats}.
1.14      anton    3351: 
                   3352: @end table
                   3355: @c ---------------------------------------------------------------------
1.15      anton    3356: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
                   3357: @subsection Ambiguous conditions
1.14      anton    3358: @c ---------------------------------------------------------------------
                   3360: @table @i
1.15      anton    3362: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
                   3363: System-dependent. Typically results in an alignment fault like other
                   3364: alignment violations.
1.14      anton    3365: 
1.15      anton    3366: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
                   3367: System-dependent. Typically results in an alignment fault like other
                   3368: alignment violations.
1.14      anton    3369: 
1.15      anton    3370: @item Floating-point result out of range:
                   3371: System-dependent. Can result in a @code{-55 THROW} (Floating-point
                   3372: unidentified fault), or can produce a special value representing, e.g.,
                   3373: Infinity.
1.14      anton    3374: 
1.15      anton    3375: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
                   3376: System-dependent. Typically results in an alignment fault like other
                   3377: alignment violations.
1.14      anton    3378: 
1.15      anton    3379: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
                   3380: The floating-point number is converted into decimal nonetheless.
1.14      anton    3381: 
1.15      anton    3382: @item Both arguments are equal to zero (@code{FATAN2}):
                   3383: System-dependent. @code{FATAN2} is implemented using the C library
                   3384: function @code{atan2()}.
1.14      anton    3385: 
1.15      anton    3386: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
                   3387: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
                   3388: because of small errors and the tan will be a very large (or very small)
                   3389: but finite number.
1.14      anton    3390: 
1.15      anton    3391: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
                   3392: The result is rounded to the nearest float.
1.14      anton    3393: 
1.15      anton    3394: @item dividing by zero:
                   3395: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3396: 
1.15      anton    3397: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
                   3398: System dependent. On IEEE-FP based systems the number is converted into
                   3399: an infinity.
1.14      anton    3400: 
1.15      anton    3401: @item @var{float}<1 (@code{facosh}):
                   3402: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3403: 
1.15      anton    3404: @item @var{float}=<-1 (@code{flnp1}):
                   3405: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3406: negative infinity is typically produced for @var{float}=-1.
1.14      anton    3407: 
1.15      anton    3408: @item @var{float}=<0 (@code{fln}, @code{flog}):
                   3409: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3410: negative infinity is typically produced for @var{float}=0.
1.14      anton    3411: 
1.15      anton    3412: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
                   3413: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
                   3414: produces values for these inputs on my Linux box (Bug in the C library?)
1.14      anton    3415: 
1.15      anton    3416: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
                   3417: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    3418: 
1.15      anton    3419: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
                   3420: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    3421: 
1.15      anton    3422: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
                   3423: This does not happen.
                   3424: @end table
1.14      anton    3425: 
                   3428: @c =====================================================================
1.15      anton    3429: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
                   3430: @section The optional Locals word set
1.14      anton    3431: @c =====================================================================
                   3433: @menu
1.15      anton    3434: * locals-idef::                 Implementation Defined Options                 
                   3435: * locals-ambcond::              Ambiguous Conditions              
1.14      anton    3436: @end menu
                   3439: @c ---------------------------------------------------------------------
1.15      anton    3440: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14      anton    3441: @subsection Implementation Defined Options
                   3442: @c ---------------------------------------------------------------------
                   3444: @table @i
1.15      anton    3446: @item maximum number of locals in a definition:
                   3447: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
                   3448: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
                   3449: characters. The number of locals in a definition is bounded by the size
                   3450: of locals-buffer, which contains the names of the locals.
1.14      anton    3451: 
                   3452: @end table
                   3455: @c ---------------------------------------------------------------------
1.15      anton    3456: @node locals-ambcond,  , locals-idef, The optional Locals word set
1.14      anton    3457: @subsection Ambiguous conditions
                   3458: @c ---------------------------------------------------------------------
                   3460: @table @i
1.15      anton    3462: @item executing a named local in interpretation state:
                   3463: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    3464: 
1.15      anton    3465: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
                   3466: @code{-32 throw} (Invalid name argument)
1.14      anton    3467: 
                   3468: @end table
                   3471: @c =====================================================================
1.15      anton    3472: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
                   3473: @section The optional Memory-Allocation word set
1.14      anton    3474: @c =====================================================================
                   3476: @menu
1.15      anton    3477: * memory-idef::                 Implementation Defined Options                 
1.14      anton    3478: @end menu
                   3481: @c ---------------------------------------------------------------------
1.15      anton    3482: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14      anton    3483: @subsection Implementation Defined Options
                   3484: @c ---------------------------------------------------------------------
                   3486: @table @i
1.15      anton    3488: @item values and meaning of @var{ior}:
                   3489: The @var{ior}s returned by the file and memory allocation words are
                   3490: intended as throw codes. They typically are in the range
                   3491: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3492: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3493: 
                   3494: @end table
                   3496: @c =====================================================================
1.15      anton    3497: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
                   3498: @section The optional Programming-Tools word set
1.14      anton    3499: @c =====================================================================
                   3501: @menu
1.15      anton    3502: * programming-idef::            Implementation Defined Options            
                   3503: * programming-ambcond::         Ambiguous Conditions         
1.14      anton    3504: @end menu
                   3507: @c ---------------------------------------------------------------------
1.15      anton    3508: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14      anton    3509: @subsection Implementation Defined Options
                   3510: @c ---------------------------------------------------------------------
                   3512: @table @i
1.15      anton    3514: @item ending sequence for input following @code{;code} and @code{code}:
                   3515: Not implemented (yet).
1.14      anton    3516: 
1.15      anton    3517: @item manner of processing input following @code{;code} and @code{code}:
                   3518: Not implemented (yet).
                   3520: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
                   3521: Not implemented (yet). If they were implemented, they would use the
                   3522: search order wordset.
                   3524: @item source and format of display by @code{SEE}:
                   3525: The source for @code{see} is the intermediate code used by the inner
                   3526: interpreter.  The current @code{see} tries to output Forth source code
                   3527: as well as possible.
1.14      anton    3529: @end table
                   3531: @c ---------------------------------------------------------------------
1.15      anton    3532: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
1.14      anton    3533: @subsection Ambiguous conditions
                   3534: @c ---------------------------------------------------------------------
                   3536: @table @i
1.15      anton    3538: @item deleting the compilation wordlist (@code{FORGET}):
                   3539: Not implemented (yet).
1.14      anton    3540: 
1.15      anton    3541: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
                   3542: This typically results in an @code{abort"} with a descriptive error
                   3543: message (may change into a @code{-22 throw} (Control structure mismatch)
                   3544: in the future). You may also get a memory access error. If you are
                   3545: unlucky, this ambiguous condition is not caught.
                   3547: @item @var{name} can't be found (@code{forget}):
                   3548: Not implemented (yet).
1.14      anton    3549: 
1.15      anton    3550: @item @var{name} not defined via @code{CREATE}:
                   3551: @code{;code} is not implemented (yet). If it were, it would behave like
                   3552: @code{DOES>} in this respect, i.e., change the execution semantics of
                   3553: the last defined word no matter how it was defined.
1.14      anton    3554: 
1.15      anton    3555: @item @code{POSTPONE} applied to @code{[IF]}:
                   3556: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
                   3557: equivalent to @code{[IF]}.
1.14      anton    3558: 
1.15      anton    3559: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
                   3560: Continue in the same state of conditional compilation in the next outer
                   3561: input source. Currently there is no warning to the user about this.
1.14      anton    3562: 
1.15      anton    3563: @item removing a needed definition (@code{FORGET}):
                   3564: Not implemented (yet).
1.14      anton    3565: 
                   3566: @end table
                   3569: @c =====================================================================
1.15      anton    3570: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
                   3571: @section The optional Search-Order word set
1.14      anton    3572: @c =====================================================================
                   3574: @menu
1.15      anton    3575: * search-idef::                 Implementation Defined Options                 
                   3576: * search-ambcond::              Ambiguous Conditions              
1.14      anton    3577: @end menu
                   3580: @c ---------------------------------------------------------------------
1.15      anton    3581: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14      anton    3582: @subsection Implementation Defined Options
                   3583: @c ---------------------------------------------------------------------
                   3585: @table @i
1.15      anton    3587: @item maximum number of word lists in search order:
                   3588: @code{s" wordlists" environment? drop .}. Currently 16.
                   3590: @item minimum search order:
                   3591: @code{root root}.
1.14      anton    3592: 
                   3593: @end table
                   3595: @c ---------------------------------------------------------------------
1.15      anton    3596: @node search-ambcond,  , search-idef, The optional Search-Order word set
1.14      anton    3597: @subsection Ambiguous conditions
                   3598: @c ---------------------------------------------------------------------
                   3600: @table @i
1.15      anton    3602: @item changing the compilation wordlist (during compilation):
1.33      anton    3603: The word is entered into the wordlist that was the compilation wordlist
                   3604: at the start of the definition. Any changes to the name field (e.g.,
                   3605: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
                   3606: are applied to the latest defined word (as reported by @code{last} or
                   3607: @code{lastxt}), if possible, irrespective of the compilation wordlist.
1.14      anton    3608: 
1.15      anton    3609: @item search order empty (@code{previous}):
                   3610: @code{abort" Vocstack empty"}.
1.14      anton    3611: 
1.15      anton    3612: @item too many word lists in search order (@code{also}):
                   3613: @code{abort" Vocstack full"}.
1.14      anton    3614: 
                   3615: @end table
1.13      anton    3616: 
1.34      anton    3617: @node Model, Integrating Gforth, ANS conformance, Top
                   3618: @chapter Model
                   3620: This chapter has yet to be written. It will contain information, on
                   3621: which internal structures you can rely.
                   3623: @node Integrating Gforth, Emacs and Gforth, Model, Top
                   3624: @chapter Integrating Gforth into C programs
                   3626: This is not yet implemented.
                   3628: Several people like to use Forth as scripting language for applications
                   3629: that are otherwise written in C, C++, or some other language.
                   3631: The Forth system ATLAST provides facilities for embedding it into
                   3632: applications; unfortunately it has several disadvantages: most
1.36    ! anton    3633: importantly, it is not based on ANS Forth, and it is apparently dead
1.34      anton    3634: (i.e., not developed further and not supported). The facilities
                   3635: provided by Gforth in this area are inspired by ATLASTs facilities, so
                   3636: making the switch should not be hard.
                   3638: We also tried to design the interface such that it can easily be
                   3639: implemented by other Forth systems, so that we may one day arrive at a
                   3640: standardized interface. Such a standard interface would allow you to
                   3641: replace the Forth system without having to rewrite C code.
                   3643: You embed the Gforth interpreter by linking with the library
                   3644: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
                   3645: global symbols in this library that belong to the interface, have the
                   3646: prefix @code{forth_}. (Global symbols that are used internally have the
                   3647: prefix @code{gforth_}).
                   3649: You can include the declarations of Forth types and the functions and
1.36    ! anton    3650: variables of the interface with @code{#include <forth.h>}.
1.34      anton    3651: 
                   3652: Types.
1.13      anton    3653: 
1.34      anton    3654: Variables.
                   3656: Data and FP Stack pointer. Area sizes.
                   3658: functions.
                   3660: forth_init(imagefile)
                   3661: forth_evaluate(string) exceptions?
                   3662: forth_goto(address) (or forth_execute(xt)?)
                   3663: forth_continue() (a corountining mechanism)
                   3665: Adding primitives.
                   3667: No checking.
                   3669: Signals?
                   3671: Accessing the Stacks
1.4       anton    3672: 
1.34      anton    3673: @node Emacs and Gforth, Internals, Integrating Gforth, Top
1.17      anton    3674: @chapter Emacs and Gforth
1.4       anton    3675: 
1.17      anton    3676: Gforth comes with @file{gforth.el}, an improved version of
1.33      anton    3677: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.4       anton    3678: improvements are a better (but still not perfect) handling of
                   3679: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8       anton    3680: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
                   3681: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
                   3682: stuff I do not use alone, even though some of it only makes sense for
                   3683: TILE. To get a description of these features, enter Forth mode and type
                   3684: @kbd{C-h m}.
1.4       anton    3685: 
1.17      anton    3686: In addition, Gforth supports Emacs quite well: The source code locations
1.4       anton    3687: given in error messages, debugging output (from @code{~~}) and failed
                   3688: assertion messages are in the right format for Emacs' compilation mode
                   3689: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
                   3690: Manual}) so the source location corresponding to an error or other
                   3691: message is only a few keystrokes away (@kbd{C-x `} for the next error,
                   3692: @kbd{C-c C-c} for the error under the cursor).
                   3694: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
                   3695: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
                   3696: contains the definitions of all words defined afterwards. You can then
                   3697: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17      anton    3698: several tags files at the same time (e.g., one for the Gforth sources
1.28      anton    3699: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
                   3700: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
                   3701: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
1.33      anton    3702: @file{/usr/local/share/gforth/0.2.0/TAGS}).
1.4       anton    3703: 
                   3704: To get all these benefits, add the following lines to your @file{.emacs}
                   3705: file:
                   3707: @example
                   3708: (autoload 'forth-mode "gforth.el")
                   3709: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
                   3710: @end example
1.17      anton    3712: @node Internals, Bugs, Emacs and Gforth, Top
1.3       anton    3713: @chapter Internals
1.17      anton    3715: Reading this section is not necessary for programming with Gforth. It
                   3716: should be helpful for finding your way in the Gforth sources.
1.3       anton    3717: 
1.24      anton    3718: The ideas in this section have also been published in the papers
                   3719: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
                   3720: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
                   3721: Ertl, presented at EuroForth '93; the latter is available at
                   3722: @*@file{}.
1.4       anton    3724: @menu
                   3725: * Portability::                 
                   3726: * Threading::                   
                   3727: * Primitives::                  
                   3728: * System Architecture::         
1.17      anton    3729: * Performance::                 
1.4       anton    3730: @end menu
                   3732: @node Portability, Threading, Internals, Internals
1.3       anton    3733: @section Portability
                   3735: One of the main goals of the effort is availability across a wide range
                   3736: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
                   3737: this goal by manually coding the engine in assembly language for several
                   3738: then-popular processors. This approach is very labor-intensive and the
                   3739: results are short-lived due to progress in computer architecture.
                   3741: Others have avoided this problem by coding in C, e.g., Mitch Bradley
                   3742: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
                   3743: particularly popular for UNIX-based Forths due to the large variety of
                   3744: architectures of UNIX machines. Unfortunately an implementation in C
                   3745: does not mix well with the goals of efficiency and with using
                   3746: traditional techniques: Indirect or direct threading cannot be expressed
                   3747: in C, and switch threading, the fastest technique available in C, is
                   3748: significantly slower. Another problem with C is that it's very
                   3749: cumbersome to express double integer arithmetic.
                   3751: Fortunately, there is a portable language that does not have these
                   3752: limitations: GNU C, the version of C processed by the GNU C compiler
                   3753: (@pxref{C Extensions, , Extensions to the C Language Family,,
                   3754: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
                   3755: Labels as Values,, GNU C Manual}) makes direct and indirect
                   3756: threading possible, its @code{long long} type (@pxref{Long Long, ,
1.33      anton    3757: Double-Word Integers,, GNU C Manual}) corresponds to Forth's
1.32      anton    3758: double numbers@footnote{Unfortunately, long longs are not implemented
                   3759: properly on all machines (e.g., on alpha-osf1, long longs are only 64
                   3760: bits, the same size as longs (and pointers), but they should be twice as
                   3761: long according to @ref{Long Long, , Double-Word Integers,, GNU
                   3762: C Manual}). So, we had to implement doubles in C after all. Still, on
                   3763: most machines we can use long longs and achieve better performance than
                   3764: with the emulation package.}. GNU C is available for free on all
                   3765: important (and many unimportant) UNIX machines, VMS, 80386s running
                   3766: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
                   3767: on all these machines.
1.3       anton    3768: 
                   3769: Writing in a portable language has the reputation of producing code that
                   3770: is slower than assembly. For our Forth engine we repeatedly looked at
                   3771: the code produced by the compiler and eliminated most compiler-induced
                   3772: inefficiencies by appropriate changes in the source-code.
                   3774: However, register allocation cannot be portably influenced by the
                   3775: programmer, leading to some inefficiencies on register-starved
                   3776: machines. We use explicit register declarations (@pxref{Explicit Reg
                   3777: Vars, , Variables in Specified Registers,, GNU C Manual}) to
                   3778: improve the speed on some machines. They are turned on by using the
                   3779: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
                   3780: only depends on the machine, but also on the compiler version: On some
                   3781: machines some compiler versions produce incorrect code when certain
                   3782: explicit register declarations are used. So by default
                   3783: @code{-DFORCE_REG} is not used.
1.4       anton    3785: @node Threading, Primitives, Portability, Internals
1.3       anton    3786: @section Threading
                   3788: GNU C's labels as values extension (available since @code{gcc-2.0},
                   3789: @pxref{Labels as Values, , Labels as Values,, GNU C Manual})
                   3790: makes it possible to take the address of @var{label} by writing
                   3791: @code{&&@var{label}}.  This address can then be used in a statement like
                   3792: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
                   3793: @code{goto x}.
                   3795: With this feature an indirect threaded NEXT looks like:
                   3796: @example
                   3797: cfa = *ip++;
                   3798: ca = *cfa;
                   3799: goto *ca;
                   3800: @end example
                   3801: For those unfamiliar with the names: @code{ip} is the Forth instruction
                   3802: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
                   3803: execution token and points to the code field of the next word to be
                   3804: executed; The @code{ca} (code address) fetched from there points to some
                   3805: executable code, e.g., a primitive or the colon definition handler
                   3806: @code{docol}.
                   3808: Direct threading is even simpler:
                   3809: @example
                   3810: ca = *ip++;
                   3811: goto *ca;
                   3812: @end example
                   3814: Of course we have packaged the whole thing neatly in macros called
                   3815: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
1.4       anton    3817: @menu
                   3818: * Scheduling::                  
                   3819: * Direct or Indirect Threaded?::  
                   3820: * DOES>::                       
                   3821: @end menu
                   3823: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3       anton    3824: @subsection Scheduling
                   3826: There is a little complication: Pipelined and superscalar processors,
                   3827: i.e., RISC and some modern CISC machines can process independent
                   3828: instructions while waiting for the results of an instruction. The
                   3829: compiler usually reorders (schedules) the instructions in a way that
                   3830: achieves good usage of these delay slots. However, on our first tries
                   3831: the compiler did not do well on scheduling primitives. E.g., for
                   3832: @code{+} implemented as
                   3833: @example
                   3834: n=sp[0]+sp[1];
                   3835: sp++;
                   3836: sp[0]=n;
                   3837: NEXT;
                   3838: @end example
                   3839: the NEXT comes strictly after the other code, i.e., there is nearly no
                   3840: scheduling. After a little thought the problem becomes clear: The
                   3841: compiler cannot know that sp and ip point to different addresses (and
1.4       anton    3842: the version of @code{gcc} we used would not know it even if it was
                   3843: possible), so it could not move the load of the cfa above the store to
                   3844: the TOS. Indeed the pointers could be the same, if code on or very near
                   3845: the top of stack were executed. In the interest of speed we chose to
                   3846: forbid this probably unused ``feature'' and helped the compiler in
                   3847: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
                   3848: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3       anton    3849: @example
                   3850: n=sp[0]+sp[1];
                   3851: sp++;
                   3852: NEXT_P1;
                   3853: sp[0]=n;
                   3854: NEXT_P2;
                   3855: @end example
1.4       anton    3856: This can be scheduled optimally by the compiler.
1.3       anton    3857: 
                   3858: This division can be turned off with the switch @code{-DCISC_NEXT}. This
                   3859: switch is on by default on machines that do not profit from scheduling
                   3860: (e.g., the 80386), in order to preserve registers.
1.4       anton    3862: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3       anton    3863: @subsection Direct or Indirect Threaded?
                   3865: Both! After packaging the nasty details in macro definitions we
                   3866: realized that we could switch between direct and indirect threading by
                   3867: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
                   3868: defining a few machine-specific macros for the direct-threading case.
                   3869: On the Forth level we also offer access words that hide the
                   3870: differences between the threading methods (@pxref{Threading Words}).
                   3872: Indirect threading is implemented completely
                   3873: machine-independently. Direct threading needs routines for creating
                   3874: jumps to the executable code (e.g. to docol or dodoes). These routines
                   3875: are inherently machine-dependent, but they do not amount to many source
                   3876: lines. I.e., even porting direct threading to a new machine is a small
                   3877: effort.
1.4       anton    3879: @node DOES>,  , Direct or Indirect Threaded?, Threading
1.3       anton    3880: @subsection DOES>
                   3881: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
                   3882: the chunk of code executed by every word defined by a
                   3883: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
                   3884: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
                   3885: DOES-code)? There are two solutions:
                   3887: In fig-Forth the code field points directly to the dodoes and the
                   3888: DOES-code address is stored in the cell after the code address
                   3889: (i.e. at cfa cell+). It may seem that this solution is illegal in the
                   3890: Forth-79 and all later standards, because in fig-Forth this address
                   3891: lies in the body (which is illegal in these standards). However, by
                   3892: making the code field larger for all words this solution becomes legal
                   3893: again. We use this approach for the indirect threaded version. Leaving
                   3894: a cell unused in most words is a bit wasteful, but on the machines we
                   3895: are targetting this is hardly a problem. The other reason for having a
                   3896: code field size of two cells is to avoid having different image files
1.4       anton    3897: for direct and indirect threaded systems (@pxref{System Architecture}).
1.3       anton    3898: 
                   3899: The other approach is that the code field points or jumps to the cell
                   3900: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
                   3901: this address. @code{dodoes} can then get the DOES-code address by
                   3902: computing the code address, i.e., the address of the jump to dodoes,
                   3903: and add the length of that jump field. A variant of this is to have a
                   3904: call to @code{dodoes} after the @code{DOES>}; then the return address
                   3905: (which can be found in the return register on RISCs) is the DOES-code
                   3906: address. Since the two cells available in the code field are usually
                   3907: used up by the jump to the code address in direct threading, we use
                   3908: this approach for direct threading. We did not want to add another
                   3909: cell to the code field.
1.4       anton    3911: @node Primitives, System Architecture, Threading, Internals
1.3       anton    3912: @section Primitives
1.4       anton    3914: @menu
                   3915: * Automatic Generation::        
                   3916: * TOS Optimization::            
                   3917: * Produced code::               
                   3918: @end menu
                   3920: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3       anton    3921: @subsection Automatic Generation
                   3923: Since the primitives are implemented in a portable language, there is no
                   3924: longer any need to minimize the number of primitives. On the contrary,
                   3925: having many primitives is an advantage: speed. In order to reduce the
                   3926: number of errors in primitives and to make programming them easier, we
                   3927: provide a tool, the primitive generator (@file{prims2x.fs}), that
                   3928: automatically generates most (and sometimes all) of the C code for a
                   3929: primitive from the stack effect notation.  The source for a primitive
                   3930: has the following form:
                   3932: @format
                   3933: @var{Forth-name}       @var{stack-effect}      @var{category}  [@var{pronounc.}]
                   3934: [@code{""}@var{glossary entry}@code{""}]
                   3935: @var{C code}
                   3936: [@code{:}
                   3937: @var{Forth code}]
                   3938: @end format
                   3940: The items in brackets are optional. The category and glossary fields
                   3941: are there for generating the documentation, the Forth code is there
                   3942: for manual implementations on machines without GNU C. E.g., the source
                   3943: for the primitive @code{+} is:
                   3944: @example
                   3945: +    n1 n2 -- n    core    plus
                   3946: n = n1+n2;
                   3947: @end example
                   3949: This looks like a specification, but in fact @code{n = n1+n2} is C
                   3950: code. Our primitive generation tool extracts a lot of information from
                   3951: the stack effect notations@footnote{We use a one-stack notation, even
                   3952: though we have separate data and floating-point stacks; The separate
                   3953: notation can be generated easily from the unified notation.}: The number
                   3954: of items popped from and pushed on the stack, their type, and by what
                   3955: name they are referred to in the C code. It then generates a C code
                   3956: prelude and postlude for each primitive. The final C code for @code{+}
                   3957: looks like this:
                   3959: @example
                   3960: I_plus:        /* + ( n1 n2 -- n ) */  /* label, stack effect */
                   3961: /*  */                          /* documentation */
1.4       anton    3962: @{
1.3       anton    3963: DEF_CA                          /* definition of variable ca (indirect threading) */
                   3964: Cell n1;                        /* definitions of variables */
                   3965: Cell n2;
                   3966: Cell n;
                   3967: n1 = (Cell) sp[1];              /* input */
                   3968: n2 = (Cell) TOS;
                   3969: sp += 1;                        /* stack adjustment */
                   3970: NAME("+")                       /* debugging output (with -DDEBUG) */
1.4       anton    3971: @{
1.3       anton    3972: n = n1+n2;                      /* C code taken from the source */
1.4       anton    3973: @}
1.3       anton    3974: NEXT_P1;                        /* NEXT part 1 */
                   3975: TOS = (Cell)n;                  /* output */
                   3976: NEXT_P2;                        /* NEXT part 2 */
1.4       anton    3977: @}
1.3       anton    3978: @end example
                   3980: This looks long and inefficient, but the GNU C compiler optimizes quite
                   3981: well and produces optimal code for @code{+} on, e.g., the R3000 and the
                   3982: HP RISC machines: Defining the @code{n}s does not produce any code, and
                   3983: using them as intermediate storage also adds no cost.
                   3985: There are also other optimizations, that are not illustrated by this
                   3986: example: Assignments between simple variables are usually for free (copy
                   3987: propagation). If one of the stack items is not used by the primitive
                   3988: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
                   3989: (dead code elimination). On the other hand, there are some things that
                   3990: the compiler does not do, therefore they are performed by
                   3991: @file{prims2x.fs}: The compiler does not optimize code away that stores
                   3992: a stack item to the place where it just came from (e.g., @code{over}).
                   3994: While programming a primitive is usually easy, there are a few cases
                   3995: where the programmer has to take the actions of the generator into
                   3996: account, most notably @code{?dup}, but also words that do not (always)
                   3997: fall through to NEXT.
1.4       anton    3999: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3       anton    4000: @subsection TOS Optimization
                   4002: An important optimization for stack machine emulators, e.g., Forth
                   4003: engines, is keeping  one or more of the top stack items in
1.4       anton    4004: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
                   4005: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.34      anton    4006: @itemize @bullet
1.3       anton    4007: @item
                   4008: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
                   4009: due to fewer loads from and stores to the stack.
                   4010: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
                   4011: @var{y<n}, due to additional moves between registers.
                   4012: @end itemize
                   4014: In particular, keeping one item in a register is never a disadvantage,
                   4015: if there are enough registers. Keeping two items in registers is a
                   4016: disadvantage for frequent words like @code{?branch}, constants,
                   4017: variables, literals and @code{i}. Therefore our generator only produces
                   4018: code that keeps zero or one items in registers. The generated C code
                   4019: covers both cases; the selection between these alternatives is made at
                   4020: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
                   4021: code for @code{+} is just a simple variable name in the one-item case,
                   4022: otherwise it is a macro that expands into @code{sp[0]}. Note that the
                   4023: GNU C compiler tries to keep simple variables like @code{TOS} in
                   4024: registers, and it usually succeeds, if there are enough registers.
                   4026: The primitive generator performs the TOS optimization for the
                   4027: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
                   4028: operations the benefit of this optimization is even larger:
                   4029: floating-point operations take quite long on most processors, but can be
                   4030: performed in parallel with other operations as long as their results are
                   4031: not used. If the FP-TOS is kept in a register, this works. If
                   4032: it is kept on the stack, i.e., in memory, the store into memory has to
                   4033: wait for the result of the floating-point operation, lengthening the
                   4034: execution time of the primitive considerably.
                   4036: The TOS optimization makes the automatic generation of primitives a
                   4037: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
                   4038: @code{TOS} is not sufficient. There are some special cases to
                   4039: consider:
1.34      anton    4040: @itemize @bullet
1.3       anton    4041: @item In the case of @code{dup ( w -- w w )} the generator must not
                   4042: eliminate the store to the original location of the item on the stack,
                   4043: if the TOS optimization is turned on.
1.4       anton    4044: @item Primitives with stack effects of the form @code{--}
                   4045: @var{out1}...@var{outy} must store the TOS to the stack at the start.
                   4046: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3       anton    4047: must load the TOS from the stack at the end. But for the null stack
                   4048: effect @code{--} no stores or loads should be generated.
                   4049: @end itemize
1.4       anton    4051: @node Produced code,  , TOS Optimization, Primitives
1.3       anton    4052: @subsection Produced code
                   4054: To see what assembly code is produced for the primitives on your machine
                   4055: with your compiler and your flag settings, type @code{make engine.s} and
1.4       anton    4056: look at the resulting file @file{engine.s}.
1.3       anton    4057: 
1.17      anton    4058: @node System Architecture, Performance, Primitives, Internals
1.3       anton    4059: @section System Architecture
                   4061: Our Forth system consists not only of primitives, but also of
                   4062: definitions written in Forth. Since the Forth compiler itself belongs
                   4063: to those definitions, it is not possible to start the system with the
                   4064: primitives and the Forth source alone. Therefore we provide the Forth
                   4065: code as an image file in nearly executable form. At the start of the
                   4066: system a C routine loads the image file into memory, sets up the
                   4067: memory (stacks etc.) according to information in the image file, and
                   4068: starts executing Forth code.
                   4070: The image file format is a compromise between the goals of making it
                   4071: easy to generate image files and making them portable. The easiest way
                   4072: to generate an image file is to just generate a memory dump. However,
                   4073: this kind of image file cannot be used on a different machine, or on
                   4074: the next version of the engine on the same machine, it even might not
                   4075: work with the same engine compiled by a different version of the C
                   4076: compiler. We would like to have as few versions of the image file as
                   4077: possible, because we do not want to distribute many versions of the
                   4078: same image file, and to make it easy for the users to use their image
                   4079: files on many machines. We currently need to create a different image
                   4080: file for machines with different cell sizes and different byte order
1.17      anton    4081: (little- or big-endian)@footnote{We are considering adding information to the
1.3       anton    4082: image file that enables the loader to change the byte order.}.
                   4084: Forth code that is going to end up in a portable image file has to
1.4       anton    4085: comply to some restrictions: addresses have to be stored in memory with
                   4086: special words (@code{A!}, @code{A,}, etc.) in order to make the code
                   4087: relocatable. Cells, floats, etc., have to be stored at the natural
                   4088: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
                   4089: dividable by~8. This happens automatically in our system when you use
                   4090: the ANS Forth alignment words.}, in order to avoid alignment faults on
                   4091: machines with stricter alignment. The image file is produced by a
                   4092: metacompiler (@file{cross.fs}).
1.3       anton    4093: 
                   4094: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
                   4095: file is not directly executable, but has to undergo some manipulations
                   4096: during loading. Address relocation is performed at image load-time, not
                   4097: at run-time. The loader also has to replace tokens standing for
                   4098: primitive calls with the appropriate code-field addresses (or code
                   4099: addresses in the case of direct threading).
1.4       anton    4100: 
1.17      anton    4101: @node  Performance,  , System Architecture, Internals
                   4102: @section Performance
                   4104: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
                   4105: impossible to write a significantly faster engine.
                   4107: On register-starved machines like the 386 architecture processors
                   4108: improvements are possible, because @code{gcc} does not utilize the
                   4109: registers as well as a human, even with explicit register declarations;
                   4110: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
                   4111: and hand-tuned it for the 486; this system is 1.19 times faster on the
                   4112: Sieve benchmark on a 486DX2/66 than Gforth compiled with
                   4113: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
                   4115: However, this potential advantage of assembly language implementations
                   4116: is not necessarily realized in complete Forth systems: We compared
1.26      anton    4117: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
                   4118: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
                   4119: 1994) and Eforth (with and without peephole (aka pinhole) optimization
                   4120: of the threaded code); all these systems were written in assembly
1.30      anton    4121: language. We also compared Gforth with three systems written in C:
1.32      anton    4122: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
                   4123: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
                   4124: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
                   4125: -fomit-frame-pointer; ThisForth employs peephole optimization of the
                   4126: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
                   4127: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
                   4128: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
                   4129: 486DX2/66 with similar memory performance under Windows NT. Marcel
                   4130: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
                   4131: added the peephole optimizer, ran the benchmarks and reported the
                   4132: results.
1.17      anton    4133:  
                   4134: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
                   4135: matrix multiplication come from the Stanford integer benchmarks and have
                   4136: been translated into Forth by Martin Fraeman; we used the versions
1.30      anton    4137: included in the TILE Forth package, but with bigger data set sizes; and
                   4138: a recursive Fibonacci number computation for benchmarking calling
                   4139: performance. The following table shows the time taken for the benchmarks
                   4140: scaled by the time taken by Gforth (in other words, it shows the speedup
                   4141: factor that Gforth achieved over the other systems).
1.17      anton    4142: 
                   4143: @example
1.30      anton    4144: relative      Win32-    NT       eforth       This-
                   4145:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
1.32      anton    4146: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
                   4147: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
                   4148: matmul    1.00  1.47  1.35   1.46  1.16  1.58        4.09
                   4149: fib       1.00  1.52  1.34   1.22  1.13  1.74  2.99  4.30
1.17      anton    4150: @end example
                   4152: You may find the good performance of Gforth compared with the systems
                   4153: written in assembly language quite surprising. One important reason for
                   4154: the disappointing performance of these systems is probably that they are
                   4155: not written optimally for the 486 (e.g., they use the @code{lods}
                   4156: instruction). In addition, Win32Forth uses a comfortable, but costly
                   4157: method for relocating the Forth image: like @code{cforth}, it computes
                   4158: the actual addresses at run time, resulting in two address computations
                   4159: per NEXT (@pxref{System Architecture}).
1.26      anton    4161: Only Eforth with the peephole optimizer performs comparable to
                   4162: Gforth. The speedups achieved with peephole optimization of threaded
                   4163: code are quite remarkable. Adding a peephole optimizer to Gforth should
                   4164: cause similar speedups.
1.30      anton    4166: The speedup of Gforth over PFE, ThisForth and TILE can be easily
                   4167: explained with the self-imposed restriction to standard C, which makes
                   4168: efficient threading impossible (however, the measured implementation of
                   4169: PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
                   4170: Register Variables,, GNU C Manual}).  Moreover, current C
                   4171: compilers have a hard time optimizing other aspects of the ThisForth
                   4172: and the TILE source.
1.17      anton    4173: 
                   4174: Note that the performance of Gforth on 386 architecture processors
                   4175: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
                   4176: failed to allocate any of the virtual machine registers into real
                   4177: machine registers by itself and would not work correctly with explicit
                   4178: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
                   4179: running the Sieve) than the one measured above.
1.26      anton    4181: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
                   4182: Maierhofer (presented at EuroForth '95), an indirect threaded version of
                   4183: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
                   4184: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
                   4185: used here. The paper available at
1.24      anton    4186: @*@file{};
                   4187: it also contains numbers for some native code systems. You can find
                   4188: numbers for Gforth on various machines in @file{Benchres}.
1.29      anton    4190: @node Bugs, Origin, Internals, Top
1.4       anton    4191: @chapter Bugs
1.17      anton    4193: Known bugs are described in the file BUGS in the Gforth distribution.
1.24      anton    4195: If you find a bug, please send a bug report to
1.32      anton    4196: @code{}. A bug report should
1.17      anton    4197: describe the Gforth version used (it is announced at the start of an
                   4198: interactive Gforth session), the machine and operating system (on Unix
                   4199: systems you can use @code{uname -a} to produce this information), the
1.24      anton    4200: installation options (send the @code{config.status} file), and a
                   4201: complete list of changes you (or your installer) have made to the Gforth
                   4202: sources (if any); it should contain a program (or a sequence of keyboard
                   4203: commands) that reproduces the bug and a description of what you think
                   4204: constitutes the buggy behaviour.
1.17      anton    4205: 
                   4206: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
                   4207: to Report Bugs,, GNU C Manual}.
1.29      anton    4210: @node Origin, Word Index, Bugs, Top
                   4211: @chapter Authors and Ancestors of Gforth
                   4213: @section Authors and Contributors
                   4215: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
1.30      anton    4216: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
                   4217: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
                   4218: with their continuous feedback. Lennart Benshop contributed
1.29      anton    4219: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
                   4220: support for calling C libraries. Helpful comments also came from Paul
                   4221: Kleinrubatscher, Christian Pirker, Dirk Zoller and Marcel Hendrix.
1.30      anton    4223: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
                   4224: and autoconf, among others), and to the creators of the Internet: Gforth
                   4225: was developed across the Internet, and its authors have not met
                   4226: physically yet.
1.29      anton    4228: @section Pedigree
1.4       anton    4229: 
1.17      anton    4230: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24      anton    4231: Dirk Zoller) will cross-fertilize each other. Of course, a significant
                   4232: part of the design of Gforth was prescribed by ANS Forth.
1.17      anton    4233: 
1.23      pazsan   4234: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
                   4235: 32 bit native code version of VolksForth for the Atari ST, written
                   4236: mostly by Dietrich Weineck.
                   4238: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
                   4239: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24      anton    4240: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17      anton    4241: 
1.34      anton    4242: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17      anton    4243: Forth-83 standard. !! Pedigree? When?
                   4245: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24      anton    4246: 1979. Robert Selzer and Bill Ragsdale developed the original
                   4247: implementation of fig-Forth for the 6502 based on microForth.
                   4249: The principal architect of microForth was Dean Sanderson. microForth was
                   4250: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
                   4251: the 1802, and subsequently implemented on the 8080, the 6800 and the
                   4252: Z80.
1.17      anton    4253: 
1.24      anton    4254: All earlier Forth systems were custom-made, usually by Charles Moore,
1.30      anton    4255: who discovered (as he puts it) Forth during the late 60s. The first full
                   4256: Forth existed in 1971.
1.17      anton    4257: 
                   4258: A part of the information in this section comes from @cite{The Evolution
                   4259: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
                   4260: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
                   4261: Notices 28(3), 1993.  You can find more historical and genealogical
                   4262: information about Forth there.
1.29      anton    4264: @node Word Index, Node Index, Origin, Top
1.4       anton    4265: @chapter Word Index
1.18      anton    4267: This index is as incomplete as the manual. Each word is listed with
                   4268: stack effect and wordset.
1.17      anton    4269: 
                   4270: @printindex fn
1.4       anton    4272: @node Node Index,  , Word Index, Top
                   4273: @chapter Node Index
1.17      anton    4274: 
                   4275: This index is even less complete than the manual.
1.1       anton    4276: 
                   4277: @contents
                   4278: @bye

FreeBSD-CVSweb <>