Annotation of gforth/gforth.ds, revision 1.46

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 gforth.info
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.)
                      8: 
                      9: @ifinfo
1.43      anton      10: This file documents Gforth 0.3
1.1       anton      11: 
1.43      anton      12: Copyright @copyright{} 1995-1997 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.
                     17:      
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).
                     23:      
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.
                     31:      
                     32:      Permission is granted to copy and distribute translations of this manual
                     33:      into another language, under the above conditions for modified versions,
                     34:      except that the sections entitled "Distribution" and "General Public
                     35:      License" may be included in a translation approved by the author instead
                     36:      of in the original English.
                     37: @end ifinfo
                     38: 
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
1.43      anton      44: @center for version 0.3
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
1.43      anton      54: Copyright @copyright{} 1995--1997 Free Software Foundation, Inc.
1.1       anton      55: 
                     56: @comment !! Published by ... or You can get a copy of this manual ...
                     57: 
                     58:      Permission is granted to make and distribute verbatim copies of
                     59:      this manual provided the copyright notice and this permission notice
                     60:      are preserved on all copies.
                     61:      
                     62:      Permission is granted to copy and distribute modified versions of this
                     63:      manual under the conditions for verbatim copying, provided also that the
                     64:      sections entitled "Distribution" and "General Public License" are
                     65:      included exactly as in the original, and provided that the entire
                     66:      resulting derived work is distributed under the terms of a permission
                     67:      notice identical to this one.
                     68:      
                     69:      Permission is granted to copy and distribute translations of this manual
                     70:      into another language, under the above conditions for modified versions,
                     71:      except that the sections entitled "Distribution" and "General Public
                     72:      License" may be included in a translation approved by the author instead
                     73:      of in the original English.
                     74: @end titlepage
                     75: 
                     76: 
                     77: @node Top, License, (dir), (dir)
                     78: @ifinfo
1.17      anton      79: Gforth is a free implementation of ANS Forth available on many
1.43      anton      80: personal machines. This manual corresponds to version 0.3.
1.1       anton      81: @end ifinfo
                     82: 
                     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
1.43      anton      87: * Invoking Gforth::             Starting Gforth
1.17      anton      88: * Words::                       Forth words available in Gforth
1.40      anton      89: * Tools::                       Programming tools
1.4       anton      90: * ANS conformance::             Implementation-defined options etc.
1.17      anton      91: * Model::                       The abstract machine of Gforth
1.43      anton      92: * Integrating Gforth::          Forth as scripting language for applications
1.17      anton      93: * Emacs and Gforth::            The Gforth Mode
1.43      anton      94: * Image Files::                 @code{.fi} files contain compiled code
                     95: * Engine::                      The inner interpreter and the primitives
1.4       anton      96: * Bugs::                        How to report them
1.29      anton      97: * Origin::                      Authors and ancestors of Gforth
1.4       anton      98: * Word Index::                  An item for each Forth word
1.43      anton      99: * Concept Index::               A menu covering many topics
1.1       anton     100: @end menu
                    101: 
1.43      anton     102: @node License, Preface, Top, Top
1.20      pazsan    103: @unnumbered GNU GENERAL PUBLIC LICENSE
                    104: @center Version 2, June 1991
                    105: 
                    106: @display
                    107: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
                    108: 675 Mass Ave, Cambridge, MA 02139, USA
                    109: 
                    110: Everyone is permitted to copy and distribute verbatim copies
                    111: of this license document, but changing it is not allowed.
                    112: @end display
                    113: 
                    114: @unnumberedsec Preamble
                    115: 
                    116:   The licenses for most software are designed to take away your
                    117: freedom to share and change it.  By contrast, the GNU General Public
                    118: License is intended to guarantee your freedom to share and change free
                    119: software---to make sure the software is free for all its users.  This
                    120: General Public License applies to most of the Free Software
                    121: Foundation's software and to any other program whose authors commit to
                    122: using it.  (Some other Free Software Foundation software is covered by
                    123: the GNU Library General Public License instead.)  You can apply it to
                    124: your programs, too.
                    125: 
                    126:   When we speak of free software, we are referring to freedom, not
                    127: price.  Our General Public Licenses are designed to make sure that you
                    128: have the freedom to distribute copies of free software (and charge for
                    129: this service if you wish), that you receive source code or can get it
                    130: if you want it, that you can change the software or use pieces of it
                    131: in new free programs; and that you know you can do these things.
                    132: 
                    133:   To protect your rights, we need to make restrictions that forbid
                    134: anyone to deny you these rights or to ask you to surrender the rights.
                    135: These restrictions translate to certain responsibilities for you if you
                    136: distribute copies of the software, or if you modify it.
                    137: 
                    138:   For example, if you distribute copies of such a program, whether
                    139: gratis or for a fee, you must give the recipients all the rights that
                    140: you have.  You must make sure that they, too, receive or can get the
                    141: source code.  And you must show them these terms so they know their
                    142: rights.
                    143: 
                    144:   We protect your rights with two steps: (1) copyright the software, and
                    145: (2) offer you this license which gives you legal permission to copy,
                    146: distribute and/or modify the software.
                    147: 
                    148:   Also, for each author's protection and ours, we want to make certain
                    149: that everyone understands that there is no warranty for this free
                    150: software.  If the software is modified by someone else and passed on, we
                    151: want its recipients to know that what they have is not the original, so
                    152: that any problems introduced by others will not reflect on the original
                    153: authors' reputations.
                    154: 
                    155:   Finally, any free program is threatened constantly by software
                    156: patents.  We wish to avoid the danger that redistributors of a free
                    157: program will individually obtain patent licenses, in effect making the
                    158: program proprietary.  To prevent this, we have made it clear that any
                    159: patent must be licensed for everyone's free use or not licensed at all.
                    160: 
                    161:   The precise terms and conditions for copying, distribution and
                    162: modification follow.
                    163: 
                    164: @iftex
                    165: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
                    166: @end iftex
                    167: @ifinfo
                    168: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
                    169: @end ifinfo
                    170: 
                    171: @enumerate 0
                    172: @item
                    173: This License applies to any program or other work which contains
                    174: a notice placed by the copyright holder saying it may be distributed
                    175: under the terms of this General Public License.  The ``Program'', below,
                    176: refers to any such program or work, and a ``work based on the Program''
                    177: means either the Program or any derivative work under copyright law:
                    178: that is to say, a work containing the Program or a portion of it,
                    179: either verbatim or with modifications and/or translated into another
                    180: language.  (Hereinafter, translation is included without limitation in
                    181: the term ``modification''.)  Each licensee is addressed as ``you''.
                    182: 
                    183: Activities other than copying, distribution and modification are not
                    184: covered by this License; they are outside its scope.  The act of
                    185: running the Program is not restricted, and the output from the Program
                    186: is covered only if its contents constitute a work based on the
                    187: Program (independent of having been made by running the Program).
                    188: Whether that is true depends on what the Program does.
                    189: 
                    190: @item
                    191: You may copy and distribute verbatim copies of the Program's
                    192: source code as you receive it, in any medium, provided that you
                    193: conspicuously and appropriately publish on each copy an appropriate
                    194: copyright notice and disclaimer of warranty; keep intact all the
                    195: notices that refer to this License and to the absence of any warranty;
                    196: and give any other recipients of the Program a copy of this License
                    197: along with the Program.
                    198: 
                    199: You may charge a fee for the physical act of transferring a copy, and
                    200: you may at your option offer warranty protection in exchange for a fee.
                    201: 
                    202: @item
                    203: You may modify your copy or copies of the Program or any portion
                    204: of it, thus forming a work based on the Program, and copy and
                    205: distribute such modifications or work under the terms of Section 1
                    206: above, provided that you also meet all of these conditions:
                    207: 
                    208: @enumerate a
                    209: @item
                    210: You must cause the modified files to carry prominent notices
                    211: stating that you changed the files and the date of any change.
                    212: 
                    213: @item
                    214: You must cause any work that you distribute or publish, that in
                    215: whole or in part contains or is derived from the Program or any
                    216: part thereof, to be licensed as a whole at no charge to all third
                    217: parties under the terms of this License.
                    218: 
                    219: @item
                    220: If the modified program normally reads commands interactively
                    221: when run, you must cause it, when started running for such
                    222: interactive use in the most ordinary way, to print or display an
                    223: announcement including an appropriate copyright notice and a
                    224: notice that there is no warranty (or else, saying that you provide
                    225: a warranty) and that users may redistribute the program under
                    226: these conditions, and telling the user how to view a copy of this
                    227: License.  (Exception: if the Program itself is interactive but
                    228: does not normally print such an announcement, your work based on
                    229: the Program is not required to print an announcement.)
                    230: @end enumerate
                    231: 
                    232: These requirements apply to the modified work as a whole.  If
                    233: identifiable sections of that work are not derived from the Program,
                    234: and can be reasonably considered independent and separate works in
                    235: themselves, then this License, and its terms, do not apply to those
                    236: sections when you distribute them as separate works.  But when you
                    237: distribute the same sections as part of a whole which is a work based
                    238: on the Program, the distribution of the whole must be on the terms of
                    239: this License, whose permissions for other licensees extend to the
                    240: entire whole, and thus to each and every part regardless of who wrote it.
                    241: 
                    242: Thus, it is not the intent of this section to claim rights or contest
                    243: your rights to work written entirely by you; rather, the intent is to
                    244: exercise the right to control the distribution of derivative or
                    245: collective works based on the Program.
                    246: 
                    247: In addition, mere aggregation of another work not based on the Program
                    248: with the Program (or with a work based on the Program) on a volume of
                    249: a storage or distribution medium does not bring the other work under
                    250: the scope of this License.
                    251: 
                    252: @item
                    253: You may copy and distribute the Program (or a work based on it,
                    254: under Section 2) in object code or executable form under the terms of
                    255: Sections 1 and 2 above provided that you also do one of the following:
                    256: 
                    257: @enumerate a
                    258: @item
                    259: Accompany it with the complete corresponding machine-readable
                    260: source code, which must be distributed under the terms of Sections
                    261: 1 and 2 above on a medium customarily used for software interchange; or,
                    262: 
                    263: @item
                    264: Accompany it with a written offer, valid for at least three
                    265: years, to give any third party, for a charge no more than your
                    266: cost of physically performing source distribution, a complete
                    267: machine-readable copy of the corresponding source code, to be
                    268: distributed under the terms of Sections 1 and 2 above on a medium
                    269: customarily used for software interchange; or,
                    270: 
                    271: @item
                    272: Accompany it with the information you received as to the offer
                    273: to distribute corresponding source code.  (This alternative is
                    274: allowed only for noncommercial distribution and only if you
                    275: received the program in object code or executable form with such
                    276: an offer, in accord with Subsection b above.)
                    277: @end enumerate
                    278: 
                    279: The source code for a work means the preferred form of the work for
                    280: making modifications to it.  For an executable work, complete source
                    281: code means all the source code for all modules it contains, plus any
                    282: associated interface definition files, plus the scripts used to
                    283: control compilation and installation of the executable.  However, as a
                    284: special exception, the source code distributed need not include
                    285: anything that is normally distributed (in either source or binary
                    286: form) with the major components (compiler, kernel, and so on) of the
                    287: operating system on which the executable runs, unless that component
                    288: itself accompanies the executable.
                    289: 
                    290: If distribution of executable or object code is made by offering
                    291: access to copy from a designated place, then offering equivalent
                    292: access to copy the source code from the same place counts as
                    293: distribution of the source code, even though third parties are not
                    294: compelled to copy the source along with the object code.
                    295: 
                    296: @item
                    297: You may not copy, modify, sublicense, or distribute the Program
                    298: except as expressly provided under this License.  Any attempt
                    299: otherwise to copy, modify, sublicense or distribute the Program is
                    300: void, and will automatically terminate your rights under this License.
                    301: However, parties who have received copies, or rights, from you under
                    302: this License will not have their licenses terminated so long as such
                    303: parties remain in full compliance.
                    304: 
                    305: @item
                    306: You are not required to accept this License, since you have not
                    307: signed it.  However, nothing else grants you permission to modify or
                    308: distribute the Program or its derivative works.  These actions are
                    309: prohibited by law if you do not accept this License.  Therefore, by
                    310: modifying or distributing the Program (or any work based on the
                    311: Program), you indicate your acceptance of this License to do so, and
                    312: all its terms and conditions for copying, distributing or modifying
                    313: the Program or works based on it.
                    314: 
                    315: @item
                    316: Each time you redistribute the Program (or any work based on the
                    317: Program), the recipient automatically receives a license from the
                    318: original licensor to copy, distribute or modify the Program subject to
                    319: these terms and conditions.  You may not impose any further
                    320: restrictions on the recipients' exercise of the rights granted herein.
                    321: You are not responsible for enforcing compliance by third parties to
                    322: this License.
                    323: 
                    324: @item
                    325: If, as a consequence of a court judgment or allegation of patent
                    326: infringement or for any other reason (not limited to patent issues),
                    327: conditions are imposed on you (whether by court order, agreement or
                    328: otherwise) that contradict the conditions of this License, they do not
                    329: excuse you from the conditions of this License.  If you cannot
                    330: distribute so as to satisfy simultaneously your obligations under this
                    331: License and any other pertinent obligations, then as a consequence you
                    332: may not distribute the Program at all.  For example, if a patent
                    333: license would not permit royalty-free redistribution of the Program by
                    334: all those who receive copies directly or indirectly through you, then
                    335: the only way you could satisfy both it and this License would be to
                    336: refrain entirely from distribution of the Program.
                    337: 
                    338: If any portion of this section is held invalid or unenforceable under
                    339: any particular circumstance, the balance of the section is intended to
                    340: apply and the section as a whole is intended to apply in other
                    341: circumstances.
                    342: 
                    343: It is not the purpose of this section to induce you to infringe any
                    344: patents or other property right claims or to contest validity of any
                    345: such claims; this section has the sole purpose of protecting the
                    346: integrity of the free software distribution system, which is
                    347: implemented by public license practices.  Many people have made
                    348: generous contributions to the wide range of software distributed
                    349: through that system in reliance on consistent application of that
                    350: system; it is up to the author/donor to decide if he or she is willing
                    351: to distribute software through any other system and a licensee cannot
                    352: impose that choice.
                    353: 
                    354: This section is intended to make thoroughly clear what is believed to
                    355: be a consequence of the rest of this License.
                    356: 
                    357: @item
                    358: If the distribution and/or use of the Program is restricted in
                    359: certain countries either by patents or by copyrighted interfaces, the
                    360: original copyright holder who places the Program under this License
                    361: may add an explicit geographical distribution limitation excluding
                    362: those countries, so that distribution is permitted only in or among
                    363: countries not thus excluded.  In such case, this License incorporates
                    364: the limitation as if written in the body of this License.
                    365: 
                    366: @item
                    367: The Free Software Foundation may publish revised and/or new versions
                    368: of the General Public License from time to time.  Such new versions will
                    369: be similar in spirit to the present version, but may differ in detail to
                    370: address new problems or concerns.
                    371: 
                    372: Each version is given a distinguishing version number.  If the Program
                    373: specifies a version number of this License which applies to it and ``any
                    374: later version'', you have the option of following the terms and conditions
                    375: either of that version or of any later version published by the Free
                    376: Software Foundation.  If the Program does not specify a version number of
                    377: this License, you may choose any version ever published by the Free Software
                    378: Foundation.
                    379: 
                    380: @item
                    381: If you wish to incorporate parts of the Program into other free
                    382: programs whose distribution conditions are different, write to the author
                    383: to ask for permission.  For software which is copyrighted by the Free
                    384: Software Foundation, write to the Free Software Foundation; we sometimes
                    385: make exceptions for this.  Our decision will be guided by the two goals
                    386: of preserving the free status of all derivatives of our free software and
                    387: of promoting the sharing and reuse of software generally.
                    388: 
                    389: @iftex
                    390: @heading NO WARRANTY
                    391: @end iftex
                    392: @ifinfo
                    393: @center NO WARRANTY
                    394: @end ifinfo
                    395: 
                    396: @item
                    397: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
                    398: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
                    399: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
                    400: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
                    401: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
                    402: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
                    403: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
                    404: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
                    405: REPAIR OR CORRECTION.
                    406: 
                    407: @item
                    408: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
                    409: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
                    410: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
                    411: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
                    412: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
                    413: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
                    414: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
                    415: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
                    416: POSSIBILITY OF SUCH DAMAGES.
                    417: @end enumerate
                    418: 
                    419: @iftex
                    420: @heading END OF TERMS AND CONDITIONS
                    421: @end iftex
                    422: @ifinfo
                    423: @center END OF TERMS AND CONDITIONS
                    424: @end ifinfo
                    425: 
                    426: @page
                    427: @unnumberedsec How to Apply These Terms to Your New Programs
                    428: 
                    429:   If you develop a new program, and you want it to be of the greatest
                    430: possible use to the public, the best way to achieve this is to make it
                    431: free software which everyone can redistribute and change under these terms.
                    432: 
                    433:   To do so, attach the following notices to the program.  It is safest
                    434: to attach them to the start of each source file to most effectively
                    435: convey the exclusion of warranty; and each file should have at least
                    436: the ``copyright'' line and a pointer to where the full notice is found.
                    437: 
                    438: @smallexample
                    439: @var{one line to give the program's name and a brief idea of what it does.}
                    440: Copyright (C) 19@var{yy}  @var{name of author}
                    441: 
                    442: This program is free software; you can redistribute it and/or modify 
                    443: it under the terms of the GNU General Public License as published by 
                    444: the Free Software Foundation; either version 2 of the License, or 
                    445: (at your option) any later version.
                    446: 
                    447: This program is distributed in the hope that it will be useful,
                    448: but WITHOUT ANY WARRANTY; without even the implied warranty of
                    449: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
                    450: GNU General Public License for more details.
                    451: 
                    452: You should have received a copy of the GNU General Public License
                    453: along with this program; if not, write to the Free Software
                    454: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
                    455: @end smallexample
                    456: 
                    457: Also add information on how to contact you by electronic and paper mail.
                    458: 
                    459: If the program is interactive, make it output a short notice like this
                    460: when it starts in an interactive mode:
                    461: 
                    462: @smallexample
                    463: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
                    464: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
                    465: type `show w'.  
                    466: This is free software, and you are welcome to redistribute it 
                    467: under certain conditions; type `show c' for details.
                    468: @end smallexample
                    469: 
                    470: The hypothetical commands @samp{show w} and @samp{show c} should show
                    471: the appropriate parts of the General Public License.  Of course, the
                    472: commands you use may be called something other than @samp{show w} and
                    473: @samp{show c}; they could even be mouse-clicks or menu items---whatever
                    474: suits your program.
                    475: 
                    476: You should also get your employer (if you work as a programmer) or your
                    477: school, if any, to sign a ``copyright disclaimer'' for the program, if
                    478: necessary.  Here is a sample; alter the names:
                    479: 
                    480: @smallexample
                    481: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
                    482: `Gnomovision' (which makes passes at compilers) written by James Hacker.
                    483: 
                    484: @var{signature of Ty Coon}, 1 April 1989
                    485: Ty Coon, President of Vice
                    486: @end smallexample
                    487: 
                    488: This General Public License does not permit incorporating your program into
                    489: proprietary programs.  If your program is a subroutine library, you may
                    490: consider it more useful to permit linking proprietary applications with the
                    491: library.  If this is what you want to do, use the GNU Library General
                    492: Public License instead of this License.
1.1       anton     493: 
                    494: @iftex
1.43      anton     495: @node    Preface, Goals, License, Top
1.23      pazsan    496: @comment node-name,     next,           previous, up
1.1       anton     497: @unnumbered Preface
1.23      pazsan    498: @cindex Preface
1.17      anton     499: This manual documents Gforth. The reader is expected to know
1.1       anton     500: Forth. This manual is primarily a reference manual. @xref{Other Books}
                    501: for introductory material.
                    502: @end iftex
                    503: 
1.43      anton     504: @node    Goals, Other Books, Preface, Top
1.1       anton     505: @comment node-name,     next,           previous, up
1.17      anton     506: @chapter Goals of Gforth
1.1       anton     507: @cindex Goals
1.17      anton     508: The goal of the Gforth Project is to develop a standard model for
1.43      anton     509: ANS Forth. This can be split into several subgoals:
1.1       anton     510: 
                    511: @itemize @bullet
                    512: @item
1.43      anton     513: Gforth should conform to the Forth standard (ANS Forth).
1.1       anton     514: @item
                    515: It should be a model, i.e. it should define all the
                    516: implementation-dependent things.
                    517: @item
                    518: It should become standard, i.e. widely accepted and used. This goal
                    519: is the most difficult one.
                    520: @end itemize
                    521: 
1.17      anton     522: To achieve these goals Gforth should be
1.1       anton     523: @itemize @bullet
                    524: @item
                    525: Similar to previous models (fig-Forth, F83)
                    526: @item
                    527: Powerful. It should provide for all the things that are considered
                    528: necessary today and even some that are not yet considered necessary.
                    529: @item
                    530: Efficient. It should not get the reputation of being exceptionally
                    531: slow.
                    532: @item
                    533: Free.
                    534: @item
                    535: Available on many machines/easy to port.
                    536: @end itemize
                    537: 
1.17      anton     538: Have we achieved these goals? Gforth conforms to the ANS Forth
                    539: standard. It may be considered a model, but we have not yet documented
1.1       anton     540: which parts of the model are stable and which parts we are likely to
1.17      anton     541: change. It certainly has not yet become a de facto standard. It has some
                    542: similarities and some differences to previous models. It has some
                    543: powerful features, but not yet everything that we envisioned. We
                    544: certainly have achieved our execution speed goals (@pxref{Performance}).
                    545: It is free and available on many machines.
1.1       anton     546: 
1.43      anton     547: @node Other Books, Invoking Gforth, Goals, Top
1.1       anton     548: @chapter Other books on ANS Forth
1.43      anton     549: @cindex books on Forth
1.1       anton     550: 
                    551: As the standard is relatively new, there are not many books out yet. It
1.17      anton     552: is not recommended to learn Forth by using Gforth and a book that is
1.1       anton     553: not written for ANS Forth, as you will not know your mistakes from the
                    554: deviations of the book.
                    555: 
1.43      anton     556: @cindex standard document for ANS Forth
                    557: @cindex ANS Forth document
1.1       anton     558: There is, of course, the standard, the definite reference if you want to
1.19      anton     559: write ANS Forth programs. It is available in printed form from the
                    560: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
                    561: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
                    562: can also get it from Global Engineering Documents (Tel.: USA (800)
                    563: 854-7179; Fax.: (303) 843-9880) for about $300.
                    564: 
                    565: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
                    566: for publication is available electronically and for free in some MS Word
                    567: format, and it has been converted to HTML. Some pointers to these
                    568: versions can be found through
1.24      anton     569: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
1.1       anton     570: 
1.43      anton     571: @cindex introductory book
                    572: @cindex book, introductory
                    573: @cindex Woehr, Jack: @cite{Forth: The New Model}
                    574: @cindex @cite{Forth: The new model} (book)
                    575: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
1.1       anton     576: introductory book based on a draft version of the standard. It does not
                    577: cover the whole standard. It also contains interesting background
1.41      anton     578: information (Jack Woehr was in the ANS Forth Technical Committee). It is
1.1       anton     579: not appropriate for complete newbies, but programmers experienced in
                    580: other languages should find it ok.
                    581: 
1.43      anton     582: @node Invoking Gforth, Words, Other Books, Top
                    583: @chapter Invoking Gforth
                    584: @cindex invoking Gforth
                    585: @cindex running Gforth
                    586: @cindex command-line options
                    587: @cindex options on the command line
                    588: @cindex flags on the command line
1.1       anton     589: 
                    590: You will usually just say @code{gforth}. In many other cases the default
1.17      anton     591: Gforth image will be invoked like this:
1.1       anton     592: 
                    593: @example
                    594: gforth [files] [-e forth-code]
                    595: @end example
                    596: 
                    597: executing the contents of the files and the Forth code in the order they
                    598: are given.
                    599: 
                    600: In general, the command line looks like this:
                    601: 
                    602: @example
                    603: gforth [initialization options] [image-specific options]
                    604: @end example
                    605: 
                    606: The initialization options must come before the rest of the command
                    607: line. They are:
                    608: 
                    609: @table @code
1.43      anton     610: @cindex -i, command-line option
                    611: @cindex --image-file, command-line option
1.1       anton     612: @item --image-file @var{file}
1.43      anton     613: @itemx -i @var{file}
1.1       anton     614: Loads the Forth image @var{file} instead of the default
1.43      anton     615: @file{gforth.fi} (@pxref{Image Files}).
1.1       anton     616: 
1.43      anton     617: @cindex --path, command-line option
                    618: @cindex -p, command-line option
1.1       anton     619: @item --path @var{path}
1.43      anton     620: @itemx -p @var{path}
1.39      anton     621: Uses @var{path} for searching the image file and Forth source code files
                    622: instead of the default in the environment variable @code{GFORTHPATH} or
                    623: the path specified at installation time (e.g.,
                    624: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
                    625: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1.1       anton     626: 
1.43      anton     627: @cindex --dictionary-size, command-line option
                    628: @cindex -m, command-line option
                    629: @cindex @var{size} parameters for command-line options
                    630: @cindex size of the dictionary and the stacks
1.1       anton     631: @item --dictionary-size @var{size}
1.43      anton     632: @itemx -m @var{size}
1.1       anton     633: Allocate @var{size} space for the Forth dictionary space instead of
                    634: using the default specified in the image (typically 256K). The
                    635: @var{size} specification consists of an integer and a unit (e.g.,
                    636: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
                    637: size, in this case Cells), @code{k} (kilobytes), and @code{M}
                    638: (Megabytes). If no unit is specified, @code{e} is used.
                    639: 
1.43      anton     640: @cindex --data-stack-size, command-line option
                    641: @cindex -d, command-line option
1.1       anton     642: @item --data-stack-size @var{size}
1.43      anton     643: @itemx -d @var{size}
1.1       anton     644: Allocate @var{size} space for the data stack instead of using the
                    645: default specified in the image (typically 16K).
                    646: 
1.43      anton     647: @cindex --return-stack-size, command-line option
                    648: @cindex -r, command-line option
1.1       anton     649: @item --return-stack-size @var{size}
1.43      anton     650: @itemx -r @var{size}
1.1       anton     651: Allocate @var{size} space for the return stack instead of using the
1.43      anton     652: default specified in the image (typically 15K).
1.1       anton     653: 
1.43      anton     654: @cindex --fp-stack-size, command-line option
                    655: @cindex -f, command-line option
1.1       anton     656: @item --fp-stack-size @var{size}
1.43      anton     657: @itemx -f @var{size}
1.1       anton     658: Allocate @var{size} space for the floating point stack instead of
1.43      anton     659: using the default specified in the image (typically 15.5K). In this case
1.1       anton     660: the unit specifier @code{e} refers to floating point numbers.
                    661: 
1.43      anton     662: @cindex --locals-stack-size, command-line option
                    663: @cindex -l, command-line option
1.1       anton     664: @item --locals-stack-size @var{size}
1.43      anton     665: @itemx -l @var{size}
1.1       anton     666: Allocate @var{size} space for the locals stack instead of using the
1.43      anton     667: default specified in the image (typically 14.5K).
1.1       anton     668: 
1.43      anton     669: @cindex -h, command-line option
                    670: @cindex --help, command-line option
                    671: @item --help
                    672: @itemx -h
                    673: Print a message about the command-line options
                    674: 
                    675: @cindex -v, command-line option
                    676: @cindex --version, command-line option
                    677: @item --version
                    678: @itemx -v
                    679: Print version and exit
                    680: 
                    681: @cindex --debug, command-line option
                    682: @item --debug
                    683: Print some information useful for debugging on startup.
                    684: 
                    685: @cindex --offset-image, command-line option
                    686: @item --offset-image
                    687: Start the dictionary at a slightly different position than would be used
                    688: otherwise (useful for creating data-relocatable images,
                    689: @pxref{Data-Relocatable Image Files}).
                    690: 
                    691: @cindex --clear-dictionary, command-line option
                    692: @item --clear-dictionary
                    693: Initialize all bytes in the dictionary to 0 before loading the image
                    694: (@pxref{Data-Relocatable Image Files}).
1.1       anton     695: @end table
                    696: 
1.43      anton     697: @cindex loading files at startup
                    698: @cindex executing code on startup
                    699: @cindex batch processing with Gforth
1.1       anton     700: As explained above, the image-specific command-line arguments for the
                    701: default image @file{gforth.fi} consist of a sequence of filenames and
1.41      anton     702: @code{-e @var{forth-code}} options that are interpreted in the sequence
1.1       anton     703: in which they are given. The @code{-e @var{forth-code}} or
                    704: @code{--evaluate @var{forth-code}} option evaluates the forth
                    705: code. This option takes only one argument; if you want to evaluate more
                    706: Forth words, you have to quote them or use several @code{-e}s. To exit
                    707: after processing the command line (instead of entering interactive mode)
                    708: append @code{-e bye} to the command line.
                    709: 
1.43      anton     710: @cindex versions, invoking other versions of Gforth
1.22      anton     711: If you have several versions of Gforth installed, @code{gforth} will
                    712: invoke the version that was installed last. @code{gforth-@var{version}}
                    713: invokes a specific version. You may want to use the option
                    714: @code{--path}, if your environment contains the variable
                    715: @code{GFORTHPATH}.
                    716: 
1.1       anton     717: Not yet implemented:
                    718: On startup the system first executes the system initialization file
                    719: (unless the option @code{--no-init-file} is given; note that the system
                    720: resulting from using this option may not be ANS Forth conformant). Then
                    721: the user initialization file @file{.gforth.fs} is executed, unless the
                    722: option @code{--no-rc} is given; this file is first searched in @file{.},
                    723: then in @file{~}, then in the normal path (see above).
                    724: 
1.43      anton     725: @node Words, Tools, Invoking Gforth, Top
1.1       anton     726: @chapter Forth Words
1.43      anton     727: @cindex Words
1.1       anton     728: 
                    729: @menu
1.4       anton     730: * Notation::                    
                    731: * Arithmetic::                  
                    732: * Stack Manipulation::          
1.43      anton     733: * Memory::               
1.4       anton     734: * Control Structures::          
                    735: * Locals::                      
                    736: * Defining Words::              
1.37      anton     737: * Tokens for Words::            
1.4       anton     738: * Wordlists::                   
                    739: * Files::                       
                    740: * Blocks::                      
                    741: * Other I/O::                   
                    742: * Programming Tools::           
1.18      anton     743: * Assembler and Code words::    
1.4       anton     744: * Threading Words::             
1.1       anton     745: @end menu
                    746: 
                    747: @node Notation, Arithmetic, Words, Words
                    748: @section Notation
1.43      anton     749: @cindex notation of glossary entries
                    750: @cindex format of glossary entries
                    751: @cindex glossary notation format
                    752: @cindex word glossary entry format
1.1       anton     753: 
                    754: The Forth words are described in this section in the glossary notation
1.43      anton     755: that has become a de-facto standard for Forth texts, i.e.,
1.1       anton     756: 
1.4       anton     757: @format
1.1       anton     758: @var{word}     @var{Stack effect}   @var{wordset}   @var{pronunciation}
1.4       anton     759: @end format
1.1       anton     760: @var{Description}
                    761: 
                    762: @table @var
                    763: @item word
1.43      anton     764: @cindex case insensitivity
1.17      anton     765: The name of the word. BTW, Gforth is case insensitive, so you can
1.14      anton     766: type the words in in lower case (However, @pxref{core-idef}).
1.1       anton     767: 
                    768: @item Stack effect
1.43      anton     769: @cindex stack effect
1.1       anton     770: The stack effect is written in the notation @code{@var{before} --
                    771: @var{after}}, where @var{before} and @var{after} describe the top of
                    772: stack entries before and after the execution of the word. The rest of
                    773: the stack is not touched by the word. The top of stack is rightmost,
1.17      anton     774: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1       anton     775: uses a separate floating point stack, but a unified stack
                    776: notation. Also, return stack effects are not shown in @var{stack
                    777: effect}, but in @var{Description}. The name of a stack item describes
                    778: the type and/or the function of the item. See below for a discussion of
                    779: the types.
                    780: 
1.19      anton     781: All words have two stack effects: A compile-time stack effect and a
                    782: run-time stack effect. The compile-time stack-effect of most words is
                    783: @var{ -- }. If the compile-time stack-effect of a word deviates from
                    784: this standard behaviour, or the word does other unusual things at
                    785: compile time, both stack effects are shown; otherwise only the run-time
                    786: stack effect is shown.
                    787: 
1.43      anton     788: @cindex pronounciation of words
1.1       anton     789: @item pronunciation
1.43      anton     790: How the word is pronounced.
1.1       anton     791: 
1.43      anton     792: @cindex wordset
1.1       anton     793: @item wordset
                    794: The ANS Forth standard is divided into several wordsets. A standard
                    795: system need not support all of them. So, the fewer wordsets your program
                    796: uses the more portable it will be in theory. However, we suspect that
                    797: most ANS Forth systems on personal machines will feature all
                    798: wordsets. Words that are not defined in the ANS standard have
1.19      anton     799: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
                    800: describes words that will work in future releases of Gforth;
                    801: @code{gforth-internal} words are more volatile. Environmental query
                    802: strings are also displayed like words; you can recognize them by the
                    803: @code{environment} in the wordset field.
1.1       anton     804: 
                    805: @item Description
                    806: A description of the behaviour of the word.
                    807: @end table
                    808: 
1.43      anton     809: @cindex types of stack items
                    810: @cindex stack item types
1.4       anton     811: The type of a stack item is specified by the character(s) the name
                    812: starts with:
1.1       anton     813: 
                    814: @table @code
                    815: @item f
1.43      anton     816: @cindex @code{f}, stack item type
                    817: Boolean flags, i.e. @code{false} or @code{true}.
1.1       anton     818: @item c
1.43      anton     819: @cindex @code{c}, stack item type
1.1       anton     820: Char
                    821: @item w
1.43      anton     822: @cindex @code{w}, stack item type
1.1       anton     823: Cell, can contain an integer or an address
                    824: @item n
1.43      anton     825: @cindex @code{n}, stack item type
1.1       anton     826: signed integer
                    827: @item u
1.43      anton     828: @cindex @code{u}, stack item type
1.1       anton     829: unsigned integer
                    830: @item d
1.43      anton     831: @cindex @code{d}, stack item type
1.1       anton     832: double sized signed integer
                    833: @item ud
1.43      anton     834: @cindex @code{ud}, stack item type
1.1       anton     835: double sized unsigned integer
                    836: @item r
1.43      anton     837: @cindex @code{r}, stack item type
1.36      anton     838: Float (on the FP stack)
1.1       anton     839: @item a_
1.43      anton     840: @cindex @code{a_}, stack item type
1.1       anton     841: Cell-aligned address
                    842: @item c_
1.43      anton     843: @cindex @code{c_}, stack item type
1.36      anton     844: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.1       anton     845: @item f_
1.43      anton     846: @cindex @code{f_}, stack item type
1.1       anton     847: Float-aligned address
                    848: @item df_
1.43      anton     849: @cindex @code{df_}, stack item type
1.1       anton     850: Address aligned for IEEE double precision float
                    851: @item sf_
1.43      anton     852: @cindex @code{sf_}, stack item type
1.1       anton     853: Address aligned for IEEE single precision float
                    854: @item xt
1.43      anton     855: @cindex @code{xt}, stack item type
1.1       anton     856: Execution token, same size as Cell
                    857: @item wid
1.43      anton     858: @cindex @code{wid}, stack item type
1.1       anton     859: Wordlist ID, same size as Cell
                    860: @item f83name
1.43      anton     861: @cindex @code{f83name}, stack item type
1.1       anton     862: Pointer to a name structure
1.36      anton     863: @item "
1.43      anton     864: @cindex @code{"}, stack item type
1.36      anton     865: string in the input stream (not the stack). The terminating character is
                    866: a blank by default. If it is not a blank, it is shown in @code{<>}
                    867: quotes.
1.1       anton     868: @end table
                    869: 
1.4       anton     870: @node Arithmetic, Stack Manipulation, Notation, Words
1.1       anton     871: @section Arithmetic
1.43      anton     872: @cindex arithmetic words
                    873: 
                    874: @cindex division with potentially negative operands
1.1       anton     875: Forth arithmetic is not checked, i.e., you will not hear about integer
                    876: overflow on addition or multiplication, you may hear about division by
                    877: zero if you are lucky. The operator is written after the operands, but
                    878: the operands are still in the original order. I.e., the infix @code{2-1}
                    879: corresponds to @code{2 1 -}. Forth offers a variety of division
                    880: operators. If you perform division with potentially negative operands,
                    881: you do not want to use @code{/} or @code{/mod} with its undefined
                    882: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4       anton     883: former, @pxref{Mixed precision}).
                    884: 
                    885: @menu
                    886: * Single precision::            
                    887: * Bitwise operations::          
                    888: * Mixed precision::             operations with single and double-cell integers
                    889: * Double precision::            Double-cell integer arithmetic
                    890: * Floating Point::              
                    891: @end menu
1.1       anton     892: 
1.4       anton     893: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1       anton     894: @subsection Single precision
1.43      anton     895: @cindex single precision arithmetic words
                    896: 
1.1       anton     897: doc-+
                    898: doc--
                    899: doc-*
                    900: doc-/
                    901: doc-mod
                    902: doc-/mod
                    903: doc-negate
                    904: doc-abs
                    905: doc-min
                    906: doc-max
                    907: 
1.4       anton     908: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1       anton     909: @subsection Bitwise operations
1.43      anton     910: @cindex bitwise operation words
                    911: 
1.1       anton     912: doc-and
                    913: doc-or
                    914: doc-xor
                    915: doc-invert
                    916: doc-2*
                    917: doc-2/
                    918: 
1.4       anton     919: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1       anton     920: @subsection Mixed precision
1.43      anton     921: @cindex mixed precision arithmetic words
                    922: 
1.1       anton     923: doc-m+
                    924: doc-*/
                    925: doc-*/mod
                    926: doc-m*
                    927: doc-um*
                    928: doc-m*/
                    929: doc-um/mod
                    930: doc-fm/mod
                    931: doc-sm/rem
                    932: 
1.4       anton     933: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1       anton     934: @subsection Double precision
1.43      anton     935: @cindex double precision arithmetic words
1.16      anton     936: 
1.43      anton     937: @cindex double-cell numbers, input format
                    938: @cindex input format for double-cell numbers
1.16      anton     939: The outer (aka text) interpreter converts numbers containing a dot into
                    940: a double precision number. Note that only numbers with the dot as last
                    941: character are standard-conforming.
                    942: 
1.1       anton     943: doc-d+
                    944: doc-d-
                    945: doc-dnegate
                    946: doc-dabs
                    947: doc-dmin
                    948: doc-dmax
                    949: 
1.4       anton     950: @node Floating Point,  , Double precision, Arithmetic
                    951: @subsection Floating Point
1.43      anton     952: @cindex floating point arithmetic words
1.16      anton     953: 
1.43      anton     954: @cindex floating-point numbers, input format
                    955: @cindex input format for floating-point numbers
1.16      anton     956: The format of floating point numbers recognized by the outer (aka text)
                    957: interpreter is: a signed decimal number, possibly containing a decimal
                    958: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
1.41      anton     959: by a signed integer (the exponent). E.g., @code{1e} is the same as
1.35      anton     960: @code{+1.0e+0}. Note that a number without @code{e}
1.16      anton     961: is not interpreted as floating-point number, but as double (if the
                    962: number contains a @code{.}) or single precision integer. Also,
                    963: conversions between string and floating point numbers always use base
                    964: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
                    965: value greater then 14, the @code{E} may be interpreted as digit and the
                    966: number will be interpreted as integer, unless it has a signed exponent
                    967: (both @code{+} and @code{-} are allowed as signs).
1.4       anton     968: 
1.43      anton     969: @cindex angles in trigonometric operations
                    970: @cindex trigonometric operations
1.4       anton     971: Angles in floating point operations are given in radians (a full circle
1.17      anton     972: has 2 pi radians). Note, that Gforth has a separate floating point
1.4       anton     973: stack, but we use the unified notation.
                    974: 
1.43      anton     975: @cindex floating-point arithmetic, pitfalls
1.4       anton     976: Floating point numbers have a number of unpleasant surprises for the
                    977: unwary (e.g., floating point addition is not associative) and even a few
                    978: for the wary. You should not use them unless you know what you are doing
                    979: or you don't care that the results you get are totally bogus. If you
                    980: want to learn about the problems of floating point numbers (and how to
1.11      anton     981: avoid them), you might start with @cite{David Goldberg, What Every
1.6       anton     982: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
                    983: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4       anton     984: 
                    985: doc-f+
                    986: doc-f-
                    987: doc-f*
                    988: doc-f/
                    989: doc-fnegate
                    990: doc-fabs
                    991: doc-fmax
                    992: doc-fmin
                    993: doc-floor
                    994: doc-fround
                    995: doc-f**
                    996: doc-fsqrt
                    997: doc-fexp
                    998: doc-fexpm1
                    999: doc-fln
                   1000: doc-flnp1
                   1001: doc-flog
1.6       anton    1002: doc-falog
1.4       anton    1003: doc-fsin
                   1004: doc-fcos
                   1005: doc-fsincos
                   1006: doc-ftan
                   1007: doc-fasin
                   1008: doc-facos
                   1009: doc-fatan
                   1010: doc-fatan2
                   1011: doc-fsinh
                   1012: doc-fcosh
                   1013: doc-ftanh
                   1014: doc-fasinh
                   1015: doc-facosh
                   1016: doc-fatanh
                   1017: 
1.43      anton    1018: @node Stack Manipulation, Memory, Arithmetic, Words
1.1       anton    1019: @section Stack Manipulation
1.43      anton    1020: @cindex stack manipulation words
1.1       anton    1021: 
1.43      anton    1022: @cindex floating-point stack in the standard
1.17      anton    1023: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1       anton    1024: addresses, and double cells, a floating point stack for floating point
                   1025: numbers, a return stack for storing the return addresses of colon
                   1026: definitions and other data, and a locals stack for storing local
                   1027: variables. Note that while every sane Forth has a separate floating
                   1028: point stack, this is not strictly required; an ANS Forth system could
                   1029: theoretically keep floating point numbers on the data stack. As an
                   1030: additional difficulty, you don't know how many cells a floating point
                   1031: number takes. It is reportedly possible to write words in a way that
                   1032: they work also for a unified stack model, but we do not recommend trying
1.4       anton    1033: it. Instead, just say that your program has an environmental dependency
                   1034: on a separate FP stack.
                   1035: 
1.43      anton    1036: @cindex return stack and locals
                   1037: @cindex locals and return stack
1.4       anton    1038: Also, a Forth system is allowed to keep the local variables on the
1.1       anton    1039: return stack. This is reasonable, as local variables usually eliminate
                   1040: the need to use the return stack explicitly. So, if you want to produce
                   1041: a standard complying program and if you are using local variables in a
                   1042: word, forget about return stack manipulations in that word (see the
                   1043: standard document for the exact rules).
                   1044: 
1.4       anton    1045: @menu
                   1046: * Data stack::                  
                   1047: * Floating point stack::        
                   1048: * Return stack::                
                   1049: * Locals stack::                
                   1050: * Stack pointer manipulation::  
                   1051: @end menu
                   1052: 
                   1053: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1       anton    1054: @subsection Data stack
1.43      anton    1055: @cindex data stack manipulation words
                   1056: @cindex stack manipulations words, data stack
                   1057: 
1.1       anton    1058: doc-drop
                   1059: doc-nip
                   1060: doc-dup
                   1061: doc-over
                   1062: doc-tuck
                   1063: doc-swap
                   1064: doc-rot
                   1065: doc--rot
                   1066: doc-?dup
                   1067: doc-pick
                   1068: doc-roll
                   1069: doc-2drop
                   1070: doc-2nip
                   1071: doc-2dup
                   1072: doc-2over
                   1073: doc-2tuck
                   1074: doc-2swap
                   1075: doc-2rot
                   1076: 
1.4       anton    1077: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1       anton    1078: @subsection Floating point stack
1.43      anton    1079: @cindex floating-point stack manipulation words
                   1080: @cindex stack manipulation words, floating-point stack
                   1081: 
1.1       anton    1082: doc-fdrop
                   1083: doc-fnip
                   1084: doc-fdup
                   1085: doc-fover
                   1086: doc-ftuck
                   1087: doc-fswap
                   1088: doc-frot
                   1089: 
1.4       anton    1090: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1       anton    1091: @subsection Return stack
1.43      anton    1092: @cindex return stack manipulation words
                   1093: @cindex stack manipulation words, return stack
                   1094: 
1.1       anton    1095: doc->r
                   1096: doc-r>
                   1097: doc-r@
                   1098: doc-rdrop
                   1099: doc-2>r
                   1100: doc-2r>
                   1101: doc-2r@
                   1102: doc-2rdrop
                   1103: 
1.4       anton    1104: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1       anton    1105: @subsection Locals stack
                   1106: 
1.4       anton    1107: @node Stack pointer manipulation,  , Locals stack, Stack Manipulation
1.1       anton    1108: @subsection Stack pointer manipulation
1.43      anton    1109: @cindex stack pointer manipulation words
                   1110: 
1.1       anton    1111: doc-sp@
                   1112: doc-sp!
                   1113: doc-fp@
                   1114: doc-fp!
                   1115: doc-rp@
                   1116: doc-rp!
                   1117: doc-lp@
                   1118: doc-lp!
                   1119: 
1.43      anton    1120: @node Memory, Control Structures, Stack Manipulation, Words
                   1121: @section Memory
                   1122: @cindex Memory words
1.1       anton    1123: 
1.4       anton    1124: @menu
1.43      anton    1125: * Memory Access::      
1.4       anton    1126: * Address arithmetic::          
1.43      anton    1127: * Memory Blocks::         
1.4       anton    1128: @end menu
                   1129: 
1.43      anton    1130: @node Memory Access, Address arithmetic, Memory, Memory
                   1131: @subsection Memory Access
                   1132: @cindex memory access words
1.1       anton    1133: 
                   1134: doc-@
                   1135: doc-!
                   1136: doc-+!
                   1137: doc-c@
                   1138: doc-c!
                   1139: doc-2@
                   1140: doc-2!
                   1141: doc-f@
                   1142: doc-f!
                   1143: doc-sf@
                   1144: doc-sf!
                   1145: doc-df@
                   1146: doc-df!
                   1147: 
1.43      anton    1148: @node Address arithmetic, Memory Blocks, Memory Access, Memory
1.1       anton    1149: @subsection Address arithmetic
1.43      anton    1150: @cindex address arithmetic words
1.1       anton    1151: 
                   1152: ANS Forth does not specify the sizes of the data types. Instead, it
                   1153: offers a number of words for computing sizes and doing address
                   1154: arithmetic. Basically, address arithmetic is performed in terms of
                   1155: address units (aus); on most systems the address unit is one byte. Note
                   1156: that a character may have more than one au, so @code{chars} is no noop
                   1157: (on systems where it is a noop, it compiles to nothing).
                   1158: 
1.43      anton    1159: @cindex alignment of addresses for types
1.1       anton    1160: ANS Forth also defines words for aligning addresses for specific
1.43      anton    1161: types. Many computers require that accesses to specific data types
1.1       anton    1162: must only occur at specific addresses; e.g., that cells may only be
                   1163: accessed at addresses divisible by 4. Even if a machine allows unaligned
                   1164: accesses, it can usually perform aligned accesses faster. 
                   1165: 
1.17      anton    1166: For the performance-conscious: alignment operations are usually only
1.1       anton    1167: necessary during the definition of a data structure, not during the
                   1168: (more frequent) accesses to it.
                   1169: 
                   1170: ANS Forth defines no words for character-aligning addresses. This is not
                   1171: an oversight, but reflects the fact that addresses that are not
                   1172: char-aligned have no use in the standard and therefore will not be
                   1173: created.
                   1174: 
1.43      anton    1175: @cindex @code{CREATE} and alignment
1.1       anton    1176: The standard guarantees that addresses returned by @code{CREATE}d words
1.17      anton    1177: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1       anton    1178: are aligned for all purposes.
                   1179: 
1.9       anton    1180: Note that the standard defines a word @code{char}, which has nothing to
                   1181: do with address arithmetic.
                   1182: 
1.1       anton    1183: doc-chars
                   1184: doc-char+
                   1185: doc-cells
                   1186: doc-cell+
1.43      anton    1187: doc-cell
1.1       anton    1188: doc-align
                   1189: doc-aligned
                   1190: doc-floats
                   1191: doc-float+
1.43      anton    1192: doc-float
1.1       anton    1193: doc-falign
                   1194: doc-faligned
                   1195: doc-sfloats
                   1196: doc-sfloat+
                   1197: doc-sfalign
                   1198: doc-sfaligned
                   1199: doc-dfloats
                   1200: doc-dfloat+
                   1201: doc-dfalign
                   1202: doc-dfaligned
1.10      anton    1203: doc-maxalign
                   1204: doc-maxaligned
                   1205: doc-cfalign
                   1206: doc-cfaligned
1.1       anton    1207: doc-address-unit-bits
                   1208: 
1.43      anton    1209: @node Memory Blocks,  , Address arithmetic, Memory
                   1210: @subsection Memory Blocks
                   1211: @cindex memory block words
1.1       anton    1212: 
                   1213: doc-move
                   1214: doc-erase
                   1215: 
                   1216: While the previous words work on address units, the rest works on
                   1217: characters.
                   1218: 
                   1219: doc-cmove
                   1220: doc-cmove>
                   1221: doc-fill
                   1222: doc-blank
                   1223: 
1.43      anton    1224: @node Control Structures, Locals, Memory, Words
1.1       anton    1225: @section Control Structures
1.43      anton    1226: @cindex control structures
1.1       anton    1227: 
                   1228: Control structures in Forth cannot be used in interpret state, only in
1.43      anton    1229: compile state@footnote{More precisely, they have no interpretation
                   1230: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
                   1231: a colon definition. We do not like this limitation, but have not seen a
                   1232: satisfying way around it yet, although many schemes have been proposed.
1.1       anton    1233: 
1.4       anton    1234: @menu
                   1235: * Selection::                   
                   1236: * Simple Loops::                
                   1237: * Counted Loops::               
                   1238: * Arbitrary control structures::  
                   1239: * Calls and returns::           
                   1240: * Exception Handling::          
                   1241: @end menu
                   1242: 
                   1243: @node Selection, Simple Loops, Control Structures, Control Structures
1.1       anton    1244: @subsection Selection
1.43      anton    1245: @cindex selection control structures
                   1246: @cindex control structures for selection
1.1       anton    1247: 
1.43      anton    1248: @cindex @code{IF} control structure
1.1       anton    1249: @example
                   1250: @var{flag}
                   1251: IF
                   1252:   @var{code}
                   1253: ENDIF
                   1254: @end example
                   1255: or
                   1256: @example
                   1257: @var{flag}
                   1258: IF
                   1259:   @var{code1}
                   1260: ELSE
                   1261:   @var{code2}
                   1262: ENDIF
                   1263: @end example
                   1264: 
1.4       anton    1265: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1       anton    1266: standard, and @code{ENDIF} is not, although it is quite popular. We
                   1267: recommend using @code{ENDIF}, because it is less confusing for people
                   1268: who also know other languages (and is not prone to reinforcing negative
                   1269: prejudices against Forth in these people). Adding @code{ENDIF} to a
                   1270: system that only supplies @code{THEN} is simple:
                   1271: @example
                   1272: : endif   POSTPONE then ; immediate
                   1273: @end example
                   1274: 
                   1275: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
                   1276: (adv.)}  has the following meanings:
                   1277: @quotation
                   1278: ... 2b: following next after in order ... 3d: as a necessary consequence
                   1279: (if you were there, then you saw them).
                   1280: @end quotation
                   1281: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
                   1282: and many other programming languages has the meaning 3d.]
                   1283: 
1.31      anton    1284: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
                   1285: you can avoid using @code{?dup}. Using these alternatives is also more
                   1286: efficient than using @code{?dup}. Definitions in plain standard Forth
                   1287: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
                   1288: @file{compat/control.fs}.
1.1       anton    1289: 
1.43      anton    1290: @cindex @code{CASE} control structure
1.1       anton    1291: @example
                   1292: @var{n}
                   1293: CASE
                   1294:   @var{n1} OF @var{code1} ENDOF
                   1295:   @var{n2} OF @var{code2} ENDOF
1.4       anton    1296:   @dots{}
1.1       anton    1297: ENDCASE
                   1298: @end example
                   1299: 
                   1300: Executes the first @var{codei}, where the @var{ni} is equal to
                   1301: @var{n}. A default case can be added by simply writing the code after
                   1302: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
                   1303: but must not consume it.
                   1304: 
1.4       anton    1305: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1       anton    1306: @subsection Simple Loops
1.43      anton    1307: @cindex simple loops
                   1308: @cindex loops without count 
1.1       anton    1309: 
1.43      anton    1310: @cindex @code{WHILE} loop
1.1       anton    1311: @example
                   1312: BEGIN
                   1313:   @var{code1}
                   1314:   @var{flag}
                   1315: WHILE
                   1316:   @var{code2}
                   1317: REPEAT
                   1318: @end example
                   1319: 
                   1320: @var{code1} is executed and @var{flag} is computed. If it is true,
1.43      anton    1321: @var{code2} is executed and the loop is restarted; If @var{flag} is
                   1322: false, execution continues after the @code{REPEAT}.
1.1       anton    1323: 
1.43      anton    1324: @cindex @code{UNTIL} loop
1.1       anton    1325: @example
                   1326: BEGIN
                   1327:   @var{code}
                   1328:   @var{flag}
                   1329: UNTIL
                   1330: @end example
                   1331: 
                   1332: @var{code} is executed. The loop is restarted if @code{flag} is false.
                   1333: 
1.43      anton    1334: @cindex endless loop
                   1335: @cindex loops, endless
1.1       anton    1336: @example
                   1337: BEGIN
                   1338:   @var{code}
                   1339: AGAIN
                   1340: @end example
                   1341: 
                   1342: This is an endless loop.
                   1343: 
1.4       anton    1344: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1       anton    1345: @subsection Counted Loops
1.43      anton    1346: @cindex counted loops
                   1347: @cindex loops, counted
                   1348: @cindex @code{DO} loops
1.1       anton    1349: 
                   1350: The basic counted loop is:
                   1351: @example
                   1352: @var{limit} @var{start}
                   1353: ?DO
                   1354:   @var{body}
                   1355: LOOP
                   1356: @end example
                   1357: 
                   1358: This performs one iteration for every integer, starting from @var{start}
                   1359: and up to, but excluding @var{limit}. The counter, aka index, can be
                   1360: accessed with @code{i}. E.g., the loop
                   1361: @example
                   1362: 10 0 ?DO
                   1363:   i .
                   1364: LOOP
                   1365: @end example
                   1366: prints
                   1367: @example
                   1368: 0 1 2 3 4 5 6 7 8 9
                   1369: @end example
                   1370: The index of the innermost loop can be accessed with @code{i}, the index
                   1371: of the next loop with @code{j}, and the index of the third loop with
                   1372: @code{k}.
                   1373: 
1.46    ! anton    1374: doc-i
        !          1375: doc-j
        !          1376: doc-k
        !          1377: 
1.1       anton    1378: The loop control data are kept on the return stack, so there are some
                   1379: restrictions on mixing return stack accesses and counted loop
                   1380: words. E.g., if you put values on the return stack outside the loop, you
                   1381: cannot read them inside the loop. If you put values on the return stack
                   1382: within a loop, you have to remove them before the end of the loop and
                   1383: before accessing the index of the loop.
                   1384: 
                   1385: There are several variations on the counted loop:
                   1386: 
                   1387: @code{LEAVE} leaves the innermost counted loop immediately.
                   1388: 
1.18      anton    1389: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
                   1390: (and @code{LOOP} iterates until they become equal by wrap-around
                   1391: arithmetic). This behaviour is usually not what you want. Therefore,
                   1392: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
                   1393: @code{?DO}), which do not enter the loop if @var{start} is greater than
                   1394: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.30      anton    1395: unsigned loop parameters.
1.18      anton    1396: 
1.1       anton    1397: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
                   1398: index by @var{n} instead of by 1. The loop is terminated when the border
                   1399: between @var{limit-1} and @var{limit} is crossed. E.g.:
                   1400: 
1.18      anton    1401: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
1.1       anton    1402: 
1.18      anton    1403: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
1.1       anton    1404: 
1.43      anton    1405: @cindex negative increment for counted loops
                   1406: @cindex counted loops with negative increment
1.1       anton    1407: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
                   1408: 
1.2       anton    1409: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
1.1       anton    1410: 
1.2       anton    1411: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
1.1       anton    1412: 
1.18      anton    1413: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
                   1414: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
                   1415: index by @var{u} each iteration. The loop is terminated when the border
                   1416: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
                   1417: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1       anton    1418: 
1.18      anton    1419: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
1.1       anton    1420: 
1.18      anton    1421: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
1.1       anton    1422: 
1.18      anton    1423: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
1.1       anton    1424: 
1.30      anton    1425: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
                   1426: @code{-LOOP} are not in the ANS Forth standard. However, an
                   1427: implementation for these words that uses only standard words is provided
                   1428: in @file{compat/loops.fs}.
1.18      anton    1429: 
                   1430: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
                   1431: the loop, independent of the loop parameters. Do not use @code{DO}, even
                   1432: if you know that the loop is entered in any case. Such knowledge tends
                   1433: to become invalid during maintenance of a program, and then the
                   1434: @code{DO} will make trouble.
1.1       anton    1435: 
                   1436: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
                   1437: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
                   1438: return stack so @code{EXIT} can get to its return address.
                   1439: 
1.43      anton    1440: @cindex @code{FOR} loops
1.1       anton    1441: Another counted loop is
                   1442: @example
                   1443: @var{n}
                   1444: FOR
                   1445:   @var{body}
                   1446: NEXT
                   1447: @end example
                   1448: This is the preferred loop of native code compiler writers who are too
1.17      anton    1449: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1       anton    1450: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
                   1451: and ending with 0. Other Forth systems may behave differently, even if
1.30      anton    1452: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
                   1453: loops.
1.1       anton    1454: 
1.4       anton    1455: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2       anton    1456: @subsection Arbitrary control structures
1.43      anton    1457: @cindex control structures, user-defined
1.2       anton    1458: 
1.43      anton    1459: @cindex control-flow stack
1.2       anton    1460: ANS Forth permits and supports using control structures in a non-nested
                   1461: way. Information about incomplete control structures is stored on the
                   1462: control-flow stack. This stack may be implemented on the Forth data
1.17      anton    1463: stack, and this is what we have done in Gforth.
1.2       anton    1464: 
1.43      anton    1465: @cindex @code{orig}, control-flow stack item
                   1466: @cindex @code{dest}, control-flow stack item
1.2       anton    1467: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
                   1468: entry represents a backward branch target. A few words are the basis for
                   1469: building any control structure possible (except control structures that
                   1470: need storage, like calls, coroutines, and backtracking).
                   1471: 
1.3       anton    1472: doc-if
                   1473: doc-ahead
                   1474: doc-then
                   1475: doc-begin
                   1476: doc-until
                   1477: doc-again
                   1478: doc-cs-pick
                   1479: doc-cs-roll
1.2       anton    1480: 
1.17      anton    1481: On many systems control-flow stack items take one word, in Gforth they
1.2       anton    1482: currently take three (this may change in the future). Therefore it is a
                   1483: really good idea to manipulate the control flow stack with
                   1484: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
                   1485: words.
                   1486: 
                   1487: Some standard control structure words are built from these words:
                   1488: 
1.3       anton    1489: doc-else
                   1490: doc-while
                   1491: doc-repeat
1.2       anton    1492: 
1.31      anton    1493: Gforth adds some more control-structure words:
                   1494: 
                   1495: doc-endif
                   1496: doc-?dup-if
                   1497: doc-?dup-0=-if
                   1498: 
1.2       anton    1499: Counted loop words constitute a separate group of words:
                   1500: 
1.3       anton    1501: doc-?do
1.18      anton    1502: doc-+do
                   1503: doc-u+do
                   1504: doc--do
                   1505: doc-u-do
1.3       anton    1506: doc-do
                   1507: doc-for
                   1508: doc-loop
                   1509: doc-+loop
1.18      anton    1510: doc--loop
1.3       anton    1511: doc-next
                   1512: doc-leave
                   1513: doc-?leave
                   1514: doc-unloop
1.10      anton    1515: doc-done
1.2       anton    1516: 
                   1517: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
                   1518: @i{do-sys}. Our system allows it, but it's your job to ensure that for
                   1519: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3       anton    1520: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
                   1521: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7       pazsan   1522: resolved (by using one of the loop-ending words or @code{DONE}).
1.2       anton    1523: 
                   1524: Another group of control structure words are
                   1525: 
1.3       anton    1526: doc-case
                   1527: doc-endcase
                   1528: doc-of
                   1529: doc-endof
1.2       anton    1530: 
                   1531: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
                   1532: @code{cs-roll}.
                   1533: 
1.3       anton    1534: @subsubsection Programming Style
                   1535: 
                   1536: In order to ensure readability we recommend that you do not create
                   1537: arbitrary control structures directly, but define new control structure
                   1538: words for the control structure you want and use these words in your
                   1539: program.
                   1540: 
                   1541: E.g., instead of writing
                   1542: 
                   1543: @example
                   1544: begin
                   1545:   ...
                   1546: if [ 1 cs-roll ]
                   1547:   ...
                   1548: again then
                   1549: @end example
                   1550: 
                   1551: we recommend defining control structure words, e.g.,
                   1552: 
                   1553: @example
                   1554: : while ( dest -- orig dest )
                   1555:  POSTPONE if
                   1556:  1 cs-roll ; immediate
                   1557: 
                   1558: : repeat ( orig dest -- )
                   1559:  POSTPONE again
                   1560:  POSTPONE then ; immediate
                   1561: @end example
                   1562: 
                   1563: and then using these to create the control structure:
                   1564: 
                   1565: @example
                   1566: begin
                   1567:   ...
                   1568: while
                   1569:   ...
                   1570: repeat
                   1571: @end example
                   1572: 
1.30      anton    1573: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1.3       anton    1574: @code{WHILE} are predefined, so in this example it would not be
                   1575: necessary to define them.
                   1576: 
1.4       anton    1577: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3       anton    1578: @subsection Calls and returns
1.43      anton    1579: @cindex calling a definition
                   1580: @cindex returning from a definition
1.3       anton    1581: 
                   1582: A definition can be called simply be writing the name of the
1.17      anton    1583: definition. When the end of the definition is reached, it returns. An
                   1584: earlier return can be forced using
1.3       anton    1585: 
                   1586: doc-exit
                   1587: 
                   1588: Don't forget to clean up the return stack and @code{UNLOOP} any
                   1589: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
                   1590: primitive compiled by @code{EXIT} is
                   1591: 
                   1592: doc-;s
                   1593: 
1.4       anton    1594: @node Exception Handling,  , Calls and returns, Control Structures
1.3       anton    1595: @subsection Exception Handling
1.43      anton    1596: @cindex Exceptions
1.3       anton    1597: 
                   1598: doc-catch
                   1599: doc-throw
                   1600: 
1.4       anton    1601: @node Locals, Defining Words, Control Structures, Words
1.1       anton    1602: @section Locals
1.43      anton    1603: @cindex locals
1.1       anton    1604: 
1.2       anton    1605: Local variables can make Forth programming more enjoyable and Forth
                   1606: programs easier to read. Unfortunately, the locals of ANS Forth are
                   1607: laden with restrictions. Therefore, we provide not only the ANS Forth
                   1608: locals wordset, but also our own, more powerful locals wordset (we
                   1609: implemented the ANS Forth locals wordset through our locals wordset).
                   1610: 
1.24      anton    1611: The ideas in this section have also been published in the paper
                   1612: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
                   1613: at EuroForth '94; it is available at
                   1614: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
                   1615: 
1.2       anton    1616: @menu
1.17      anton    1617: * Gforth locals::               
1.4       anton    1618: * ANS Forth locals::            
1.2       anton    1619: @end menu
                   1620: 
1.17      anton    1621: @node Gforth locals, ANS Forth locals, Locals, Locals
                   1622: @subsection Gforth locals
1.43      anton    1623: @cindex Gforth locals
                   1624: @cindex locals, Gforth style
1.2       anton    1625: 
                   1626: Locals can be defined with
                   1627: 
                   1628: @example
                   1629: @{ local1 local2 ... -- comment @}
                   1630: @end example
                   1631: or
                   1632: @example
                   1633: @{ local1 local2 ... @}
                   1634: @end example
                   1635: 
                   1636: E.g.,
                   1637: @example
                   1638: : max @{ n1 n2 -- n3 @}
                   1639:  n1 n2 > if
                   1640:    n1
                   1641:  else
                   1642:    n2
                   1643:  endif ;
                   1644: @end example
                   1645: 
                   1646: The similarity of locals definitions with stack comments is intended. A
                   1647: locals definition often replaces the stack comment of a word. The order
                   1648: of the locals corresponds to the order in a stack comment and everything
                   1649: after the @code{--} is really a comment.
                   1650: 
                   1651: This similarity has one disadvantage: It is too easy to confuse locals
                   1652: declarations with stack comments, causing bugs and making them hard to
                   1653: find. However, this problem can be avoided by appropriate coding
                   1654: conventions: Do not use both notations in the same program. If you do,
                   1655: they should be distinguished using additional means, e.g. by position.
                   1656: 
1.43      anton    1657: @cindex types of locals
                   1658: @cindex locals types
1.2       anton    1659: The name of the local may be preceded by a type specifier, e.g.,
                   1660: @code{F:} for a floating point value:
                   1661: 
                   1662: @example
                   1663: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
                   1664: \ complex multiplication
                   1665:  Ar Br f* Ai Bi f* f-
                   1666:  Ar Bi f* Ai Br f* f+ ;
                   1667: @end example
                   1668: 
1.43      anton    1669: @cindex flavours of locals
                   1670: @cindex locals flavours
                   1671: @cindex value-flavoured locals
                   1672: @cindex variable-flavoured locals
1.17      anton    1673: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2       anton    1674: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
                   1675: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
                   1676: with @code{W:}, @code{D:} etc.) produces its value and can be changed
                   1677: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
                   1678: produces its address (which becomes invalid when the variable's scope is
1.41      anton    1679: left). E.g., the standard word @code{emit} can be defined in terms of
1.2       anton    1680: @code{type} like this:
                   1681: 
                   1682: @example
                   1683: : emit @{ C^ char* -- @}
                   1684:     char* 1 type ;
                   1685: @end example
                   1686: 
1.43      anton    1687: @cindex default type of locals
                   1688: @cindex locals, default type
1.2       anton    1689: A local without type specifier is a @code{W:} local. Both flavours of
                   1690: locals are initialized with values from the data or FP stack.
                   1691: 
                   1692: Currently there is no way to define locals with user-defined data
                   1693: structures, but we are working on it.
                   1694: 
1.17      anton    1695: Gforth allows defining locals everywhere in a colon definition. This
1.7       pazsan   1696: poses the following questions:
1.2       anton    1697: 
1.4       anton    1698: @menu
                   1699: * Where are locals visible by name?::  
1.14      anton    1700: * How long do locals live?::    
1.4       anton    1701: * Programming Style::           
                   1702: * Implementation::              
                   1703: @end menu
                   1704: 
1.17      anton    1705: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2       anton    1706: @subsubsection Where are locals visible by name?
1.43      anton    1707: @cindex locals visibility
                   1708: @cindex visibility of locals
                   1709: @cindex scope of locals
1.2       anton    1710: 
                   1711: Basically, the answer is that locals are visible where you would expect
                   1712: it in block-structured languages, and sometimes a little longer. If you
                   1713: want to restrict the scope of a local, enclose its definition in
                   1714: @code{SCOPE}...@code{ENDSCOPE}.
                   1715: 
                   1716: doc-scope
                   1717: doc-endscope
                   1718: 
                   1719: These words behave like control structure words, so you can use them
                   1720: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
                   1721: arbitrary ways.
                   1722: 
                   1723: If you want a more exact answer to the visibility question, here's the
                   1724: basic principle: A local is visible in all places that can only be
                   1725: reached through the definition of the local@footnote{In compiler
                   1726: construction terminology, all places dominated by the definition of the
                   1727: local.}. In other words, it is not visible in places that can be reached
                   1728: without going through the definition of the local. E.g., locals defined
                   1729: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
                   1730: defined in @code{BEGIN}...@code{UNTIL} are visible after the
                   1731: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
                   1732: 
                   1733: The reasoning behind this solution is: We want to have the locals
                   1734: visible as long as it is meaningful. The user can always make the
                   1735: visibility shorter by using explicit scoping. In a place that can
                   1736: only be reached through the definition of a local, the meaning of a
                   1737: local name is clear. In other places it is not: How is the local
                   1738: initialized at the control flow path that does not contain the
                   1739: definition? Which local is meant, if the same name is defined twice in
                   1740: two independent control flow paths?
                   1741: 
                   1742: This should be enough detail for nearly all users, so you can skip the
1.41      anton    1743: rest of this section. If you really must know all the gory details and
1.2       anton    1744: options, read on.
                   1745: 
                   1746: In order to implement this rule, the compiler has to know which places
                   1747: are unreachable. It knows this automatically after @code{AHEAD},
                   1748: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
                   1749: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
                   1750: compiler that the control flow never reaches that place. If
                   1751: @code{UNREACHABLE} is not used where it could, the only consequence is
                   1752: that the visibility of some locals is more limited than the rule above
                   1753: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
                   1754: lie to the compiler), buggy code will be produced.
                   1755: 
1.43      anton    1756: doc-unreachable
                   1757: 
1.2       anton    1758: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3       anton    1759: does not know which locals will be visible on the incoming
                   1760: back-edge. All problems discussed in the following are due to this
                   1761: ignorance of the compiler (we discuss the problems using @code{BEGIN}
                   1762: loops as examples; the discussion also applies to @code{?DO} and other
1.2       anton    1763: loops). Perhaps the most insidious example is:
                   1764: @example
                   1765: AHEAD
                   1766: BEGIN
                   1767:   x
                   1768: [ 1 CS-ROLL ] THEN
1.4       anton    1769:   @{ x @}
1.2       anton    1770:   ...
                   1771: UNTIL
                   1772: @end example
                   1773: 
                   1774: This should be legal according to the visibility rule. The use of
                   1775: @code{x} can only be reached through the definition; but that appears
                   1776: textually below the use.
                   1777: 
                   1778: From this example it is clear that the visibility rules cannot be fully
                   1779: implemented without major headaches. Our implementation treats common
                   1780: cases as advertised and the exceptions are treated in a safe way: The
                   1781: compiler makes a reasonable guess about the locals visible after a
                   1782: @code{BEGIN}; if it is too pessimistic, the
                   1783: user will get a spurious error about the local not being defined; if the
                   1784: compiler is too optimistic, it will notice this later and issue a
                   1785: warning. In the case above the compiler would complain about @code{x}
                   1786: being undefined at its use. You can see from the obscure examples in
                   1787: this section that it takes quite unusual control structures to get the
                   1788: compiler into trouble, and even then it will often do fine.
                   1789: 
                   1790: If the @code{BEGIN} is reachable from above, the most optimistic guess
                   1791: is that all locals visible before the @code{BEGIN} will also be
                   1792: visible after the @code{BEGIN}. This guess is valid for all loops that
                   1793: are entered only through the @code{BEGIN}, in particular, for normal
                   1794: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
                   1795: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
                   1796: compiler. When the branch to the @code{BEGIN} is finally generated by
                   1797: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1.41      anton    1798: warns the user if it was too optimistic:
1.2       anton    1799: @example
                   1800: IF
1.4       anton    1801:   @{ x @}
1.2       anton    1802: BEGIN
                   1803:   \ x ? 
                   1804: [ 1 cs-roll ] THEN
                   1805:   ...
                   1806: UNTIL
                   1807: @end example
                   1808: 
                   1809: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
                   1810: optimistically assumes that it lives until the @code{THEN}. It notices
                   1811: this difference when it compiles the @code{UNTIL} and issues a
                   1812: warning. The user can avoid the warning, and make sure that @code{x}
                   1813: is not used in the wrong area by using explicit scoping:
                   1814: @example
                   1815: IF
                   1816:   SCOPE
1.4       anton    1817:   @{ x @}
1.2       anton    1818:   ENDSCOPE
                   1819: BEGIN
                   1820: [ 1 cs-roll ] THEN
                   1821:   ...
                   1822: UNTIL
                   1823: @end example
                   1824: 
                   1825: Since the guess is optimistic, there will be no spurious error messages
                   1826: about undefined locals.
                   1827: 
                   1828: If the @code{BEGIN} is not reachable from above (e.g., after
                   1829: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
                   1830: optimistic guess, as the locals visible after the @code{BEGIN} may be
                   1831: defined later. Therefore, the compiler assumes that no locals are
1.17      anton    1832: visible after the @code{BEGIN}. However, the user can use
1.2       anton    1833: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17      anton    1834: visible at the BEGIN as at the point where the top control-flow stack
                   1835: item was created.
1.2       anton    1836: 
                   1837: doc-assume-live
                   1838: 
                   1839: E.g.,
                   1840: @example
1.4       anton    1841: @{ x @}
1.2       anton    1842: AHEAD
                   1843: ASSUME-LIVE
                   1844: BEGIN
                   1845:   x
                   1846: [ 1 CS-ROLL ] THEN
                   1847:   ...
                   1848: UNTIL
                   1849: @end example
                   1850: 
                   1851: Other cases where the locals are defined before the @code{BEGIN} can be
                   1852: handled by inserting an appropriate @code{CS-ROLL} before the
                   1853: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
                   1854: behind the @code{ASSUME-LIVE}).
                   1855: 
                   1856: Cases where locals are defined after the @code{BEGIN} (but should be
                   1857: visible immediately after the @code{BEGIN}) can only be handled by
                   1858: rearranging the loop. E.g., the ``most insidious'' example above can be
                   1859: arranged into:
                   1860: @example
                   1861: BEGIN
1.4       anton    1862:   @{ x @}
1.2       anton    1863:   ... 0=
                   1864: WHILE
                   1865:   x
                   1866: REPEAT
                   1867: @end example
                   1868: 
1.17      anton    1869: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2       anton    1870: @subsubsection How long do locals live?
1.43      anton    1871: @cindex locals lifetime
                   1872: @cindex lifetime of locals
1.2       anton    1873: 
                   1874: The right answer for the lifetime question would be: A local lives at
                   1875: least as long as it can be accessed. For a value-flavoured local this
                   1876: means: until the end of its visibility. However, a variable-flavoured
                   1877: local could be accessed through its address far beyond its visibility
                   1878: scope. Ultimately, this would mean that such locals would have to be
                   1879: garbage collected. Since this entails un-Forth-like implementation
                   1880: complexities, I adopted the same cowardly solution as some other
                   1881: languages (e.g., C): The local lives only as long as it is visible;
                   1882: afterwards its address is invalid (and programs that access it
                   1883: afterwards are erroneous).
                   1884: 
1.17      anton    1885: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2       anton    1886: @subsubsection Programming Style
1.43      anton    1887: @cindex locals programming style
                   1888: @cindex programming style, locals
1.2       anton    1889: 
                   1890: The freedom to define locals anywhere has the potential to change
                   1891: programming styles dramatically. In particular, the need to use the
                   1892: return stack for intermediate storage vanishes. Moreover, all stack
                   1893: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
                   1894: determined arguments) can be eliminated: If the stack items are in the
                   1895: wrong order, just write a locals definition for all of them; then
                   1896: write the items in the order you want.
                   1897: 
                   1898: This seems a little far-fetched and eliminating stack manipulations is
1.4       anton    1899: unlikely to become a conscious programming objective. Still, the number
                   1900: of stack manipulations will be reduced dramatically if local variables
1.17      anton    1901: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4       anton    1902: a traditional implementation of @code{max}).
1.2       anton    1903: 
                   1904: This shows one potential benefit of locals: making Forth programs more
                   1905: readable. Of course, this benefit will only be realized if the
                   1906: programmers continue to honour the principle of factoring instead of
                   1907: using the added latitude to make the words longer.
                   1908: 
1.43      anton    1909: @cindex single-assignment style for locals
1.2       anton    1910: Using @code{TO} can and should be avoided.  Without @code{TO},
                   1911: every value-flavoured local has only a single assignment and many
                   1912: advantages of functional languages apply to Forth. I.e., programs are
                   1913: easier to analyse, to optimize and to read: It is clear from the
                   1914: definition what the local stands for, it does not turn into something
                   1915: different later.
                   1916: 
                   1917: E.g., a definition using @code{TO} might look like this:
                   1918: @example
                   1919: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1920:  u1 u2 min 0
                   1921:  ?do
1.36      anton    1922:    addr1 c@@ addr2 c@@ -
1.31      anton    1923:    ?dup-if
1.2       anton    1924:      unloop exit
                   1925:    then
                   1926:    addr1 char+ TO addr1
                   1927:    addr2 char+ TO addr2
                   1928:  loop
                   1929:  u1 u2 - ;
                   1930: @end example
                   1931: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
                   1932: every loop iteration. @code{strcmp} is a typical example of the
                   1933: readability problems of using @code{TO}. When you start reading
                   1934: @code{strcmp}, you think that @code{addr1} refers to the start of the
                   1935: string. Only near the end of the loop you realize that it is something
                   1936: else.
                   1937: 
                   1938: This can be avoided by defining two locals at the start of the loop that
                   1939: are initialized with the right value for the current iteration.
                   1940: @example
                   1941: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1942:  addr1 addr2
                   1943:  u1 u2 min 0 
                   1944:  ?do @{ s1 s2 @}
1.36      anton    1945:    s1 c@@ s2 c@@ -
1.31      anton    1946:    ?dup-if
1.2       anton    1947:      unloop exit
                   1948:    then
                   1949:    s1 char+ s2 char+
                   1950:  loop
                   1951:  2drop
                   1952:  u1 u2 - ;
                   1953: @end example
                   1954: Here it is clear from the start that @code{s1} has a different value
                   1955: in every loop iteration.
                   1956: 
1.17      anton    1957: @node Implementation,  , Programming Style, Gforth locals
1.2       anton    1958: @subsubsection Implementation
1.43      anton    1959: @cindex locals implementation
                   1960: @cindex implementation of locals
1.2       anton    1961: 
1.43      anton    1962: @cindex locals stack
1.17      anton    1963: Gforth uses an extra locals stack. The most compelling reason for
1.2       anton    1964: this is that the return stack is not float-aligned; using an extra stack
                   1965: also eliminates the problems and restrictions of using the return stack
                   1966: as locals stack. Like the other stacks, the locals stack grows toward
                   1967: lower addresses. A few primitives allow an efficient implementation:
                   1968: 
                   1969: doc-@local#
                   1970: doc-f@local#
                   1971: doc-laddr#
                   1972: doc-lp+!#
                   1973: doc-lp!
                   1974: doc->l
                   1975: doc-f>l
                   1976: 
                   1977: In addition to these primitives, some specializations of these
                   1978: primitives for commonly occurring inline arguments are provided for
                   1979: efficiency reasons, e.g., @code{@@local0} as specialization of
                   1980: @code{@@local#} for the inline argument 0. The following compiling words
                   1981: compile the right specialized version, or the general version, as
                   1982: appropriate:
                   1983: 
1.12      anton    1984: doc-compile-@local
                   1985: doc-compile-f@local
1.2       anton    1986: doc-compile-lp+!
                   1987: 
                   1988: Combinations of conditional branches and @code{lp+!#} like
                   1989: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
                   1990: is taken) are provided for efficiency and correctness in loops.
                   1991: 
                   1992: A special area in the dictionary space is reserved for keeping the
                   1993: local variable names. @code{@{} switches the dictionary pointer to this
                   1994: area and @code{@}} switches it back and generates the locals
                   1995: initializing code. @code{W:} etc.@ are normal defining words. This
                   1996: special area is cleared at the start of every colon definition.
                   1997: 
1.43      anton    1998: @cindex wordlist for defining locals
1.17      anton    1999: A special feature of Gforth's dictionary is used to implement the
1.2       anton    2000: definition of locals without type specifiers: every wordlist (aka
                   2001: vocabulary) has its own methods for searching
1.4       anton    2002: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2       anton    2003: with a special search method: When it is searched for a word, it
                   2004: actually creates that word using @code{W:}. @code{@{} changes the search
                   2005: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
                   2006: and then the wordlist for defining locals without type specifiers.
                   2007: 
                   2008: The lifetime rules support a stack discipline within a colon
                   2009: definition: The lifetime of a local is either nested with other locals
                   2010: lifetimes or it does not overlap them.
                   2011: 
                   2012: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
                   2013: pointer manipulation is generated. Between control structure words
                   2014: locals definitions can push locals onto the locals stack. @code{AGAIN}
                   2015: is the simplest of the other three control flow words. It has to
                   2016: restore the locals stack depth of the corresponding @code{BEGIN}
                   2017: before branching. The code looks like this:
                   2018: @format
                   2019: @code{lp+!#} current-locals-size @minus{} dest-locals-size
                   2020: @code{branch} <begin>
                   2021: @end format
                   2022: 
                   2023: @code{UNTIL} is a little more complicated: If it branches back, it
                   2024: must adjust the stack just like @code{AGAIN}. But if it falls through,
                   2025: the locals stack must not be changed. The compiler generates the
                   2026: following code:
                   2027: @format
                   2028: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
                   2029: @end format
                   2030: The locals stack pointer is only adjusted if the branch is taken.
                   2031: 
                   2032: @code{THEN} can produce somewhat inefficient code:
                   2033: @format
                   2034: @code{lp+!#} current-locals-size @minus{} orig-locals-size
                   2035: <orig target>:
                   2036: @code{lp+!#} orig-locals-size @minus{} new-locals-size
                   2037: @end format
                   2038: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4       anton    2039: level at the @var{orig} point to the level after the @code{THEN}. The
1.2       anton    2040: first @code{lp+!#} adjusts the locals stack pointer from the current
                   2041: level to the level at the orig point, so the complete effect is an
                   2042: adjustment from the current level to the right level after the
                   2043: @code{THEN}.
                   2044: 
1.43      anton    2045: @cindex locals information on the control-flow stack
                   2046: @cindex control-flow stack items, locals information
1.2       anton    2047: In a conventional Forth implementation a dest control-flow stack entry
                   2048: is just the target address and an orig entry is just the address to be
                   2049: patched. Our locals implementation adds a wordlist to every orig or dest
                   2050: item. It is the list of locals visible (or assumed visible) at the point
                   2051: described by the entry. Our implementation also adds a tag to identify
                   2052: the kind of entry, in particular to differentiate between live and dead
                   2053: (reachable and unreachable) orig entries.
                   2054: 
                   2055: A few unusual operations have to be performed on locals wordlists:
                   2056: 
                   2057: doc-common-list
                   2058: doc-sub-list?
                   2059: doc-list-size
                   2060: 
                   2061: Several features of our locals wordlist implementation make these
                   2062: operations easy to implement: The locals wordlists are organised as
                   2063: linked lists; the tails of these lists are shared, if the lists
                   2064: contain some of the same locals; and the address of a name is greater
                   2065: than the address of the names behind it in the list.
                   2066: 
                   2067: Another important implementation detail is the variable
                   2068: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
                   2069: determine if they can be reached directly or only through the branch
                   2070: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
                   2071: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
                   2072: definition, by @code{BEGIN} and usually by @code{THEN}.
                   2073: 
                   2074: Counted loops are similar to other loops in most respects, but
                   2075: @code{LEAVE} requires special attention: It performs basically the same
                   2076: service as @code{AHEAD}, but it does not create a control-flow stack
                   2077: entry. Therefore the information has to be stored elsewhere;
                   2078: traditionally, the information was stored in the target fields of the
                   2079: branches created by the @code{LEAVE}s, by organizing these fields into a
                   2080: linked list. Unfortunately, this clever trick does not provide enough
                   2081: space for storing our extended control flow information. Therefore, we
                   2082: introduce another stack, the leave stack. It contains the control-flow
                   2083: stack entries for all unresolved @code{LEAVE}s.
                   2084: 
                   2085: Local names are kept until the end of the colon definition, even if
                   2086: they are no longer visible in any control-flow path. In a few cases
                   2087: this may lead to increased space needs for the locals name area, but
                   2088: usually less than reclaiming this space would cost in code size.
                   2089: 
                   2090: 
1.17      anton    2091: @node ANS Forth locals,  , Gforth locals, Locals
1.2       anton    2092: @subsection ANS Forth locals
1.43      anton    2093: @cindex locals, ANS Forth style
1.2       anton    2094: 
                   2095: The ANS Forth locals wordset does not define a syntax for locals, but
                   2096: words that make it possible to define various syntaxes. One of the
1.17      anton    2097: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2       anton    2098: wordset, i.e.:
                   2099: 
                   2100: @example
                   2101: @{ local1 local2 ... -- comment @}
                   2102: @end example
                   2103: or
                   2104: @example
                   2105: @{ local1 local2 ... @}
                   2106: @end example
                   2107: 
                   2108: The order of the locals corresponds to the order in a stack comment. The
                   2109: restrictions are:
1.1       anton    2110: 
1.2       anton    2111: @itemize @bullet
                   2112: @item
1.17      anton    2113: Locals can only be cell-sized values (no type specifiers are allowed).
1.2       anton    2114: @item
                   2115: Locals can be defined only outside control structures.
                   2116: @item
                   2117: Locals can interfere with explicit usage of the return stack. For the
                   2118: exact (and long) rules, see the standard. If you don't use return stack
1.17      anton    2119: accessing words in a definition using locals, you will be all right. The
1.2       anton    2120: purpose of this rule is to make locals implementation on the return
                   2121: stack easier.
                   2122: @item
                   2123: The whole definition must be in one line.
                   2124: @end itemize
                   2125: 
1.35      anton    2126: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
                   2127: Defining Words}). I.e., they are initialized from the stack. Using their
1.2       anton    2128: name produces their value. Their value can be changed using @code{TO}.
                   2129: 
1.17      anton    2130: Since this syntax is supported by Gforth directly, you need not do
1.2       anton    2131: anything to use it. If you want to port a program using this syntax to
1.30      anton    2132: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
                   2133: syntax on the other system.
1.2       anton    2134: 
                   2135: Note that a syntax shown in the standard, section A.13 looks
                   2136: similar, but is quite different in having the order of locals
                   2137: reversed. Beware!
                   2138: 
                   2139: The ANS Forth locals wordset itself consists of the following word
                   2140: 
                   2141: doc-(local)
                   2142: 
                   2143: The ANS Forth locals extension wordset defines a syntax, but it is so
                   2144: awful that we strongly recommend not to use it. We have implemented this
1.17      anton    2145: syntax to make porting to Gforth easy, but do not document it here. The
1.2       anton    2146: problem with this syntax is that the locals are defined in an order
                   2147: reversed with respect to the standard stack comment notation, making
                   2148: programs harder to read, and easier to misread and miswrite. The only
                   2149: merit of this syntax is that it is easy to implement using the ANS Forth
                   2150: locals wordset.
1.3       anton    2151: 
1.37      anton    2152: @node Defining Words, Tokens for Words, Locals, Words
1.4       anton    2153: @section Defining Words
1.43      anton    2154: @cindex defining words
1.4       anton    2155: 
1.14      anton    2156: @menu
1.35      anton    2157: * Simple Defining Words::       
                   2158: * Colon Definitions::           
                   2159: * User-defined Defining Words::  
                   2160: * Supplying names::             
                   2161: * Interpretation and Compilation Semantics::  
1.14      anton    2162: @end menu
                   2163: 
1.35      anton    2164: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
                   2165: @subsection Simple Defining Words
1.43      anton    2166: @cindex simple defining words
                   2167: @cindex defining words, simple
1.35      anton    2168: 
                   2169: doc-constant
                   2170: doc-2constant
                   2171: doc-fconstant
                   2172: doc-variable
                   2173: doc-2variable
                   2174: doc-fvariable
                   2175: doc-create
                   2176: doc-user
                   2177: doc-value
                   2178: doc-to
                   2179: doc-defer
                   2180: doc-is
                   2181: 
                   2182: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
                   2183: @subsection Colon Definitions
1.43      anton    2184: @cindex colon definitions
1.35      anton    2185: 
                   2186: @example
                   2187: : name ( ... -- ... )
                   2188:     word1 word2 word3 ;
                   2189: @end example
                   2190: 
                   2191: creates a word called @code{name}, that, upon execution, executes
                   2192: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
                   2193: 
                   2194: The explanation above is somewhat superficial. @xref{Interpretation and
                   2195: Compilation Semantics} for an in-depth discussion of some of the issues
                   2196: involved.
                   2197: 
                   2198: doc-:
                   2199: doc-;
                   2200: 
                   2201: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
                   2202: @subsection User-defined Defining Words
1.43      anton    2203: @cindex user-defined defining words
                   2204: @cindex defining words, user-defined
1.35      anton    2205: 
                   2206: You can create new defining words simply by wrapping defining-time code
                   2207: around existing defining words and putting the sequence in a colon
                   2208: definition.
                   2209: 
1.43      anton    2210: @cindex @code{CREATE} ... @code{DOES>}
1.36      anton    2211: If you want the words defined with your defining words to behave
                   2212: differently from words defined with standard defining words, you can
1.35      anton    2213: write your defining word like this:
                   2214: 
                   2215: @example
                   2216: : def-word ( "name" -- )
                   2217:     Create @var{code1}
                   2218: DOES> ( ... -- ... )
                   2219:     @var{code2} ;
                   2220: 
                   2221: def-word name
                   2222: @end example
                   2223: 
                   2224: Technically, this fragment defines a defining word @code{def-word}, and
                   2225: a word @code{name}; when you execute @code{name}, the address of the
                   2226: body of @code{name} is put on the data stack and @var{code2} is executed
                   2227: (the address of the body of @code{name} is the address @code{HERE}
1.36      anton    2228: returns immediately after the @code{CREATE}).
                   2229: 
                   2230: In other words, if you make the following definitions:
                   2231: 
                   2232: @example
                   2233: : def-word1 ( "name" -- )
                   2234:     Create @var{code1} ;
                   2235: 
                   2236: : action1 ( ... -- ... )
                   2237:     @var{code2} ;
                   2238: 
                   2239: def-word name1
                   2240: @end example
                   2241: 
                   2242: Using @code{name1 action1} is equivalent to using @code{name}.
                   2243: 
                   2244: E.g., you can implement @code{Constant} in this way:
1.35      anton    2245: 
                   2246: @example
                   2247: : constant ( w "name" -- )
                   2248:     create ,
                   2249: DOES> ( -- w )
1.36      anton    2250:     @@ ;
1.35      anton    2251: @end example
                   2252: 
                   2253: When you create a constant with @code{5 constant five}, first a new word
                   2254: @code{five} is created, then the value 5 is laid down in the body of
                   2255: @code{five} with @code{,}. When @code{five} is invoked, the address of
                   2256: the body is put on the stack, and @code{@@} retrieves the value 5.
                   2257: 
1.43      anton    2258: @cindex stack effect of @code{DOES>}-parts
                   2259: @cindex @code{DOES>}-parts, stack effect
1.35      anton    2260: In the example above the stack comment after the @code{DOES>} specifies
                   2261: the stack effect of the defined words, not the stack effect of the
                   2262: following code (the following code expects the address of the body on
                   2263: the top of stack, which is not reflected in the stack comment). This is
                   2264: the convention that I use and recommend (it clashes a bit with using
                   2265: locals declarations for stack effect specification, though).
                   2266: 
                   2267: @subsubsection Applications of @code{CREATE..DOES>}
1.43      anton    2268: @cindex @code{CREATE} ... @code{DOES>}, applications
1.35      anton    2269: 
1.36      anton    2270: You may wonder how to use this feature. Here are some usage patterns:
1.35      anton    2271: 
1.43      anton    2272: @cindex factoring similar colon definitions
1.35      anton    2273: When you see a sequence of code occurring several times, and you can
                   2274: identify a meaning, you will factor it out as a colon definition. When
                   2275: you see similar colon definitions, you can factor them using
                   2276: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
                   2277: that look very similar:
                   2278: @example
1.41      anton    2279: : ori, ( reg-target reg-source n -- )
1.35      anton    2280:     0 asm-reg-reg-imm ;
1.41      anton    2281: : andi, ( reg-target reg-source n -- )
1.35      anton    2282:     1 asm-reg-reg-imm ;
                   2283: @end example
                   2284: 
                   2285: This could be factored with:
                   2286: @example
                   2287: : reg-reg-imm ( op-code -- )
                   2288:     create ,
1.41      anton    2289: DOES> ( reg-target reg-source n -- )
1.36      anton    2290:     @@ asm-reg-reg-imm ;
1.35      anton    2291: 
                   2292: 0 reg-reg-imm ori,
                   2293: 1 reg-reg-imm andi,
                   2294: @end example
                   2295: 
1.43      anton    2296: @cindex currying
1.35      anton    2297: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
                   2298: supply a part of the parameters for a word (known as @dfn{currying} in
                   2299: the functional language community). E.g., @code{+} needs two
                   2300: parameters. Creating versions of @code{+} with one parameter fixed can
                   2301: be done like this:
                   2302: @example
                   2303: : curry+ ( n1 -- )
                   2304:     create ,
                   2305: DOES> ( n2 -- n1+n2 )
1.36      anton    2306:     @@ + ;
1.35      anton    2307: 
                   2308:  3 curry+ 3+
                   2309: -2 curry+ 2-
                   2310: @end example
                   2311: 
                   2312: @subsubsection The gory details of @code{CREATE..DOES>}
1.43      anton    2313: @cindex @code{CREATE} ... @code{DOES>}, details
1.35      anton    2314: 
                   2315: doc-does>
                   2316: 
1.43      anton    2317: @cindex @code{DOES>} in a separate definition
1.35      anton    2318: This means that you need not use @code{CREATE} and @code{DOES>} in the
                   2319: same definition; E.g., you can put the @code{DOES>}-part in a separate
                   2320: definition. This allows us to, e.g., select among different DOES>-parts:
                   2321: @example
                   2322: : does1 
                   2323: DOES> ( ... -- ... )
                   2324:     ... ;
                   2325: 
                   2326: : does2
                   2327: DOES> ( ... -- ... )
                   2328:     ... ;
                   2329: 
                   2330: : def-word ( ... -- ... )
                   2331:     create ...
                   2332:     IF
                   2333:        does1
                   2334:     ELSE
                   2335:        does2
                   2336:     ENDIF ;
                   2337: @end example
                   2338: 
1.43      anton    2339: @cindex @code{DOES>} in interpretation state
1.35      anton    2340: In a standard program you can apply a @code{DOES>}-part only if the last
                   2341: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
                   2342: will override the behaviour of the last word defined in any case. In a
                   2343: standard program, you can use @code{DOES>} only in a colon
                   2344: definition. In Gforth, you can also use it in interpretation state, in a
                   2345: kind of one-shot mode:
                   2346: @example
                   2347: CREATE name ( ... -- ... )
                   2348:   @var{initialization}
                   2349: DOES>
                   2350:   @var{code} ;
                   2351: @end example
1.41      anton    2352: This is equivalent to the standard
1.35      anton    2353: @example
                   2354: :noname
                   2355: DOES>
                   2356:     @var{code} ;
                   2357: CREATE name EXECUTE ( ... -- ... )
                   2358:     @var{initialization}
                   2359: @end example
                   2360: 
                   2361: You can get the address of the body of a word with
                   2362: 
                   2363: doc->body
                   2364: 
                   2365: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
                   2366: @subsection Supplying names for the defined words
1.43      anton    2367: @cindex names for defined words
                   2368: @cindex defining words, name parameter
1.35      anton    2369: 
1.43      anton    2370: @cindex defining words, name given in a string
1.35      anton    2371: By default, defining words take the names for the defined words from the
                   2372: input stream. Sometimes you want to supply the name from a string. You
                   2373: can do this with
                   2374: 
                   2375: doc-nextname
                   2376: 
                   2377: E.g.,
                   2378: 
                   2379: @example
                   2380: s" foo" nextname create
                   2381: @end example
                   2382: is equivalent to
                   2383: @example
                   2384: create foo
                   2385: @end example
                   2386: 
1.43      anton    2387: @cindex defining words without name
1.35      anton    2388: Sometimes you want to define a word without a name. You can do this with
                   2389: 
                   2390: doc-noname
                   2391: 
1.43      anton    2392: @cindex execution token of last defined word
1.35      anton    2393: To make any use of the newly defined word, you need its execution
                   2394: token. You can get it with
                   2395: 
                   2396: doc-lastxt
                   2397: 
                   2398: E.g., you can initialize a deferred word with an anonymous colon
                   2399: definition:
                   2400: @example
                   2401: Defer deferred
                   2402: noname : ( ... -- ... )
                   2403:   ... ;
                   2404: lastxt IS deferred
                   2405: @end example
                   2406: 
                   2407: @code{lastxt} also works when the last word was not defined as
                   2408: @code{noname}. 
                   2409: 
                   2410: The standard has also recognized the need for anonymous words and
                   2411: provides
                   2412: 
                   2413: doc-:noname
                   2414: 
                   2415: This leaves the execution token for the word on the stack after the
                   2416: closing @code{;}. You can rewrite the last example with @code{:noname}:
                   2417: @example
                   2418: Defer deferred
                   2419: :noname ( ... -- ... )
                   2420:   ... ;
                   2421: IS deferred
                   2422: @end example
                   2423: 
                   2424: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
                   2425: @subsection Interpretation and Compilation Semantics
1.43      anton    2426: @cindex semantics, interpretation and compilation
1.35      anton    2427: 
1.43      anton    2428: @cindex interpretation semantics
1.36      anton    2429: The @dfn{interpretation semantics} of a word are what the text
                   2430: interpreter does when it encounters the word in interpret state. It also
                   2431: appears in some other contexts, e.g., the execution token returned by
                   2432: @code{' @var{word}} identifies the interpretation semantics of
                   2433: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
                   2434: interpret-state text interpretation of @code{@var{word}}).
                   2435: 
1.43      anton    2436: @cindex compilation semantics
1.36      anton    2437: The @dfn{compilation semantics} of a word are what the text interpreter
                   2438: does when it encounters the word in compile state. It also appears in
                   2439: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
                   2440: standard terminology, ``appends to the current definition''.} the
                   2441: compilation semantics of @var{word}.
                   2442: 
1.43      anton    2443: @cindex execution semantics
1.36      anton    2444: The standard also talks about @dfn{execution semantics}. They are used
                   2445: only for defining the interpretation and compilation semantics of many
                   2446: words. By default, the interpretation semantics of a word are to
                   2447: @code{execute} its execution semantics, and the compilation semantics of
                   2448: a word are to @code{compile,} its execution semantics.@footnote{In
                   2449: standard terminology: The default interpretation semantics are its
                   2450: execution semantics; the default compilation semantics are to append its
                   2451: execution semantics to the execution semantics of the current
                   2452: definition.}
                   2453: 
1.43      anton    2454: @cindex immediate words
1.36      anton    2455: You can change the compilation semantics into @code{execute}ing the
                   2456: execution semantics with
                   2457: 
1.35      anton    2458: doc-immediate
1.36      anton    2459: 
1.43      anton    2460: @cindex compile-only words
1.36      anton    2461: You can remove the interpretation semantics of a word with
                   2462: 
                   2463: doc-compile-only
                   2464: doc-restrict
                   2465: 
                   2466: Note that ticking (@code{'}) compile-only words gives an error
                   2467: (``Interpreting a compile-only word'').
                   2468: 
                   2469: Gforth also allows you to define words with arbitrary combinations of
                   2470: interpretation and compilation semantics.
                   2471: 
1.35      anton    2472: doc-interpret/compile:
                   2473: 
1.36      anton    2474: This feature was introduced for implementing @code{TO} and @code{S"}. I
                   2475: recommend that you do not define such words, as cute as they may be:
                   2476: they make it hard to get at both parts of the word in some contexts.
                   2477: E.g., assume you want to get an execution token for the compilation
                   2478: part. Instead, define two words, one that embodies the interpretation
                   2479: part, and one that embodies the compilation part.
                   2480: 
                   2481: There is, however, a potentially useful application of this feature:
                   2482: Providing differing implementations for the default semantics. While
                   2483: this introduces redundancy and is therefore usually a bad idea, a
                   2484: performance improvement may be worth the trouble. E.g., consider the
                   2485: word @code{foobar}:
                   2486: 
                   2487: @example
                   2488: : foobar
                   2489:     foo bar ;
                   2490: @end example
                   2491: 
                   2492: Let us assume that @code{foobar} is called so frequently that the
                   2493: calling overhead would take a significant amount of the run-time. We can
                   2494: optimize it with @code{interpret/compile:}:
1.35      anton    2495: 
1.36      anton    2496: @example
                   2497: :noname
                   2498:    foo bar ;
                   2499: :noname
                   2500:    POSTPONE foo POSTPONE bar ;
                   2501: interpret/compile: foobar
                   2502: @end example
                   2503: 
                   2504: This definition has the same interpretation semantics and essentially
                   2505: the same compilation semantics as the simple definition of
                   2506: @code{foobar}, but the implementation of the compilation semantics is
                   2507: more efficient with respect to run-time.
                   2508: 
1.43      anton    2509: @cindex state-smart words are a bad idea
1.36      anton    2510: Some people try to use state-smart words to emulate the feature provided
                   2511: by @code{interpret/compile:} (words are state-smart if they check
                   2512: @code{STATE} during execution). E.g., they would try to code
                   2513: @code{foobar} like this:
                   2514: 
                   2515: @example
                   2516: : foobar
                   2517:   STATE @@
                   2518:   IF ( compilation state )
                   2519:     POSTPONE foo POSTPONE bar
                   2520:   ELSE
                   2521:     foo bar
                   2522:   ENDIF ; immediate
                   2523: @end example
                   2524: 
                   2525: While this works if @code{foobar} is processed only by the text
                   2526: interpreter, it does not work in other contexts (like @code{'} or
                   2527: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
                   2528: for a state-smart word, not for the interpretation semantics of the
                   2529: original @code{foobar}; when you execute this execution token (directly
                   2530: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
                   2531: state, the result will not be what you expected (i.e., it will not
                   2532: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
                   2533: write them!
                   2534: 
1.43      anton    2535: @cindex defining words with arbitrary semantics combinations
1.36      anton    2536: It is also possible to write defining words that define words with
                   2537: arbitrary combinations of interpretation and compilation semantics (or,
                   2538: preferably, arbitrary combinations of implementations of the default
                   2539: semantics). In general, this looks like:
                   2540: 
                   2541: @example
                   2542: : def-word
                   2543:     create-interpret/compile
                   2544:     @var{code1}
                   2545: interpretation>
                   2546:     @var{code2}
                   2547: <interpretation
                   2548: compilation>
                   2549:     @var{code3}
                   2550: <compilation ;
                   2551: @end example
                   2552: 
                   2553: For a @var{word} defined with @code{def-word}, the interpretation
                   2554: semantics are to push the address of the body of @var{word} and perform
                   2555: @var{code2}, and the compilation semantics are to push the address of
                   2556: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
                   2557: can also be defined like this:
                   2558: 
                   2559: @example
                   2560: : constant ( n "name" -- )
                   2561:     create-interpret/compile
                   2562:     ,
                   2563: interpretation> ( -- n )
                   2564:     @@
                   2565: <interpretation
                   2566: compilation> ( compilation. -- ; run-time. -- n )
                   2567:     @@ postpone literal
                   2568: <compilation ;
                   2569: @end example
                   2570: 
                   2571: doc-create-interpret/compile
                   2572: doc-interpretation>
                   2573: doc-<interpretation
                   2574: doc-compilation>
                   2575: doc-<compilation
                   2576: 
                   2577: Note that words defined with @code{interpret/compile:} and
                   2578: @code{create-interpret/compile} have an extended header structure that
                   2579: differs from other words; however, unless you try to access them with
                   2580: plain address arithmetic, you should not notice this. Words for
                   2581: accessing the header structure usually know how to deal with this; e.g.,
                   2582: @code{' word >body} also gives you the body of a word created with
                   2583: @code{create-interpret/compile}.
1.4       anton    2584: 
1.37      anton    2585: @node Tokens for Words, Wordlists, Defining Words, Words
                   2586: @section Tokens for Words
1.43      anton    2587: @cindex tokens for words
1.37      anton    2588: 
                   2589: This chapter describes the creation and use of tokens that represent
                   2590: words on the stack (and in data space).
                   2591: 
                   2592: Named words have interpretation and compilation semantics. Unnamed words
                   2593: just have execution semantics.
                   2594: 
1.43      anton    2595: @cindex execution token
1.37      anton    2596: An @dfn{execution token} represents the execution semantics of an
                   2597: unnamed word. An execution token occupies one cell. As explained in
                   2598: section @ref{Supplying names}, the execution token of the last words
                   2599: defined can be produced with
                   2600: 
                   2601: short-lastxt
                   2602: 
                   2603: You can perform the semantics represented by an execution token with
                   2604: doc-execute
                   2605: You can compile the word with
                   2606: doc-compile,
                   2607: 
1.43      anton    2608: @cindex code field address
                   2609: @cindex CFA
1.37      anton    2610: In Gforth, the abstract data type @emph{execution token} is implemented
                   2611: as CFA (code field address).
                   2612: 
                   2613: The interpretation semantics of a named word are also represented by an
                   2614: execution token. You can get it with
                   2615: 
                   2616: doc-[']
                   2617: doc-'
                   2618: 
                   2619: For literals, you use @code{'} in interpreted code and @code{[']} in
                   2620: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
                   2621: by complaining about compile-only words. To get an execution token for a
                   2622: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
                   2623: @var{X} drop}.
                   2624: 
1.43      anton    2625: @cindex compilation token
1.37      anton    2626: The compilation semantics are represented by a @dfn{compilation token}
                   2627: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
                   2628: execution token. The compilation semantics represented by the
                   2629: compilation token can be performed with @code{execute}, which consumes
                   2630: the whole compilation token, with an additional stack effect determined
                   2631: by the represented compilation semantics.
                   2632: 
                   2633: doc-[comp']
                   2634: doc-comp'
                   2635: 
1.38      anton    2636: You can compile the compilation semantics with @code{postpone,}. I.e.,
                   2637: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
                   2638: @var{word}}.
                   2639: 
                   2640: doc-postpone,
                   2641: 
1.37      anton    2642: At present, the @var{w} part of a compilation token is an execution
                   2643: token, and the @var{xt} part represents either @code{execute} or
1.41      anton    2644: @code{compile,}. However, don't rely on that knowledge, unless necessary;
1.37      anton    2645: we may introduce unusual compilation tokens in the future (e.g.,
                   2646: compilation tokens representing the compilation semantics of literals).
                   2647: 
1.43      anton    2648: @cindex name token
                   2649: @cindex name field address
                   2650: @cindex NFA
1.37      anton    2651: Named words are also represented by the @dfn{name token}. The abstract
                   2652: data type @emph{name token} is implemented as NFA (name field address).
                   2653: 
                   2654: doc-find-name
                   2655: doc-name>int
                   2656: doc-name?int
                   2657: doc-name>comp
                   2658: doc-name>string
                   2659: 
                   2660: @node Wordlists, Files, Tokens for Words, Words
1.4       anton    2661: @section Wordlists
                   2662: 
                   2663: @node Files, Blocks, Wordlists, Words
                   2664: @section Files
                   2665: 
                   2666: @node Blocks, Other I/O, Files, Words
                   2667: @section Blocks
                   2668: 
                   2669: @node Other I/O, Programming Tools, Blocks, Words
                   2670: @section Other I/O
                   2671: 
1.18      anton    2672: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4       anton    2673: @section Programming Tools
1.43      anton    2674: @cindex programming tools
1.4       anton    2675: 
1.5       anton    2676: @menu
                   2677: * Debugging::                   Simple and quick.
                   2678: * Assertions::                  Making your programs self-checking.
                   2679: @end menu
                   2680: 
                   2681: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4       anton    2682: @subsection Debugging
1.43      anton    2683: @cindex debugging
1.4       anton    2684: 
                   2685: The simple debugging aids provided in @file{debugging.fs}
                   2686: are meant to support a different style of debugging than the
                   2687: tracing/stepping debuggers used in languages with long turn-around
                   2688: times.
                   2689: 
1.41      anton    2690: A much better (faster) way in fast-compiling languages is to add
1.4       anton    2691: printing code at well-selected places, let the program run, look at
                   2692: the output, see where things went wrong, add more printing code, etc.,
                   2693: until the bug is found.
                   2694: 
                   2695: The word @code{~~} is easy to insert. It just prints debugging
                   2696: information (by default the source location and the stack contents). It
                   2697: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
                   2698: query-replace them with nothing). The deferred words
                   2699: @code{printdebugdata} and @code{printdebugline} control the output of
                   2700: @code{~~}. The default source location output format works well with
                   2701: Emacs' compilation mode, so you can step through the program at the
1.5       anton    2702: source level using @kbd{C-x `} (the advantage over a stepping debugger
                   2703: is that you can step in any direction and you know where the crash has
                   2704: happened or where the strange data has occurred).
1.4       anton    2705: 
                   2706: Note that the default actions clobber the contents of the pictured
                   2707: numeric output string, so you should not use @code{~~}, e.g., between
                   2708: @code{<#} and @code{#>}.
                   2709: 
                   2710: doc-~~
                   2711: doc-printdebugdata
                   2712: doc-printdebugline
                   2713: 
1.5       anton    2714: @node Assertions,  , Debugging, Programming Tools
1.4       anton    2715: @subsection Assertions
1.43      anton    2716: @cindex assertions
1.4       anton    2717: 
1.5       anton    2718: It is a good idea to make your programs self-checking, in particular, if
                   2719: you use an assumption (e.g., that a certain field of a data structure is
1.17      anton    2720: never zero) that may become wrong during maintenance. Gforth supports
1.5       anton    2721: assertions for this purpose. They are used like this:
                   2722: 
                   2723: @example
                   2724: assert( @var{flag} )
                   2725: @end example
                   2726: 
                   2727: The code between @code{assert(} and @code{)} should compute a flag, that
                   2728: should be true if everything is alright and false otherwise. It should
                   2729: not change anything else on the stack. The overall stack effect of the
                   2730: assertion is @code{( -- )}. E.g.
                   2731: 
                   2732: @example
                   2733: assert( 1 1 + 2 = ) \ what we learn in school
                   2734: assert( dup 0<> ) \ assert that the top of stack is not zero
                   2735: assert( false ) \ this code should not be reached
                   2736: @end example
                   2737: 
                   2738: The need for assertions is different at different times. During
                   2739: debugging, we want more checking, in production we sometimes care more
                   2740: for speed. Therefore, assertions can be turned off, i.e., the assertion
                   2741: becomes a comment. Depending on the importance of an assertion and the
                   2742: time it takes to check it, you may want to turn off some assertions and
1.17      anton    2743: keep others turned on. Gforth provides several levels of assertions for
1.5       anton    2744: this purpose:
                   2745: 
                   2746: doc-assert0(
                   2747: doc-assert1(
                   2748: doc-assert2(
                   2749: doc-assert3(
                   2750: doc-assert(
                   2751: doc-)
                   2752: 
                   2753: @code{Assert(} is the same as @code{assert1(}. The variable
                   2754: @code{assert-level} specifies the highest assertions that are turned
                   2755: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
                   2756: @code{assert1(} assertions perform checking, while @code{assert2(} and
                   2757: @code{assert3(} assertions are treated as comments.
                   2758: 
                   2759: Note that the @code{assert-level} is evaluated at compile-time, not at
                   2760: run-time. I.e., you cannot turn assertions on or off at run-time, you
                   2761: have to set the @code{assert-level} appropriately before compiling a
                   2762: piece of code. You can compile several pieces of code at several
                   2763: @code{assert-level}s (e.g., a trusted library at level 1 and newly
                   2764: written code at level 3).
                   2765: 
                   2766: doc-assert-level
                   2767: 
                   2768: If an assertion fails, a message compatible with Emacs' compilation mode
                   2769: is produced and the execution is aborted (currently with @code{ABORT"}.
                   2770: If there is interest, we will introduce a special throw code. But if you
                   2771: intend to @code{catch} a specific condition, using @code{throw} is
                   2772: probably more appropriate than an assertion).
                   2773: 
1.18      anton    2774: @node Assembler and Code words, Threading Words, Programming Tools, Words
                   2775: @section Assembler and Code words
1.43      anton    2776: @cindex assembler
                   2777: @cindex code words
1.18      anton    2778: 
                   2779: Gforth provides some words for defining primitives (words written in
                   2780: machine code), and for defining the the machine-code equivalent of
                   2781: @code{DOES>}-based defining words. However, the machine-independent
1.40      anton    2782: nature of Gforth poses a few problems: First of all, Gforth runs on
1.18      anton    2783: several architectures, so it can provide no standard assembler. What's
                   2784: worse is that the register allocation not only depends on the processor,
1.25      anton    2785: but also on the @code{gcc} version and options used.
1.18      anton    2786: 
1.25      anton    2787: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18      anton    2788: header structure), so a system-independent assembler may be used in
                   2789: Gforth. If you do not have an assembler, you can compile machine code
                   2790: directly with @code{,} and @code{c,}.
                   2791: 
                   2792: doc-assembler
                   2793: doc-code
                   2794: doc-end-code
                   2795: doc-;code
                   2796: doc-flush-icache
                   2797: 
                   2798: If @code{flush-icache} does not work correctly, @code{code} words
                   2799: etc. will not work (reliably), either.
                   2800: 
                   2801: These words are rarely used. Therefore they reside in @code{code.fs},
                   2802: which is usually not loaded (except @code{flush-icache}, which is always
1.19      anton    2803: present). You can load them with @code{require code.fs}.
1.18      anton    2804: 
1.43      anton    2805: @cindex registers of the inner interpreter
1.25      anton    2806: In the assembly code you will want to refer to the inner interpreter's
                   2807: registers (e.g., the data stack pointer) and you may want to use other
                   2808: registers for temporary storage. Unfortunately, the register allocation
                   2809: is installation-dependent.
                   2810: 
                   2811: The easiest solution is to use explicit register declarations
                   2812: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
                   2813: GNU C Manual}) for all of the inner interpreter's registers: You have to
                   2814: compile Gforth with @code{-DFORCE_REG} (configure option
                   2815: @code{--enable-force-reg}) and the appropriate declarations must be
                   2816: present in the @code{machine.h} file (see @code{mips.h} for an example;
                   2817: you can find a full list of all declarable register symbols with
                   2818: @code{grep register engine.c}). If you give explicit registers to all
                   2819: variables that are declared at the beginning of @code{engine()}, you
                   2820: should be able to use the other caller-saved registers for temporary
                   2821: storage. Alternatively, you can use the @code{gcc} option
                   2822: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
                   2823: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
                   2824: (however, this restriction on register allocation may slow Gforth
                   2825: significantly).
                   2826: 
                   2827: If this solution is not viable (e.g., because @code{gcc} does not allow
                   2828: you to explicitly declare all the registers you need), you have to find
                   2829: out by looking at the code where the inner interpreter's registers
                   2830: reside and which registers can be used for temporary storage. You can
                   2831: get an assembly listing of the engine's code with @code{make engine.s}.
                   2832: 
                   2833: In any case, it is good practice to abstract your assembly code from the
                   2834: actual register allocation. E.g., if the data stack pointer resides in
                   2835: register @code{$17}, create an alias for this register called @code{sp},
                   2836: and use that in your assembly code.
                   2837: 
1.43      anton    2838: @cindex code words, portable
1.18      anton    2839: Another option for implementing normal and defining words efficiently
                   2840: is: adding the wanted functionality to the source of Gforth. For normal
1.35      anton    2841: words you just have to edit @file{primitives} (@pxref{Automatic
                   2842: Generation}), defining words (equivalent to @code{;CODE} words, for fast
                   2843: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
                   2844: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.18      anton    2845: 
                   2846: 
                   2847: @node Threading Words,  , Assembler and Code words, Words
1.4       anton    2848: @section Threading Words
1.43      anton    2849: @cindex threading words
1.4       anton    2850: 
1.43      anton    2851: @cindex code address
1.4       anton    2852: These words provide access to code addresses and other threading stuff
1.17      anton    2853: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4       anton    2854: abstracts away the differences between direct and indirect threading
                   2855: (and, for direct threading, the machine dependences). However, at
1.43      anton    2856: present this wordset is still incomplete. It is also pretty low-level;
                   2857: some day it will hopefully be made unnecessary by an internals wordset
1.4       anton    2858: that abstracts implementation details away completely.
                   2859: 
                   2860: doc->code-address
                   2861: doc->does-code
                   2862: doc-code-address!
                   2863: doc-does-code!
                   2864: doc-does-handler!
                   2865: doc-/does-handler
                   2866: 
1.18      anton    2867: The code addresses produced by various defining words are produced by
                   2868: the following words:
1.14      anton    2869: 
1.18      anton    2870: doc-docol:
                   2871: doc-docon:
                   2872: doc-dovar:
                   2873: doc-douser:
                   2874: doc-dodefer:
                   2875: doc-dofield:
                   2876: 
1.35      anton    2877: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
                   2878: with @code{>DOES-CODE}. If the word was defined in that way, the value
                   2879: returned is different from 0 and identifies the @code{DOES>} used by the
                   2880: defining word.
1.14      anton    2881: 
1.40      anton    2882: @node Tools, ANS conformance, Words, Top
                   2883: @chapter Tools
                   2884: 
                   2885: @menu
1.43      anton    2886: * ANS Report::                  Report the words used, sorted by wordset.
1.40      anton    2887: @end menu
                   2888: 
                   2889: See also @ref{Emacs and Gforth}.
                   2890: 
                   2891: @node ANS Report,  , Tools, Tools
                   2892: @section @file{ans-report.fs}: Report the words used, sorted by wordset
1.43      anton    2893: @cindex @file{ans-report.fs}
                   2894: @cindex report the words used in your program
                   2895: @cindex words used in your program
1.40      anton    2896: 
                   2897: If you want to label a Forth program as ANS Forth Program, you must
                   2898: document which wordsets the program uses; for extension wordsets, it is
                   2899: helpful to list the words the program requires from these wordsets
                   2900: (because Forth systems are allowed to provide only some words of them).
                   2901: 
                   2902: The @file{ans-report.fs} tool makes it easy for you to determine which
                   2903: words from which wordset and which non-ANS words your application
                   2904: uses. You simply have to include @file{ans-report.fs} before loading the
                   2905: program you want to check. After loading your program, you can get the
                   2906: report with @code{print-ans-report}. A typical use is to run this as
                   2907: batch job like this:
                   2908: @example
                   2909: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
                   2910: @end example
                   2911: 
                   2912: The output looks like this (for @file{compat/control.fs}):
                   2913: @example
                   2914: The program uses the following words
                   2915: from CORE :
                   2916: : POSTPONE THEN ; immediate ?dup IF 0= 
                   2917: from BLOCK-EXT :
                   2918: \ 
                   2919: from FILE :
                   2920: ( 
                   2921: @end example
                   2922: 
                   2923: @subsection Caveats
                   2924: 
                   2925: Note that @file{ans-report.fs} just checks which words are used, not whether
                   2926: they are used in an ANS Forth conforming way!
                   2927: 
                   2928: Some words are defined in several wordsets in the
                   2929: standard. @file{ans-report.fs} reports them for only one of the
                   2930: wordsets, and not necessarily the one you expect. It depends on usage
                   2931: which wordset is the right one to specify. E.g., if you only use the
                   2932: compilation semantics of @code{S"}, it is a Core word; if you also use
                   2933: its interpretation semantics, it is a File word.
                   2934: 
1.43      anton    2935: @c ******************************************************************
1.40      anton    2936: @node ANS conformance, Model, Tools, Top
1.4       anton    2937: @chapter ANS conformance
1.43      anton    2938: @cindex ANS conformance of Gforth
1.4       anton    2939: 
1.17      anton    2940: To the best of our knowledge, Gforth is an
1.14      anton    2941: 
1.15      anton    2942: ANS Forth System
1.34      anton    2943: @itemize @bullet
1.15      anton    2944: @item providing the Core Extensions word set
                   2945: @item providing the Block word set
                   2946: @item providing the Block Extensions word set
                   2947: @item providing the Double-Number word set
                   2948: @item providing the Double-Number Extensions word set
                   2949: @item providing the Exception word set
                   2950: @item providing the Exception Extensions word set
                   2951: @item providing the Facility word set
                   2952: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
                   2953: @item providing the File Access word set
                   2954: @item providing the File Access Extensions word set
                   2955: @item providing the Floating-Point word set
                   2956: @item providing the Floating-Point Extensions word set
                   2957: @item providing the Locals word set
                   2958: @item providing the Locals Extensions word set
                   2959: @item providing the Memory-Allocation word set
                   2960: @item providing the Memory-Allocation Extensions word set (that one's easy)
                   2961: @item providing the Programming-Tools word set
1.34      anton    2962: @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    2963: @item providing the Search-Order word set
                   2964: @item providing the Search-Order Extensions word set
                   2965: @item providing the String word set
                   2966: @item providing the String Extensions word set (another easy one)
                   2967: @end itemize
                   2968: 
1.43      anton    2969: @cindex system documentation
1.15      anton    2970: In addition, ANS Forth systems are required to document certain
                   2971: implementation choices. This chapter tries to meet these
                   2972: requirements. In many cases it gives a way to ask the system for the
                   2973: information instead of providing the information directly, in
                   2974: particular, if the information depends on the processor, the operating
                   2975: system or the installation options chosen, or if they are likely to
1.17      anton    2976: change during the maintenance of Gforth.
1.15      anton    2977: 
1.14      anton    2978: @comment The framework for the rest has been taken from pfe.
                   2979: 
                   2980: @menu
                   2981: * The Core Words::              
                   2982: * The optional Block word set::  
                   2983: * The optional Double Number word set::  
                   2984: * The optional Exception word set::  
                   2985: * The optional Facility word set::  
                   2986: * The optional File-Access word set::  
                   2987: * The optional Floating-Point word set::  
                   2988: * The optional Locals word set::  
                   2989: * The optional Memory-Allocation word set::  
                   2990: * The optional Programming-Tools word set::  
                   2991: * The optional Search-Order word set::  
                   2992: @end menu
                   2993: 
                   2994: 
                   2995: @c =====================================================================
                   2996: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
                   2997: @comment  node-name,  next,  previous,  up
                   2998: @section The Core Words
                   2999: @c =====================================================================
1.43      anton    3000: @cindex core words, system documentation
                   3001: @cindex system documentation, core words
1.14      anton    3002: 
                   3003: @menu
1.15      anton    3004: * core-idef::                   Implementation Defined Options                   
                   3005: * core-ambcond::                Ambiguous Conditions                
                   3006: * core-other::                  Other System Documentation                  
1.14      anton    3007: @end menu
                   3008: 
                   3009: @c ---------------------------------------------------------------------
                   3010: @node core-idef, core-ambcond, The Core Words, The Core Words
                   3011: @subsection Implementation Defined Options
                   3012: @c ---------------------------------------------------------------------
1.43      anton    3013: @cindex core words, implementation-defined options
                   3014: @cindex implementation-defined options, core words
                   3015: 
1.14      anton    3016: 
                   3017: @table @i
                   3018: @item (Cell) aligned addresses:
1.43      anton    3019: @cindex cell-aligned addresses
                   3020: @cindex aligned addresses
1.17      anton    3021: processor-dependent. Gforth's alignment words perform natural alignment
1.14      anton    3022: (e.g., an address aligned for a datum of size 8 is divisible by
                   3023: 8). Unaligned accesses usually result in a @code{-23 THROW}.
                   3024: 
                   3025: @item @code{EMIT} and non-graphic characters:
1.43      anton    3026: @cindex @code{EMIT} and non-graphic characters
                   3027: @cindex non-graphic characters and @code{EMIT}
1.14      anton    3028: The character is output using the C library function (actually, macro)
1.36      anton    3029: @code{putc}.
1.14      anton    3030: 
                   3031: @item character editing of @code{ACCEPT} and @code{EXPECT}:
1.43      anton    3032: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
                   3033: @cindex editing in @code{ACCEPT} and @code{EXPECT}
                   3034: @cindex @code{ACCEPT}, editing
                   3035: @cindex @code{EXPECT}, editing
1.14      anton    3036: This is modeled on the GNU readline library (@pxref{Readline
                   3037: Interaction, , Command Line Editing, readline, The GNU Readline
                   3038: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
                   3039: producing a full word completion every time you type it (instead of
                   3040: producing the common prefix of all completions).
                   3041: 
                   3042: @item character set:
1.43      anton    3043: @cindex character set
1.14      anton    3044: The character set of your computer and display device. Gforth is
                   3045: 8-bit-clean (but some other component in your system may make trouble).
                   3046: 
                   3047: @item Character-aligned address requirements:
1.43      anton    3048: @cindex character-aligned address requirements
1.14      anton    3049: installation-dependent. Currently a character is represented by a C
                   3050: @code{unsigned char}; in the future we might switch to @code{wchar_t}
                   3051: (Comments on that requested).
                   3052: 
                   3053: @item character-set extensions and matching of names:
1.43      anton    3054: @cindex character-set extensions and matching of names
                   3055: @cindex case sensitivity for name lookup
                   3056: @cindex name lookup, case sensitivity
                   3057: @cindex locale and case sensitivity
1.17      anton    3058: Any character except the ASCII NUL charcter can be used in a
1.43      anton    3059: name. Matching is case-insensitive (except in @code{TABLE}s). The
1.36      anton    3060: matching is performed using the C function @code{strncasecmp}, whose
                   3061: function is probably influenced by the locale. E.g., the @code{C} locale
                   3062: does not know about accents and umlauts, so they are matched
                   3063: case-sensitively in that locale. For portability reasons it is best to
                   3064: write programs such that they work in the @code{C} locale. Then one can
                   3065: use libraries written by a Polish programmer (who might use words
                   3066: containing ISO Latin-2 encoded characters) and by a French programmer
                   3067: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
                   3068: funny results for some of the words (which ones, depends on the font you
                   3069: are using)). Also, the locale you prefer may not be available in other
                   3070: operating systems. Hopefully, Unicode will solve these problems one day.
1.14      anton    3071: 
                   3072: @item conditions under which control characters match a space delimiter:
1.43      anton    3073: @cindex space delimiters
                   3074: @cindex control characters as delimiters
1.14      anton    3075: If @code{WORD} is called with the space character as a delimiter, all
                   3076: white-space characters (as identified by the C macro @code{isspace()})
                   3077: are delimiters. @code{PARSE}, on the other hand, treats space like other
                   3078: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
                   3079: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
                   3080: interpreter (aka text interpreter) by default, treats all white-space
                   3081: characters as delimiters.
                   3082: 
                   3083: @item format of the control flow stack:
1.43      anton    3084: @cindex control flow stack, format
1.14      anton    3085: The data stack is used as control flow stack. The size of a control flow
                   3086: stack item in cells is given by the constant @code{cs-item-size}. At the
                   3087: time of this writing, an item consists of a (pointer to a) locals list
                   3088: (third), an address in the code (second), and a tag for identifying the
                   3089: item (TOS). The following tags are used: @code{defstart},
                   3090: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
                   3091: @code{scopestart}.
                   3092: 
                   3093: @item conversion of digits > 35
1.43      anton    3094: @cindex digits > 35
1.14      anton    3095: The characters @code{[\]^_'} are the digits with the decimal value
                   3096: 36@minus{}41. There is no way to input many of the larger digits.
                   3097: 
                   3098: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
1.43      anton    3099: @cindex @code{EXPECT}, display after end of input
                   3100: @cindex @code{ACCEPT}, display after end of input
1.14      anton    3101: The cursor is moved to the end of the entered string. If the input is
                   3102: terminated using the @kbd{Return} key, a space is typed.
                   3103: 
                   3104: @item exception abort sequence of @code{ABORT"}:
1.43      anton    3105: @cindex exception abort sequence of @code{ABORT"}
                   3106: @cindex @code{ABORT"}, exception abort sequence
1.14      anton    3107: The error string is stored into the variable @code{"error} and a
                   3108: @code{-2 throw} is performed.
                   3109: 
                   3110: @item input line terminator:
1.43      anton    3111: @cindex input line terminator
                   3112: @cindex line terminator on input
                   3113: @cindex newline charcter on input
1.36      anton    3114: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
                   3115: lines. One of these characters is typically produced when you type the
                   3116: @kbd{Enter} or @kbd{Return} key.
1.14      anton    3117: 
                   3118: @item maximum size of a counted string:
1.43      anton    3119: @cindex maximum size of a counted string
                   3120: @cindex counted string, maximum size
1.14      anton    3121: @code{s" /counted-string" environment? drop .}. Currently 255 characters
                   3122: on all ports, but this may change.
                   3123: 
                   3124: @item maximum size of a parsed string:
1.43      anton    3125: @cindex maximum size of a parsed string
                   3126: @cindex parsed string, maximum size
1.14      anton    3127: Given by the constant @code{/line}. Currently 255 characters.
                   3128: 
                   3129: @item maximum size of a definition name, in characters:
1.43      anton    3130: @cindex maximum size of a definition name, in characters
                   3131: @cindex name, maximum length
1.14      anton    3132: 31
                   3133: 
                   3134: @item maximum string length for @code{ENVIRONMENT?}, in characters:
1.43      anton    3135: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
                   3136: @cindex @code{ENVIRONMENT?} string length, maximum
1.14      anton    3137: 31
                   3138: 
                   3139: @item method of selecting the user input device:
1.43      anton    3140: @cindex user input device, method of selecting
1.17      anton    3141: The user input device is the standard input. There is currently no way to
                   3142: change it from within Gforth. However, the input can typically be
                   3143: redirected in the command line that starts Gforth.
1.14      anton    3144: 
                   3145: @item method of selecting the user output device:
1.43      anton    3146: @cindex user output device, method of selecting
1.36      anton    3147: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
                   3148: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
                   3149: output, so output on a terminal does not become visible before the next
                   3150: newline or buffer overflow. Output on non-terminals is invisible until
                   3151: the buffer overflows.
1.14      anton    3152: 
                   3153: @item methods of dictionary compilation:
1.17      anton    3154: What are we expected to document here?
1.14      anton    3155: 
                   3156: @item number of bits in one address unit:
1.43      anton    3157: @cindex number of bits in one address unit
                   3158: @cindex address unit, size in bits
1.14      anton    3159: @code{s" address-units-bits" environment? drop .}. 8 in all current
                   3160: ports.
                   3161: 
                   3162: @item number representation and arithmetic:
1.43      anton    3163: @cindex number representation and arithmetic
1.14      anton    3164: Processor-dependent. Binary two's complement on all current ports.
                   3165: 
                   3166: @item ranges for integer types:
1.43      anton    3167: @cindex ranges for integer types
                   3168: @cindex integer types, ranges
1.14      anton    3169: Installation-dependent. Make environmental queries for @code{MAX-N},
                   3170: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
                   3171: unsigned (and positive) types is 0. The lower bound for signed types on
                   3172: two's complement and one's complement machines machines can be computed
                   3173: by adding 1 to the upper bound.
                   3174: 
                   3175: @item read-only data space regions:
1.43      anton    3176: @cindex read-only data space regions
                   3177: @cindex data-space, read-only regions
1.14      anton    3178: The whole Forth data space is writable.
                   3179: 
                   3180: @item size of buffer at @code{WORD}:
1.43      anton    3181: @cindex size of buffer at @code{WORD}
                   3182: @cindex @code{WORD} buffer size
1.14      anton    3183: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   3184: shared with the pictured numeric output string. If overwriting
                   3185: @code{PAD} is acceptable, it is as large as the remaining dictionary
                   3186: space, although only as much can be sensibly used as fits in a counted
                   3187: string.
                   3188: 
                   3189: @item size of one cell in address units:
1.43      anton    3190: @cindex cell size
1.14      anton    3191: @code{1 cells .}.
                   3192: 
                   3193: @item size of one character in address units:
1.43      anton    3194: @cindex char size
1.14      anton    3195: @code{1 chars .}. 1 on all current ports.
                   3196: 
                   3197: @item size of the keyboard terminal buffer:
1.43      anton    3198: @cindex size of the keyboard terminal buffer
                   3199: @cindex terminal buffer, size
1.36      anton    3200: Varies. You can determine the size at a specific time using @code{lp@@
1.14      anton    3201: tib - .}. It is shared with the locals stack and TIBs of files that
                   3202: include the current file. You can change the amount of space for TIBs
1.17      anton    3203: and locals stack at Gforth startup with the command line option
1.14      anton    3204: @code{-l}.
                   3205: 
                   3206: @item size of the pictured numeric output buffer:
1.43      anton    3207: @cindex size of the pictured numeric output buffer
                   3208: @cindex pictured numeric output buffer, size
1.14      anton    3209: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   3210: shared with @code{WORD}.
                   3211: 
                   3212: @item size of the scratch area returned by @code{PAD}:
1.43      anton    3213: @cindex size of the scratch area returned by @code{PAD}
                   3214: @cindex @code{PAD} size
                   3215: The remainder of dictionary space. @code{unused pad here - - .}.
1.14      anton    3216: 
                   3217: @item system case-sensitivity characteristics:
1.43      anton    3218: @cindex case-sensitivity characteristics
1.36      anton    3219: Dictionary searches are case insensitive (except in
                   3220: @code{TABLE}s). However, as explained above under @i{character-set
                   3221: extensions}, the matching for non-ASCII characters is determined by the
                   3222: locale you are using. In the default @code{C} locale all non-ASCII
                   3223: characters are matched case-sensitively.
1.14      anton    3224: 
                   3225: @item system prompt:
1.43      anton    3226: @cindex system prompt
                   3227: @cindex prompt
1.14      anton    3228: @code{ ok} in interpret state, @code{ compiled} in compile state.
                   3229: 
                   3230: @item division rounding:
1.43      anton    3231: @cindex division rounding
1.14      anton    3232: installation dependent. @code{s" floored" environment? drop .}. We leave
1.43      anton    3233: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
                   3234: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
1.14      anton    3235: 
                   3236: @item values of @code{STATE} when true:
1.43      anton    3237: @cindex @code{STATE} values
1.14      anton    3238: -1.
                   3239: 
                   3240: @item values returned after arithmetic overflow:
                   3241: On two's complement machines, arithmetic is performed modulo
                   3242: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   3243: arithmetic (with appropriate mapping for signed types). Division by zero
1.36      anton    3244: typically results in a @code{-55 throw} (Floating-point unidentified
1.14      anton    3245: fault), although a @code{-10 throw} (divide by zero) would be more
                   3246: appropriate.
                   3247: 
                   3248: @item whether the current definition can be found after @t{DOES>}:
1.43      anton    3249: @cindex @t{DOES>}, visibility of current definition
1.14      anton    3250: No.
                   3251: 
                   3252: @end table
                   3253: 
                   3254: @c ---------------------------------------------------------------------
                   3255: @node core-ambcond, core-other, core-idef, The Core Words
                   3256: @subsection Ambiguous conditions
                   3257: @c ---------------------------------------------------------------------
1.43      anton    3258: @cindex core words, ambiguous conditions
                   3259: @cindex ambiguous conditions, core words
1.14      anton    3260: 
                   3261: @table @i
                   3262: 
                   3263: @item a name is neither a word nor a number:
1.43      anton    3264: @cindex name not found
                   3265: @cindex Undefined word
1.36      anton    3266: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
                   3267: preserves the data and FP stack, so you don't lose more work than
                   3268: necessary.
1.14      anton    3269: 
                   3270: @item a definition name exceeds the maximum length allowed:
1.43      anton    3271: @cindex Word name too long
1.14      anton    3272: @code{-19 throw} (Word name too long)
                   3273: 
                   3274: @item addressing a region not inside the various data spaces of the forth system:
1.43      anton    3275: @cindex Invalid memory address
1.14      anton    3276: The stacks, code space and name space are accessible. Machine code space is
                   3277: typically readable. Accessing other addresses gives results dependent on
                   3278: the operating system. On decent systems: @code{-9 throw} (Invalid memory
                   3279: address).
                   3280: 
                   3281: @item argument type incompatible with parameter:
1.43      anton    3282: @cindex Argument type mismatch
1.14      anton    3283: This is usually not caught. Some words perform checks, e.g., the control
                   3284: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
                   3285: mismatch).
                   3286: 
                   3287: @item attempting to obtain the execution token of a word with undefined execution semantics:
1.43      anton    3288: @cindex Interpreting a compile-only word, for @code{'} etc.
                   3289: @cindex execution token of words with undefined execution semantics
1.36      anton    3290: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
                   3291: get an execution token for @code{compile-only-error} (which performs a
                   3292: @code{-14 throw} when executed).
1.14      anton    3293: 
                   3294: @item dividing by zero:
1.43      anton    3295: @cindex dividing by zero
                   3296: @cindex floating point unidentified fault, integer division
                   3297: @cindex divide by zero
1.14      anton    3298: typically results in a @code{-55 throw} (floating point unidentified
                   3299: fault), although a @code{-10 throw} (divide by zero) would be more
                   3300: appropriate.
                   3301: 
                   3302: @item insufficient data stack or return stack space:
1.43      anton    3303: @cindex insufficient data stack or return stack space
                   3304: @cindex stack overflow
                   3305: @cindex Address alignment exception, stack overflow
                   3306: @cindex Invalid memory address, stack overflow
                   3307: Depending on the operating system, the installation, and the invocation
                   3308: of Gforth, this is either checked by the memory management hardware, or
                   3309: it is not checked. If it is checked, you typically get a @code{-9 throw}
                   3310: (Invalid memory address) as soon as the overflow happens. If it is not
                   3311: check, overflows typically result in mysterious illegal memory accesses,
                   3312: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
                   3313: (Address alignment exception); they might also destroy the internal data
                   3314: structure of @code{ALLOCATE} and friends, resulting in various errors in
                   3315: these words.
1.14      anton    3316: 
                   3317: @item insufficient space for loop control parameters:
1.43      anton    3318: @cindex insufficient space for loop control parameters
1.14      anton    3319: like other return stack overflows.
                   3320: 
                   3321: @item insufficient space in the dictionary:
1.43      anton    3322: @cindex insufficient space in the dictionary
                   3323: @cindex dictionary overflow
                   3324: Depending on the operating system, the installation, and the invocation
                   3325: of Gforth, this is either checked by the memory management hardware, or
                   3326: it is not checked. Similar results as stack overflows. However,
                   3327: typically the error appears at a different place when one inserts or
                   3328: removes code. Also, the @code{THROW} does not relieve the situation (it
                   3329: does for stack overflows).
1.14      anton    3330: 
                   3331: @item interpreting a word with undefined interpretation semantics:
1.43      anton    3332: @cindex interpreting a word with undefined interpretation semantics
                   3333: @cindex Interpreting a compile-only word
                   3334: For some words, we have defined interpretation semantics. For the
                   3335: others: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    3336: 
                   3337: @item modifying the contents of the input buffer or a string literal:
1.43      anton    3338: @cindex modifying the contents of the input buffer or a string literal
1.14      anton    3339: These are located in writable memory and can be modified.
                   3340: 
                   3341: @item overflow of the pictured numeric output string:
1.43      anton    3342: @cindex overflow of the pictured numeric output string
                   3343: @cindex pictured numeric output string, overflow
                   3344: Not checked. Runs into the dictionary and destroys it (at least,
                   3345: partially).
1.14      anton    3346: 
                   3347: @item parsed string overflow:
1.43      anton    3348: @cindex parsed string overflow
1.14      anton    3349: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
                   3350: 
                   3351: @item producing a result out of range:
1.43      anton    3352: @cindex result out of range
1.14      anton    3353: On two's complement machines, arithmetic is performed modulo
                   3354: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   3355: arithmetic (with appropriate mapping for signed types). Division by zero
                   3356: typically results in a @code{-55 throw} (floatingpoint unidentified
                   3357: fault), although a @code{-10 throw} (divide by zero) would be more
                   3358: appropriate. @code{convert} and @code{>number} currently overflow
                   3359: silently.
                   3360: 
                   3361: @item reading from an empty data or return stack:
1.43      anton    3362: @cindex stack empty
                   3363: @cindex stack underflow
1.14      anton    3364: The data stack is checked by the outer (aka text) interpreter after
                   3365: every word executed. If it has underflowed, a @code{-4 throw} (Stack
1.43      anton    3366: underflow) is performed. Apart from that, stacks may be checked or not,
                   3367: depending on operating system, installation, and invocation. The
                   3368: consequences of stack underflows are similar to the consequences of
                   3369: stack overflows. Note that even if the system uses checking (through the
                   3370: MMU), your program may have to underflow by a significant number of
                   3371: stack items to trigger the reaction (the reason for this is that the
                   3372: MMU, and therefore the checking, works with a page-size granularity).
1.14      anton    3373: 
1.36      anton    3374: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
1.43      anton    3375: @cindex unexpected end of the input buffer
                   3376: @cindex zero-length string as a name
                   3377: @cindex Attempt to use zero-length string as a name
1.14      anton    3378: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
                   3379: use zero-length string as a name). Words like @code{'} probably will not
                   3380: find what they search. Note that it is possible to create zero-length
                   3381: names with @code{nextname} (should it not?).
                   3382: 
                   3383: @item @code{>IN} greater than input buffer:
1.43      anton    3384: @cindex @code{>IN} greater than input buffer
1.41      anton    3385: The next invocation of a parsing word returns a string with length 0.
1.14      anton    3386: 
                   3387: @item @code{RECURSE} appears after @code{DOES>}:
1.43      anton    3388: @cindex @code{RECURSE} appears after @code{DOES>}
1.36      anton    3389: Compiles a recursive call to the defining word, not to the defined word.
1.14      anton    3390: 
                   3391: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.43      anton    3392: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
                   3393: @cindex Argument type mismatch, @code{RESTORE-INPUT}
                   3394: @cindex @code{RESTORE-INPUT}, Argument type mismatch
1.27      anton    3395: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
                   3396: the end of the file was reached), its source-id may be
                   3397: reused. Therefore, restoring an input source specification referencing a
                   3398: closed file may lead to unpredictable results instead of a @code{-12
                   3399: THROW}.
                   3400: 
1.36      anton    3401: In the future, Gforth may be able to restore input source specifications
1.41      anton    3402: from other than the current input source.
1.14      anton    3403: 
                   3404: @item data space containing definitions gets de-allocated:
1.43      anton    3405: @cindex data space containing definitions gets de-allocated
1.41      anton    3406: Deallocation with @code{allot} is not checked. This typically results in
1.14      anton    3407: memory access faults or execution of illegal instructions.
                   3408: 
                   3409: @item data space read/write with incorrect alignment:
1.43      anton    3410: @cindex data space read/write with incorrect alignment
                   3411: @cindex alignment faults
                   3412: @cindex Address alignment exception
1.14      anton    3413: Processor-dependent. Typically results in a @code{-23 throw} (Address
                   3414: alignment exception). Under Linux on a 486 or later processor with
                   3415: alignment turned on, incorrect alignment results in a @code{-9 throw}
                   3416: (Invalid memory address). There are reportedly some processors with
                   3417: alignment restrictions that do not report them.
                   3418: 
                   3419: @item data space pointer not properly aligned, @code{,}, @code{C,}:
1.43      anton    3420: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
1.14      anton    3421: Like other alignment errors.
                   3422: 
                   3423: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
1.43      anton    3424: Like other stack underflows.
1.14      anton    3425: 
                   3426: @item loop control parameters not available:
1.43      anton    3427: @cindex loop control parameters not available
1.14      anton    3428: Not checked. The counted loop words simply assume that the top of return
                   3429: stack items are loop control parameters and behave accordingly.
                   3430: 
                   3431: @item most recent definition does not have a name (@code{IMMEDIATE}):
1.43      anton    3432: @cindex most recent definition does not have a name (@code{IMMEDIATE})
                   3433: @cindex last word was headerless
1.14      anton    3434: @code{abort" last word was headerless"}.
                   3435: 
                   3436: @item name not defined by @code{VALUE} used by @code{TO}:
1.43      anton    3437: @cindex name not defined by @code{VALUE} used by @code{TO}
                   3438: @cindex @code{TO} on non-@code{VALUE}s
                   3439: @cindex Invalid name argument, @code{TO}
                   3440: @code{-32 throw} (Invalid name argument) (unless name is a local or was
                   3441: defined by @code{CONSTANT}; in the latter case it just changes the constant).
1.14      anton    3442: 
1.15      anton    3443: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.43      anton    3444: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
                   3445: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
1.14      anton    3446: @code{-13 throw} (Undefined word)
                   3447: 
                   3448: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
1.43      anton    3449: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
1.14      anton    3450: Gforth behaves as if they were of the same type. I.e., you can predict
                   3451: the behaviour by interpreting all parameters as, e.g., signed.
                   3452: 
                   3453: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
1.43      anton    3454: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
1.36      anton    3455: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
                   3456: compilation semantics of @code{TO}.
1.14      anton    3457: 
                   3458: @item String longer than a counted string returned by @code{WORD}:
1.43      anton    3459: @cindex String longer than a counted string returned by @code{WORD}
                   3460: @cindex @code{WORD}, string overflow
1.14      anton    3461: Not checked. The string will be ok, but the count will, of course,
                   3462: contain only the least significant bits of the length.
                   3463: 
1.15      anton    3464: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.43      anton    3465: @cindex @code{LSHIFT}, large shift counts
                   3466: @cindex @code{RSHIFT}, large shift counts
1.14      anton    3467: Processor-dependent. Typical behaviours are returning 0 and using only
                   3468: the low bits of the shift count.
                   3469: 
                   3470: @item word not defined via @code{CREATE}:
1.43      anton    3471: @cindex @code{>BODY} of non-@code{CREATE}d words
1.14      anton    3472: @code{>BODY} produces the PFA of the word no matter how it was defined.
                   3473: 
1.43      anton    3474: @cindex @code{DOES>} of non-@code{CREATE}d words
1.14      anton    3475: @code{DOES>} changes the execution semantics of the last defined word no
                   3476: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
                   3477: @code{CREATE , DOES>}.
                   3478: 
                   3479: @item words improperly used outside @code{<#} and @code{#>}:
                   3480: Not checked. As usual, you can expect memory faults.
                   3481: 
                   3482: @end table
                   3483: 
                   3484: 
                   3485: @c ---------------------------------------------------------------------
                   3486: @node core-other,  , core-ambcond, The Core Words
                   3487: @subsection Other system documentation
                   3488: @c ---------------------------------------------------------------------
1.43      anton    3489: @cindex other system documentation, core words
                   3490: @cindex core words, other system documentation
1.14      anton    3491: 
                   3492: @table @i
                   3493: @item nonstandard words using @code{PAD}:
1.43      anton    3494: @cindex @code{PAD} use by nonstandard words
1.14      anton    3495: None.
                   3496: 
                   3497: @item operator's terminal facilities available:
1.43      anton    3498: @cindex operator's terminal facilities available
1.26      anton    3499: After processing the command line, Gforth goes into interactive mode,
                   3500: and you can give commands to Gforth interactively. The actual facilities
                   3501: available depend on how you invoke Gforth.
1.14      anton    3502: 
                   3503: @item program data space available:
1.43      anton    3504: @cindex program data space available
                   3505: @cindex data space available
1.42      anton    3506: @code{UNUSED .} gives the remaining dictionary space. The total
                   3507: dictionary space can be specified with the @code{-m} switch
1.43      anton    3508: (@pxref{Invoking Gforth}) when Gforth starts up.
1.14      anton    3509: 
                   3510: @item return stack space available:
1.43      anton    3511: @cindex return stack space available
1.42      anton    3512: You can compute the total return stack space in cells with
                   3513: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
1.43      anton    3514: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
1.14      anton    3515: 
                   3516: @item stack space available:
1.43      anton    3517: @cindex stack space available
1.42      anton    3518: You can compute the total data stack space in cells with
                   3519: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
1.43      anton    3520: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
1.14      anton    3521: 
                   3522: @item system dictionary space required, in address units:
1.43      anton    3523: @cindex system dictionary space required, in address units
1.14      anton    3524: Type @code{here forthstart - .} after startup. At the time of this
1.42      anton    3525: writing, this gives 80080 (bytes) on a 32-bit system.
1.14      anton    3526: @end table
                   3527: 
                   3528: 
                   3529: @c =====================================================================
                   3530: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
                   3531: @section The optional Block word set
                   3532: @c =====================================================================
1.43      anton    3533: @cindex system documentation, block words
                   3534: @cindex block words, system documentation
1.14      anton    3535: 
                   3536: @menu
1.43      anton    3537: * block-idef::                  Implementation Defined Options
1.15      anton    3538: * block-ambcond::               Ambiguous Conditions               
                   3539: * block-other::                 Other System Documentation                 
1.14      anton    3540: @end menu
                   3541: 
                   3542: 
                   3543: @c ---------------------------------------------------------------------
                   3544: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
                   3545: @subsection Implementation Defined Options
                   3546: @c ---------------------------------------------------------------------
1.43      anton    3547: @cindex implementation-defined options, block words
                   3548: @cindex block words, implementation-defined options
1.14      anton    3549: 
                   3550: @table @i
                   3551: @item the format for display by @code{LIST}:
1.43      anton    3552: @cindex @code{LIST} display format
1.14      anton    3553: First the screen number is displayed, then 16 lines of 64 characters,
                   3554: each line preceded by the line number.
                   3555: 
                   3556: @item the length of a line affected by @code{\}:
1.43      anton    3557: @cindex length of a line affected by @code{\}
                   3558: @cindex @code{\}, line length in blocks
1.14      anton    3559: 64 characters.
                   3560: @end table
                   3561: 
                   3562: 
                   3563: @c ---------------------------------------------------------------------
                   3564: @node block-ambcond, block-other, block-idef, The optional Block word set
                   3565: @subsection Ambiguous conditions
                   3566: @c ---------------------------------------------------------------------
1.43      anton    3567: @cindex block words, ambiguous conditions
                   3568: @cindex ambiguous conditions, block words
1.14      anton    3569: 
                   3570: @table @i
                   3571: @item correct block read was not possible:
1.43      anton    3572: @cindex block read not possible
1.14      anton    3573: Typically results in a @code{throw} of some OS-derived value (between
                   3574: -512 and -2048). If the blocks file was just not long enough, blanks are
                   3575: supplied for the missing portion.
                   3576: 
                   3577: @item I/O exception in block transfer:
1.43      anton    3578: @cindex I/O exception in block transfer
                   3579: @cindex block transfer, I/O exception
1.14      anton    3580: Typically results in a @code{throw} of some OS-derived value (between
                   3581: -512 and -2048).
                   3582: 
                   3583: @item invalid block number:
1.43      anton    3584: @cindex invalid block number
                   3585: @cindex block number invalid
1.14      anton    3586: @code{-35 throw} (Invalid block number)
                   3587: 
                   3588: @item a program directly alters the contents of @code{BLK}:
1.43      anton    3589: @cindex @code{BLK}, altering @code{BLK}
1.14      anton    3590: The input stream is switched to that other block, at the same
                   3591: position. If the storing to @code{BLK} happens when interpreting
                   3592: non-block input, the system will get quite confused when the block ends.
                   3593: 
                   3594: @item no current block buffer for @code{UPDATE}:
1.43      anton    3595: @cindex @code{UPDATE}, no current block buffer
1.14      anton    3596: @code{UPDATE} has no effect.
                   3597: 
                   3598: @end table
                   3599: 
                   3600: @c ---------------------------------------------------------------------
                   3601: @node block-other,  , block-ambcond, The optional Block word set
                   3602: @subsection Other system documentation
                   3603: @c ---------------------------------------------------------------------
1.43      anton    3604: @cindex other system documentation, block words
                   3605: @cindex block words, other system documentation
1.14      anton    3606: 
                   3607: @table @i
                   3608: @item any restrictions a multiprogramming system places on the use of buffer addresses:
                   3609: No restrictions (yet).
                   3610: 
                   3611: @item the number of blocks available for source and data:
                   3612: depends on your disk space.
                   3613: 
                   3614: @end table
                   3615: 
                   3616: 
                   3617: @c =====================================================================
                   3618: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
                   3619: @section The optional Double Number word set
                   3620: @c =====================================================================
1.43      anton    3621: @cindex system documentation, double words
                   3622: @cindex double words, system documentation
1.14      anton    3623: 
                   3624: @menu
1.15      anton    3625: * double-ambcond::              Ambiguous Conditions              
1.14      anton    3626: @end menu
                   3627: 
                   3628: 
                   3629: @c ---------------------------------------------------------------------
1.15      anton    3630: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
1.14      anton    3631: @subsection Ambiguous conditions
                   3632: @c ---------------------------------------------------------------------
1.43      anton    3633: @cindex double words, ambiguous conditions
                   3634: @cindex ambiguous conditions, double words
1.14      anton    3635: 
                   3636: @table @i
1.15      anton    3637: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.43      anton    3638: @cindex @code{D>S}, @var{d} out of range of @var{n} 
1.14      anton    3639: The least significant cell of @var{d} is produced.
                   3640: 
                   3641: @end table
                   3642: 
                   3643: 
                   3644: @c =====================================================================
                   3645: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
                   3646: @section The optional Exception word set
                   3647: @c =====================================================================
1.43      anton    3648: @cindex system documentation, exception words
                   3649: @cindex exception words, system documentation
1.14      anton    3650: 
                   3651: @menu
1.15      anton    3652: * exception-idef::              Implementation Defined Options              
1.14      anton    3653: @end menu
                   3654: 
                   3655: 
                   3656: @c ---------------------------------------------------------------------
1.15      anton    3657: @node exception-idef,  , The optional Exception word set, The optional Exception word set
1.14      anton    3658: @subsection Implementation Defined Options
                   3659: @c ---------------------------------------------------------------------
1.43      anton    3660: @cindex implementation-defined options, exception words
                   3661: @cindex exception words, implementation-defined options
1.14      anton    3662: 
                   3663: @table @i
                   3664: @item @code{THROW}-codes used in the system:
1.43      anton    3665: @cindex @code{THROW}-codes used in the system
                   3666: The codes -256@minus{}-511 are used for reporting signals. The mapping
                   3667: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
                   3668: codes -512@minus{}-2047 are used for OS errors (for file and memory
                   3669: allocation operations). The mapping from OS error numbers to throw codes
                   3670: is -512@minus{}@code{errno}. One side effect of this mapping is that
                   3671: undefined OS errors produce a message with a strange number; e.g.,
                   3672: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
1.14      anton    3673: @end table
                   3674: 
                   3675: @c =====================================================================
                   3676: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
                   3677: @section The optional Facility word set
                   3678: @c =====================================================================
1.43      anton    3679: @cindex system documentation, facility words
                   3680: @cindex facility words, system documentation
1.14      anton    3681: 
                   3682: @menu
1.15      anton    3683: * facility-idef::               Implementation Defined Options               
                   3684: * facility-ambcond::            Ambiguous Conditions            
1.14      anton    3685: @end menu
                   3686: 
                   3687: 
                   3688: @c ---------------------------------------------------------------------
                   3689: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
                   3690: @subsection Implementation Defined Options
                   3691: @c ---------------------------------------------------------------------
1.43      anton    3692: @cindex implementation-defined options, facility words
                   3693: @cindex facility words, implementation-defined options
1.14      anton    3694: 
                   3695: @table @i
                   3696: @item encoding of keyboard events (@code{EKEY}):
1.43      anton    3697: @cindex keyboard events, encoding in @code{EKEY}
                   3698: @cindex @code{EKEY}, encoding of keyboard events
1.41      anton    3699: Not yet implemented.
1.14      anton    3700: 
1.43      anton    3701: @item duration of a system clock tick:
                   3702: @cindex duration of a system clock tick
                   3703: @cindex clock tick duration
1.14      anton    3704: System dependent. With respect to @code{MS}, the time is specified in
                   3705: microseconds. How well the OS and the hardware implement this, is
                   3706: another question.
                   3707: 
                   3708: @item repeatability to be expected from the execution of @code{MS}:
1.43      anton    3709: @cindex repeatability to be expected from the execution of @code{MS}
                   3710: @cindex @code{MS}, repeatability to be expected
1.14      anton    3711: System dependent. On Unix, a lot depends on load. If the system is
1.17      anton    3712: lightly loaded, and the delay is short enough that Gforth does not get
1.14      anton    3713: swapped out, the performance should be acceptable. Under MS-DOS and
                   3714: other single-tasking systems, it should be good.
                   3715: 
                   3716: @end table
                   3717: 
                   3718: 
                   3719: @c ---------------------------------------------------------------------
1.15      anton    3720: @node facility-ambcond,  , facility-idef, The optional Facility word set
1.14      anton    3721: @subsection Ambiguous conditions
                   3722: @c ---------------------------------------------------------------------
1.43      anton    3723: @cindex facility words, ambiguous conditions
                   3724: @cindex ambiguous conditions, facility words
1.14      anton    3725: 
                   3726: @table @i
                   3727: @item @code{AT-XY} can't be performed on user output device:
1.43      anton    3728: @cindex @code{AT-XY} can't be performed on user output device
1.41      anton    3729: Largely terminal dependent. No range checks are done on the arguments.
1.14      anton    3730: No errors are reported. You may see some garbage appearing, you may see
                   3731: simply nothing happen.
                   3732: 
                   3733: @end table
                   3734: 
                   3735: 
                   3736: @c =====================================================================
                   3737: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
                   3738: @section The optional File-Access word set
                   3739: @c =====================================================================
1.43      anton    3740: @cindex system documentation, file words
                   3741: @cindex file words, system documentation
1.14      anton    3742: 
                   3743: @menu
1.43      anton    3744: * file-idef::                   Implementation Defined Options
1.15      anton    3745: * file-ambcond::                Ambiguous Conditions                
1.14      anton    3746: @end menu
                   3747: 
                   3748: @c ---------------------------------------------------------------------
                   3749: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
                   3750: @subsection Implementation Defined Options
                   3751: @c ---------------------------------------------------------------------
1.43      anton    3752: @cindex implementation-defined options, file words
                   3753: @cindex file words, implementation-defined options
1.14      anton    3754: 
                   3755: @table @i
1.43      anton    3756: @item file access methods used:
                   3757: @cindex file access methods used
1.14      anton    3758: @code{R/O}, @code{R/W} and @code{BIN} work as you would
                   3759: expect. @code{W/O} translates into the C file opening mode @code{w} (or
                   3760: @code{wb}): The file is cleared, if it exists, and created, if it does
1.43      anton    3761: not (with both @code{open-file} and @code{create-file}).  Under Unix
1.14      anton    3762: @code{create-file} creates a file with 666 permissions modified by your
                   3763: umask.
                   3764: 
                   3765: @item file exceptions:
1.43      anton    3766: @cindex file exceptions
1.14      anton    3767: The file words do not raise exceptions (except, perhaps, memory access
                   3768: faults when you pass illegal addresses or file-ids).
                   3769: 
                   3770: @item file line terminator:
1.43      anton    3771: @cindex file line terminator
1.14      anton    3772: System-dependent. Gforth uses C's newline character as line
                   3773: terminator. What the actual character code(s) of this are is
                   3774: system-dependent.
                   3775: 
1.43      anton    3776: @item file name format:
                   3777: @cindex file name format
1.14      anton    3778: System dependent. Gforth just uses the file name format of your OS.
                   3779: 
                   3780: @item information returned by @code{FILE-STATUS}:
1.43      anton    3781: @cindex @code{FILE-STATUS}, returned information
1.14      anton    3782: @code{FILE-STATUS} returns the most powerful file access mode allowed
                   3783: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
                   3784: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
1.41      anton    3785: along with the returned mode.
1.14      anton    3786: 
                   3787: @item input file state after an exception when including source:
1.43      anton    3788: @cindex exception when including source
1.14      anton    3789: All files that are left via the exception are closed.
                   3790: 
                   3791: @item @var{ior} values and meaning:
1.43      anton    3792: @cindex @var{ior} values and meaning
1.15      anton    3793: The @var{ior}s returned by the file and memory allocation words are
                   3794: intended as throw codes. They typically are in the range
                   3795: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3796: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3797: 
                   3798: @item maximum depth of file input nesting:
1.43      anton    3799: @cindex maximum depth of file input nesting
                   3800: @cindex file input nesting, maximum depth
1.14      anton    3801: limited by the amount of return stack, locals/TIB stack, and the number
                   3802: of open files available. This should not give you troubles.
                   3803: 
                   3804: @item maximum size of input line:
1.43      anton    3805: @cindex maximum size of input line
                   3806: @cindex input line size, maximum
1.14      anton    3807: @code{/line}. Currently 255.
                   3808: 
                   3809: @item methods of mapping block ranges to files:
1.43      anton    3810: @cindex mapping block ranges to files
                   3811: @cindex files containing blocks
                   3812: @cindex blocks in files
1.37      anton    3813: By default, blocks are accessed in the file @file{blocks.fb} in the
                   3814: current working directory. The file can be switched with @code{USE}.
1.14      anton    3815: 
                   3816: @item number of string buffers provided by @code{S"}:
1.43      anton    3817: @cindex @code{S"}, number of string buffers
1.14      anton    3818: 1
                   3819: 
                   3820: @item size of string buffer used by @code{S"}:
1.43      anton    3821: @cindex @code{S"}, size of string buffer
1.14      anton    3822: @code{/line}. currently 255.
                   3823: 
                   3824: @end table
                   3825: 
                   3826: @c ---------------------------------------------------------------------
1.15      anton    3827: @node file-ambcond,  , file-idef, The optional File-Access word set
1.14      anton    3828: @subsection Ambiguous conditions
                   3829: @c ---------------------------------------------------------------------
1.43      anton    3830: @cindex file words, ambiguous conditions
                   3831: @cindex ambiguous conditions, file words
1.14      anton    3832: 
                   3833: @table @i
1.43      anton    3834: @item attempting to position a file outside its boundaries:
                   3835: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
1.14      anton    3836: @code{REPOSITION-FILE} is performed as usual: Afterwards,
                   3837: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
                   3838: 
                   3839: @item attempting to read from file positions not yet written:
1.43      anton    3840: @cindex reading from file positions not yet written
1.14      anton    3841: End-of-file, i.e., zero characters are read and no error is reported.
                   3842: 
                   3843: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
1.43      anton    3844: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid 
1.14      anton    3845: An appropriate exception may be thrown, but a memory fault or other
                   3846: problem is more probable.
                   3847: 
1.43      anton    3848: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
                   3849: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
                   3850: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
1.14      anton    3851: The @var{ior} produced by the operation, that discovered the problem, is
                   3852: thrown.
                   3853: 
1.43      anton    3854: @item named file cannot be opened (@code{INCLUDED}):
                   3855: @cindex @code{INCLUDED}, named file cannot be opened
1.14      anton    3856: The @var{ior} produced by @code{open-file} is thrown.
                   3857: 
                   3858: @item requesting an unmapped block number:
1.43      anton    3859: @cindex unmapped block numbers
1.14      anton    3860: There are no unmapped legal block numbers. On some operating systems,
                   3861: writing a block with a large number may overflow the file system and
                   3862: have an error message as consequence.
                   3863: 
                   3864: @item using @code{source-id} when @code{blk} is non-zero:
1.43      anton    3865: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
1.14      anton    3866: @code{source-id} performs its function. Typically it will give the id of
                   3867: the source which loaded the block. (Better ideas?)
                   3868: 
                   3869: @end table
                   3870: 
                   3871: 
                   3872: @c =====================================================================
                   3873: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15      anton    3874: @section The optional Floating-Point word set
1.14      anton    3875: @c =====================================================================
1.43      anton    3876: @cindex system documentation, floating-point words
                   3877: @cindex floating-point words, system documentation
1.14      anton    3878: 
                   3879: @menu
1.15      anton    3880: * floating-idef::               Implementation Defined Options
                   3881: * floating-ambcond::            Ambiguous Conditions            
1.14      anton    3882: @end menu
                   3883: 
                   3884: 
                   3885: @c ---------------------------------------------------------------------
                   3886: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
                   3887: @subsection Implementation Defined Options
                   3888: @c ---------------------------------------------------------------------
1.43      anton    3889: @cindex implementation-defined options, floating-point words
                   3890: @cindex floating-point words, implementation-defined options
1.14      anton    3891: 
                   3892: @table @i
1.15      anton    3893: @item format and range of floating point numbers:
1.43      anton    3894: @cindex format and range of floating point numbers
                   3895: @cindex floating point numbers, format and range
1.15      anton    3896: System-dependent; the @code{double} type of C.
1.14      anton    3897: 
1.15      anton    3898: @item results of @code{REPRESENT} when @var{float} is out of range:
1.43      anton    3899: @cindex  @code{REPRESENT}, results when @var{float} is out of range
1.15      anton    3900: System dependent; @code{REPRESENT} is implemented using the C library
                   3901: function @code{ecvt()} and inherits its behaviour in this respect.
1.14      anton    3902: 
1.15      anton    3903: @item rounding or truncation of floating-point numbers:
1.43      anton    3904: @cindex rounding of floating-point numbers
                   3905: @cindex truncation of floating-point numbers
                   3906: @cindex floating-point numbers, rounding or truncation
1.26      anton    3907: System dependent; the rounding behaviour is inherited from the hosting C
                   3908: compiler. IEEE-FP-based (i.e., most) systems by default round to
                   3909: nearest, and break ties by rounding to even (i.e., such that the last
                   3910: bit of the mantissa is 0).
1.14      anton    3911: 
1.15      anton    3912: @item size of floating-point stack:
1.43      anton    3913: @cindex floating-point stack size
1.42      anton    3914: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
                   3915: the floating-point stack (in floats). You can specify this on startup
1.43      anton    3916: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
1.14      anton    3917: 
1.15      anton    3918: @item width of floating-point stack:
1.43      anton    3919: @cindex floating-point stack width 
1.15      anton    3920: @code{1 floats}.
1.14      anton    3921: 
                   3922: @end table
                   3923: 
                   3924: 
                   3925: @c ---------------------------------------------------------------------
1.15      anton    3926: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
                   3927: @subsection Ambiguous conditions
1.14      anton    3928: @c ---------------------------------------------------------------------
1.43      anton    3929: @cindex floating-point words, ambiguous conditions
                   3930: @cindex ambiguous conditions, floating-point words
1.14      anton    3931: 
                   3932: @table @i
1.15      anton    3933: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
1.43      anton    3934: @cindex @code{df@@} or @code{df!} used with an address that is not double-float  aligned
1.37      anton    3935: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3936: alignment violations.
1.14      anton    3937: 
1.15      anton    3938: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
1.43      anton    3939: @cindex @code{f@@} used with an address that is not float aligned
                   3940: @cindex @code{f!} used with an address that is not float aligned
1.37      anton    3941: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3942: alignment violations.
1.14      anton    3943: 
1.43      anton    3944: @item floating-point result out of range:
                   3945: @cindex floating-point result out of range
1.15      anton    3946: System-dependent. Can result in a @code{-55 THROW} (Floating-point
                   3947: unidentified fault), or can produce a special value representing, e.g.,
                   3948: Infinity.
1.14      anton    3949: 
1.15      anton    3950: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
1.43      anton    3951: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned
1.15      anton    3952: System-dependent. Typically results in an alignment fault like other
                   3953: alignment violations.
1.14      anton    3954: 
1.43      anton    3955: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
                   3956: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
1.15      anton    3957: The floating-point number is converted into decimal nonetheless.
1.14      anton    3958: 
1.15      anton    3959: @item Both arguments are equal to zero (@code{FATAN2}):
1.43      anton    3960: @cindex @code{FATAN2}, both arguments are equal to zero
1.15      anton    3961: System-dependent. @code{FATAN2} is implemented using the C library
                   3962: function @code{atan2()}.
1.14      anton    3963: 
1.43      anton    3964: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
                   3965: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
1.15      anton    3966: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
                   3967: because of small errors and the tan will be a very large (or very small)
                   3968: but finite number.
1.14      anton    3969: 
1.15      anton    3970: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
1.43      anton    3971: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
1.15      anton    3972: The result is rounded to the nearest float.
1.14      anton    3973: 
1.15      anton    3974: @item dividing by zero:
1.43      anton    3975: @cindex dividing by zero, floating-point
                   3976: @cindex floating-point dividing by zero
                   3977: @cindex floating-point unidentified fault, FP divide-by-zero
1.15      anton    3978: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3979: 
1.15      anton    3980: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
1.43      anton    3981: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
1.15      anton    3982: System dependent. On IEEE-FP based systems the number is converted into
                   3983: an infinity.
1.14      anton    3984: 
1.43      anton    3985: @item @var{float}<1 (@code{FACOSH}):
                   3986: @cindex @code{FACOSH}, @var{float}<1
                   3987: @cindex floating-point unidentified fault, @code{FACOSH}
1.15      anton    3988: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3989: 
1.43      anton    3990: @item @var{float}=<-1 (@code{FLNP1}):
                   3991: @cindex @code{FLNP1}, @var{float}=<-1
                   3992: @cindex floating-point unidentified fault, @code{FLNP1}
1.15      anton    3993: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3994: negative infinity is typically produced for @var{float}=-1.
1.14      anton    3995: 
1.43      anton    3996: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
                   3997: @cindex @code{FLN}, @var{float}=<0
                   3998: @cindex @code{FLOG}, @var{float}=<0
                   3999: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
1.15      anton    4000: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   4001: negative infinity is typically produced for @var{float}=0.
1.14      anton    4002: 
1.43      anton    4003: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
                   4004: @cindex @code{FASINH}, @var{float}<0
                   4005: @cindex @code{FSQRT}, @var{float}<0
                   4006: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
1.15      anton    4007: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
                   4008: produces values for these inputs on my Linux box (Bug in the C library?)
1.14      anton    4009: 
1.43      anton    4010: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
                   4011: @cindex @code{FACOS}, |@var{float}|>1
                   4012: @cindex @code{FASIN}, |@var{float}|>1
                   4013: @cindex @code{FATANH}, |@var{float}|>1
                   4014: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
1.15      anton    4015: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    4016: 
1.43      anton    4017: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
                   4018: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
                   4019: @cindex floating-point unidentified fault, @code{F>D}
1.15      anton    4020: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    4021: 
1.15      anton    4022: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
1.43      anton    4023: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
1.15      anton    4024: This does not happen.
                   4025: @end table
1.14      anton    4026: 
                   4027: @c =====================================================================
1.15      anton    4028: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
                   4029: @section The optional Locals word set
1.14      anton    4030: @c =====================================================================
1.43      anton    4031: @cindex system documentation, locals words
                   4032: @cindex locals words, system documentation
1.14      anton    4033: 
                   4034: @menu
1.15      anton    4035: * locals-idef::                 Implementation Defined Options                 
                   4036: * locals-ambcond::              Ambiguous Conditions              
1.14      anton    4037: @end menu
                   4038: 
                   4039: 
                   4040: @c ---------------------------------------------------------------------
1.15      anton    4041: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14      anton    4042: @subsection Implementation Defined Options
                   4043: @c ---------------------------------------------------------------------
1.43      anton    4044: @cindex implementation-defined options, locals words
                   4045: @cindex locals words, implementation-defined options
1.14      anton    4046: 
                   4047: @table @i
1.15      anton    4048: @item maximum number of locals in a definition:
1.43      anton    4049: @cindex maximum number of locals in a definition
                   4050: @cindex locals, maximum number in a definition
1.15      anton    4051: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
                   4052: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
                   4053: characters. The number of locals in a definition is bounded by the size
                   4054: of locals-buffer, which contains the names of the locals.
1.14      anton    4055: 
                   4056: @end table
                   4057: 
                   4058: 
                   4059: @c ---------------------------------------------------------------------
1.15      anton    4060: @node locals-ambcond,  , locals-idef, The optional Locals word set
1.14      anton    4061: @subsection Ambiguous conditions
                   4062: @c ---------------------------------------------------------------------
1.43      anton    4063: @cindex locals words, ambiguous conditions
                   4064: @cindex ambiguous conditions, locals words
1.14      anton    4065: 
                   4066: @table @i
1.15      anton    4067: @item executing a named local in interpretation state:
1.43      anton    4068: @cindex local in interpretation state
                   4069: @cindex Interpreting a compile-only word, for a local
                   4070: Locals have no interpretation semantics. If you try to perform the
                   4071: interpretation semantics, you will get a @code{-14 throw} somewhere
                   4072: (Interpreting a compile-only word). If you perform the compilation
                   4073: semantics, the locals access will be compiled (irrespective of state).
1.14      anton    4074: 
1.15      anton    4075: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
1.43      anton    4076: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
                   4077: @cindex @code{TO} on non-@code{VALUE}s and non-locals
                   4078: @cindex Invalid name argument, @code{TO}
1.15      anton    4079: @code{-32 throw} (Invalid name argument)
1.14      anton    4080: 
                   4081: @end table
                   4082: 
                   4083: 
                   4084: @c =====================================================================
1.15      anton    4085: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
                   4086: @section The optional Memory-Allocation word set
1.14      anton    4087: @c =====================================================================
1.43      anton    4088: @cindex system documentation, memory-allocation words
                   4089: @cindex memory-allocation words, system documentation
1.14      anton    4090: 
                   4091: @menu
1.15      anton    4092: * memory-idef::                 Implementation Defined Options                 
1.14      anton    4093: @end menu
                   4094: 
                   4095: 
                   4096: @c ---------------------------------------------------------------------
1.15      anton    4097: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14      anton    4098: @subsection Implementation Defined Options
                   4099: @c ---------------------------------------------------------------------
1.43      anton    4100: @cindex implementation-defined options, memory-allocation words
                   4101: @cindex memory-allocation words, implementation-defined options
1.14      anton    4102: 
                   4103: @table @i
1.15      anton    4104: @item values and meaning of @var{ior}:
1.43      anton    4105: @cindex  @var{ior} values and meaning
1.15      anton    4106: The @var{ior}s returned by the file and memory allocation words are
                   4107: intended as throw codes. They typically are in the range
                   4108: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   4109: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    4110: 
                   4111: @end table
                   4112: 
                   4113: @c =====================================================================
1.15      anton    4114: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
                   4115: @section The optional Programming-Tools word set
1.14      anton    4116: @c =====================================================================
1.43      anton    4117: @cindex system documentation, programming-tools words
                   4118: @cindex programming-tools words, system documentation
1.14      anton    4119: 
                   4120: @menu
1.15      anton    4121: * programming-idef::            Implementation Defined Options            
                   4122: * programming-ambcond::         Ambiguous Conditions         
1.14      anton    4123: @end menu
                   4124: 
                   4125: 
                   4126: @c ---------------------------------------------------------------------
1.15      anton    4127: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14      anton    4128: @subsection Implementation Defined Options
                   4129: @c ---------------------------------------------------------------------
1.43      anton    4130: @cindex implementation-defined options, programming-tools words
                   4131: @cindex programming-tools words, implementation-defined options
1.14      anton    4132: 
                   4133: @table @i
1.43      anton    4134: @item ending sequence for input following @code{;CODE} and @code{CODE}:
                   4135: @cindex @code{;CODE} ending sequence
                   4136: @cindex @code{CODE} ending sequence
                   4137: @code{END-CODE}
                   4138: 
                   4139: @item manner of processing input following @code{;CODE} and @code{CODE}:
                   4140: @cindex @code{;CODE}, processing input
                   4141: @cindex @code{CODE}, processing input
                   4142: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
1.37      anton    4143: the input is processed by the text interpreter, (starting) in interpret
                   4144: state.
1.15      anton    4145: 
                   4146: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
1.43      anton    4147: @cindex @code{ASSEMBLER}, search order capability
1.37      anton    4148: The ANS Forth search order word set.
1.15      anton    4149: 
                   4150: @item source and format of display by @code{SEE}:
1.43      anton    4151: @cindex @code{SEE}, source and format of output
1.15      anton    4152: The source for @code{see} is the intermediate code used by the inner
                   4153: interpreter.  The current @code{see} tries to output Forth source code
                   4154: as well as possible.
                   4155: 
1.14      anton    4156: @end table
                   4157: 
                   4158: @c ---------------------------------------------------------------------
1.15      anton    4159: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
1.14      anton    4160: @subsection Ambiguous conditions
                   4161: @c ---------------------------------------------------------------------
1.43      anton    4162: @cindex programming-tools words, ambiguous conditions
                   4163: @cindex ambiguous conditions, programming-tools words
1.14      anton    4164: 
                   4165: @table @i
                   4166: 
1.15      anton    4167: @item deleting the compilation wordlist (@code{FORGET}):
1.43      anton    4168: @cindex @code{FORGET}, deleting the compilation wordlist
1.15      anton    4169: Not implemented (yet).
1.14      anton    4170: 
1.15      anton    4171: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
1.43      anton    4172: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
                   4173: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
                   4174: @cindex control-flow stack underflow
1.15      anton    4175: This typically results in an @code{abort"} with a descriptive error
                   4176: message (may change into a @code{-22 throw} (Control structure mismatch)
                   4177: in the future). You may also get a memory access error. If you are
                   4178: unlucky, this ambiguous condition is not caught.
                   4179: 
1.43      anton    4180: @item @var{name} can't be found (@code{FORGET}):
                   4181: @cindex @code{FORGET}, @var{name} can't be found
1.15      anton    4182: Not implemented (yet).
1.14      anton    4183: 
1.15      anton    4184: @item @var{name} not defined via @code{CREATE}:
1.43      anton    4185: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
                   4186: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
1.37      anton    4187: the execution semantics of the last defined word no matter how it was
                   4188: defined.
1.14      anton    4189: 
1.15      anton    4190: @item @code{POSTPONE} applied to @code{[IF]}:
1.43      anton    4191: @cindex @code{POSTPONE} applied to @code{[IF]}
                   4192: @cindex @code{[IF]} and @code{POSTPONE}
1.15      anton    4193: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
                   4194: equivalent to @code{[IF]}.
1.14      anton    4195: 
1.15      anton    4196: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
1.43      anton    4197: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
1.15      anton    4198: Continue in the same state of conditional compilation in the next outer
                   4199: input source. Currently there is no warning to the user about this.
1.14      anton    4200: 
1.15      anton    4201: @item removing a needed definition (@code{FORGET}):
1.43      anton    4202: @cindex @code{FORGET}, removing a needed definition
1.15      anton    4203: Not implemented (yet).
1.14      anton    4204: 
                   4205: @end table
                   4206: 
                   4207: 
                   4208: @c =====================================================================
1.15      anton    4209: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
                   4210: @section The optional Search-Order word set
1.14      anton    4211: @c =====================================================================
1.43      anton    4212: @cindex system documentation, search-order words
                   4213: @cindex search-order words, system documentation
1.14      anton    4214: 
                   4215: @menu
1.15      anton    4216: * search-idef::                 Implementation Defined Options                 
                   4217: * search-ambcond::              Ambiguous Conditions              
1.14      anton    4218: @end menu
                   4219: 
                   4220: 
                   4221: @c ---------------------------------------------------------------------
1.15      anton    4222: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14      anton    4223: @subsection Implementation Defined Options
                   4224: @c ---------------------------------------------------------------------
1.43      anton    4225: @cindex implementation-defined options, search-order words
                   4226: @cindex search-order words, implementation-defined options
1.14      anton    4227: 
                   4228: @table @i
1.15      anton    4229: @item maximum number of word lists in search order:
1.43      anton    4230: @cindex maximum number of word lists in search order
                   4231: @cindex search order, maximum depth
1.15      anton    4232: @code{s" wordlists" environment? drop .}. Currently 16.
                   4233: 
                   4234: @item minimum search order:
1.43      anton    4235: @cindex minimum search order
                   4236: @cindex search order, minimum
1.15      anton    4237: @code{root root}.
1.14      anton    4238: 
                   4239: @end table
                   4240: 
                   4241: @c ---------------------------------------------------------------------
1.15      anton    4242: @node search-ambcond,  , search-idef, The optional Search-Order word set
1.14      anton    4243: @subsection Ambiguous conditions
                   4244: @c ---------------------------------------------------------------------
1.43      anton    4245: @cindex search-order words, ambiguous conditions
                   4246: @cindex ambiguous conditions, search-order words
1.14      anton    4247: 
                   4248: @table @i
1.15      anton    4249: @item changing the compilation wordlist (during compilation):
1.43      anton    4250: @cindex changing the compilation wordlist (during compilation)
                   4251: @cindex compilation wordlist, change before definition ends
1.33      anton    4252: The word is entered into the wordlist that was the compilation wordlist
                   4253: at the start of the definition. Any changes to the name field (e.g.,
                   4254: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
                   4255: are applied to the latest defined word (as reported by @code{last} or
                   4256: @code{lastxt}), if possible, irrespective of the compilation wordlist.
1.14      anton    4257: 
1.15      anton    4258: @item search order empty (@code{previous}):
1.43      anton    4259: @cindex @code{previous}, search order empty
                   4260: @cindex Vocstack empty, @code{previous}
1.15      anton    4261: @code{abort" Vocstack empty"}.
1.14      anton    4262: 
1.15      anton    4263: @item too many word lists in search order (@code{also}):
1.43      anton    4264: @cindex @code{also}, too many word lists in search order
                   4265: @cindex Vocstack full, @code{also}
1.15      anton    4266: @code{abort" Vocstack full"}.
1.14      anton    4267: 
                   4268: @end table
1.13      anton    4269: 
1.43      anton    4270: @c ***************************************************************
1.34      anton    4271: @node Model, Integrating Gforth, ANS conformance, Top
                   4272: @chapter Model
                   4273: 
                   4274: This chapter has yet to be written. It will contain information, on
                   4275: which internal structures you can rely.
                   4276: 
1.43      anton    4277: @c ***************************************************************
1.34      anton    4278: @node Integrating Gforth, Emacs and Gforth, Model, Top
                   4279: @chapter Integrating Gforth into C programs
                   4280: 
                   4281: This is not yet implemented.
                   4282: 
                   4283: Several people like to use Forth as scripting language for applications
                   4284: that are otherwise written in C, C++, or some other language.
                   4285: 
                   4286: The Forth system ATLAST provides facilities for embedding it into
                   4287: applications; unfortunately it has several disadvantages: most
1.36      anton    4288: importantly, it is not based on ANS Forth, and it is apparently dead
1.34      anton    4289: (i.e., not developed further and not supported). The facilities
                   4290: provided by Gforth in this area are inspired by ATLASTs facilities, so
                   4291: making the switch should not be hard.
                   4292: 
                   4293: We also tried to design the interface such that it can easily be
                   4294: implemented by other Forth systems, so that we may one day arrive at a
                   4295: standardized interface. Such a standard interface would allow you to
                   4296: replace the Forth system without having to rewrite C code.
                   4297: 
                   4298: You embed the Gforth interpreter by linking with the library
                   4299: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
                   4300: global symbols in this library that belong to the interface, have the
                   4301: prefix @code{forth_}. (Global symbols that are used internally have the
                   4302: prefix @code{gforth_}).
                   4303: 
                   4304: You can include the declarations of Forth types and the functions and
1.36      anton    4305: variables of the interface with @code{#include <forth.h>}.
1.34      anton    4306: 
                   4307: Types.
1.13      anton    4308: 
1.34      anton    4309: Variables.
                   4310: 
                   4311: Data and FP Stack pointer. Area sizes.
                   4312: 
                   4313: functions.
                   4314: 
                   4315: forth_init(imagefile)
                   4316: forth_evaluate(string) exceptions?
                   4317: forth_goto(address) (or forth_execute(xt)?)
                   4318: forth_continue() (a corountining mechanism)
                   4319: 
                   4320: Adding primitives.
                   4321: 
                   4322: No checking.
                   4323: 
                   4324: Signals?
                   4325: 
                   4326: Accessing the Stacks
1.4       anton    4327: 
1.43      anton    4328: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
1.17      anton    4329: @chapter Emacs and Gforth
1.43      anton    4330: @cindex Emacs and Gforth
1.4       anton    4331: 
1.43      anton    4332: @cindex @file{gforth.el}
                   4333: @cindex @file{forth.el}
                   4334: @cindex Rydqvist, Goran
                   4335: @cindex comment editing commands
                   4336: @cindex @code{\}, editing with Emacs
                   4337: @cindex debug tracer editing commands
                   4338: @cindex @code{~~}, removal with Emacs
                   4339: @cindex Forth mode in Emacs
1.17      anton    4340: Gforth comes with @file{gforth.el}, an improved version of
1.33      anton    4341: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.4       anton    4342: improvements are a better (but still not perfect) handling of
                   4343: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8       anton    4344: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
                   4345: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
                   4346: stuff I do not use alone, even though some of it only makes sense for
                   4347: TILE. To get a description of these features, enter Forth mode and type
                   4348: @kbd{C-h m}.
1.4       anton    4349: 
1.43      anton    4350: @cindex source location of error or debugging output in Emacs
                   4351: @cindex error output, finding the source location in Emacs
                   4352: @cindex debugging output, finding the source location in Emacs
1.17      anton    4353: In addition, Gforth supports Emacs quite well: The source code locations
1.4       anton    4354: given in error messages, debugging output (from @code{~~}) and failed
                   4355: assertion messages are in the right format for Emacs' compilation mode
                   4356: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
                   4357: Manual}) so the source location corresponding to an error or other
                   4358: message is only a few keystrokes away (@kbd{C-x `} for the next error,
                   4359: @kbd{C-c C-c} for the error under the cursor).
                   4360: 
1.43      anton    4361: @cindex @file{TAGS} file
                   4362: @cindex @file{etags.fs}
                   4363: @cindex viewing the source of a word in Emacs
1.4       anton    4364: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
                   4365: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
                   4366: contains the definitions of all words defined afterwards. You can then
                   4367: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17      anton    4368: several tags files at the same time (e.g., one for the Gforth sources
1.28      anton    4369: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
                   4370: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
                   4371: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
1.33      anton    4372: @file{/usr/local/share/gforth/0.2.0/TAGS}).
1.4       anton    4373: 
1.43      anton    4374: @cindex @file{.emacs}
1.4       anton    4375: To get all these benefits, add the following lines to your @file{.emacs}
                   4376: file:
                   4377: 
                   4378: @example
                   4379: (autoload 'forth-mode "gforth.el")
                   4380: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
                   4381: @end example
                   4382: 
1.43      anton    4383: @node Image Files, Engine, Emacs and Gforth, Top
                   4384: @chapter Image Files
                   4385: @cindex image files
                   4386: @cindex @code{.fi} files
                   4387: @cindex precompiled Forth code
                   4388: @cindex dictionary in persistent form
                   4389: @cindex persistent form of dictionary
                   4390: 
                   4391: An image file is a file containing an image of the Forth dictionary,
                   4392: i.e., compiled Forth code and data residing in the dictionary.  By
                   4393: convention, we use the extension @code{.fi} for image files.
                   4394: 
                   4395: @menu
                   4396: * Image File Background::          Why have image files?
                   4397: * Non-Relocatable Image Files::    don't always work.
                   4398: * Data-Relocatable Image Files::   are better.
1.45      anton    4399: * Fully Relocatable Image Files::  better yet.
1.43      anton    4400: * Stack and Dictionary Sizes::     Setting the default sizes for an image.
                   4401: * Running Image Files::            @code{gforth -i @var{file}} or @var{file}.
                   4402: * Modifying the Startup Sequence:: and turnkey applications.
                   4403: @end menu
                   4404: 
                   4405: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
                   4406: @section Image File Background
                   4407: @cindex image file background
                   4408: 
                   4409: Our Forth system consists not only of primitives, but also of
1.45      anton    4410: definitions written in Forth. Since the Forth compiler itself belongs to
                   4411: those definitions, it is not possible to start the system with the
1.43      anton    4412: primitives and the Forth source alone. Therefore we provide the Forth
                   4413: code as an image file in nearly executable form. At the start of the
1.45      anton    4414: system a C routine loads the image file into memory, optionally
                   4415: relocates the addresses, then sets up the memory (stacks etc.) according
                   4416: to information in the image file, and starts executing Forth code.
1.43      anton    4417: 
                   4418: The image file variants represent different compromises between the
                   4419: goals of making it easy to generate image files and making them
                   4420: portable.
                   4421: 
                   4422: @cindex relocation at run-time
                   4423: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
                   4424: run-time. This avoids many of the complications discussed below (image
                   4425: files are data relocatable without further ado), but costs performance
                   4426: (one addition per memory access).
                   4427: 
                   4428: @cindex relocation at load-time
1.45      anton    4429: By contrast, our loader performs relocation at image load time. The
                   4430: loader also has to replace tokens standing for primitive calls with the
                   4431: appropriate code-field addresses (or code addresses in the case of
                   4432: direct threading).
1.43      anton    4433: 
                   4434: There are three kinds of image files, with different degrees of
                   4435: relocatability: non-relocatable, data-relocatable, and fully relocatable
                   4436: image files.
                   4437: 
                   4438: @cindex image file loader
                   4439: @cindex relocating loader
                   4440: @cindex loader for image files
                   4441: These image file variants have several restrictions in common; they are
                   4442: caused by the design of the image file loader:
                   4443: 
                   4444: @itemize @bullet
                   4445: @item
                   4446: There is only one segment; in particular, this means, that an image file
                   4447: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
                   4448: them). And the contents of the stacks are not represented, either.
                   4449: 
                   4450: @item
1.44      anton    4451: The only kinds of relocation supported are: adding the same offset to
                   4452: all cells that represent data addresses; and replacing special tokens
                   4453: with code addresses or with pieces of machine code.
1.43      anton    4454: 
                   4455: If any complex computations involving addresses are performed, the
                   4456: results cannot be represented in the image file. Several applications that
                   4457: use such computations come to mind:
                   4458: @itemize @minus
                   4459: @item
                   4460: Hashing addresses (or data structures which contain addresses) for table
                   4461: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
                   4462: purpose, you will have no problem, because the hash tables are
                   4463: recomputed automatically when the system is started. If you use your own
                   4464: hash tables, you will have to do something similar.
                   4465: 
                   4466: @item
                   4467: There's a cute implementation of doubly-linked lists that uses
                   4468: @code{XOR}ed addresses. You could represent such lists as singly-linked
                   4469: in the image file, and restore the doubly-linked representation on
                   4470: startup.@footnote{In my opinion, though, you should think thrice before
                   4471: using a doubly-linked list (whatever implementation).}
1.44      anton    4472: 
                   4473: @item
                   4474: The code addresses of run-time routines like @code{docol:} cannot be
                   4475: represented in the image file (because their tokens would be replaced by
                   4476: machine code in direct threaded implementations). As a workaround,
                   4477: compute these addresses at run-time with @code{>code-address} from the
1.45      anton    4478: executions tokens of appropriate words (see the definitions of
1.44      anton    4479: @code{docol:} and friends in @file{kernel.fs}).
1.43      anton    4480: 
                   4481: @item
                   4482: On many architectures addresses are represented in machine code in some
                   4483: shifted or mangled form. You cannot put @code{CODE} words that contain
                   4484: absolute addresses in this form in a relocatable image file. Workarounds
                   4485: are representing the address in some relative form (e.g., relative to
                   4486: the CFA, which is present in some register), or loading the address from
                   4487: a place where it is stored in a non-mangled form.
                   4488: @end itemize
                   4489: @end itemize
                   4490: 
                   4491: @node  Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
                   4492: @section Non-Relocatable Image Files
                   4493: @cindex non-relocatable image files
                   4494: @cindex image files, non-relocatable
                   4495: 
                   4496: These files are simple memory dumps of the dictionary. They are specific
                   4497: to the executable (i.e., @file{gforth} file) they were created
                   4498: with. What's worse, they are specific to the place on which the
                   4499: dictionary resided when the image was created. Now, there is no
                   4500: guarantee that the dictionary will reside at the same place the next
                   4501: time you start Gforth, so there's no guarantee that a non-relocatable
                   4502: image will work the next time (Gforth will complain instead of crashing,
                   4503: though).
                   4504: 
                   4505: You can create a non-relocatable image file with
                   4506: 
                   4507: doc-savesystem
                   4508: 
                   4509: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
                   4510: @section Data-Relocatable Image Files
                   4511: @cindex data-relocatable image files
                   4512: @cindex image files, data-relocatable
                   4513: 
                   4514: These files contain relocatable data addresses, but fixed code addresses
                   4515: (instead of tokens). They are specific to the executable (i.e.,
1.45      anton    4516: @file{gforth} file) they were created with. For direct threading on some
                   4517: architectures (e.g., the i386), data-relocatable images do not work. You
                   4518: get a data-relocatable image, if you use @file{gforth-makeimage} with a
                   4519: Gforth binary that is not doubly indirect threaded (@pxref{Fully
                   4520: Relocatable Image Files}).
                   4521: 
                   4522: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
                   4523: @section Fully Relocatable Image Files
                   4524: @cindex fully relocatable image files
                   4525: @cindex image files, fully relocatable
1.43      anton    4526: 
1.45      anton    4527: @cindex @file{kern*.fi}, relocatability
                   4528: @cindex @file{gforth.fi}, relocatability
                   4529: These image files have relocatable data addresses, and tokens for code
                   4530: addresses. They can be used with different binaries (e.g., with and
                   4531: without debugging) on the same machine, and even across machines with
                   4532: the same data formats (byte order, cell size, floating point
                   4533: format). However, they are usually specific to the version of Gforth
                   4534: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
                   4535: are fully relocatable.
                   4536: 
                   4537: There are two ways to create a fully relocatable image file:
                   4538: 
                   4539: @menu
                   4540: * gforth-makeimage::            The normal way
                   4541: * cross.fs::                    The hard way
                   4542: @end menu
                   4543: 
                   4544: @node gforth-makeimage, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
                   4545: @subsection @file{gforth-makeimage}
1.43      anton    4546: @cindex @file{comp-image.fs}
1.45      anton    4547: @cindex @file{gforth-makeimage}
1.43      anton    4548: 
1.45      anton    4549: You will usually use @file{gforth-makeimage}. If you want to create an
                   4550: image @var{file} that contains everything you would load by invoking
                   4551: Gforth with @code{gforth @var{options}}, you simply say
1.43      anton    4552: @example
1.45      anton    4553: gforth-makeimage @var{file} @var{options}
1.43      anton    4554: @end example
                   4555: 
                   4556: E.g., if you want to create an image @file{asm.fi} that has the file
                   4557: @file{asm.fs} loaded in addition to the usual stuff, you could do it
                   4558: like this:
                   4559: 
                   4560: @example
1.45      anton    4561: gforth-makeimage asm.fi asm.fs
1.43      anton    4562: @end example
                   4563: 
1.45      anton    4564: @file{gforth-makeimage} works like this: It produces two non-relocatable
                   4565: images for different addresses and then compares them. Its output
                   4566: reflects this: first you see the output (if any) of the two Gforth
                   4567: invocations that produce the nonrelocatable image files, then you see
                   4568: the output of the comparing program: It displays the offset used for
                   4569: data addresses and the offset used for code addresses;
                   4570: moreover, for each cell that cannot be represented correctly in the
                   4571: image files, it displays a line like the following one:
1.43      anton    4572: 
1.45      anton    4573: @example
                   4574:      78DC         BFFFFA50         BFFFFA40
                   4575: @end example
1.43      anton    4576: 
1.45      anton    4577: This means that at offset $78dc from @code{forthstart}, one input image
                   4578: contains $bffffa50, and the other contains $bffffa40. Since these cells
                   4579: cannot be represented correctly in the output image, you should examine
                   4580: these places in the dictionary and verify that these cells are dead
                   4581: (i.e., not read before they are written).
1.43      anton    4582: 
1.46    ! anton    4583: \cindex @code{savesystem} during @file{gforth-makeimage}
        !          4584: \cindex @code{bye} during @file{gforth-makeimage}
        !          4585: \cindex doubly indirect threaded code
        !          4586: \cindex environment variable @code{GFORTHD}
        !          4587: \cindex @code{GFORTHD} environment variable
        !          4588: \cindex @code{gforth-ditc}
1.45      anton    4589: There are a few wrinkles: After processing the passed @var{options}, the
                   4590: words @code{savesystem} and @code{bye} must be visible. A special doubly
                   4591: indirect threaded version of the @file{gforth} executable is used for
                   4592: creating the nonrelocatable images; you can pass the exact filename of
                   4593: this executable through the environment variable @code{GFORTHD}
                   4594: (default: @file{gforth-ditc}); if you pass a version that is not doubly
                   4595: indirect threaded, you will not get a fully relocatable image, but a
                   4596: data-relocatable image (because there is no code address offset).
                   4597: 
                   4598: @node cross.fs,  , gforth-makeimage, Fully Relocatable Image Files
                   4599: @subsection @file{cross.fs}
                   4600: @cindex @file{cross.fs}
1.43      anton    4601: @cindex cross-compiler
                   4602: @cindex metacompiler
1.45      anton    4603: 
                   4604: You can also use @code{cross}, a batch compiler that accepts a Forth-like
                   4605: programming language. This @code{cross} language has to be documented
1.43      anton    4606: yet.
                   4607: 
                   4608: @cindex target compiler
1.45      anton    4609: @code{cross} also allows you to create image files for machines with
                   4610: different data sizes and data formats than the one used for generating
                   4611: the image file. You can also use it to create an application image that
                   4612: does not contain a Forth compiler. These features are bought with
                   4613: restrictions and inconveniences in programming. E.g., addresses have to
                   4614: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
                   4615: order to make the code relocatable.
                   4616: 
1.43      anton    4617: 
                   4618: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
                   4619: @section Stack and Dictionary Sizes
                   4620: @cindex image file, stack and dictionary sizes
                   4621: @cindex dictionary size default
                   4622: @cindex stack size default
                   4623: 
                   4624: If you invoke Gforth with a command line flag for the size
                   4625: (@pxref{Invoking Gforth}), the size you specify is stored in the
1.45      anton    4626: dictionary. If you save the dictionary with @code{savesystem} or create
                   4627: an image with @file{gforth-makeimage}, this size will become the default
                   4628: for the resulting image file. E.g., the following will create a
                   4629: fully relocatable version of gforth.fi with a 1MB dictionary:
1.43      anton    4630: 
                   4631: @example
1.45      anton    4632: gforth-makeimage gforth.fi -m 1M
1.43      anton    4633: @end example
                   4634: 
                   4635: In other words, if you want to set the default size for the dictionary
1.45      anton    4636: and the stacks of an image, just invoke @file{gforth-makeimage} with the
                   4637: appropriate options when creating the image.
1.43      anton    4638: 
                   4639: @cindex stack size, cache-friendly
                   4640: Note: For cache-friendly behaviour (i.e., good performance), you should
                   4641: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
                   4642: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
                   4643: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
                   4644: 
                   4645: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
                   4646: @section Running Image Files
                   4647: @cindex running image files
                   4648: @cindex invoking image files
                   4649: @cindex image file invocation
                   4650: 
                   4651: @cindex -i, invoke image file
                   4652: @cindex --image file, invoke image file
                   4653: You can invoke Gforth with an image file @var{image} instead of the
                   4654: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
                   4655: @example
                   4656: gforth -i @var{image}
                   4657: @end example
                   4658: 
                   4659: @cindex executable image file
                   4660: @cindex image files, executable
                   4661: If your operating system supports starting scripts with a line of the
                   4662: form @code{#! ...}, you can make your image file executable, and you'll
                   4663: just have to type the image file name to start Gforth with this image
                   4664: file (note that the file extension @code{.fi} is just a convention).
                   4665: 
                   4666: I.e., in most Unix systems, you just have to make the image file
                   4667: @var{image} executable with
                   4668: 
                   4669: @example
                   4670: chmod +x @var{image}
                   4671: @end example
                   4672: 
                   4673: and then you can invoke it by simply typing @var{image} instead of
                   4674: @code{gforth -i @var{image}}.
                   4675: 
                   4676: @node Modifying the Startup Sequence,  , Running Image Files, Image Files
                   4677: @section Modifying the Startup Sequence
                   4678: @cindex startup sequence for image file
                   4679: @cindex image file initialization sequence
                   4680: @cindex initialization sequence of image file
                   4681: 
                   4682: You can add your own initialization to the startup sequence through the
                   4683: deferred word
                   4684: 
                   4685: doc-'cold
                   4686: 
                   4687: @code{'cold} is invoked just before the image-specific command line
1.45      anton    4688: processing (by default, loading files and evaluating (@code{-e}) strings)
1.43      anton    4689: starts.
                   4690: 
                   4691: A sequence for adding your initialization usually looks like this:
                   4692: 
                   4693: @example
                   4694: :noname
                   4695:     Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
                   4696:     ... \ your stuff
                   4697: ; IS 'cold
                   4698: @end example
                   4699: 
                   4700: @cindex turnkey image files
                   4701: @cindex image files, turnkey applications
                   4702: You can make a turnkey image by letting @code{'cold} execute a word
                   4703: (your turnkey application) that never returns; instead, it exits Gforth
                   4704: via @code{bye} or @code{throw}.
                   4705: 
                   4706: @c ******************************************************************
                   4707: @node Engine, Bugs, Image Files, Top
                   4708: @chapter Engine
                   4709: @cindex engine
                   4710: @cindex virtual machine
1.3       anton    4711: 
1.17      anton    4712: Reading this section is not necessary for programming with Gforth. It
1.43      anton    4713: may be helpful for finding your way in the Gforth sources.
1.3       anton    4714: 
1.24      anton    4715: The ideas in this section have also been published in the papers
                   4716: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
                   4717: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
                   4718: Ertl, presented at EuroForth '93; the latter is available at
                   4719: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
                   4720: 
1.4       anton    4721: @menu
                   4722: * Portability::                 
                   4723: * Threading::                   
                   4724: * Primitives::                  
1.17      anton    4725: * Performance::                 
1.4       anton    4726: @end menu
                   4727: 
1.43      anton    4728: @node Portability, Threading, Engine, Engine
1.3       anton    4729: @section Portability
1.43      anton    4730: @cindex engine portability
1.3       anton    4731: 
                   4732: One of the main goals of the effort is availability across a wide range
                   4733: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
                   4734: this goal by manually coding the engine in assembly language for several
                   4735: then-popular processors. This approach is very labor-intensive and the
                   4736: results are short-lived due to progress in computer architecture.
                   4737: 
1.43      anton    4738: @cindex C, using C for the engine
1.3       anton    4739: Others have avoided this problem by coding in C, e.g., Mitch Bradley
                   4740: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
                   4741: particularly popular for UNIX-based Forths due to the large variety of
                   4742: architectures of UNIX machines. Unfortunately an implementation in C
                   4743: does not mix well with the goals of efficiency and with using
                   4744: traditional techniques: Indirect or direct threading cannot be expressed
                   4745: in C, and switch threading, the fastest technique available in C, is
1.43      anton    4746: significantly slower. Another problem with C is that it is very
1.3       anton    4747: cumbersome to express double integer arithmetic.
                   4748: 
1.43      anton    4749: @cindex GNU C for the engine
                   4750: @cindex long long
1.3       anton    4751: Fortunately, there is a portable language that does not have these
                   4752: limitations: GNU C, the version of C processed by the GNU C compiler
                   4753: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
                   4754: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
                   4755: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
                   4756: threading possible, its @code{long long} type (@pxref{Long Long, ,
1.33      anton    4757: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
1.32      anton    4758: double numbers@footnote{Unfortunately, long longs are not implemented
                   4759: properly on all machines (e.g., on alpha-osf1, long longs are only 64
                   4760: bits, the same size as longs (and pointers), but they should be twice as
                   4761: long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
                   4762: C Manual}). So, we had to implement doubles in C after all. Still, on
                   4763: most machines we can use long longs and achieve better performance than
                   4764: with the emulation package.}. GNU C is available for free on all
                   4765: important (and many unimportant) UNIX machines, VMS, 80386s running
                   4766: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
                   4767: on all these machines.
1.3       anton    4768: 
                   4769: Writing in a portable language has the reputation of producing code that
                   4770: is slower than assembly. For our Forth engine we repeatedly looked at
                   4771: the code produced by the compiler and eliminated most compiler-induced
1.43      anton    4772: inefficiencies by appropriate changes in the source code.
1.3       anton    4773: 
1.43      anton    4774: @cindex explicit register declarations
                   4775: @cindex --enable-force-reg, configuration flag
                   4776: @cindex -DFORCE_REG
1.3       anton    4777: However, register allocation cannot be portably influenced by the
                   4778: programmer, leading to some inefficiencies on register-starved
                   4779: machines. We use explicit register declarations (@pxref{Explicit Reg
                   4780: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
                   4781: improve the speed on some machines. They are turned on by using the
1.43      anton    4782: configuration flag @code{--enable-force-reg} (@code{gcc} switch
                   4783: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
                   4784: machine, but also on the compiler version: On some machines some
                   4785: compiler versions produce incorrect code when certain explicit register
                   4786: declarations are used. So by default @code{-DFORCE_REG} is not used.
1.3       anton    4787: 
1.43      anton    4788: @node Threading, Primitives, Portability, Engine
1.3       anton    4789: @section Threading
1.43      anton    4790: @cindex inner interpreter implementation
                   4791: @cindex threaded code implementation
1.3       anton    4792: 
1.43      anton    4793: @cindex labels as values
1.3       anton    4794: GNU C's labels as values extension (available since @code{gcc-2.0},
                   4795: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
                   4796: makes it possible to take the address of @var{label} by writing
                   4797: @code{&&@var{label}}.  This address can then be used in a statement like
                   4798: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
                   4799: @code{goto x}.
                   4800: 
1.43      anton    4801: @cindex NEXT, indirect threaded
                   4802: @cindex indirect threaded inner interpreter
                   4803: @cindex inner interpreter, indirect threaded
1.3       anton    4804: With this feature an indirect threaded NEXT looks like:
                   4805: @example
                   4806: cfa = *ip++;
                   4807: ca = *cfa;
                   4808: goto *ca;
                   4809: @end example
1.43      anton    4810: @cindex instruction pointer
1.3       anton    4811: For those unfamiliar with the names: @code{ip} is the Forth instruction
                   4812: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
                   4813: execution token and points to the code field of the next word to be
                   4814: executed; The @code{ca} (code address) fetched from there points to some
                   4815: executable code, e.g., a primitive or the colon definition handler
                   4816: @code{docol}.
                   4817: 
1.43      anton    4818: @cindex NEXT, direct threaded
                   4819: @cindex direct threaded inner interpreter
                   4820: @cindex inner interpreter, direct threaded
1.3       anton    4821: Direct threading is even simpler:
                   4822: @example
                   4823: ca = *ip++;
                   4824: goto *ca;
                   4825: @end example
                   4826: 
                   4827: Of course we have packaged the whole thing neatly in macros called
                   4828: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
                   4829: 
1.4       anton    4830: @menu
                   4831: * Scheduling::                  
                   4832: * Direct or Indirect Threaded?::  
                   4833: * DOES>::                       
                   4834: @end menu
                   4835: 
                   4836: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3       anton    4837: @subsection Scheduling
1.43      anton    4838: @cindex inner interpreter optimization
1.3       anton    4839: 
                   4840: There is a little complication: Pipelined and superscalar processors,
                   4841: i.e., RISC and some modern CISC machines can process independent
                   4842: instructions while waiting for the results of an instruction. The
                   4843: compiler usually reorders (schedules) the instructions in a way that
                   4844: achieves good usage of these delay slots. However, on our first tries
                   4845: the compiler did not do well on scheduling primitives. E.g., for
                   4846: @code{+} implemented as
                   4847: @example
                   4848: n=sp[0]+sp[1];
                   4849: sp++;
                   4850: sp[0]=n;
                   4851: NEXT;
                   4852: @end example
                   4853: the NEXT comes strictly after the other code, i.e., there is nearly no
                   4854: scheduling. After a little thought the problem becomes clear: The
                   4855: compiler cannot know that sp and ip point to different addresses (and
1.4       anton    4856: the version of @code{gcc} we used would not know it even if it was
                   4857: possible), so it could not move the load of the cfa above the store to
                   4858: the TOS. Indeed the pointers could be the same, if code on or very near
                   4859: the top of stack were executed. In the interest of speed we chose to
                   4860: forbid this probably unused ``feature'' and helped the compiler in
                   4861: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
                   4862: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3       anton    4863: @example
                   4864: n=sp[0]+sp[1];
                   4865: sp++;
                   4866: NEXT_P1;
                   4867: sp[0]=n;
                   4868: NEXT_P2;
                   4869: @end example
1.4       anton    4870: This can be scheduled optimally by the compiler.
1.3       anton    4871: 
                   4872: This division can be turned off with the switch @code{-DCISC_NEXT}. This
                   4873: switch is on by default on machines that do not profit from scheduling
                   4874: (e.g., the 80386), in order to preserve registers.
                   4875: 
1.4       anton    4876: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3       anton    4877: @subsection Direct or Indirect Threaded?
1.43      anton    4878: @cindex threading, direct or indirect?
1.3       anton    4879: 
1.43      anton    4880: @cindex -DDIRECT_THREADED
1.3       anton    4881: Both! After packaging the nasty details in macro definitions we
                   4882: realized that we could switch between direct and indirect threading by
                   4883: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
                   4884: defining a few machine-specific macros for the direct-threading case.
                   4885: On the Forth level we also offer access words that hide the
                   4886: differences between the threading methods (@pxref{Threading Words}).
                   4887: 
1.43      anton    4888: Indirect threading is implemented completely machine-independently.
                   4889: Direct threading needs routines for creating jumps to the executable
                   4890: code (e.g. to docol or dodoes). These routines are inherently
                   4891: machine-dependent, but they do not amount to many source lines. I.e.,
                   4892: even porting direct threading to a new machine is a small effort.
                   4893: 
                   4894: @cindex --enable-indirect-threaded, configuration flag
                   4895: @cindex --enable-direct-threaded, configuration flag
                   4896: The default threading method is machine-dependent. You can enforce a
                   4897: specific threading method when building Gforth with the configuration
                   4898: flag @code{--enable-direct-threaded} or
                   4899: @code{--enable-indirect-threaded}. Note that direct threading is not
                   4900: supported on all machines.
1.3       anton    4901: 
1.4       anton    4902: @node DOES>,  , Direct or Indirect Threaded?, Threading
1.3       anton    4903: @subsection DOES>
1.43      anton    4904: @cindex @code{DOES>} implementation
                   4905: 
                   4906: @cindex dodoes routine
                   4907: @cindex DOES-code
1.3       anton    4908: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
                   4909: the chunk of code executed by every word defined by a
                   4910: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
1.43      anton    4911: the Forth code to be executed, i.e. the code after the
                   4912: @code{DOES>} (the DOES-code)? There are two solutions:
1.3       anton    4913: 
                   4914: In fig-Forth the code field points directly to the dodoes and the
1.43      anton    4915: DOES-code address is stored in the cell after the code address (i.e. at
                   4916: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
                   4917: the Forth-79 and all later standards, because in fig-Forth this address
1.3       anton    4918: lies in the body (which is illegal in these standards). However, by
                   4919: making the code field larger for all words this solution becomes legal
1.43      anton    4920: again. We use this approach for the indirect threaded version and for
                   4921: direct threading on some machines. Leaving a cell unused in most words
                   4922: is a bit wasteful, but on the machines we are targeting this is hardly a
                   4923: problem. The other reason for having a code field size of two cells is
                   4924: to avoid having different image files for direct and indirect threaded
                   4925: systems (direct threaded systems require two-cell code fields on many
                   4926: machines).
1.3       anton    4927: 
1.43      anton    4928: @cindex DOES-handler
1.3       anton    4929: The other approach is that the code field points or jumps to the cell
                   4930: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
1.43      anton    4931: this address (the DOES-handler). @code{dodoes} can then get the
                   4932: DOES-code address by computing the code address, i.e., the address of
                   4933: the jump to dodoes, and add the length of that jump field. A variant of
                   4934: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
                   4935: return address (which can be found in the return register on RISCs) is
                   4936: the DOES-code address. Since the two cells available in the code field
                   4937: are used up by the jump to the code address in direct threading on many
                   4938: architectures, we use this approach for direct threading on these
                   4939: architectures. We did not want to add another cell to the code field.
1.3       anton    4940: 
1.43      anton    4941: @node Primitives, Performance, Threading, Engine
1.3       anton    4942: @section Primitives
1.43      anton    4943: @cindex primitives, implementation
                   4944: @cindex virtual machine instructions, implementation
1.3       anton    4945: 
1.4       anton    4946: @menu
                   4947: * Automatic Generation::        
                   4948: * TOS Optimization::            
                   4949: * Produced code::               
                   4950: @end menu
                   4951: 
                   4952: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3       anton    4953: @subsection Automatic Generation
1.43      anton    4954: @cindex primitives, automatic generation
1.3       anton    4955: 
1.43      anton    4956: @cindex @file{prims2x.fs}
1.3       anton    4957: Since the primitives are implemented in a portable language, there is no
                   4958: longer any need to minimize the number of primitives. On the contrary,
1.43      anton    4959: having many primitives has an advantage: speed. In order to reduce the
1.3       anton    4960: number of errors in primitives and to make programming them easier, we
                   4961: provide a tool, the primitive generator (@file{prims2x.fs}), that
                   4962: automatically generates most (and sometimes all) of the C code for a
                   4963: primitive from the stack effect notation.  The source for a primitive
                   4964: has the following form:
                   4965: 
1.43      anton    4966: @cindex primitive source format
1.3       anton    4967: @format
                   4968: @var{Forth-name}       @var{stack-effect}      @var{category}  [@var{pronounc.}]
                   4969: [@code{""}@var{glossary entry}@code{""}]
                   4970: @var{C code}
                   4971: [@code{:}
                   4972: @var{Forth code}]
                   4973: @end format
                   4974: 
                   4975: The items in brackets are optional. The category and glossary fields
                   4976: are there for generating the documentation, the Forth code is there
                   4977: for manual implementations on machines without GNU C. E.g., the source
                   4978: for the primitive @code{+} is:
                   4979: @example
                   4980: +    n1 n2 -- n    core    plus
                   4981: n = n1+n2;
                   4982: @end example
                   4983: 
                   4984: This looks like a specification, but in fact @code{n = n1+n2} is C
                   4985: code. Our primitive generation tool extracts a lot of information from
                   4986: the stack effect notations@footnote{We use a one-stack notation, even
                   4987: though we have separate data and floating-point stacks; The separate
                   4988: notation can be generated easily from the unified notation.}: The number
                   4989: of items popped from and pushed on the stack, their type, and by what
                   4990: name they are referred to in the C code. It then generates a C code
                   4991: prelude and postlude for each primitive. The final C code for @code{+}
                   4992: looks like this:
                   4993: 
                   4994: @example
                   4995: I_plus:        /* + ( n1 n2 -- n ) */  /* label, stack effect */
                   4996: /*  */                          /* documentation */
1.4       anton    4997: @{
1.3       anton    4998: DEF_CA                          /* definition of variable ca (indirect threading) */
                   4999: Cell n1;                        /* definitions of variables */
                   5000: Cell n2;
                   5001: Cell n;
                   5002: n1 = (Cell) sp[1];              /* input */
                   5003: n2 = (Cell) TOS;
                   5004: sp += 1;                        /* stack adjustment */
                   5005: NAME("+")                       /* debugging output (with -DDEBUG) */
1.4       anton    5006: @{
1.3       anton    5007: n = n1+n2;                      /* C code taken from the source */
1.4       anton    5008: @}
1.3       anton    5009: NEXT_P1;                        /* NEXT part 1 */
                   5010: TOS = (Cell)n;                  /* output */
                   5011: NEXT_P2;                        /* NEXT part 2 */
1.4       anton    5012: @}
1.3       anton    5013: @end example
                   5014: 
                   5015: This looks long and inefficient, but the GNU C compiler optimizes quite
                   5016: well and produces optimal code for @code{+} on, e.g., the R3000 and the
                   5017: HP RISC machines: Defining the @code{n}s does not produce any code, and
                   5018: using them as intermediate storage also adds no cost.
                   5019: 
                   5020: There are also other optimizations, that are not illustrated by this
                   5021: example: Assignments between simple variables are usually for free (copy
                   5022: propagation). If one of the stack items is not used by the primitive
                   5023: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
                   5024: (dead code elimination). On the other hand, there are some things that
                   5025: the compiler does not do, therefore they are performed by
                   5026: @file{prims2x.fs}: The compiler does not optimize code away that stores
                   5027: a stack item to the place where it just came from (e.g., @code{over}).
                   5028: 
                   5029: While programming a primitive is usually easy, there are a few cases
                   5030: where the programmer has to take the actions of the generator into
                   5031: account, most notably @code{?dup}, but also words that do not (always)
                   5032: fall through to NEXT.
                   5033: 
1.4       anton    5034: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3       anton    5035: @subsection TOS Optimization
1.43      anton    5036: @cindex TOS optimization for primitives
                   5037: @cindex primitives, keeping the TOS in a register
1.3       anton    5038: 
                   5039: An important optimization for stack machine emulators, e.g., Forth
                   5040: engines, is keeping  one or more of the top stack items in
1.4       anton    5041: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
                   5042: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.34      anton    5043: @itemize @bullet
1.3       anton    5044: @item
                   5045: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
                   5046: due to fewer loads from and stores to the stack.
                   5047: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
                   5048: @var{y<n}, due to additional moves between registers.
                   5049: @end itemize
                   5050: 
1.43      anton    5051: @cindex -DUSE_TOS
                   5052: @cindex -DUSE_NO_TOS
1.3       anton    5053: In particular, keeping one item in a register is never a disadvantage,
                   5054: if there are enough registers. Keeping two items in registers is a
                   5055: disadvantage for frequent words like @code{?branch}, constants,
                   5056: variables, literals and @code{i}. Therefore our generator only produces
                   5057: code that keeps zero or one items in registers. The generated C code
                   5058: covers both cases; the selection between these alternatives is made at
                   5059: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
                   5060: code for @code{+} is just a simple variable name in the one-item case,
                   5061: otherwise it is a macro that expands into @code{sp[0]}. Note that the
                   5062: GNU C compiler tries to keep simple variables like @code{TOS} in
                   5063: registers, and it usually succeeds, if there are enough registers.
                   5064: 
1.43      anton    5065: @cindex -DUSE_FTOS
                   5066: @cindex -DUSE_NO_FTOS
1.3       anton    5067: The primitive generator performs the TOS optimization for the
                   5068: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
                   5069: operations the benefit of this optimization is even larger:
                   5070: floating-point operations take quite long on most processors, but can be
                   5071: performed in parallel with other operations as long as their results are
                   5072: not used. If the FP-TOS is kept in a register, this works. If
                   5073: it is kept on the stack, i.e., in memory, the store into memory has to
                   5074: wait for the result of the floating-point operation, lengthening the
                   5075: execution time of the primitive considerably.
                   5076: 
                   5077: The TOS optimization makes the automatic generation of primitives a
                   5078: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
                   5079: @code{TOS} is not sufficient. There are some special cases to
                   5080: consider:
1.34      anton    5081: @itemize @bullet
1.3       anton    5082: @item In the case of @code{dup ( w -- w w )} the generator must not
                   5083: eliminate the store to the original location of the item on the stack,
                   5084: if the TOS optimization is turned on.
1.4       anton    5085: @item Primitives with stack effects of the form @code{--}
                   5086: @var{out1}...@var{outy} must store the TOS to the stack at the start.
                   5087: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3       anton    5088: must load the TOS from the stack at the end. But for the null stack
                   5089: effect @code{--} no stores or loads should be generated.
                   5090: @end itemize
                   5091: 
1.4       anton    5092: @node Produced code,  , TOS Optimization, Primitives
1.3       anton    5093: @subsection Produced code
1.43      anton    5094: @cindex primitives, assembly code listing
1.3       anton    5095: 
1.43      anton    5096: @cindex @file{engine.s}
1.3       anton    5097: To see what assembly code is produced for the primitives on your machine
                   5098: with your compiler and your flag settings, type @code{make engine.s} and
1.4       anton    5099: look at the resulting file @file{engine.s}.
1.3       anton    5100: 
1.43      anton    5101: @node  Performance,  , Primitives, Engine
1.17      anton    5102: @section Performance
1.43      anton    5103: @cindex performance of some Forth interpreters
                   5104: @cindex engine performance
                   5105: @cindex benchmarking Forth systems
                   5106: @cindex Gforth performance
1.17      anton    5107: 
                   5108: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
                   5109: impossible to write a significantly faster engine.
                   5110: 
                   5111: On register-starved machines like the 386 architecture processors
                   5112: improvements are possible, because @code{gcc} does not utilize the
                   5113: registers as well as a human, even with explicit register declarations;
                   5114: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
                   5115: and hand-tuned it for the 486; this system is 1.19 times faster on the
                   5116: Sieve benchmark on a 486DX2/66 than Gforth compiled with
                   5117: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
                   5118: 
1.43      anton    5119: @cindex Win32Forth performance
                   5120: @cindex NT Forth performance
                   5121: @cindex eforth performance
                   5122: @cindex ThisForth performance
                   5123: @cindex PFE performance
                   5124: @cindex TILE performance
1.17      anton    5125: However, this potential advantage of assembly language implementations
                   5126: is not necessarily realized in complete Forth systems: We compared
1.26      anton    5127: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
                   5128: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
                   5129: 1994) and Eforth (with and without peephole (aka pinhole) optimization
                   5130: of the threaded code); all these systems were written in assembly
1.30      anton    5131: language. We also compared Gforth with three systems written in C:
1.32      anton    5132: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
                   5133: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
                   5134: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
                   5135: -fomit-frame-pointer; ThisForth employs peephole optimization of the
                   5136: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
                   5137: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
                   5138: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
                   5139: 486DX2/66 with similar memory performance under Windows NT. Marcel
                   5140: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
                   5141: added the peephole optimizer, ran the benchmarks and reported the
                   5142: results.
1.17      anton    5143:  
                   5144: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
                   5145: matrix multiplication come from the Stanford integer benchmarks and have
                   5146: been translated into Forth by Martin Fraeman; we used the versions
1.30      anton    5147: included in the TILE Forth package, but with bigger data set sizes; and
                   5148: a recursive Fibonacci number computation for benchmarking calling
                   5149: performance. The following table shows the time taken for the benchmarks
                   5150: scaled by the time taken by Gforth (in other words, it shows the speedup
                   5151: factor that Gforth achieved over the other systems).
1.17      anton    5152: 
                   5153: @example
1.30      anton    5154: relative      Win32-    NT       eforth       This-
                   5155:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
1.32      anton    5156: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
                   5157: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
1.38      anton    5158: matmul    1.00  1.47  1.35   1.46  0.74  1.58        4.09
                   5159: fib       1.00  1.52  1.34   1.22  0.86  1.74  2.99  4.30
1.17      anton    5160: @end example
                   5161: 
                   5162: You may find the good performance of Gforth compared with the systems
                   5163: written in assembly language quite surprising. One important reason for
                   5164: the disappointing performance of these systems is probably that they are
                   5165: not written optimally for the 486 (e.g., they use the @code{lods}
                   5166: instruction). In addition, Win32Forth uses a comfortable, but costly
                   5167: method for relocating the Forth image: like @code{cforth}, it computes
                   5168: the actual addresses at run time, resulting in two address computations
1.43      anton    5169: per NEXT (@pxref{Image File Background}).
1.17      anton    5170: 
1.26      anton    5171: Only Eforth with the peephole optimizer performs comparable to
                   5172: Gforth. The speedups achieved with peephole optimization of threaded
                   5173: code are quite remarkable. Adding a peephole optimizer to Gforth should
                   5174: cause similar speedups.
                   5175: 
1.30      anton    5176: The speedup of Gforth over PFE, ThisForth and TILE can be easily
1.43      anton    5177: explained with the self-imposed restriction of the latter systems to
                   5178: standard C, which makes efficient threading impossible (however, the
                   5179: measured implementation of PFE uses a GNU C extension: @ref{Global Reg
                   5180: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
                   5181: Moreover, current C compilers have a hard time optimizing other aspects
                   5182: of the ThisForth and the TILE source.
1.17      anton    5183: 
                   5184: Note that the performance of Gforth on 386 architecture processors
                   5185: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
                   5186: failed to allocate any of the virtual machine registers into real
                   5187: machine registers by itself and would not work correctly with explicit
                   5188: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
                   5189: running the Sieve) than the one measured above.
                   5190: 
1.43      anton    5191: Note also that there have been several releases of Win32Forth since the
                   5192: release presented here, so the results presented here may have little
                   5193: predictive value for the performance of Win32Forth today.
                   5194: 
                   5195: @cindex @file{Benchres}
1.26      anton    5196: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
                   5197: Maierhofer (presented at EuroForth '95), an indirect threaded version of
                   5198: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
1.46    ! anton    5199: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
        !          5200: threaded version used here. The paper available at
1.24      anton    5201: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
1.43      anton    5202: it also contains numbers for some native code systems. You can find a
                   5203: newer version of these measurements at
                   5204: @file{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
                   5205: find numbers for Gforth on various machines in @file{Benchres}.
1.24      anton    5206: 
1.43      anton    5207: @node Bugs, Origin, Engine, Top
1.4       anton    5208: @chapter Bugs
1.43      anton    5209: @cindex bug reporting
1.4       anton    5210: 
1.17      anton    5211: Known bugs are described in the file BUGS in the Gforth distribution.
                   5212: 
1.24      anton    5213: If you find a bug, please send a bug report to
1.32      anton    5214: @code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
1.17      anton    5215: describe the Gforth version used (it is announced at the start of an
                   5216: interactive Gforth session), the machine and operating system (on Unix
                   5217: systems you can use @code{uname -a} to produce this information), the
1.43      anton    5218: installation options (send the @file{config.status} file), and a
1.24      anton    5219: complete list of changes you (or your installer) have made to the Gforth
                   5220: sources (if any); it should contain a program (or a sequence of keyboard
                   5221: commands) that reproduces the bug and a description of what you think
                   5222: constitutes the buggy behaviour.
1.17      anton    5223: 
                   5224: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
                   5225: to Report Bugs, gcc.info, GNU C Manual}.
                   5226: 
                   5227: 
1.29      anton    5228: @node Origin, Word Index, Bugs, Top
                   5229: @chapter Authors and Ancestors of Gforth
                   5230: 
                   5231: @section Authors and Contributors
1.43      anton    5232: @cindex authors of Gforth
                   5233: @cindex contributors to Gforth
1.29      anton    5234: 
                   5235: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
1.30      anton    5236: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
                   5237: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
                   5238: with their continuous feedback. Lennart Benshop contributed
1.29      anton    5239: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
                   5240: support for calling C libraries. Helpful comments also came from Paul
1.37      anton    5241: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.39      anton    5242: Wavrik, Barrie Stott and Marc de Groot.
1.29      anton    5243: 
1.30      anton    5244: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
                   5245: and autoconf, among others), and to the creators of the Internet: Gforth
                   5246: was developed across the Internet, and its authors have not met
                   5247: physically yet.
                   5248: 
1.29      anton    5249: @section Pedigree
1.43      anton    5250: @cindex Pedigree of Gforth
1.4       anton    5251: 
1.17      anton    5252: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24      anton    5253: Dirk Zoller) will cross-fertilize each other. Of course, a significant
                   5254: part of the design of Gforth was prescribed by ANS Forth.
1.17      anton    5255: 
1.23      pazsan   5256: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
                   5257: 32 bit native code version of VolksForth for the Atari ST, written
                   5258: mostly by Dietrich Weineck.
                   5259: 
                   5260: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
                   5261: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24      anton    5262: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17      anton    5263: 
1.34      anton    5264: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17      anton    5265: Forth-83 standard. !! Pedigree? When?
                   5266: 
                   5267: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24      anton    5268: 1979. Robert Selzer and Bill Ragsdale developed the original
                   5269: implementation of fig-Forth for the 6502 based on microForth.
                   5270: 
                   5271: The principal architect of microForth was Dean Sanderson. microForth was
1.41      anton    5272: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
1.24      anton    5273: the 1802, and subsequently implemented on the 8080, the 6800 and the
                   5274: Z80.
1.17      anton    5275: 
1.24      anton    5276: All earlier Forth systems were custom-made, usually by Charles Moore,
1.30      anton    5277: who discovered (as he puts it) Forth during the late 60s. The first full
                   5278: Forth existed in 1971.
1.17      anton    5279: 
                   5280: A part of the information in this section comes from @cite{The Evolution
                   5281: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
                   5282: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
                   5283: Notices 28(3), 1993.  You can find more historical and genealogical
                   5284: information about Forth there.
                   5285: 
1.43      anton    5286: @node Word Index, Concept Index, Origin, Top
                   5287: @unnumbered Word Index
1.4       anton    5288: 
1.18      anton    5289: This index is as incomplete as the manual. Each word is listed with
                   5290: stack effect and wordset.
1.17      anton    5291: 
                   5292: @printindex fn
                   5293: 
1.43      anton    5294: @node Concept Index,  , Word Index, Top
                   5295: @unnumbered Concept and Word Index
                   5296: 
                   5297: This index is as incomplete as the manual. Not all entries listed are
                   5298: present verbatim in the text. Only the names are listed for the words
                   5299: here.
1.17      anton    5300: 
1.43      anton    5301: @printindex cp
1.1       anton    5302: 
                   5303: @contents
                   5304: @bye
                   5305: 

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