Annotation of gforth/gforth.ds, revision 1.43

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: 
                   1374: The loop control data are kept on the return stack, so there are some
                   1375: restrictions on mixing return stack accesses and counted loop
                   1376: words. E.g., if you put values on the return stack outside the loop, you
                   1377: cannot read them inside the loop. If you put values on the return stack
                   1378: within a loop, you have to remove them before the end of the loop and
                   1379: before accessing the index of the loop.
                   1380: 
                   1381: There are several variations on the counted loop:
                   1382: 
                   1383: @code{LEAVE} leaves the innermost counted loop immediately.
                   1384: 
1.18      anton    1385: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
                   1386: (and @code{LOOP} iterates until they become equal by wrap-around
                   1387: arithmetic). This behaviour is usually not what you want. Therefore,
                   1388: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
                   1389: @code{?DO}), which do not enter the loop if @var{start} is greater than
                   1390: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.30      anton    1391: unsigned loop parameters.
1.18      anton    1392: 
1.1       anton    1393: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
                   1394: index by @var{n} instead of by 1. The loop is terminated when the border
                   1395: between @var{limit-1} and @var{limit} is crossed. E.g.:
                   1396: 
1.18      anton    1397: @code{4 0 +DO  i .  2 +LOOP}   prints @code{0 2}
1.1       anton    1398: 
1.18      anton    1399: @code{4 1 +DO  i .  2 +LOOP}   prints @code{1 3}
1.1       anton    1400: 
1.43    ! anton    1401: @cindex negative increment for counted loops
        !          1402: @cindex counted loops with negative increment
1.1       anton    1403: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
                   1404: 
1.2       anton    1405: @code{-1 0 ?DO  i .  -1 +LOOP}  prints @code{0 -1}
1.1       anton    1406: 
1.2       anton    1407: @code{ 0 0 ?DO  i .  -1 +LOOP}  prints nothing
1.1       anton    1408: 
1.18      anton    1409: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
                   1410: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
                   1411: index by @var{u} each iteration. The loop is terminated when the border
                   1412: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
                   1413: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1       anton    1414: 
1.18      anton    1415: @code{-2 0 -DO  i .  1 -LOOP}  prints @code{0 -1}
1.1       anton    1416: 
1.18      anton    1417: @code{-1 0 -DO  i .  1 -LOOP}  prints @code{0}
1.1       anton    1418: 
1.18      anton    1419: @code{ 0 0 -DO  i .  1 -LOOP}  prints nothing
1.1       anton    1420: 
1.30      anton    1421: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
                   1422: @code{-LOOP} are not in the ANS Forth standard. However, an
                   1423: implementation for these words that uses only standard words is provided
                   1424: in @file{compat/loops.fs}.
1.18      anton    1425: 
                   1426: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
                   1427: the loop, independent of the loop parameters. Do not use @code{DO}, even
                   1428: if you know that the loop is entered in any case. Such knowledge tends
                   1429: to become invalid during maintenance of a program, and then the
                   1430: @code{DO} will make trouble.
1.1       anton    1431: 
                   1432: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
                   1433: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
                   1434: return stack so @code{EXIT} can get to its return address.
                   1435: 
1.43    ! anton    1436: @cindex @code{FOR} loops
1.1       anton    1437: Another counted loop is
                   1438: @example
                   1439: @var{n}
                   1440: FOR
                   1441:   @var{body}
                   1442: NEXT
                   1443: @end example
                   1444: This is the preferred loop of native code compiler writers who are too
1.17      anton    1445: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1       anton    1446: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
                   1447: and ending with 0. Other Forth systems may behave differently, even if
1.30      anton    1448: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
                   1449: loops.
1.1       anton    1450: 
1.4       anton    1451: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2       anton    1452: @subsection Arbitrary control structures
1.43    ! anton    1453: @cindex control structures, user-defined
1.2       anton    1454: 
1.43    ! anton    1455: @cindex control-flow stack
1.2       anton    1456: ANS Forth permits and supports using control structures in a non-nested
                   1457: way. Information about incomplete control structures is stored on the
                   1458: control-flow stack. This stack may be implemented on the Forth data
1.17      anton    1459: stack, and this is what we have done in Gforth.
1.2       anton    1460: 
1.43    ! anton    1461: @cindex @code{orig}, control-flow stack item
        !          1462: @cindex @code{dest}, control-flow stack item
1.2       anton    1463: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
                   1464: entry represents a backward branch target. A few words are the basis for
                   1465: building any control structure possible (except control structures that
                   1466: need storage, like calls, coroutines, and backtracking).
                   1467: 
1.3       anton    1468: doc-if
                   1469: doc-ahead
                   1470: doc-then
                   1471: doc-begin
                   1472: doc-until
                   1473: doc-again
                   1474: doc-cs-pick
                   1475: doc-cs-roll
1.2       anton    1476: 
1.17      anton    1477: On many systems control-flow stack items take one word, in Gforth they
1.2       anton    1478: currently take three (this may change in the future). Therefore it is a
                   1479: really good idea to manipulate the control flow stack with
                   1480: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
                   1481: words.
                   1482: 
                   1483: Some standard control structure words are built from these words:
                   1484: 
1.3       anton    1485: doc-else
                   1486: doc-while
                   1487: doc-repeat
1.2       anton    1488: 
1.31      anton    1489: Gforth adds some more control-structure words:
                   1490: 
                   1491: doc-endif
                   1492: doc-?dup-if
                   1493: doc-?dup-0=-if
                   1494: 
1.2       anton    1495: Counted loop words constitute a separate group of words:
                   1496: 
1.3       anton    1497: doc-?do
1.18      anton    1498: doc-+do
                   1499: doc-u+do
                   1500: doc--do
                   1501: doc-u-do
1.3       anton    1502: doc-do
                   1503: doc-for
                   1504: doc-loop
                   1505: doc-+loop
1.18      anton    1506: doc--loop
1.3       anton    1507: doc-next
                   1508: doc-leave
                   1509: doc-?leave
                   1510: doc-unloop
1.10      anton    1511: doc-done
1.2       anton    1512: 
                   1513: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
                   1514: @i{do-sys}. Our system allows it, but it's your job to ensure that for
                   1515: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3       anton    1516: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
                   1517: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7       pazsan   1518: resolved (by using one of the loop-ending words or @code{DONE}).
1.2       anton    1519: 
                   1520: Another group of control structure words are
                   1521: 
1.3       anton    1522: doc-case
                   1523: doc-endcase
                   1524: doc-of
                   1525: doc-endof
1.2       anton    1526: 
                   1527: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
                   1528: @code{cs-roll}.
                   1529: 
1.3       anton    1530: @subsubsection Programming Style
                   1531: 
                   1532: In order to ensure readability we recommend that you do not create
                   1533: arbitrary control structures directly, but define new control structure
                   1534: words for the control structure you want and use these words in your
                   1535: program.
                   1536: 
                   1537: E.g., instead of writing
                   1538: 
                   1539: @example
                   1540: begin
                   1541:   ...
                   1542: if [ 1 cs-roll ]
                   1543:   ...
                   1544: again then
                   1545: @end example
                   1546: 
                   1547: we recommend defining control structure words, e.g.,
                   1548: 
                   1549: @example
                   1550: : while ( dest -- orig dest )
                   1551:  POSTPONE if
                   1552:  1 cs-roll ; immediate
                   1553: 
                   1554: : repeat ( orig dest -- )
                   1555:  POSTPONE again
                   1556:  POSTPONE then ; immediate
                   1557: @end example
                   1558: 
                   1559: and then using these to create the control structure:
                   1560: 
                   1561: @example
                   1562: begin
                   1563:   ...
                   1564: while
                   1565:   ...
                   1566: repeat
                   1567: @end example
                   1568: 
1.30      anton    1569: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1.3       anton    1570: @code{WHILE} are predefined, so in this example it would not be
                   1571: necessary to define them.
                   1572: 
1.4       anton    1573: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3       anton    1574: @subsection Calls and returns
1.43    ! anton    1575: @cindex calling a definition
        !          1576: @cindex returning from a definition
1.3       anton    1577: 
                   1578: A definition can be called simply be writing the name of the
1.17      anton    1579: definition. When the end of the definition is reached, it returns. An
                   1580: earlier return can be forced using
1.3       anton    1581: 
                   1582: doc-exit
                   1583: 
                   1584: Don't forget to clean up the return stack and @code{UNLOOP} any
                   1585: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
                   1586: primitive compiled by @code{EXIT} is
                   1587: 
                   1588: doc-;s
                   1589: 
1.4       anton    1590: @node Exception Handling,  , Calls and returns, Control Structures
1.3       anton    1591: @subsection Exception Handling
1.43    ! anton    1592: @cindex Exceptions
1.3       anton    1593: 
                   1594: doc-catch
                   1595: doc-throw
                   1596: 
1.4       anton    1597: @node Locals, Defining Words, Control Structures, Words
1.1       anton    1598: @section Locals
1.43    ! anton    1599: @cindex locals
1.1       anton    1600: 
1.2       anton    1601: Local variables can make Forth programming more enjoyable and Forth
                   1602: programs easier to read. Unfortunately, the locals of ANS Forth are
                   1603: laden with restrictions. Therefore, we provide not only the ANS Forth
                   1604: locals wordset, but also our own, more powerful locals wordset (we
                   1605: implemented the ANS Forth locals wordset through our locals wordset).
                   1606: 
1.24      anton    1607: The ideas in this section have also been published in the paper
                   1608: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
                   1609: at EuroForth '94; it is available at
                   1610: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
                   1611: 
1.2       anton    1612: @menu
1.17      anton    1613: * Gforth locals::               
1.4       anton    1614: * ANS Forth locals::            
1.2       anton    1615: @end menu
                   1616: 
1.17      anton    1617: @node Gforth locals, ANS Forth locals, Locals, Locals
                   1618: @subsection Gforth locals
1.43    ! anton    1619: @cindex Gforth locals
        !          1620: @cindex locals, Gforth style
1.2       anton    1621: 
                   1622: Locals can be defined with
                   1623: 
                   1624: @example
                   1625: @{ local1 local2 ... -- comment @}
                   1626: @end example
                   1627: or
                   1628: @example
                   1629: @{ local1 local2 ... @}
                   1630: @end example
                   1631: 
                   1632: E.g.,
                   1633: @example
                   1634: : max @{ n1 n2 -- n3 @}
                   1635:  n1 n2 > if
                   1636:    n1
                   1637:  else
                   1638:    n2
                   1639:  endif ;
                   1640: @end example
                   1641: 
                   1642: The similarity of locals definitions with stack comments is intended. A
                   1643: locals definition often replaces the stack comment of a word. The order
                   1644: of the locals corresponds to the order in a stack comment and everything
                   1645: after the @code{--} is really a comment.
                   1646: 
                   1647: This similarity has one disadvantage: It is too easy to confuse locals
                   1648: declarations with stack comments, causing bugs and making them hard to
                   1649: find. However, this problem can be avoided by appropriate coding
                   1650: conventions: Do not use both notations in the same program. If you do,
                   1651: they should be distinguished using additional means, e.g. by position.
                   1652: 
1.43    ! anton    1653: @cindex types of locals
        !          1654: @cindex locals types
1.2       anton    1655: The name of the local may be preceded by a type specifier, e.g.,
                   1656: @code{F:} for a floating point value:
                   1657: 
                   1658: @example
                   1659: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
                   1660: \ complex multiplication
                   1661:  Ar Br f* Ai Bi f* f-
                   1662:  Ar Bi f* Ai Br f* f+ ;
                   1663: @end example
                   1664: 
1.43    ! anton    1665: @cindex flavours of locals
        !          1666: @cindex locals flavours
        !          1667: @cindex value-flavoured locals
        !          1668: @cindex variable-flavoured locals
1.17      anton    1669: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2       anton    1670: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
                   1671: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
                   1672: with @code{W:}, @code{D:} etc.) produces its value and can be changed
                   1673: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
                   1674: produces its address (which becomes invalid when the variable's scope is
1.41      anton    1675: left). E.g., the standard word @code{emit} can be defined in terms of
1.2       anton    1676: @code{type} like this:
                   1677: 
                   1678: @example
                   1679: : emit @{ C^ char* -- @}
                   1680:     char* 1 type ;
                   1681: @end example
                   1682: 
1.43    ! anton    1683: @cindex default type of locals
        !          1684: @cindex locals, default type
1.2       anton    1685: A local without type specifier is a @code{W:} local. Both flavours of
                   1686: locals are initialized with values from the data or FP stack.
                   1687: 
                   1688: Currently there is no way to define locals with user-defined data
                   1689: structures, but we are working on it.
                   1690: 
1.17      anton    1691: Gforth allows defining locals everywhere in a colon definition. This
1.7       pazsan   1692: poses the following questions:
1.2       anton    1693: 
1.4       anton    1694: @menu
                   1695: * Where are locals visible by name?::  
1.14      anton    1696: * How long do locals live?::    
1.4       anton    1697: * Programming Style::           
                   1698: * Implementation::              
                   1699: @end menu
                   1700: 
1.17      anton    1701: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2       anton    1702: @subsubsection Where are locals visible by name?
1.43    ! anton    1703: @cindex locals visibility
        !          1704: @cindex visibility of locals
        !          1705: @cindex scope of locals
1.2       anton    1706: 
                   1707: Basically, the answer is that locals are visible where you would expect
                   1708: it in block-structured languages, and sometimes a little longer. If you
                   1709: want to restrict the scope of a local, enclose its definition in
                   1710: @code{SCOPE}...@code{ENDSCOPE}.
                   1711: 
                   1712: doc-scope
                   1713: doc-endscope
                   1714: 
                   1715: These words behave like control structure words, so you can use them
                   1716: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
                   1717: arbitrary ways.
                   1718: 
                   1719: If you want a more exact answer to the visibility question, here's the
                   1720: basic principle: A local is visible in all places that can only be
                   1721: reached through the definition of the local@footnote{In compiler
                   1722: construction terminology, all places dominated by the definition of the
                   1723: local.}. In other words, it is not visible in places that can be reached
                   1724: without going through the definition of the local. E.g., locals defined
                   1725: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
                   1726: defined in @code{BEGIN}...@code{UNTIL} are visible after the
                   1727: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
                   1728: 
                   1729: The reasoning behind this solution is: We want to have the locals
                   1730: visible as long as it is meaningful. The user can always make the
                   1731: visibility shorter by using explicit scoping. In a place that can
                   1732: only be reached through the definition of a local, the meaning of a
                   1733: local name is clear. In other places it is not: How is the local
                   1734: initialized at the control flow path that does not contain the
                   1735: definition? Which local is meant, if the same name is defined twice in
                   1736: two independent control flow paths?
                   1737: 
                   1738: This should be enough detail for nearly all users, so you can skip the
1.41      anton    1739: rest of this section. If you really must know all the gory details and
1.2       anton    1740: options, read on.
                   1741: 
                   1742: In order to implement this rule, the compiler has to know which places
                   1743: are unreachable. It knows this automatically after @code{AHEAD},
                   1744: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
                   1745: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
                   1746: compiler that the control flow never reaches that place. If
                   1747: @code{UNREACHABLE} is not used where it could, the only consequence is
                   1748: that the visibility of some locals is more limited than the rule above
                   1749: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
                   1750: lie to the compiler), buggy code will be produced.
                   1751: 
1.43    ! anton    1752: doc-unreachable
        !          1753: 
1.2       anton    1754: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3       anton    1755: does not know which locals will be visible on the incoming
                   1756: back-edge. All problems discussed in the following are due to this
                   1757: ignorance of the compiler (we discuss the problems using @code{BEGIN}
                   1758: loops as examples; the discussion also applies to @code{?DO} and other
1.2       anton    1759: loops). Perhaps the most insidious example is:
                   1760: @example
                   1761: AHEAD
                   1762: BEGIN
                   1763:   x
                   1764: [ 1 CS-ROLL ] THEN
1.4       anton    1765:   @{ x @}
1.2       anton    1766:   ...
                   1767: UNTIL
                   1768: @end example
                   1769: 
                   1770: This should be legal according to the visibility rule. The use of
                   1771: @code{x} can only be reached through the definition; but that appears
                   1772: textually below the use.
                   1773: 
                   1774: From this example it is clear that the visibility rules cannot be fully
                   1775: implemented without major headaches. Our implementation treats common
                   1776: cases as advertised and the exceptions are treated in a safe way: The
                   1777: compiler makes a reasonable guess about the locals visible after a
                   1778: @code{BEGIN}; if it is too pessimistic, the
                   1779: user will get a spurious error about the local not being defined; if the
                   1780: compiler is too optimistic, it will notice this later and issue a
                   1781: warning. In the case above the compiler would complain about @code{x}
                   1782: being undefined at its use. You can see from the obscure examples in
                   1783: this section that it takes quite unusual control structures to get the
                   1784: compiler into trouble, and even then it will often do fine.
                   1785: 
                   1786: If the @code{BEGIN} is reachable from above, the most optimistic guess
                   1787: is that all locals visible before the @code{BEGIN} will also be
                   1788: visible after the @code{BEGIN}. This guess is valid for all loops that
                   1789: are entered only through the @code{BEGIN}, in particular, for normal
                   1790: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
                   1791: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
                   1792: compiler. When the branch to the @code{BEGIN} is finally generated by
                   1793: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1.41      anton    1794: warns the user if it was too optimistic:
1.2       anton    1795: @example
                   1796: IF
1.4       anton    1797:   @{ x @}
1.2       anton    1798: BEGIN
                   1799:   \ x ? 
                   1800: [ 1 cs-roll ] THEN
                   1801:   ...
                   1802: UNTIL
                   1803: @end example
                   1804: 
                   1805: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
                   1806: optimistically assumes that it lives until the @code{THEN}. It notices
                   1807: this difference when it compiles the @code{UNTIL} and issues a
                   1808: warning. The user can avoid the warning, and make sure that @code{x}
                   1809: is not used in the wrong area by using explicit scoping:
                   1810: @example
                   1811: IF
                   1812:   SCOPE
1.4       anton    1813:   @{ x @}
1.2       anton    1814:   ENDSCOPE
                   1815: BEGIN
                   1816: [ 1 cs-roll ] THEN
                   1817:   ...
                   1818: UNTIL
                   1819: @end example
                   1820: 
                   1821: Since the guess is optimistic, there will be no spurious error messages
                   1822: about undefined locals.
                   1823: 
                   1824: If the @code{BEGIN} is not reachable from above (e.g., after
                   1825: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
                   1826: optimistic guess, as the locals visible after the @code{BEGIN} may be
                   1827: defined later. Therefore, the compiler assumes that no locals are
1.17      anton    1828: visible after the @code{BEGIN}. However, the user can use
1.2       anton    1829: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17      anton    1830: visible at the BEGIN as at the point where the top control-flow stack
                   1831: item was created.
1.2       anton    1832: 
                   1833: doc-assume-live
                   1834: 
                   1835: E.g.,
                   1836: @example
1.4       anton    1837: @{ x @}
1.2       anton    1838: AHEAD
                   1839: ASSUME-LIVE
                   1840: BEGIN
                   1841:   x
                   1842: [ 1 CS-ROLL ] THEN
                   1843:   ...
                   1844: UNTIL
                   1845: @end example
                   1846: 
                   1847: Other cases where the locals are defined before the @code{BEGIN} can be
                   1848: handled by inserting an appropriate @code{CS-ROLL} before the
                   1849: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
                   1850: behind the @code{ASSUME-LIVE}).
                   1851: 
                   1852: Cases where locals are defined after the @code{BEGIN} (but should be
                   1853: visible immediately after the @code{BEGIN}) can only be handled by
                   1854: rearranging the loop. E.g., the ``most insidious'' example above can be
                   1855: arranged into:
                   1856: @example
                   1857: BEGIN
1.4       anton    1858:   @{ x @}
1.2       anton    1859:   ... 0=
                   1860: WHILE
                   1861:   x
                   1862: REPEAT
                   1863: @end example
                   1864: 
1.17      anton    1865: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2       anton    1866: @subsubsection How long do locals live?
1.43    ! anton    1867: @cindex locals lifetime
        !          1868: @cindex lifetime of locals
1.2       anton    1869: 
                   1870: The right answer for the lifetime question would be: A local lives at
                   1871: least as long as it can be accessed. For a value-flavoured local this
                   1872: means: until the end of its visibility. However, a variable-flavoured
                   1873: local could be accessed through its address far beyond its visibility
                   1874: scope. Ultimately, this would mean that such locals would have to be
                   1875: garbage collected. Since this entails un-Forth-like implementation
                   1876: complexities, I adopted the same cowardly solution as some other
                   1877: languages (e.g., C): The local lives only as long as it is visible;
                   1878: afterwards its address is invalid (and programs that access it
                   1879: afterwards are erroneous).
                   1880: 
1.17      anton    1881: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2       anton    1882: @subsubsection Programming Style
1.43    ! anton    1883: @cindex locals programming style
        !          1884: @cindex programming style, locals
1.2       anton    1885: 
                   1886: The freedom to define locals anywhere has the potential to change
                   1887: programming styles dramatically. In particular, the need to use the
                   1888: return stack for intermediate storage vanishes. Moreover, all stack
                   1889: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
                   1890: determined arguments) can be eliminated: If the stack items are in the
                   1891: wrong order, just write a locals definition for all of them; then
                   1892: write the items in the order you want.
                   1893: 
                   1894: This seems a little far-fetched and eliminating stack manipulations is
1.4       anton    1895: unlikely to become a conscious programming objective. Still, the number
                   1896: of stack manipulations will be reduced dramatically if local variables
1.17      anton    1897: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4       anton    1898: a traditional implementation of @code{max}).
1.2       anton    1899: 
                   1900: This shows one potential benefit of locals: making Forth programs more
                   1901: readable. Of course, this benefit will only be realized if the
                   1902: programmers continue to honour the principle of factoring instead of
                   1903: using the added latitude to make the words longer.
                   1904: 
1.43    ! anton    1905: @cindex single-assignment style for locals
1.2       anton    1906: Using @code{TO} can and should be avoided.  Without @code{TO},
                   1907: every value-flavoured local has only a single assignment and many
                   1908: advantages of functional languages apply to Forth. I.e., programs are
                   1909: easier to analyse, to optimize and to read: It is clear from the
                   1910: definition what the local stands for, it does not turn into something
                   1911: different later.
                   1912: 
                   1913: E.g., a definition using @code{TO} might look like this:
                   1914: @example
                   1915: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1916:  u1 u2 min 0
                   1917:  ?do
1.36      anton    1918:    addr1 c@@ addr2 c@@ -
1.31      anton    1919:    ?dup-if
1.2       anton    1920:      unloop exit
                   1921:    then
                   1922:    addr1 char+ TO addr1
                   1923:    addr2 char+ TO addr2
                   1924:  loop
                   1925:  u1 u2 - ;
                   1926: @end example
                   1927: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
                   1928: every loop iteration. @code{strcmp} is a typical example of the
                   1929: readability problems of using @code{TO}. When you start reading
                   1930: @code{strcmp}, you think that @code{addr1} refers to the start of the
                   1931: string. Only near the end of the loop you realize that it is something
                   1932: else.
                   1933: 
                   1934: This can be avoided by defining two locals at the start of the loop that
                   1935: are initialized with the right value for the current iteration.
                   1936: @example
                   1937: : strcmp @{ addr1 u1 addr2 u2 -- n @}
                   1938:  addr1 addr2
                   1939:  u1 u2 min 0 
                   1940:  ?do @{ s1 s2 @}
1.36      anton    1941:    s1 c@@ s2 c@@ -
1.31      anton    1942:    ?dup-if
1.2       anton    1943:      unloop exit
                   1944:    then
                   1945:    s1 char+ s2 char+
                   1946:  loop
                   1947:  2drop
                   1948:  u1 u2 - ;
                   1949: @end example
                   1950: Here it is clear from the start that @code{s1} has a different value
                   1951: in every loop iteration.
                   1952: 
1.17      anton    1953: @node Implementation,  , Programming Style, Gforth locals
1.2       anton    1954: @subsubsection Implementation
1.43    ! anton    1955: @cindex locals implementation
        !          1956: @cindex implementation of locals
1.2       anton    1957: 
1.43    ! anton    1958: @cindex locals stack
1.17      anton    1959: Gforth uses an extra locals stack. The most compelling reason for
1.2       anton    1960: this is that the return stack is not float-aligned; using an extra stack
                   1961: also eliminates the problems and restrictions of using the return stack
                   1962: as locals stack. Like the other stacks, the locals stack grows toward
                   1963: lower addresses. A few primitives allow an efficient implementation:
                   1964: 
                   1965: doc-@local#
                   1966: doc-f@local#
                   1967: doc-laddr#
                   1968: doc-lp+!#
                   1969: doc-lp!
                   1970: doc->l
                   1971: doc-f>l
                   1972: 
                   1973: In addition to these primitives, some specializations of these
                   1974: primitives for commonly occurring inline arguments are provided for
                   1975: efficiency reasons, e.g., @code{@@local0} as specialization of
                   1976: @code{@@local#} for the inline argument 0. The following compiling words
                   1977: compile the right specialized version, or the general version, as
                   1978: appropriate:
                   1979: 
1.12      anton    1980: doc-compile-@local
                   1981: doc-compile-f@local
1.2       anton    1982: doc-compile-lp+!
                   1983: 
                   1984: Combinations of conditional branches and @code{lp+!#} like
                   1985: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
                   1986: is taken) are provided for efficiency and correctness in loops.
                   1987: 
                   1988: A special area in the dictionary space is reserved for keeping the
                   1989: local variable names. @code{@{} switches the dictionary pointer to this
                   1990: area and @code{@}} switches it back and generates the locals
                   1991: initializing code. @code{W:} etc.@ are normal defining words. This
                   1992: special area is cleared at the start of every colon definition.
                   1993: 
1.43    ! anton    1994: @cindex wordlist for defining locals
1.17      anton    1995: A special feature of Gforth's dictionary is used to implement the
1.2       anton    1996: definition of locals without type specifiers: every wordlist (aka
                   1997: vocabulary) has its own methods for searching
1.4       anton    1998: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2       anton    1999: with a special search method: When it is searched for a word, it
                   2000: actually creates that word using @code{W:}. @code{@{} changes the search
                   2001: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
                   2002: and then the wordlist for defining locals without type specifiers.
                   2003: 
                   2004: The lifetime rules support a stack discipline within a colon
                   2005: definition: The lifetime of a local is either nested with other locals
                   2006: lifetimes or it does not overlap them.
                   2007: 
                   2008: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
                   2009: pointer manipulation is generated. Between control structure words
                   2010: locals definitions can push locals onto the locals stack. @code{AGAIN}
                   2011: is the simplest of the other three control flow words. It has to
                   2012: restore the locals stack depth of the corresponding @code{BEGIN}
                   2013: before branching. The code looks like this:
                   2014: @format
                   2015: @code{lp+!#} current-locals-size @minus{} dest-locals-size
                   2016: @code{branch} <begin>
                   2017: @end format
                   2018: 
                   2019: @code{UNTIL} is a little more complicated: If it branches back, it
                   2020: must adjust the stack just like @code{AGAIN}. But if it falls through,
                   2021: the locals stack must not be changed. The compiler generates the
                   2022: following code:
                   2023: @format
                   2024: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
                   2025: @end format
                   2026: The locals stack pointer is only adjusted if the branch is taken.
                   2027: 
                   2028: @code{THEN} can produce somewhat inefficient code:
                   2029: @format
                   2030: @code{lp+!#} current-locals-size @minus{} orig-locals-size
                   2031: <orig target>:
                   2032: @code{lp+!#} orig-locals-size @minus{} new-locals-size
                   2033: @end format
                   2034: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4       anton    2035: level at the @var{orig} point to the level after the @code{THEN}. The
1.2       anton    2036: first @code{lp+!#} adjusts the locals stack pointer from the current
                   2037: level to the level at the orig point, so the complete effect is an
                   2038: adjustment from the current level to the right level after the
                   2039: @code{THEN}.
                   2040: 
1.43    ! anton    2041: @cindex locals information on the control-flow stack
        !          2042: @cindex control-flow stack items, locals information
1.2       anton    2043: In a conventional Forth implementation a dest control-flow stack entry
                   2044: is just the target address and an orig entry is just the address to be
                   2045: patched. Our locals implementation adds a wordlist to every orig or dest
                   2046: item. It is the list of locals visible (or assumed visible) at the point
                   2047: described by the entry. Our implementation also adds a tag to identify
                   2048: the kind of entry, in particular to differentiate between live and dead
                   2049: (reachable and unreachable) orig entries.
                   2050: 
                   2051: A few unusual operations have to be performed on locals wordlists:
                   2052: 
                   2053: doc-common-list
                   2054: doc-sub-list?
                   2055: doc-list-size
                   2056: 
                   2057: Several features of our locals wordlist implementation make these
                   2058: operations easy to implement: The locals wordlists are organised as
                   2059: linked lists; the tails of these lists are shared, if the lists
                   2060: contain some of the same locals; and the address of a name is greater
                   2061: than the address of the names behind it in the list.
                   2062: 
                   2063: Another important implementation detail is the variable
                   2064: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
                   2065: determine if they can be reached directly or only through the branch
                   2066: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
                   2067: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
                   2068: definition, by @code{BEGIN} and usually by @code{THEN}.
                   2069: 
                   2070: Counted loops are similar to other loops in most respects, but
                   2071: @code{LEAVE} requires special attention: It performs basically the same
                   2072: service as @code{AHEAD}, but it does not create a control-flow stack
                   2073: entry. Therefore the information has to be stored elsewhere;
                   2074: traditionally, the information was stored in the target fields of the
                   2075: branches created by the @code{LEAVE}s, by organizing these fields into a
                   2076: linked list. Unfortunately, this clever trick does not provide enough
                   2077: space for storing our extended control flow information. Therefore, we
                   2078: introduce another stack, the leave stack. It contains the control-flow
                   2079: stack entries for all unresolved @code{LEAVE}s.
                   2080: 
                   2081: Local names are kept until the end of the colon definition, even if
                   2082: they are no longer visible in any control-flow path. In a few cases
                   2083: this may lead to increased space needs for the locals name area, but
                   2084: usually less than reclaiming this space would cost in code size.
                   2085: 
                   2086: 
1.17      anton    2087: @node ANS Forth locals,  , Gforth locals, Locals
1.2       anton    2088: @subsection ANS Forth locals
1.43    ! anton    2089: @cindex locals, ANS Forth style
1.2       anton    2090: 
                   2091: The ANS Forth locals wordset does not define a syntax for locals, but
                   2092: words that make it possible to define various syntaxes. One of the
1.17      anton    2093: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2       anton    2094: wordset, i.e.:
                   2095: 
                   2096: @example
                   2097: @{ local1 local2 ... -- comment @}
                   2098: @end example
                   2099: or
                   2100: @example
                   2101: @{ local1 local2 ... @}
                   2102: @end example
                   2103: 
                   2104: The order of the locals corresponds to the order in a stack comment. The
                   2105: restrictions are:
1.1       anton    2106: 
1.2       anton    2107: @itemize @bullet
                   2108: @item
1.17      anton    2109: Locals can only be cell-sized values (no type specifiers are allowed).
1.2       anton    2110: @item
                   2111: Locals can be defined only outside control structures.
                   2112: @item
                   2113: Locals can interfere with explicit usage of the return stack. For the
                   2114: exact (and long) rules, see the standard. If you don't use return stack
1.17      anton    2115: accessing words in a definition using locals, you will be all right. The
1.2       anton    2116: purpose of this rule is to make locals implementation on the return
                   2117: stack easier.
                   2118: @item
                   2119: The whole definition must be in one line.
                   2120: @end itemize
                   2121: 
1.35      anton    2122: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
                   2123: Defining Words}). I.e., they are initialized from the stack. Using their
1.2       anton    2124: name produces their value. Their value can be changed using @code{TO}.
                   2125: 
1.17      anton    2126: Since this syntax is supported by Gforth directly, you need not do
1.2       anton    2127: anything to use it. If you want to port a program using this syntax to
1.30      anton    2128: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
                   2129: syntax on the other system.
1.2       anton    2130: 
                   2131: Note that a syntax shown in the standard, section A.13 looks
                   2132: similar, but is quite different in having the order of locals
                   2133: reversed. Beware!
                   2134: 
                   2135: The ANS Forth locals wordset itself consists of the following word
                   2136: 
                   2137: doc-(local)
                   2138: 
                   2139: The ANS Forth locals extension wordset defines a syntax, but it is so
                   2140: awful that we strongly recommend not to use it. We have implemented this
1.17      anton    2141: syntax to make porting to Gforth easy, but do not document it here. The
1.2       anton    2142: problem with this syntax is that the locals are defined in an order
                   2143: reversed with respect to the standard stack comment notation, making
                   2144: programs harder to read, and easier to misread and miswrite. The only
                   2145: merit of this syntax is that it is easy to implement using the ANS Forth
                   2146: locals wordset.
1.3       anton    2147: 
1.37      anton    2148: @node Defining Words, Tokens for Words, Locals, Words
1.4       anton    2149: @section Defining Words
1.43    ! anton    2150: @cindex defining words
1.4       anton    2151: 
1.14      anton    2152: @menu
1.35      anton    2153: * Simple Defining Words::       
                   2154: * Colon Definitions::           
                   2155: * User-defined Defining Words::  
                   2156: * Supplying names::             
                   2157: * Interpretation and Compilation Semantics::  
1.14      anton    2158: @end menu
                   2159: 
1.35      anton    2160: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
                   2161: @subsection Simple Defining Words
1.43    ! anton    2162: @cindex simple defining words
        !          2163: @cindex defining words, simple
1.35      anton    2164: 
                   2165: doc-constant
                   2166: doc-2constant
                   2167: doc-fconstant
                   2168: doc-variable
                   2169: doc-2variable
                   2170: doc-fvariable
                   2171: doc-create
                   2172: doc-user
                   2173: doc-value
                   2174: doc-to
                   2175: doc-defer
                   2176: doc-is
                   2177: 
                   2178: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
                   2179: @subsection Colon Definitions
1.43    ! anton    2180: @cindex colon definitions
1.35      anton    2181: 
                   2182: @example
                   2183: : name ( ... -- ... )
                   2184:     word1 word2 word3 ;
                   2185: @end example
                   2186: 
                   2187: creates a word called @code{name}, that, upon execution, executes
                   2188: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
                   2189: 
                   2190: The explanation above is somewhat superficial. @xref{Interpretation and
                   2191: Compilation Semantics} for an in-depth discussion of some of the issues
                   2192: involved.
                   2193: 
                   2194: doc-:
                   2195: doc-;
                   2196: 
                   2197: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
                   2198: @subsection User-defined Defining Words
1.43    ! anton    2199: @cindex user-defined defining words
        !          2200: @cindex defining words, user-defined
1.35      anton    2201: 
                   2202: You can create new defining words simply by wrapping defining-time code
                   2203: around existing defining words and putting the sequence in a colon
                   2204: definition.
                   2205: 
1.43    ! anton    2206: @cindex @code{CREATE} ... @code{DOES>}
1.36      anton    2207: If you want the words defined with your defining words to behave
                   2208: differently from words defined with standard defining words, you can
1.35      anton    2209: write your defining word like this:
                   2210: 
                   2211: @example
                   2212: : def-word ( "name" -- )
                   2213:     Create @var{code1}
                   2214: DOES> ( ... -- ... )
                   2215:     @var{code2} ;
                   2216: 
                   2217: def-word name
                   2218: @end example
                   2219: 
                   2220: Technically, this fragment defines a defining word @code{def-word}, and
                   2221: a word @code{name}; when you execute @code{name}, the address of the
                   2222: body of @code{name} is put on the data stack and @var{code2} is executed
                   2223: (the address of the body of @code{name} is the address @code{HERE}
1.36      anton    2224: returns immediately after the @code{CREATE}).
                   2225: 
                   2226: In other words, if you make the following definitions:
                   2227: 
                   2228: @example
                   2229: : def-word1 ( "name" -- )
                   2230:     Create @var{code1} ;
                   2231: 
                   2232: : action1 ( ... -- ... )
                   2233:     @var{code2} ;
                   2234: 
                   2235: def-word name1
                   2236: @end example
                   2237: 
                   2238: Using @code{name1 action1} is equivalent to using @code{name}.
                   2239: 
                   2240: E.g., you can implement @code{Constant} in this way:
1.35      anton    2241: 
                   2242: @example
                   2243: : constant ( w "name" -- )
                   2244:     create ,
                   2245: DOES> ( -- w )
1.36      anton    2246:     @@ ;
1.35      anton    2247: @end example
                   2248: 
                   2249: When you create a constant with @code{5 constant five}, first a new word
                   2250: @code{five} is created, then the value 5 is laid down in the body of
                   2251: @code{five} with @code{,}. When @code{five} is invoked, the address of
                   2252: the body is put on the stack, and @code{@@} retrieves the value 5.
                   2253: 
1.43    ! anton    2254: @cindex stack effect of @code{DOES>}-parts
        !          2255: @cindex @code{DOES>}-parts, stack effect
1.35      anton    2256: In the example above the stack comment after the @code{DOES>} specifies
                   2257: the stack effect of the defined words, not the stack effect of the
                   2258: following code (the following code expects the address of the body on
                   2259: the top of stack, which is not reflected in the stack comment). This is
                   2260: the convention that I use and recommend (it clashes a bit with using
                   2261: locals declarations for stack effect specification, though).
                   2262: 
                   2263: @subsubsection Applications of @code{CREATE..DOES>}
1.43    ! anton    2264: @cindex @code{CREATE} ... @code{DOES>}, applications
1.35      anton    2265: 
1.36      anton    2266: You may wonder how to use this feature. Here are some usage patterns:
1.35      anton    2267: 
1.43    ! anton    2268: @cindex factoring similar colon definitions
1.35      anton    2269: When you see a sequence of code occurring several times, and you can
                   2270: identify a meaning, you will factor it out as a colon definition. When
                   2271: you see similar colon definitions, you can factor them using
                   2272: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
                   2273: that look very similar:
                   2274: @example
1.41      anton    2275: : ori, ( reg-target reg-source n -- )
1.35      anton    2276:     0 asm-reg-reg-imm ;
1.41      anton    2277: : andi, ( reg-target reg-source n -- )
1.35      anton    2278:     1 asm-reg-reg-imm ;
                   2279: @end example
                   2280: 
                   2281: This could be factored with:
                   2282: @example
                   2283: : reg-reg-imm ( op-code -- )
                   2284:     create ,
1.41      anton    2285: DOES> ( reg-target reg-source n -- )
1.36      anton    2286:     @@ asm-reg-reg-imm ;
1.35      anton    2287: 
                   2288: 0 reg-reg-imm ori,
                   2289: 1 reg-reg-imm andi,
                   2290: @end example
                   2291: 
1.43    ! anton    2292: @cindex currying
1.35      anton    2293: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
                   2294: supply a part of the parameters for a word (known as @dfn{currying} in
                   2295: the functional language community). E.g., @code{+} needs two
                   2296: parameters. Creating versions of @code{+} with one parameter fixed can
                   2297: be done like this:
                   2298: @example
                   2299: : curry+ ( n1 -- )
                   2300:     create ,
                   2301: DOES> ( n2 -- n1+n2 )
1.36      anton    2302:     @@ + ;
1.35      anton    2303: 
                   2304:  3 curry+ 3+
                   2305: -2 curry+ 2-
                   2306: @end example
                   2307: 
                   2308: @subsubsection The gory details of @code{CREATE..DOES>}
1.43    ! anton    2309: @cindex @code{CREATE} ... @code{DOES>}, details
1.35      anton    2310: 
                   2311: doc-does>
                   2312: 
1.43    ! anton    2313: @cindex @code{DOES>} in a separate definition
1.35      anton    2314: This means that you need not use @code{CREATE} and @code{DOES>} in the
                   2315: same definition; E.g., you can put the @code{DOES>}-part in a separate
                   2316: definition. This allows us to, e.g., select among different DOES>-parts:
                   2317: @example
                   2318: : does1 
                   2319: DOES> ( ... -- ... )
                   2320:     ... ;
                   2321: 
                   2322: : does2
                   2323: DOES> ( ... -- ... )
                   2324:     ... ;
                   2325: 
                   2326: : def-word ( ... -- ... )
                   2327:     create ...
                   2328:     IF
                   2329:        does1
                   2330:     ELSE
                   2331:        does2
                   2332:     ENDIF ;
                   2333: @end example
                   2334: 
1.43    ! anton    2335: @cindex @code{DOES>} in interpretation state
1.35      anton    2336: In a standard program you can apply a @code{DOES>}-part only if the last
                   2337: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
                   2338: will override the behaviour of the last word defined in any case. In a
                   2339: standard program, you can use @code{DOES>} only in a colon
                   2340: definition. In Gforth, you can also use it in interpretation state, in a
                   2341: kind of one-shot mode:
                   2342: @example
                   2343: CREATE name ( ... -- ... )
                   2344:   @var{initialization}
                   2345: DOES>
                   2346:   @var{code} ;
                   2347: @end example
1.41      anton    2348: This is equivalent to the standard
1.35      anton    2349: @example
                   2350: :noname
                   2351: DOES>
                   2352:     @var{code} ;
                   2353: CREATE name EXECUTE ( ... -- ... )
                   2354:     @var{initialization}
                   2355: @end example
                   2356: 
                   2357: You can get the address of the body of a word with
                   2358: 
                   2359: doc->body
                   2360: 
                   2361: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
                   2362: @subsection Supplying names for the defined words
1.43    ! anton    2363: @cindex names for defined words
        !          2364: @cindex defining words, name parameter
1.35      anton    2365: 
1.43    ! anton    2366: @cindex defining words, name given in a string
1.35      anton    2367: By default, defining words take the names for the defined words from the
                   2368: input stream. Sometimes you want to supply the name from a string. You
                   2369: can do this with
                   2370: 
                   2371: doc-nextname
                   2372: 
                   2373: E.g.,
                   2374: 
                   2375: @example
                   2376: s" foo" nextname create
                   2377: @end example
                   2378: is equivalent to
                   2379: @example
                   2380: create foo
                   2381: @end example
                   2382: 
1.43    ! anton    2383: @cindex defining words without name
1.35      anton    2384: Sometimes you want to define a word without a name. You can do this with
                   2385: 
                   2386: doc-noname
                   2387: 
1.43    ! anton    2388: @cindex execution token of last defined word
1.35      anton    2389: To make any use of the newly defined word, you need its execution
                   2390: token. You can get it with
                   2391: 
                   2392: doc-lastxt
                   2393: 
                   2394: E.g., you can initialize a deferred word with an anonymous colon
                   2395: definition:
                   2396: @example
                   2397: Defer deferred
                   2398: noname : ( ... -- ... )
                   2399:   ... ;
                   2400: lastxt IS deferred
                   2401: @end example
                   2402: 
                   2403: @code{lastxt} also works when the last word was not defined as
                   2404: @code{noname}. 
                   2405: 
                   2406: The standard has also recognized the need for anonymous words and
                   2407: provides
                   2408: 
                   2409: doc-:noname
                   2410: 
                   2411: This leaves the execution token for the word on the stack after the
                   2412: closing @code{;}. You can rewrite the last example with @code{:noname}:
                   2413: @example
                   2414: Defer deferred
                   2415: :noname ( ... -- ... )
                   2416:   ... ;
                   2417: IS deferred
                   2418: @end example
                   2419: 
                   2420: @node Interpretation and Compilation Semantics,  , Supplying names, Defining Words
                   2421: @subsection Interpretation and Compilation Semantics
1.43    ! anton    2422: @cindex semantics, interpretation and compilation
1.35      anton    2423: 
1.43    ! anton    2424: @cindex interpretation semantics
1.36      anton    2425: The @dfn{interpretation semantics} of a word are what the text
                   2426: interpreter does when it encounters the word in interpret state. It also
                   2427: appears in some other contexts, e.g., the execution token returned by
                   2428: @code{' @var{word}} identifies the interpretation semantics of
                   2429: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
                   2430: interpret-state text interpretation of @code{@var{word}}).
                   2431: 
1.43    ! anton    2432: @cindex compilation semantics
1.36      anton    2433: The @dfn{compilation semantics} of a word are what the text interpreter
                   2434: does when it encounters the word in compile state. It also appears in
                   2435: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
                   2436: standard terminology, ``appends to the current definition''.} the
                   2437: compilation semantics of @var{word}.
                   2438: 
1.43    ! anton    2439: @cindex execution semantics
1.36      anton    2440: The standard also talks about @dfn{execution semantics}. They are used
                   2441: only for defining the interpretation and compilation semantics of many
                   2442: words. By default, the interpretation semantics of a word are to
                   2443: @code{execute} its execution semantics, and the compilation semantics of
                   2444: a word are to @code{compile,} its execution semantics.@footnote{In
                   2445: standard terminology: The default interpretation semantics are its
                   2446: execution semantics; the default compilation semantics are to append its
                   2447: execution semantics to the execution semantics of the current
                   2448: definition.}
                   2449: 
1.43    ! anton    2450: @cindex immediate words
1.36      anton    2451: You can change the compilation semantics into @code{execute}ing the
                   2452: execution semantics with
                   2453: 
1.35      anton    2454: doc-immediate
1.36      anton    2455: 
1.43    ! anton    2456: @cindex compile-only words
1.36      anton    2457: You can remove the interpretation semantics of a word with
                   2458: 
                   2459: doc-compile-only
                   2460: doc-restrict
                   2461: 
                   2462: Note that ticking (@code{'}) compile-only words gives an error
                   2463: (``Interpreting a compile-only word'').
                   2464: 
                   2465: Gforth also allows you to define words with arbitrary combinations of
                   2466: interpretation and compilation semantics.
                   2467: 
1.35      anton    2468: doc-interpret/compile:
                   2469: 
1.36      anton    2470: This feature was introduced for implementing @code{TO} and @code{S"}. I
                   2471: recommend that you do not define such words, as cute as they may be:
                   2472: they make it hard to get at both parts of the word in some contexts.
                   2473: E.g., assume you want to get an execution token for the compilation
                   2474: part. Instead, define two words, one that embodies the interpretation
                   2475: part, and one that embodies the compilation part.
                   2476: 
                   2477: There is, however, a potentially useful application of this feature:
                   2478: Providing differing implementations for the default semantics. While
                   2479: this introduces redundancy and is therefore usually a bad idea, a
                   2480: performance improvement may be worth the trouble. E.g., consider the
                   2481: word @code{foobar}:
                   2482: 
                   2483: @example
                   2484: : foobar
                   2485:     foo bar ;
                   2486: @end example
                   2487: 
                   2488: Let us assume that @code{foobar} is called so frequently that the
                   2489: calling overhead would take a significant amount of the run-time. We can
                   2490: optimize it with @code{interpret/compile:}:
1.35      anton    2491: 
1.36      anton    2492: @example
                   2493: :noname
                   2494:    foo bar ;
                   2495: :noname
                   2496:    POSTPONE foo POSTPONE bar ;
                   2497: interpret/compile: foobar
                   2498: @end example
                   2499: 
                   2500: This definition has the same interpretation semantics and essentially
                   2501: the same compilation semantics as the simple definition of
                   2502: @code{foobar}, but the implementation of the compilation semantics is
                   2503: more efficient with respect to run-time.
                   2504: 
1.43    ! anton    2505: @cindex state-smart words are a bad idea
1.36      anton    2506: Some people try to use state-smart words to emulate the feature provided
                   2507: by @code{interpret/compile:} (words are state-smart if they check
                   2508: @code{STATE} during execution). E.g., they would try to code
                   2509: @code{foobar} like this:
                   2510: 
                   2511: @example
                   2512: : foobar
                   2513:   STATE @@
                   2514:   IF ( compilation state )
                   2515:     POSTPONE foo POSTPONE bar
                   2516:   ELSE
                   2517:     foo bar
                   2518:   ENDIF ; immediate
                   2519: @end example
                   2520: 
                   2521: While this works if @code{foobar} is processed only by the text
                   2522: interpreter, it does not work in other contexts (like @code{'} or
                   2523: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
                   2524: for a state-smart word, not for the interpretation semantics of the
                   2525: original @code{foobar}; when you execute this execution token (directly
                   2526: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
                   2527: state, the result will not be what you expected (i.e., it will not
                   2528: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
                   2529: write them!
                   2530: 
1.43    ! anton    2531: @cindex defining words with arbitrary semantics combinations
1.36      anton    2532: It is also possible to write defining words that define words with
                   2533: arbitrary combinations of interpretation and compilation semantics (or,
                   2534: preferably, arbitrary combinations of implementations of the default
                   2535: semantics). In general, this looks like:
                   2536: 
                   2537: @example
                   2538: : def-word
                   2539:     create-interpret/compile
                   2540:     @var{code1}
                   2541: interpretation>
                   2542:     @var{code2}
                   2543: <interpretation
                   2544: compilation>
                   2545:     @var{code3}
                   2546: <compilation ;
                   2547: @end example
                   2548: 
                   2549: For a @var{word} defined with @code{def-word}, the interpretation
                   2550: semantics are to push the address of the body of @var{word} and perform
                   2551: @var{code2}, and the compilation semantics are to push the address of
                   2552: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
                   2553: can also be defined like this:
                   2554: 
                   2555: @example
                   2556: : constant ( n "name" -- )
                   2557:     create-interpret/compile
                   2558:     ,
                   2559: interpretation> ( -- n )
                   2560:     @@
                   2561: <interpretation
                   2562: compilation> ( compilation. -- ; run-time. -- n )
                   2563:     @@ postpone literal
                   2564: <compilation ;
                   2565: @end example
                   2566: 
                   2567: doc-create-interpret/compile
                   2568: doc-interpretation>
                   2569: doc-<interpretation
                   2570: doc-compilation>
                   2571: doc-<compilation
                   2572: 
                   2573: Note that words defined with @code{interpret/compile:} and
                   2574: @code{create-interpret/compile} have an extended header structure that
                   2575: differs from other words; however, unless you try to access them with
                   2576: plain address arithmetic, you should not notice this. Words for
                   2577: accessing the header structure usually know how to deal with this; e.g.,
                   2578: @code{' word >body} also gives you the body of a word created with
                   2579: @code{create-interpret/compile}.
1.4       anton    2580: 
1.37      anton    2581: @node Tokens for Words, Wordlists, Defining Words, Words
                   2582: @section Tokens for Words
1.43    ! anton    2583: @cindex tokens for words
1.37      anton    2584: 
                   2585: This chapter describes the creation and use of tokens that represent
                   2586: words on the stack (and in data space).
                   2587: 
                   2588: Named words have interpretation and compilation semantics. Unnamed words
                   2589: just have execution semantics.
                   2590: 
1.43    ! anton    2591: @cindex execution token
1.37      anton    2592: An @dfn{execution token} represents the execution semantics of an
                   2593: unnamed word. An execution token occupies one cell. As explained in
                   2594: section @ref{Supplying names}, the execution token of the last words
                   2595: defined can be produced with
                   2596: 
                   2597: short-lastxt
                   2598: 
                   2599: You can perform the semantics represented by an execution token with
                   2600: doc-execute
                   2601: You can compile the word with
                   2602: doc-compile,
                   2603: 
1.43    ! anton    2604: @cindex code field address
        !          2605: @cindex CFA
1.37      anton    2606: In Gforth, the abstract data type @emph{execution token} is implemented
                   2607: as CFA (code field address).
                   2608: 
                   2609: The interpretation semantics of a named word are also represented by an
                   2610: execution token. You can get it with
                   2611: 
                   2612: doc-[']
                   2613: doc-'
                   2614: 
                   2615: For literals, you use @code{'} in interpreted code and @code{[']} in
                   2616: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
                   2617: by complaining about compile-only words. To get an execution token for a
                   2618: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
                   2619: @var{X} drop}.
                   2620: 
1.43    ! anton    2621: @cindex compilation token
1.37      anton    2622: The compilation semantics are represented by a @dfn{compilation token}
                   2623: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
                   2624: execution token. The compilation semantics represented by the
                   2625: compilation token can be performed with @code{execute}, which consumes
                   2626: the whole compilation token, with an additional stack effect determined
                   2627: by the represented compilation semantics.
                   2628: 
                   2629: doc-[comp']
                   2630: doc-comp'
                   2631: 
1.38      anton    2632: You can compile the compilation semantics with @code{postpone,}. I.e.,
                   2633: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
                   2634: @var{word}}.
                   2635: 
                   2636: doc-postpone,
                   2637: 
1.37      anton    2638: At present, the @var{w} part of a compilation token is an execution
                   2639: token, and the @var{xt} part represents either @code{execute} or
1.41      anton    2640: @code{compile,}. However, don't rely on that knowledge, unless necessary;
1.37      anton    2641: we may introduce unusual compilation tokens in the future (e.g.,
                   2642: compilation tokens representing the compilation semantics of literals).
                   2643: 
1.43    ! anton    2644: @cindex name token
        !          2645: @cindex name field address
        !          2646: @cindex NFA
1.37      anton    2647: Named words are also represented by the @dfn{name token}. The abstract
                   2648: data type @emph{name token} is implemented as NFA (name field address).
                   2649: 
                   2650: doc-find-name
                   2651: doc-name>int
                   2652: doc-name?int
                   2653: doc-name>comp
                   2654: doc-name>string
                   2655: 
                   2656: @node Wordlists, Files, Tokens for Words, Words
1.4       anton    2657: @section Wordlists
                   2658: 
                   2659: @node Files, Blocks, Wordlists, Words
                   2660: @section Files
                   2661: 
                   2662: @node Blocks, Other I/O, Files, Words
                   2663: @section Blocks
                   2664: 
                   2665: @node Other I/O, Programming Tools, Blocks, Words
                   2666: @section Other I/O
                   2667: 
1.18      anton    2668: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4       anton    2669: @section Programming Tools
1.43    ! anton    2670: @cindex programming tools
1.4       anton    2671: 
1.5       anton    2672: @menu
                   2673: * Debugging::                   Simple and quick.
                   2674: * Assertions::                  Making your programs self-checking.
                   2675: @end menu
                   2676: 
                   2677: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4       anton    2678: @subsection Debugging
1.43    ! anton    2679: @cindex debugging
1.4       anton    2680: 
                   2681: The simple debugging aids provided in @file{debugging.fs}
                   2682: are meant to support a different style of debugging than the
                   2683: tracing/stepping debuggers used in languages with long turn-around
                   2684: times.
                   2685: 
1.41      anton    2686: A much better (faster) way in fast-compiling languages is to add
1.4       anton    2687: printing code at well-selected places, let the program run, look at
                   2688: the output, see where things went wrong, add more printing code, etc.,
                   2689: until the bug is found.
                   2690: 
                   2691: The word @code{~~} is easy to insert. It just prints debugging
                   2692: information (by default the source location and the stack contents). It
                   2693: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
                   2694: query-replace them with nothing). The deferred words
                   2695: @code{printdebugdata} and @code{printdebugline} control the output of
                   2696: @code{~~}. The default source location output format works well with
                   2697: Emacs' compilation mode, so you can step through the program at the
1.5       anton    2698: source level using @kbd{C-x `} (the advantage over a stepping debugger
                   2699: is that you can step in any direction and you know where the crash has
                   2700: happened or where the strange data has occurred).
1.4       anton    2701: 
                   2702: Note that the default actions clobber the contents of the pictured
                   2703: numeric output string, so you should not use @code{~~}, e.g., between
                   2704: @code{<#} and @code{#>}.
                   2705: 
                   2706: doc-~~
                   2707: doc-printdebugdata
                   2708: doc-printdebugline
                   2709: 
1.5       anton    2710: @node Assertions,  , Debugging, Programming Tools
1.4       anton    2711: @subsection Assertions
1.43    ! anton    2712: @cindex assertions
1.4       anton    2713: 
1.5       anton    2714: It is a good idea to make your programs self-checking, in particular, if
                   2715: you use an assumption (e.g., that a certain field of a data structure is
1.17      anton    2716: never zero) that may become wrong during maintenance. Gforth supports
1.5       anton    2717: assertions for this purpose. They are used like this:
                   2718: 
                   2719: @example
                   2720: assert( @var{flag} )
                   2721: @end example
                   2722: 
                   2723: The code between @code{assert(} and @code{)} should compute a flag, that
                   2724: should be true if everything is alright and false otherwise. It should
                   2725: not change anything else on the stack. The overall stack effect of the
                   2726: assertion is @code{( -- )}. E.g.
                   2727: 
                   2728: @example
                   2729: assert( 1 1 + 2 = ) \ what we learn in school
                   2730: assert( dup 0<> ) \ assert that the top of stack is not zero
                   2731: assert( false ) \ this code should not be reached
                   2732: @end example
                   2733: 
                   2734: The need for assertions is different at different times. During
                   2735: debugging, we want more checking, in production we sometimes care more
                   2736: for speed. Therefore, assertions can be turned off, i.e., the assertion
                   2737: becomes a comment. Depending on the importance of an assertion and the
                   2738: time it takes to check it, you may want to turn off some assertions and
1.17      anton    2739: keep others turned on. Gforth provides several levels of assertions for
1.5       anton    2740: this purpose:
                   2741: 
                   2742: doc-assert0(
                   2743: doc-assert1(
                   2744: doc-assert2(
                   2745: doc-assert3(
                   2746: doc-assert(
                   2747: doc-)
                   2748: 
                   2749: @code{Assert(} is the same as @code{assert1(}. The variable
                   2750: @code{assert-level} specifies the highest assertions that are turned
                   2751: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
                   2752: @code{assert1(} assertions perform checking, while @code{assert2(} and
                   2753: @code{assert3(} assertions are treated as comments.
                   2754: 
                   2755: Note that the @code{assert-level} is evaluated at compile-time, not at
                   2756: run-time. I.e., you cannot turn assertions on or off at run-time, you
                   2757: have to set the @code{assert-level} appropriately before compiling a
                   2758: piece of code. You can compile several pieces of code at several
                   2759: @code{assert-level}s (e.g., a trusted library at level 1 and newly
                   2760: written code at level 3).
                   2761: 
                   2762: doc-assert-level
                   2763: 
                   2764: If an assertion fails, a message compatible with Emacs' compilation mode
                   2765: is produced and the execution is aborted (currently with @code{ABORT"}.
                   2766: If there is interest, we will introduce a special throw code. But if you
                   2767: intend to @code{catch} a specific condition, using @code{throw} is
                   2768: probably more appropriate than an assertion).
                   2769: 
1.18      anton    2770: @node Assembler and Code words, Threading Words, Programming Tools, Words
                   2771: @section Assembler and Code words
1.43    ! anton    2772: @cindex assembler
        !          2773: @cindex code words
1.18      anton    2774: 
                   2775: Gforth provides some words for defining primitives (words written in
                   2776: machine code), and for defining the the machine-code equivalent of
                   2777: @code{DOES>}-based defining words. However, the machine-independent
1.40      anton    2778: nature of Gforth poses a few problems: First of all, Gforth runs on
1.18      anton    2779: several architectures, so it can provide no standard assembler. What's
                   2780: worse is that the register allocation not only depends on the processor,
1.25      anton    2781: but also on the @code{gcc} version and options used.
1.18      anton    2782: 
1.25      anton    2783: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18      anton    2784: header structure), so a system-independent assembler may be used in
                   2785: Gforth. If you do not have an assembler, you can compile machine code
                   2786: directly with @code{,} and @code{c,}.
                   2787: 
                   2788: doc-assembler
                   2789: doc-code
                   2790: doc-end-code
                   2791: doc-;code
                   2792: doc-flush-icache
                   2793: 
                   2794: If @code{flush-icache} does not work correctly, @code{code} words
                   2795: etc. will not work (reliably), either.
                   2796: 
                   2797: These words are rarely used. Therefore they reside in @code{code.fs},
                   2798: which is usually not loaded (except @code{flush-icache}, which is always
1.19      anton    2799: present). You can load them with @code{require code.fs}.
1.18      anton    2800: 
1.43    ! anton    2801: @cindex registers of the inner interpreter
1.25      anton    2802: In the assembly code you will want to refer to the inner interpreter's
                   2803: registers (e.g., the data stack pointer) and you may want to use other
                   2804: registers for temporary storage. Unfortunately, the register allocation
                   2805: is installation-dependent.
                   2806: 
                   2807: The easiest solution is to use explicit register declarations
                   2808: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
                   2809: GNU C Manual}) for all of the inner interpreter's registers: You have to
                   2810: compile Gforth with @code{-DFORCE_REG} (configure option
                   2811: @code{--enable-force-reg}) and the appropriate declarations must be
                   2812: present in the @code{machine.h} file (see @code{mips.h} for an example;
                   2813: you can find a full list of all declarable register symbols with
                   2814: @code{grep register engine.c}). If you give explicit registers to all
                   2815: variables that are declared at the beginning of @code{engine()}, you
                   2816: should be able to use the other caller-saved registers for temporary
                   2817: storage. Alternatively, you can use the @code{gcc} option
                   2818: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
                   2819: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
                   2820: (however, this restriction on register allocation may slow Gforth
                   2821: significantly).
                   2822: 
                   2823: If this solution is not viable (e.g., because @code{gcc} does not allow
                   2824: you to explicitly declare all the registers you need), you have to find
                   2825: out by looking at the code where the inner interpreter's registers
                   2826: reside and which registers can be used for temporary storage. You can
                   2827: get an assembly listing of the engine's code with @code{make engine.s}.
                   2828: 
                   2829: In any case, it is good practice to abstract your assembly code from the
                   2830: actual register allocation. E.g., if the data stack pointer resides in
                   2831: register @code{$17}, create an alias for this register called @code{sp},
                   2832: and use that in your assembly code.
                   2833: 
1.43    ! anton    2834: @cindex code words, portable
1.18      anton    2835: Another option for implementing normal and defining words efficiently
                   2836: is: adding the wanted functionality to the source of Gforth. For normal
1.35      anton    2837: words you just have to edit @file{primitives} (@pxref{Automatic
                   2838: Generation}), defining words (equivalent to @code{;CODE} words, for fast
                   2839: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
                   2840: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.18      anton    2841: 
                   2842: 
                   2843: @node Threading Words,  , Assembler and Code words, Words
1.4       anton    2844: @section Threading Words
1.43    ! anton    2845: @cindex threading words
1.4       anton    2846: 
1.43    ! anton    2847: @cindex code address
1.4       anton    2848: These words provide access to code addresses and other threading stuff
1.17      anton    2849: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4       anton    2850: abstracts away the differences between direct and indirect threading
                   2851: (and, for direct threading, the machine dependences). However, at
1.43    ! anton    2852: present this wordset is still incomplete. It is also pretty low-level;
        !          2853: some day it will hopefully be made unnecessary by an internals wordset
1.4       anton    2854: that abstracts implementation details away completely.
                   2855: 
                   2856: doc->code-address
                   2857: doc->does-code
                   2858: doc-code-address!
                   2859: doc-does-code!
                   2860: doc-does-handler!
                   2861: doc-/does-handler
                   2862: 
1.18      anton    2863: The code addresses produced by various defining words are produced by
                   2864: the following words:
1.14      anton    2865: 
1.18      anton    2866: doc-docol:
                   2867: doc-docon:
                   2868: doc-dovar:
                   2869: doc-douser:
                   2870: doc-dodefer:
                   2871: doc-dofield:
                   2872: 
1.35      anton    2873: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
                   2874: with @code{>DOES-CODE}. If the word was defined in that way, the value
                   2875: returned is different from 0 and identifies the @code{DOES>} used by the
                   2876: defining word.
1.14      anton    2877: 
1.40      anton    2878: @node Tools, ANS conformance, Words, Top
                   2879: @chapter Tools
                   2880: 
                   2881: @menu
1.43    ! anton    2882: * ANS Report::                  Report the words used, sorted by wordset.
1.40      anton    2883: @end menu
                   2884: 
                   2885: See also @ref{Emacs and Gforth}.
                   2886: 
                   2887: @node ANS Report,  , Tools, Tools
                   2888: @section @file{ans-report.fs}: Report the words used, sorted by wordset
1.43    ! anton    2889: @cindex @file{ans-report.fs}
        !          2890: @cindex report the words used in your program
        !          2891: @cindex words used in your program
1.40      anton    2892: 
                   2893: If you want to label a Forth program as ANS Forth Program, you must
                   2894: document which wordsets the program uses; for extension wordsets, it is
                   2895: helpful to list the words the program requires from these wordsets
                   2896: (because Forth systems are allowed to provide only some words of them).
                   2897: 
                   2898: The @file{ans-report.fs} tool makes it easy for you to determine which
                   2899: words from which wordset and which non-ANS words your application
                   2900: uses. You simply have to include @file{ans-report.fs} before loading the
                   2901: program you want to check. After loading your program, you can get the
                   2902: report with @code{print-ans-report}. A typical use is to run this as
                   2903: batch job like this:
                   2904: @example
                   2905: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
                   2906: @end example
                   2907: 
                   2908: The output looks like this (for @file{compat/control.fs}):
                   2909: @example
                   2910: The program uses the following words
                   2911: from CORE :
                   2912: : POSTPONE THEN ; immediate ?dup IF 0= 
                   2913: from BLOCK-EXT :
                   2914: \ 
                   2915: from FILE :
                   2916: ( 
                   2917: @end example
                   2918: 
                   2919: @subsection Caveats
                   2920: 
                   2921: Note that @file{ans-report.fs} just checks which words are used, not whether
                   2922: they are used in an ANS Forth conforming way!
                   2923: 
                   2924: Some words are defined in several wordsets in the
                   2925: standard. @file{ans-report.fs} reports them for only one of the
                   2926: wordsets, and not necessarily the one you expect. It depends on usage
                   2927: which wordset is the right one to specify. E.g., if you only use the
                   2928: compilation semantics of @code{S"}, it is a Core word; if you also use
                   2929: its interpretation semantics, it is a File word.
                   2930: 
1.43    ! anton    2931: @c ******************************************************************
1.40      anton    2932: @node ANS conformance, Model, Tools, Top
1.4       anton    2933: @chapter ANS conformance
1.43    ! anton    2934: @cindex ANS conformance of Gforth
1.4       anton    2935: 
1.17      anton    2936: To the best of our knowledge, Gforth is an
1.14      anton    2937: 
1.15      anton    2938: ANS Forth System
1.34      anton    2939: @itemize @bullet
1.15      anton    2940: @item providing the Core Extensions word set
                   2941: @item providing the Block word set
                   2942: @item providing the Block Extensions word set
                   2943: @item providing the Double-Number word set
                   2944: @item providing the Double-Number Extensions word set
                   2945: @item providing the Exception word set
                   2946: @item providing the Exception Extensions word set
                   2947: @item providing the Facility word set
                   2948: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
                   2949: @item providing the File Access word set
                   2950: @item providing the File Access Extensions word set
                   2951: @item providing the Floating-Point word set
                   2952: @item providing the Floating-Point Extensions word set
                   2953: @item providing the Locals word set
                   2954: @item providing the Locals Extensions word set
                   2955: @item providing the Memory-Allocation word set
                   2956: @item providing the Memory-Allocation Extensions word set (that one's easy)
                   2957: @item providing the Programming-Tools word set
1.34      anton    2958: @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    2959: @item providing the Search-Order word set
                   2960: @item providing the Search-Order Extensions word set
                   2961: @item providing the String word set
                   2962: @item providing the String Extensions word set (another easy one)
                   2963: @end itemize
                   2964: 
1.43    ! anton    2965: @cindex system documentation
1.15      anton    2966: In addition, ANS Forth systems are required to document certain
                   2967: implementation choices. This chapter tries to meet these
                   2968: requirements. In many cases it gives a way to ask the system for the
                   2969: information instead of providing the information directly, in
                   2970: particular, if the information depends on the processor, the operating
                   2971: system or the installation options chosen, or if they are likely to
1.17      anton    2972: change during the maintenance of Gforth.
1.15      anton    2973: 
1.14      anton    2974: @comment The framework for the rest has been taken from pfe.
                   2975: 
                   2976: @menu
                   2977: * The Core Words::              
                   2978: * The optional Block word set::  
                   2979: * The optional Double Number word set::  
                   2980: * The optional Exception word set::  
                   2981: * The optional Facility word set::  
                   2982: * The optional File-Access word set::  
                   2983: * The optional Floating-Point word set::  
                   2984: * The optional Locals word set::  
                   2985: * The optional Memory-Allocation word set::  
                   2986: * The optional Programming-Tools word set::  
                   2987: * The optional Search-Order word set::  
                   2988: @end menu
                   2989: 
                   2990: 
                   2991: @c =====================================================================
                   2992: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
                   2993: @comment  node-name,  next,  previous,  up
                   2994: @section The Core Words
                   2995: @c =====================================================================
1.43    ! anton    2996: @cindex core words, system documentation
        !          2997: @cindex system documentation, core words
1.14      anton    2998: 
                   2999: @menu
1.15      anton    3000: * core-idef::                   Implementation Defined Options                   
                   3001: * core-ambcond::                Ambiguous Conditions                
                   3002: * core-other::                  Other System Documentation                  
1.14      anton    3003: @end menu
                   3004: 
                   3005: @c ---------------------------------------------------------------------
                   3006: @node core-idef, core-ambcond, The Core Words, The Core Words
                   3007: @subsection Implementation Defined Options
                   3008: @c ---------------------------------------------------------------------
1.43    ! anton    3009: @cindex core words, implementation-defined options
        !          3010: @cindex implementation-defined options, core words
        !          3011: 
1.14      anton    3012: 
                   3013: @table @i
                   3014: @item (Cell) aligned addresses:
1.43    ! anton    3015: @cindex cell-aligned addresses
        !          3016: @cindex aligned addresses
1.17      anton    3017: processor-dependent. Gforth's alignment words perform natural alignment
1.14      anton    3018: (e.g., an address aligned for a datum of size 8 is divisible by
                   3019: 8). Unaligned accesses usually result in a @code{-23 THROW}.
                   3020: 
                   3021: @item @code{EMIT} and non-graphic characters:
1.43    ! anton    3022: @cindex @code{EMIT} and non-graphic characters
        !          3023: @cindex non-graphic characters and @code{EMIT}
1.14      anton    3024: The character is output using the C library function (actually, macro)
1.36      anton    3025: @code{putc}.
1.14      anton    3026: 
                   3027: @item character editing of @code{ACCEPT} and @code{EXPECT}:
1.43    ! anton    3028: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
        !          3029: @cindex editing in @code{ACCEPT} and @code{EXPECT}
        !          3030: @cindex @code{ACCEPT}, editing
        !          3031: @cindex @code{EXPECT}, editing
1.14      anton    3032: This is modeled on the GNU readline library (@pxref{Readline
                   3033: Interaction, , Command Line Editing, readline, The GNU Readline
                   3034: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
                   3035: producing a full word completion every time you type it (instead of
                   3036: producing the common prefix of all completions).
                   3037: 
                   3038: @item character set:
1.43    ! anton    3039: @cindex character set
1.14      anton    3040: The character set of your computer and display device. Gforth is
                   3041: 8-bit-clean (but some other component in your system may make trouble).
                   3042: 
                   3043: @item Character-aligned address requirements:
1.43    ! anton    3044: @cindex character-aligned address requirements
1.14      anton    3045: installation-dependent. Currently a character is represented by a C
                   3046: @code{unsigned char}; in the future we might switch to @code{wchar_t}
                   3047: (Comments on that requested).
                   3048: 
                   3049: @item character-set extensions and matching of names:
1.43    ! anton    3050: @cindex character-set extensions and matching of names
        !          3051: @cindex case sensitivity for name lookup
        !          3052: @cindex name lookup, case sensitivity
        !          3053: @cindex locale and case sensitivity
1.17      anton    3054: Any character except the ASCII NUL charcter can be used in a
1.43    ! anton    3055: name. Matching is case-insensitive (except in @code{TABLE}s). The
1.36      anton    3056: matching is performed using the C function @code{strncasecmp}, whose
                   3057: function is probably influenced by the locale. E.g., the @code{C} locale
                   3058: does not know about accents and umlauts, so they are matched
                   3059: case-sensitively in that locale. For portability reasons it is best to
                   3060: write programs such that they work in the @code{C} locale. Then one can
                   3061: use libraries written by a Polish programmer (who might use words
                   3062: containing ISO Latin-2 encoded characters) and by a French programmer
                   3063: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
                   3064: funny results for some of the words (which ones, depends on the font you
                   3065: are using)). Also, the locale you prefer may not be available in other
                   3066: operating systems. Hopefully, Unicode will solve these problems one day.
1.14      anton    3067: 
                   3068: @item conditions under which control characters match a space delimiter:
1.43    ! anton    3069: @cindex space delimiters
        !          3070: @cindex control characters as delimiters
1.14      anton    3071: If @code{WORD} is called with the space character as a delimiter, all
                   3072: white-space characters (as identified by the C macro @code{isspace()})
                   3073: are delimiters. @code{PARSE}, on the other hand, treats space like other
                   3074: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
                   3075: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
                   3076: interpreter (aka text interpreter) by default, treats all white-space
                   3077: characters as delimiters.
                   3078: 
                   3079: @item format of the control flow stack:
1.43    ! anton    3080: @cindex control flow stack, format
1.14      anton    3081: The data stack is used as control flow stack. The size of a control flow
                   3082: stack item in cells is given by the constant @code{cs-item-size}. At the
                   3083: time of this writing, an item consists of a (pointer to a) locals list
                   3084: (third), an address in the code (second), and a tag for identifying the
                   3085: item (TOS). The following tags are used: @code{defstart},
                   3086: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
                   3087: @code{scopestart}.
                   3088: 
                   3089: @item conversion of digits > 35
1.43    ! anton    3090: @cindex digits > 35
1.14      anton    3091: The characters @code{[\]^_'} are the digits with the decimal value
                   3092: 36@minus{}41. There is no way to input many of the larger digits.
                   3093: 
                   3094: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
1.43    ! anton    3095: @cindex @code{EXPECT}, display after end of input
        !          3096: @cindex @code{ACCEPT}, display after end of input
1.14      anton    3097: The cursor is moved to the end of the entered string. If the input is
                   3098: terminated using the @kbd{Return} key, a space is typed.
                   3099: 
                   3100: @item exception abort sequence of @code{ABORT"}:
1.43    ! anton    3101: @cindex exception abort sequence of @code{ABORT"}
        !          3102: @cindex @code{ABORT"}, exception abort sequence
1.14      anton    3103: The error string is stored into the variable @code{"error} and a
                   3104: @code{-2 throw} is performed.
                   3105: 
                   3106: @item input line terminator:
1.43    ! anton    3107: @cindex input line terminator
        !          3108: @cindex line terminator on input
        !          3109: @cindex newline charcter on input
1.36      anton    3110: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
                   3111: lines. One of these characters is typically produced when you type the
                   3112: @kbd{Enter} or @kbd{Return} key.
1.14      anton    3113: 
                   3114: @item maximum size of a counted string:
1.43    ! anton    3115: @cindex maximum size of a counted string
        !          3116: @cindex counted string, maximum size
1.14      anton    3117: @code{s" /counted-string" environment? drop .}. Currently 255 characters
                   3118: on all ports, but this may change.
                   3119: 
                   3120: @item maximum size of a parsed string:
1.43    ! anton    3121: @cindex maximum size of a parsed string
        !          3122: @cindex parsed string, maximum size
1.14      anton    3123: Given by the constant @code{/line}. Currently 255 characters.
                   3124: 
                   3125: @item maximum size of a definition name, in characters:
1.43    ! anton    3126: @cindex maximum size of a definition name, in characters
        !          3127: @cindex name, maximum length
1.14      anton    3128: 31
                   3129: 
                   3130: @item maximum string length for @code{ENVIRONMENT?}, in characters:
1.43    ! anton    3131: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
        !          3132: @cindex @code{ENVIRONMENT?} string length, maximum
1.14      anton    3133: 31
                   3134: 
                   3135: @item method of selecting the user input device:
1.43    ! anton    3136: @cindex user input device, method of selecting
1.17      anton    3137: The user input device is the standard input. There is currently no way to
                   3138: change it from within Gforth. However, the input can typically be
                   3139: redirected in the command line that starts Gforth.
1.14      anton    3140: 
                   3141: @item method of selecting the user output device:
1.43    ! anton    3142: @cindex user output device, method of selecting
1.36      anton    3143: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
                   3144: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
                   3145: output, so output on a terminal does not become visible before the next
                   3146: newline or buffer overflow. Output on non-terminals is invisible until
                   3147: the buffer overflows.
1.14      anton    3148: 
                   3149: @item methods of dictionary compilation:
1.17      anton    3150: What are we expected to document here?
1.14      anton    3151: 
                   3152: @item number of bits in one address unit:
1.43    ! anton    3153: @cindex number of bits in one address unit
        !          3154: @cindex address unit, size in bits
1.14      anton    3155: @code{s" address-units-bits" environment? drop .}. 8 in all current
                   3156: ports.
                   3157: 
                   3158: @item number representation and arithmetic:
1.43    ! anton    3159: @cindex number representation and arithmetic
1.14      anton    3160: Processor-dependent. Binary two's complement on all current ports.
                   3161: 
                   3162: @item ranges for integer types:
1.43    ! anton    3163: @cindex ranges for integer types
        !          3164: @cindex integer types, ranges
1.14      anton    3165: Installation-dependent. Make environmental queries for @code{MAX-N},
                   3166: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
                   3167: unsigned (and positive) types is 0. The lower bound for signed types on
                   3168: two's complement and one's complement machines machines can be computed
                   3169: by adding 1 to the upper bound.
                   3170: 
                   3171: @item read-only data space regions:
1.43    ! anton    3172: @cindex read-only data space regions
        !          3173: @cindex data-space, read-only regions
1.14      anton    3174: The whole Forth data space is writable.
                   3175: 
                   3176: @item size of buffer at @code{WORD}:
1.43    ! anton    3177: @cindex size of buffer at @code{WORD}
        !          3178: @cindex @code{WORD} buffer size
1.14      anton    3179: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   3180: shared with the pictured numeric output string. If overwriting
                   3181: @code{PAD} is acceptable, it is as large as the remaining dictionary
                   3182: space, although only as much can be sensibly used as fits in a counted
                   3183: string.
                   3184: 
                   3185: @item size of one cell in address units:
1.43    ! anton    3186: @cindex cell size
1.14      anton    3187: @code{1 cells .}.
                   3188: 
                   3189: @item size of one character in address units:
1.43    ! anton    3190: @cindex char size
1.14      anton    3191: @code{1 chars .}. 1 on all current ports.
                   3192: 
                   3193: @item size of the keyboard terminal buffer:
1.43    ! anton    3194: @cindex size of the keyboard terminal buffer
        !          3195: @cindex terminal buffer, size
1.36      anton    3196: Varies. You can determine the size at a specific time using @code{lp@@
1.14      anton    3197: tib - .}. It is shared with the locals stack and TIBs of files that
                   3198: include the current file. You can change the amount of space for TIBs
1.17      anton    3199: and locals stack at Gforth startup with the command line option
1.14      anton    3200: @code{-l}.
                   3201: 
                   3202: @item size of the pictured numeric output buffer:
1.43    ! anton    3203: @cindex size of the pictured numeric output buffer
        !          3204: @cindex pictured numeric output buffer, size
1.14      anton    3205: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
                   3206: shared with @code{WORD}.
                   3207: 
                   3208: @item size of the scratch area returned by @code{PAD}:
1.43    ! anton    3209: @cindex size of the scratch area returned by @code{PAD}
        !          3210: @cindex @code{PAD} size
        !          3211: The remainder of dictionary space. @code{unused pad here - - .}.
1.14      anton    3212: 
                   3213: @item system case-sensitivity characteristics:
1.43    ! anton    3214: @cindex case-sensitivity characteristics
1.36      anton    3215: Dictionary searches are case insensitive (except in
                   3216: @code{TABLE}s). However, as explained above under @i{character-set
                   3217: extensions}, the matching for non-ASCII characters is determined by the
                   3218: locale you are using. In the default @code{C} locale all non-ASCII
                   3219: characters are matched case-sensitively.
1.14      anton    3220: 
                   3221: @item system prompt:
1.43    ! anton    3222: @cindex system prompt
        !          3223: @cindex prompt
1.14      anton    3224: @code{ ok} in interpret state, @code{ compiled} in compile state.
                   3225: 
                   3226: @item division rounding:
1.43    ! anton    3227: @cindex division rounding
1.14      anton    3228: installation dependent. @code{s" floored" environment? drop .}. We leave
1.43    ! anton    3229: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
        !          3230: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
1.14      anton    3231: 
                   3232: @item values of @code{STATE} when true:
1.43    ! anton    3233: @cindex @code{STATE} values
1.14      anton    3234: -1.
                   3235: 
                   3236: @item values returned after arithmetic overflow:
                   3237: On two's complement machines, arithmetic is performed modulo
                   3238: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   3239: arithmetic (with appropriate mapping for signed types). Division by zero
1.36      anton    3240: typically results in a @code{-55 throw} (Floating-point unidentified
1.14      anton    3241: fault), although a @code{-10 throw} (divide by zero) would be more
                   3242: appropriate.
                   3243: 
                   3244: @item whether the current definition can be found after @t{DOES>}:
1.43    ! anton    3245: @cindex @t{DOES>}, visibility of current definition
1.14      anton    3246: No.
                   3247: 
                   3248: @end table
                   3249: 
                   3250: @c ---------------------------------------------------------------------
                   3251: @node core-ambcond, core-other, core-idef, The Core Words
                   3252: @subsection Ambiguous conditions
                   3253: @c ---------------------------------------------------------------------
1.43    ! anton    3254: @cindex core words, ambiguous conditions
        !          3255: @cindex ambiguous conditions, core words
1.14      anton    3256: 
                   3257: @table @i
                   3258: 
                   3259: @item a name is neither a word nor a number:
1.43    ! anton    3260: @cindex name not found
        !          3261: @cindex Undefined word
1.36      anton    3262: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
                   3263: preserves the data and FP stack, so you don't lose more work than
                   3264: necessary.
1.14      anton    3265: 
                   3266: @item a definition name exceeds the maximum length allowed:
1.43    ! anton    3267: @cindex Word name too long
1.14      anton    3268: @code{-19 throw} (Word name too long)
                   3269: 
                   3270: @item addressing a region not inside the various data spaces of the forth system:
1.43    ! anton    3271: @cindex Invalid memory address
1.14      anton    3272: The stacks, code space and name space are accessible. Machine code space is
                   3273: typically readable. Accessing other addresses gives results dependent on
                   3274: the operating system. On decent systems: @code{-9 throw} (Invalid memory
                   3275: address).
                   3276: 
                   3277: @item argument type incompatible with parameter:
1.43    ! anton    3278: @cindex Argument type mismatch
1.14      anton    3279: This is usually not caught. Some words perform checks, e.g., the control
                   3280: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
                   3281: mismatch).
                   3282: 
                   3283: @item attempting to obtain the execution token of a word with undefined execution semantics:
1.43    ! anton    3284: @cindex Interpreting a compile-only word, for @code{'} etc.
        !          3285: @cindex execution token of words with undefined execution semantics
1.36      anton    3286: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
                   3287: get an execution token for @code{compile-only-error} (which performs a
                   3288: @code{-14 throw} when executed).
1.14      anton    3289: 
                   3290: @item dividing by zero:
1.43    ! anton    3291: @cindex dividing by zero
        !          3292: @cindex floating point unidentified fault, integer division
        !          3293: @cindex divide by zero
1.14      anton    3294: typically results in a @code{-55 throw} (floating point unidentified
                   3295: fault), although a @code{-10 throw} (divide by zero) would be more
                   3296: appropriate.
                   3297: 
                   3298: @item insufficient data stack or return stack space:
1.43    ! anton    3299: @cindex insufficient data stack or return stack space
        !          3300: @cindex stack overflow
        !          3301: @cindex Address alignment exception, stack overflow
        !          3302: @cindex Invalid memory address, stack overflow
        !          3303: Depending on the operating system, the installation, and the invocation
        !          3304: of Gforth, this is either checked by the memory management hardware, or
        !          3305: it is not checked. If it is checked, you typically get a @code{-9 throw}
        !          3306: (Invalid memory address) as soon as the overflow happens. If it is not
        !          3307: check, overflows typically result in mysterious illegal memory accesses,
        !          3308: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
        !          3309: (Address alignment exception); they might also destroy the internal data
        !          3310: structure of @code{ALLOCATE} and friends, resulting in various errors in
        !          3311: these words.
1.14      anton    3312: 
                   3313: @item insufficient space for loop control parameters:
1.43    ! anton    3314: @cindex insufficient space for loop control parameters
1.14      anton    3315: like other return stack overflows.
                   3316: 
                   3317: @item insufficient space in the dictionary:
1.43    ! anton    3318: @cindex insufficient space in the dictionary
        !          3319: @cindex dictionary overflow
        !          3320: Depending on the operating system, the installation, and the invocation
        !          3321: of Gforth, this is either checked by the memory management hardware, or
        !          3322: it is not checked. Similar results as stack overflows. However,
        !          3323: typically the error appears at a different place when one inserts or
        !          3324: removes code. Also, the @code{THROW} does not relieve the situation (it
        !          3325: does for stack overflows).
1.14      anton    3326: 
                   3327: @item interpreting a word with undefined interpretation semantics:
1.43    ! anton    3328: @cindex interpreting a word with undefined interpretation semantics
        !          3329: @cindex Interpreting a compile-only word
        !          3330: For some words, we have defined interpretation semantics. For the
        !          3331: others: @code{-14 throw} (Interpreting a compile-only word).
1.14      anton    3332: 
                   3333: @item modifying the contents of the input buffer or a string literal:
1.43    ! anton    3334: @cindex modifying the contents of the input buffer or a string literal
1.14      anton    3335: These are located in writable memory and can be modified.
                   3336: 
                   3337: @item overflow of the pictured numeric output string:
1.43    ! anton    3338: @cindex overflow of the pictured numeric output string
        !          3339: @cindex pictured numeric output string, overflow
        !          3340: Not checked. Runs into the dictionary and destroys it (at least,
        !          3341: partially).
1.14      anton    3342: 
                   3343: @item parsed string overflow:
1.43    ! anton    3344: @cindex parsed string overflow
1.14      anton    3345: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
                   3346: 
                   3347: @item producing a result out of range:
1.43    ! anton    3348: @cindex result out of range
1.14      anton    3349: On two's complement machines, arithmetic is performed modulo
                   3350: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
                   3351: arithmetic (with appropriate mapping for signed types). Division by zero
                   3352: typically results in a @code{-55 throw} (floatingpoint unidentified
                   3353: fault), although a @code{-10 throw} (divide by zero) would be more
                   3354: appropriate. @code{convert} and @code{>number} currently overflow
                   3355: silently.
                   3356: 
                   3357: @item reading from an empty data or return stack:
1.43    ! anton    3358: @cindex stack empty
        !          3359: @cindex stack underflow
1.14      anton    3360: The data stack is checked by the outer (aka text) interpreter after
                   3361: every word executed. If it has underflowed, a @code{-4 throw} (Stack
1.43    ! anton    3362: underflow) is performed. Apart from that, stacks may be checked or not,
        !          3363: depending on operating system, installation, and invocation. The
        !          3364: consequences of stack underflows are similar to the consequences of
        !          3365: stack overflows. Note that even if the system uses checking (through the
        !          3366: MMU), your program may have to underflow by a significant number of
        !          3367: stack items to trigger the reaction (the reason for this is that the
        !          3368: MMU, and therefore the checking, works with a page-size granularity).
1.14      anton    3369: 
1.36      anton    3370: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
1.43    ! anton    3371: @cindex unexpected end of the input buffer
        !          3372: @cindex zero-length string as a name
        !          3373: @cindex Attempt to use zero-length string as a name
1.14      anton    3374: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
                   3375: use zero-length string as a name). Words like @code{'} probably will not
                   3376: find what they search. Note that it is possible to create zero-length
                   3377: names with @code{nextname} (should it not?).
                   3378: 
                   3379: @item @code{>IN} greater than input buffer:
1.43    ! anton    3380: @cindex @code{>IN} greater than input buffer
1.41      anton    3381: The next invocation of a parsing word returns a string with length 0.
1.14      anton    3382: 
                   3383: @item @code{RECURSE} appears after @code{DOES>}:
1.43    ! anton    3384: @cindex @code{RECURSE} appears after @code{DOES>}
1.36      anton    3385: Compiles a recursive call to the defining word, not to the defined word.
1.14      anton    3386: 
                   3387: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.43    ! anton    3388: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
        !          3389: @cindex Argument type mismatch, @code{RESTORE-INPUT}
        !          3390: @cindex @code{RESTORE-INPUT}, Argument type mismatch
1.27      anton    3391: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
                   3392: the end of the file was reached), its source-id may be
                   3393: reused. Therefore, restoring an input source specification referencing a
                   3394: closed file may lead to unpredictable results instead of a @code{-12
                   3395: THROW}.
                   3396: 
1.36      anton    3397: In the future, Gforth may be able to restore input source specifications
1.41      anton    3398: from other than the current input source.
1.14      anton    3399: 
                   3400: @item data space containing definitions gets de-allocated:
1.43    ! anton    3401: @cindex data space containing definitions gets de-allocated
1.41      anton    3402: Deallocation with @code{allot} is not checked. This typically results in
1.14      anton    3403: memory access faults or execution of illegal instructions.
                   3404: 
                   3405: @item data space read/write with incorrect alignment:
1.43    ! anton    3406: @cindex data space read/write with incorrect alignment
        !          3407: @cindex alignment faults
        !          3408: @cindex Address alignment exception
1.14      anton    3409: Processor-dependent. Typically results in a @code{-23 throw} (Address
                   3410: alignment exception). Under Linux on a 486 or later processor with
                   3411: alignment turned on, incorrect alignment results in a @code{-9 throw}
                   3412: (Invalid memory address). There are reportedly some processors with
                   3413: alignment restrictions that do not report them.
                   3414: 
                   3415: @item data space pointer not properly aligned, @code{,}, @code{C,}:
1.43    ! anton    3416: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
1.14      anton    3417: Like other alignment errors.
                   3418: 
                   3419: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
1.43    ! anton    3420: Like other stack underflows.
1.14      anton    3421: 
                   3422: @item loop control parameters not available:
1.43    ! anton    3423: @cindex loop control parameters not available
1.14      anton    3424: Not checked. The counted loop words simply assume that the top of return
                   3425: stack items are loop control parameters and behave accordingly.
                   3426: 
                   3427: @item most recent definition does not have a name (@code{IMMEDIATE}):
1.43    ! anton    3428: @cindex most recent definition does not have a name (@code{IMMEDIATE})
        !          3429: @cindex last word was headerless
1.14      anton    3430: @code{abort" last word was headerless"}.
                   3431: 
                   3432: @item name not defined by @code{VALUE} used by @code{TO}:
1.43    ! anton    3433: @cindex name not defined by @code{VALUE} used by @code{TO}
        !          3434: @cindex @code{TO} on non-@code{VALUE}s
        !          3435: @cindex Invalid name argument, @code{TO}
        !          3436: @code{-32 throw} (Invalid name argument) (unless name is a local or was
        !          3437: defined by @code{CONSTANT}; in the latter case it just changes the constant).
1.14      anton    3438: 
1.15      anton    3439: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.43    ! anton    3440: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
        !          3441: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
1.14      anton    3442: @code{-13 throw} (Undefined word)
                   3443: 
                   3444: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
1.43    ! anton    3445: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
1.14      anton    3446: Gforth behaves as if they were of the same type. I.e., you can predict
                   3447: the behaviour by interpreting all parameters as, e.g., signed.
                   3448: 
                   3449: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
1.43    ! anton    3450: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
1.36      anton    3451: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
                   3452: compilation semantics of @code{TO}.
1.14      anton    3453: 
                   3454: @item String longer than a counted string returned by @code{WORD}:
1.43    ! anton    3455: @cindex String longer than a counted string returned by @code{WORD}
        !          3456: @cindex @code{WORD}, string overflow
1.14      anton    3457: Not checked. The string will be ok, but the count will, of course,
                   3458: contain only the least significant bits of the length.
                   3459: 
1.15      anton    3460: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.43    ! anton    3461: @cindex @code{LSHIFT}, large shift counts
        !          3462: @cindex @code{RSHIFT}, large shift counts
1.14      anton    3463: Processor-dependent. Typical behaviours are returning 0 and using only
                   3464: the low bits of the shift count.
                   3465: 
                   3466: @item word not defined via @code{CREATE}:
1.43    ! anton    3467: @cindex @code{>BODY} of non-@code{CREATE}d words
1.14      anton    3468: @code{>BODY} produces the PFA of the word no matter how it was defined.
                   3469: 
1.43    ! anton    3470: @cindex @code{DOES>} of non-@code{CREATE}d words
1.14      anton    3471: @code{DOES>} changes the execution semantics of the last defined word no
                   3472: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
                   3473: @code{CREATE , DOES>}.
                   3474: 
                   3475: @item words improperly used outside @code{<#} and @code{#>}:
                   3476: Not checked. As usual, you can expect memory faults.
                   3477: 
                   3478: @end table
                   3479: 
                   3480: 
                   3481: @c ---------------------------------------------------------------------
                   3482: @node core-other,  , core-ambcond, The Core Words
                   3483: @subsection Other system documentation
                   3484: @c ---------------------------------------------------------------------
1.43    ! anton    3485: @cindex other system documentation, core words
        !          3486: @cindex core words, other system documentation
1.14      anton    3487: 
                   3488: @table @i
                   3489: @item nonstandard words using @code{PAD}:
1.43    ! anton    3490: @cindex @code{PAD} use by nonstandard words
1.14      anton    3491: None.
                   3492: 
                   3493: @item operator's terminal facilities available:
1.43    ! anton    3494: @cindex operator's terminal facilities available
1.26      anton    3495: After processing the command line, Gforth goes into interactive mode,
                   3496: and you can give commands to Gforth interactively. The actual facilities
                   3497: available depend on how you invoke Gforth.
1.14      anton    3498: 
                   3499: @item program data space available:
1.43    ! anton    3500: @cindex program data space available
        !          3501: @cindex data space available
1.42      anton    3502: @code{UNUSED .} gives the remaining dictionary space. The total
                   3503: dictionary space can be specified with the @code{-m} switch
1.43    ! anton    3504: (@pxref{Invoking Gforth}) when Gforth starts up.
1.14      anton    3505: 
                   3506: @item return stack space available:
1.43    ! anton    3507: @cindex return stack space available
1.42      anton    3508: You can compute the total return stack space in cells with
                   3509: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
1.43    ! anton    3510: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
1.14      anton    3511: 
                   3512: @item stack space available:
1.43    ! anton    3513: @cindex stack space available
1.42      anton    3514: You can compute the total data stack space in cells with
                   3515: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
1.43    ! anton    3516: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
1.14      anton    3517: 
                   3518: @item system dictionary space required, in address units:
1.43    ! anton    3519: @cindex system dictionary space required, in address units
1.14      anton    3520: Type @code{here forthstart - .} after startup. At the time of this
1.42      anton    3521: writing, this gives 80080 (bytes) on a 32-bit system.
1.14      anton    3522: @end table
                   3523: 
                   3524: 
                   3525: @c =====================================================================
                   3526: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
                   3527: @section The optional Block word set
                   3528: @c =====================================================================
1.43    ! anton    3529: @cindex system documentation, block words
        !          3530: @cindex block words, system documentation
1.14      anton    3531: 
                   3532: @menu
1.43    ! anton    3533: * block-idef::                  Implementation Defined Options
1.15      anton    3534: * block-ambcond::               Ambiguous Conditions               
                   3535: * block-other::                 Other System Documentation                 
1.14      anton    3536: @end menu
                   3537: 
                   3538: 
                   3539: @c ---------------------------------------------------------------------
                   3540: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
                   3541: @subsection Implementation Defined Options
                   3542: @c ---------------------------------------------------------------------
1.43    ! anton    3543: @cindex implementation-defined options, block words
        !          3544: @cindex block words, implementation-defined options
1.14      anton    3545: 
                   3546: @table @i
                   3547: @item the format for display by @code{LIST}:
1.43    ! anton    3548: @cindex @code{LIST} display format
1.14      anton    3549: First the screen number is displayed, then 16 lines of 64 characters,
                   3550: each line preceded by the line number.
                   3551: 
                   3552: @item the length of a line affected by @code{\}:
1.43    ! anton    3553: @cindex length of a line affected by @code{\}
        !          3554: @cindex @code{\}, line length in blocks
1.14      anton    3555: 64 characters.
                   3556: @end table
                   3557: 
                   3558: 
                   3559: @c ---------------------------------------------------------------------
                   3560: @node block-ambcond, block-other, block-idef, The optional Block word set
                   3561: @subsection Ambiguous conditions
                   3562: @c ---------------------------------------------------------------------
1.43    ! anton    3563: @cindex block words, ambiguous conditions
        !          3564: @cindex ambiguous conditions, block words
1.14      anton    3565: 
                   3566: @table @i
                   3567: @item correct block read was not possible:
1.43    ! anton    3568: @cindex block read not possible
1.14      anton    3569: Typically results in a @code{throw} of some OS-derived value (between
                   3570: -512 and -2048). If the blocks file was just not long enough, blanks are
                   3571: supplied for the missing portion.
                   3572: 
                   3573: @item I/O exception in block transfer:
1.43    ! anton    3574: @cindex I/O exception in block transfer
        !          3575: @cindex block transfer, I/O exception
1.14      anton    3576: Typically results in a @code{throw} of some OS-derived value (between
                   3577: -512 and -2048).
                   3578: 
                   3579: @item invalid block number:
1.43    ! anton    3580: @cindex invalid block number
        !          3581: @cindex block number invalid
1.14      anton    3582: @code{-35 throw} (Invalid block number)
                   3583: 
                   3584: @item a program directly alters the contents of @code{BLK}:
1.43    ! anton    3585: @cindex @code{BLK}, altering @code{BLK}
1.14      anton    3586: The input stream is switched to that other block, at the same
                   3587: position. If the storing to @code{BLK} happens when interpreting
                   3588: non-block input, the system will get quite confused when the block ends.
                   3589: 
                   3590: @item no current block buffer for @code{UPDATE}:
1.43    ! anton    3591: @cindex @code{UPDATE}, no current block buffer
1.14      anton    3592: @code{UPDATE} has no effect.
                   3593: 
                   3594: @end table
                   3595: 
                   3596: @c ---------------------------------------------------------------------
                   3597: @node block-other,  , block-ambcond, The optional Block word set
                   3598: @subsection Other system documentation
                   3599: @c ---------------------------------------------------------------------
1.43    ! anton    3600: @cindex other system documentation, block words
        !          3601: @cindex block words, other system documentation
1.14      anton    3602: 
                   3603: @table @i
                   3604: @item any restrictions a multiprogramming system places on the use of buffer addresses:
                   3605: No restrictions (yet).
                   3606: 
                   3607: @item the number of blocks available for source and data:
                   3608: depends on your disk space.
                   3609: 
                   3610: @end table
                   3611: 
                   3612: 
                   3613: @c =====================================================================
                   3614: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
                   3615: @section The optional Double Number word set
                   3616: @c =====================================================================
1.43    ! anton    3617: @cindex system documentation, double words
        !          3618: @cindex double words, system documentation
1.14      anton    3619: 
                   3620: @menu
1.15      anton    3621: * double-ambcond::              Ambiguous Conditions              
1.14      anton    3622: @end menu
                   3623: 
                   3624: 
                   3625: @c ---------------------------------------------------------------------
1.15      anton    3626: @node double-ambcond,  , The optional Double Number word set, The optional Double Number word set
1.14      anton    3627: @subsection Ambiguous conditions
                   3628: @c ---------------------------------------------------------------------
1.43    ! anton    3629: @cindex double words, ambiguous conditions
        !          3630: @cindex ambiguous conditions, double words
1.14      anton    3631: 
                   3632: @table @i
1.15      anton    3633: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.43    ! anton    3634: @cindex @code{D>S}, @var{d} out of range of @var{n} 
1.14      anton    3635: The least significant cell of @var{d} is produced.
                   3636: 
                   3637: @end table
                   3638: 
                   3639: 
                   3640: @c =====================================================================
                   3641: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
                   3642: @section The optional Exception word set
                   3643: @c =====================================================================
1.43    ! anton    3644: @cindex system documentation, exception words
        !          3645: @cindex exception words, system documentation
1.14      anton    3646: 
                   3647: @menu
1.15      anton    3648: * exception-idef::              Implementation Defined Options              
1.14      anton    3649: @end menu
                   3650: 
                   3651: 
                   3652: @c ---------------------------------------------------------------------
1.15      anton    3653: @node exception-idef,  , The optional Exception word set, The optional Exception word set
1.14      anton    3654: @subsection Implementation Defined Options
                   3655: @c ---------------------------------------------------------------------
1.43    ! anton    3656: @cindex implementation-defined options, exception words
        !          3657: @cindex exception words, implementation-defined options
1.14      anton    3658: 
                   3659: @table @i
                   3660: @item @code{THROW}-codes used in the system:
1.43    ! anton    3661: @cindex @code{THROW}-codes used in the system
        !          3662: The codes -256@minus{}-511 are used for reporting signals. The mapping
        !          3663: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
        !          3664: codes -512@minus{}-2047 are used for OS errors (for file and memory
        !          3665: allocation operations). The mapping from OS error numbers to throw codes
        !          3666: is -512@minus{}@code{errno}. One side effect of this mapping is that
        !          3667: undefined OS errors produce a message with a strange number; e.g.,
        !          3668: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
1.14      anton    3669: @end table
                   3670: 
                   3671: @c =====================================================================
                   3672: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
                   3673: @section The optional Facility word set
                   3674: @c =====================================================================
1.43    ! anton    3675: @cindex system documentation, facility words
        !          3676: @cindex facility words, system documentation
1.14      anton    3677: 
                   3678: @menu
1.15      anton    3679: * facility-idef::               Implementation Defined Options               
                   3680: * facility-ambcond::            Ambiguous Conditions            
1.14      anton    3681: @end menu
                   3682: 
                   3683: 
                   3684: @c ---------------------------------------------------------------------
                   3685: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
                   3686: @subsection Implementation Defined Options
                   3687: @c ---------------------------------------------------------------------
1.43    ! anton    3688: @cindex implementation-defined options, facility words
        !          3689: @cindex facility words, implementation-defined options
1.14      anton    3690: 
                   3691: @table @i
                   3692: @item encoding of keyboard events (@code{EKEY}):
1.43    ! anton    3693: @cindex keyboard events, encoding in @code{EKEY}
        !          3694: @cindex @code{EKEY}, encoding of keyboard events
1.41      anton    3695: Not yet implemented.
1.14      anton    3696: 
1.43    ! anton    3697: @item duration of a system clock tick:
        !          3698: @cindex duration of a system clock tick
        !          3699: @cindex clock tick duration
1.14      anton    3700: System dependent. With respect to @code{MS}, the time is specified in
                   3701: microseconds. How well the OS and the hardware implement this, is
                   3702: another question.
                   3703: 
                   3704: @item repeatability to be expected from the execution of @code{MS}:
1.43    ! anton    3705: @cindex repeatability to be expected from the execution of @code{MS}
        !          3706: @cindex @code{MS}, repeatability to be expected
1.14      anton    3707: System dependent. On Unix, a lot depends on load. If the system is
1.17      anton    3708: lightly loaded, and the delay is short enough that Gforth does not get
1.14      anton    3709: swapped out, the performance should be acceptable. Under MS-DOS and
                   3710: other single-tasking systems, it should be good.
                   3711: 
                   3712: @end table
                   3713: 
                   3714: 
                   3715: @c ---------------------------------------------------------------------
1.15      anton    3716: @node facility-ambcond,  , facility-idef, The optional Facility word set
1.14      anton    3717: @subsection Ambiguous conditions
                   3718: @c ---------------------------------------------------------------------
1.43    ! anton    3719: @cindex facility words, ambiguous conditions
        !          3720: @cindex ambiguous conditions, facility words
1.14      anton    3721: 
                   3722: @table @i
                   3723: @item @code{AT-XY} can't be performed on user output device:
1.43    ! anton    3724: @cindex @code{AT-XY} can't be performed on user output device
1.41      anton    3725: Largely terminal dependent. No range checks are done on the arguments.
1.14      anton    3726: No errors are reported. You may see some garbage appearing, you may see
                   3727: simply nothing happen.
                   3728: 
                   3729: @end table
                   3730: 
                   3731: 
                   3732: @c =====================================================================
                   3733: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
                   3734: @section The optional File-Access word set
                   3735: @c =====================================================================
1.43    ! anton    3736: @cindex system documentation, file words
        !          3737: @cindex file words, system documentation
1.14      anton    3738: 
                   3739: @menu
1.43    ! anton    3740: * file-idef::                   Implementation Defined Options
1.15      anton    3741: * file-ambcond::                Ambiguous Conditions                
1.14      anton    3742: @end menu
                   3743: 
                   3744: @c ---------------------------------------------------------------------
                   3745: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
                   3746: @subsection Implementation Defined Options
                   3747: @c ---------------------------------------------------------------------
1.43    ! anton    3748: @cindex implementation-defined options, file words
        !          3749: @cindex file words, implementation-defined options
1.14      anton    3750: 
                   3751: @table @i
1.43    ! anton    3752: @item file access methods used:
        !          3753: @cindex file access methods used
1.14      anton    3754: @code{R/O}, @code{R/W} and @code{BIN} work as you would
                   3755: expect. @code{W/O} translates into the C file opening mode @code{w} (or
                   3756: @code{wb}): The file is cleared, if it exists, and created, if it does
1.43    ! anton    3757: not (with both @code{open-file} and @code{create-file}).  Under Unix
1.14      anton    3758: @code{create-file} creates a file with 666 permissions modified by your
                   3759: umask.
                   3760: 
                   3761: @item file exceptions:
1.43    ! anton    3762: @cindex file exceptions
1.14      anton    3763: The file words do not raise exceptions (except, perhaps, memory access
                   3764: faults when you pass illegal addresses or file-ids).
                   3765: 
                   3766: @item file line terminator:
1.43    ! anton    3767: @cindex file line terminator
1.14      anton    3768: System-dependent. Gforth uses C's newline character as line
                   3769: terminator. What the actual character code(s) of this are is
                   3770: system-dependent.
                   3771: 
1.43    ! anton    3772: @item file name format:
        !          3773: @cindex file name format
1.14      anton    3774: System dependent. Gforth just uses the file name format of your OS.
                   3775: 
                   3776: @item information returned by @code{FILE-STATUS}:
1.43    ! anton    3777: @cindex @code{FILE-STATUS}, returned information
1.14      anton    3778: @code{FILE-STATUS} returns the most powerful file access mode allowed
                   3779: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
                   3780: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
1.41      anton    3781: along with the returned mode.
1.14      anton    3782: 
                   3783: @item input file state after an exception when including source:
1.43    ! anton    3784: @cindex exception when including source
1.14      anton    3785: All files that are left via the exception are closed.
                   3786: 
                   3787: @item @var{ior} values and meaning:
1.43    ! anton    3788: @cindex @var{ior} values and meaning
1.15      anton    3789: The @var{ior}s returned by the file and memory allocation words are
                   3790: intended as throw codes. They typically are in the range
                   3791: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   3792: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    3793: 
                   3794: @item maximum depth of file input nesting:
1.43    ! anton    3795: @cindex maximum depth of file input nesting
        !          3796: @cindex file input nesting, maximum depth
1.14      anton    3797: limited by the amount of return stack, locals/TIB stack, and the number
                   3798: of open files available. This should not give you troubles.
                   3799: 
                   3800: @item maximum size of input line:
1.43    ! anton    3801: @cindex maximum size of input line
        !          3802: @cindex input line size, maximum
1.14      anton    3803: @code{/line}. Currently 255.
                   3804: 
                   3805: @item methods of mapping block ranges to files:
1.43    ! anton    3806: @cindex mapping block ranges to files
        !          3807: @cindex files containing blocks
        !          3808: @cindex blocks in files
1.37      anton    3809: By default, blocks are accessed in the file @file{blocks.fb} in the
                   3810: current working directory. The file can be switched with @code{USE}.
1.14      anton    3811: 
                   3812: @item number of string buffers provided by @code{S"}:
1.43    ! anton    3813: @cindex @code{S"}, number of string buffers
1.14      anton    3814: 1
                   3815: 
                   3816: @item size of string buffer used by @code{S"}:
1.43    ! anton    3817: @cindex @code{S"}, size of string buffer
1.14      anton    3818: @code{/line}. currently 255.
                   3819: 
                   3820: @end table
                   3821: 
                   3822: @c ---------------------------------------------------------------------
1.15      anton    3823: @node file-ambcond,  , file-idef, The optional File-Access word set
1.14      anton    3824: @subsection Ambiguous conditions
                   3825: @c ---------------------------------------------------------------------
1.43    ! anton    3826: @cindex file words, ambiguous conditions
        !          3827: @cindex ambiguous conditions, file words
1.14      anton    3828: 
                   3829: @table @i
1.43    ! anton    3830: @item attempting to position a file outside its boundaries:
        !          3831: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
1.14      anton    3832: @code{REPOSITION-FILE} is performed as usual: Afterwards,
                   3833: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
                   3834: 
                   3835: @item attempting to read from file positions not yet written:
1.43    ! anton    3836: @cindex reading from file positions not yet written
1.14      anton    3837: End-of-file, i.e., zero characters are read and no error is reported.
                   3838: 
                   3839: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
1.43    ! anton    3840: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid 
1.14      anton    3841: An appropriate exception may be thrown, but a memory fault or other
                   3842: problem is more probable.
                   3843: 
1.43    ! anton    3844: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
        !          3845: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
        !          3846: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
1.14      anton    3847: The @var{ior} produced by the operation, that discovered the problem, is
                   3848: thrown.
                   3849: 
1.43    ! anton    3850: @item named file cannot be opened (@code{INCLUDED}):
        !          3851: @cindex @code{INCLUDED}, named file cannot be opened
1.14      anton    3852: The @var{ior} produced by @code{open-file} is thrown.
                   3853: 
                   3854: @item requesting an unmapped block number:
1.43    ! anton    3855: @cindex unmapped block numbers
1.14      anton    3856: There are no unmapped legal block numbers. On some operating systems,
                   3857: writing a block with a large number may overflow the file system and
                   3858: have an error message as consequence.
                   3859: 
                   3860: @item using @code{source-id} when @code{blk} is non-zero:
1.43    ! anton    3861: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
1.14      anton    3862: @code{source-id} performs its function. Typically it will give the id of
                   3863: the source which loaded the block. (Better ideas?)
                   3864: 
                   3865: @end table
                   3866: 
                   3867: 
                   3868: @c =====================================================================
                   3869: @node  The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15      anton    3870: @section The optional Floating-Point word set
1.14      anton    3871: @c =====================================================================
1.43    ! anton    3872: @cindex system documentation, floating-point words
        !          3873: @cindex floating-point words, system documentation
1.14      anton    3874: 
                   3875: @menu
1.15      anton    3876: * floating-idef::               Implementation Defined Options
                   3877: * floating-ambcond::            Ambiguous Conditions            
1.14      anton    3878: @end menu
                   3879: 
                   3880: 
                   3881: @c ---------------------------------------------------------------------
                   3882: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
                   3883: @subsection Implementation Defined Options
                   3884: @c ---------------------------------------------------------------------
1.43    ! anton    3885: @cindex implementation-defined options, floating-point words
        !          3886: @cindex floating-point words, implementation-defined options
1.14      anton    3887: 
                   3888: @table @i
1.15      anton    3889: @item format and range of floating point numbers:
1.43    ! anton    3890: @cindex format and range of floating point numbers
        !          3891: @cindex floating point numbers, format and range
1.15      anton    3892: System-dependent; the @code{double} type of C.
1.14      anton    3893: 
1.15      anton    3894: @item results of @code{REPRESENT} when @var{float} is out of range:
1.43    ! anton    3895: @cindex  @code{REPRESENT}, results when @var{float} is out of range
1.15      anton    3896: System dependent; @code{REPRESENT} is implemented using the C library
                   3897: function @code{ecvt()} and inherits its behaviour in this respect.
1.14      anton    3898: 
1.15      anton    3899: @item rounding or truncation of floating-point numbers:
1.43    ! anton    3900: @cindex rounding of floating-point numbers
        !          3901: @cindex truncation of floating-point numbers
        !          3902: @cindex floating-point numbers, rounding or truncation
1.26      anton    3903: System dependent; the rounding behaviour is inherited from the hosting C
                   3904: compiler. IEEE-FP-based (i.e., most) systems by default round to
                   3905: nearest, and break ties by rounding to even (i.e., such that the last
                   3906: bit of the mantissa is 0).
1.14      anton    3907: 
1.15      anton    3908: @item size of floating-point stack:
1.43    ! anton    3909: @cindex floating-point stack size
1.42      anton    3910: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
                   3911: the floating-point stack (in floats). You can specify this on startup
1.43    ! anton    3912: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
1.14      anton    3913: 
1.15      anton    3914: @item width of floating-point stack:
1.43    ! anton    3915: @cindex floating-point stack width 
1.15      anton    3916: @code{1 floats}.
1.14      anton    3917: 
                   3918: @end table
                   3919: 
                   3920: 
                   3921: @c ---------------------------------------------------------------------
1.15      anton    3922: @node floating-ambcond,  , floating-idef, The optional Floating-Point word set
                   3923: @subsection Ambiguous conditions
1.14      anton    3924: @c ---------------------------------------------------------------------
1.43    ! anton    3925: @cindex floating-point words, ambiguous conditions
        !          3926: @cindex ambiguous conditions, floating-point words
1.14      anton    3927: 
                   3928: @table @i
1.15      anton    3929: @item @code{df@@} or @code{df!} used with an address that is not double-float  aligned:
1.43    ! anton    3930: @cindex @code{df@@} or @code{df!} used with an address that is not double-float  aligned
1.37      anton    3931: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3932: alignment violations.
1.14      anton    3933: 
1.15      anton    3934: @item @code{f@@} or @code{f!} used with an address that is not float  aligned:
1.43    ! anton    3935: @cindex @code{f@@} used with an address that is not float aligned
        !          3936: @cindex @code{f!} used with an address that is not float aligned
1.37      anton    3937: System-dependent. Typically results in a @code{-23 THROW} like other
1.15      anton    3938: alignment violations.
1.14      anton    3939: 
1.43    ! anton    3940: @item floating-point result out of range:
        !          3941: @cindex floating-point result out of range
1.15      anton    3942: System-dependent. Can result in a @code{-55 THROW} (Floating-point
                   3943: unidentified fault), or can produce a special value representing, e.g.,
                   3944: Infinity.
1.14      anton    3945: 
1.15      anton    3946: @item @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned:
1.43    ! anton    3947: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float  aligned
1.15      anton    3948: System-dependent. Typically results in an alignment fault like other
                   3949: alignment violations.
1.14      anton    3950: 
1.43    ! anton    3951: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
        !          3952: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
1.15      anton    3953: The floating-point number is converted into decimal nonetheless.
1.14      anton    3954: 
1.15      anton    3955: @item Both arguments are equal to zero (@code{FATAN2}):
1.43    ! anton    3956: @cindex @code{FATAN2}, both arguments are equal to zero
1.15      anton    3957: System-dependent. @code{FATAN2} is implemented using the C library
                   3958: function @code{atan2()}.
1.14      anton    3959: 
1.43    ! anton    3960: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
        !          3961: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
1.15      anton    3962: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
                   3963: because of small errors and the tan will be a very large (or very small)
                   3964: but finite number.
1.14      anton    3965: 
1.15      anton    3966: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
1.43    ! anton    3967: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
1.15      anton    3968: The result is rounded to the nearest float.
1.14      anton    3969: 
1.15      anton    3970: @item dividing by zero:
1.43    ! anton    3971: @cindex dividing by zero, floating-point
        !          3972: @cindex floating-point dividing by zero
        !          3973: @cindex floating-point unidentified fault, FP divide-by-zero
1.15      anton    3974: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3975: 
1.15      anton    3976: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
1.43    ! anton    3977: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
1.15      anton    3978: System dependent. On IEEE-FP based systems the number is converted into
                   3979: an infinity.
1.14      anton    3980: 
1.43    ! anton    3981: @item @var{float}<1 (@code{FACOSH}):
        !          3982: @cindex @code{FACOSH}, @var{float}<1
        !          3983: @cindex floating-point unidentified fault, @code{FACOSH}
1.15      anton    3984: @code{-55 throw} (Floating-point unidentified fault)
1.14      anton    3985: 
1.43    ! anton    3986: @item @var{float}=<-1 (@code{FLNP1}):
        !          3987: @cindex @code{FLNP1}, @var{float}=<-1
        !          3988: @cindex floating-point unidentified fault, @code{FLNP1}
1.15      anton    3989: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3990: negative infinity is typically produced for @var{float}=-1.
1.14      anton    3991: 
1.43    ! anton    3992: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
        !          3993: @cindex @code{FLN}, @var{float}=<0
        !          3994: @cindex @code{FLOG}, @var{float}=<0
        !          3995: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
1.15      anton    3996: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
                   3997: negative infinity is typically produced for @var{float}=0.
1.14      anton    3998: 
1.43    ! anton    3999: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
        !          4000: @cindex @code{FASINH}, @var{float}<0
        !          4001: @cindex @code{FSQRT}, @var{float}<0
        !          4002: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
1.15      anton    4003: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
                   4004: produces values for these inputs on my Linux box (Bug in the C library?)
1.14      anton    4005: 
1.43    ! anton    4006: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
        !          4007: @cindex @code{FACOS}, |@var{float}|>1
        !          4008: @cindex @code{FASIN}, |@var{float}|>1
        !          4009: @cindex @code{FATANH}, |@var{float}|>1
        !          4010: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
1.15      anton    4011: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    4012: 
1.43    ! anton    4013: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
        !          4014: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
        !          4015: @cindex floating-point unidentified fault, @code{F>D}
1.15      anton    4016: @code{-55 throw} (Floating-point unidentified fault).
1.14      anton    4017: 
1.15      anton    4018: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
1.43    ! anton    4019: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
1.15      anton    4020: This does not happen.
                   4021: @end table
1.14      anton    4022: 
                   4023: @c =====================================================================
1.15      anton    4024: @node  The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
                   4025: @section The optional Locals word set
1.14      anton    4026: @c =====================================================================
1.43    ! anton    4027: @cindex system documentation, locals words
        !          4028: @cindex locals words, system documentation
1.14      anton    4029: 
                   4030: @menu
1.15      anton    4031: * locals-idef::                 Implementation Defined Options                 
                   4032: * locals-ambcond::              Ambiguous Conditions              
1.14      anton    4033: @end menu
                   4034: 
                   4035: 
                   4036: @c ---------------------------------------------------------------------
1.15      anton    4037: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14      anton    4038: @subsection Implementation Defined Options
                   4039: @c ---------------------------------------------------------------------
1.43    ! anton    4040: @cindex implementation-defined options, locals words
        !          4041: @cindex locals words, implementation-defined options
1.14      anton    4042: 
                   4043: @table @i
1.15      anton    4044: @item maximum number of locals in a definition:
1.43    ! anton    4045: @cindex maximum number of locals in a definition
        !          4046: @cindex locals, maximum number in a definition
1.15      anton    4047: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
                   4048: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
                   4049: characters. The number of locals in a definition is bounded by the size
                   4050: of locals-buffer, which contains the names of the locals.
1.14      anton    4051: 
                   4052: @end table
                   4053: 
                   4054: 
                   4055: @c ---------------------------------------------------------------------
1.15      anton    4056: @node locals-ambcond,  , locals-idef, The optional Locals word set
1.14      anton    4057: @subsection Ambiguous conditions
                   4058: @c ---------------------------------------------------------------------
1.43    ! anton    4059: @cindex locals words, ambiguous conditions
        !          4060: @cindex ambiguous conditions, locals words
1.14      anton    4061: 
                   4062: @table @i
1.15      anton    4063: @item executing a named local in interpretation state:
1.43    ! anton    4064: @cindex local in interpretation state
        !          4065: @cindex Interpreting a compile-only word, for a local
        !          4066: Locals have no interpretation semantics. If you try to perform the
        !          4067: interpretation semantics, you will get a @code{-14 throw} somewhere
        !          4068: (Interpreting a compile-only word). If you perform the compilation
        !          4069: semantics, the locals access will be compiled (irrespective of state).
1.14      anton    4070: 
1.15      anton    4071: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
1.43    ! anton    4072: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
        !          4073: @cindex @code{TO} on non-@code{VALUE}s and non-locals
        !          4074: @cindex Invalid name argument, @code{TO}
1.15      anton    4075: @code{-32 throw} (Invalid name argument)
1.14      anton    4076: 
                   4077: @end table
                   4078: 
                   4079: 
                   4080: @c =====================================================================
1.15      anton    4081: @node  The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
                   4082: @section The optional Memory-Allocation word set
1.14      anton    4083: @c =====================================================================
1.43    ! anton    4084: @cindex system documentation, memory-allocation words
        !          4085: @cindex memory-allocation words, system documentation
1.14      anton    4086: 
                   4087: @menu
1.15      anton    4088: * memory-idef::                 Implementation Defined Options                 
1.14      anton    4089: @end menu
                   4090: 
                   4091: 
                   4092: @c ---------------------------------------------------------------------
1.15      anton    4093: @node memory-idef,  , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14      anton    4094: @subsection Implementation Defined Options
                   4095: @c ---------------------------------------------------------------------
1.43    ! anton    4096: @cindex implementation-defined options, memory-allocation words
        !          4097: @cindex memory-allocation words, implementation-defined options
1.14      anton    4098: 
                   4099: @table @i
1.15      anton    4100: @item values and meaning of @var{ior}:
1.43    ! anton    4101: @cindex  @var{ior} values and meaning
1.15      anton    4102: The @var{ior}s returned by the file and memory allocation words are
                   4103: intended as throw codes. They typically are in the range
                   4104: -512@minus{}-2047 of OS errors.  The mapping from OS error numbers to
                   4105: @var{ior}s is -512@minus{}@var{errno}.
1.14      anton    4106: 
                   4107: @end table
                   4108: 
                   4109: @c =====================================================================
1.15      anton    4110: @node  The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
                   4111: @section The optional Programming-Tools word set
1.14      anton    4112: @c =====================================================================
1.43    ! anton    4113: @cindex system documentation, programming-tools words
        !          4114: @cindex programming-tools words, system documentation
1.14      anton    4115: 
                   4116: @menu
1.15      anton    4117: * programming-idef::            Implementation Defined Options            
                   4118: * programming-ambcond::         Ambiguous Conditions         
1.14      anton    4119: @end menu
                   4120: 
                   4121: 
                   4122: @c ---------------------------------------------------------------------
1.15      anton    4123: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14      anton    4124: @subsection Implementation Defined Options
                   4125: @c ---------------------------------------------------------------------
1.43    ! anton    4126: @cindex implementation-defined options, programming-tools words
        !          4127: @cindex programming-tools words, implementation-defined options
1.14      anton    4128: 
                   4129: @table @i
1.43    ! anton    4130: @item ending sequence for input following @code{;CODE} and @code{CODE}:
        !          4131: @cindex @code{;CODE} ending sequence
        !          4132: @cindex @code{CODE} ending sequence
        !          4133: @code{END-CODE}
        !          4134: 
        !          4135: @item manner of processing input following @code{;CODE} and @code{CODE}:
        !          4136: @cindex @code{;CODE}, processing input
        !          4137: @cindex @code{CODE}, processing input
        !          4138: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
1.37      anton    4139: the input is processed by the text interpreter, (starting) in interpret
                   4140: state.
1.15      anton    4141: 
                   4142: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
1.43    ! anton    4143: @cindex @code{ASSEMBLER}, search order capability
1.37      anton    4144: The ANS Forth search order word set.
1.15      anton    4145: 
                   4146: @item source and format of display by @code{SEE}:
1.43    ! anton    4147: @cindex @code{SEE}, source and format of output
1.15      anton    4148: The source for @code{see} is the intermediate code used by the inner
                   4149: interpreter.  The current @code{see} tries to output Forth source code
                   4150: as well as possible.
                   4151: 
1.14      anton    4152: @end table
                   4153: 
                   4154: @c ---------------------------------------------------------------------
1.15      anton    4155: @node programming-ambcond,  , programming-idef, The optional Programming-Tools word set
1.14      anton    4156: @subsection Ambiguous conditions
                   4157: @c ---------------------------------------------------------------------
1.43    ! anton    4158: @cindex programming-tools words, ambiguous conditions
        !          4159: @cindex ambiguous conditions, programming-tools words
1.14      anton    4160: 
                   4161: @table @i
                   4162: 
1.15      anton    4163: @item deleting the compilation wordlist (@code{FORGET}):
1.43    ! anton    4164: @cindex @code{FORGET}, deleting the compilation wordlist
1.15      anton    4165: Not implemented (yet).
1.14      anton    4166: 
1.15      anton    4167: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
1.43    ! anton    4168: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
        !          4169: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
        !          4170: @cindex control-flow stack underflow
1.15      anton    4171: This typically results in an @code{abort"} with a descriptive error
                   4172: message (may change into a @code{-22 throw} (Control structure mismatch)
                   4173: in the future). You may also get a memory access error. If you are
                   4174: unlucky, this ambiguous condition is not caught.
                   4175: 
1.43    ! anton    4176: @item @var{name} can't be found (@code{FORGET}):
        !          4177: @cindex @code{FORGET}, @var{name} can't be found
1.15      anton    4178: Not implemented (yet).
1.14      anton    4179: 
1.15      anton    4180: @item @var{name} not defined via @code{CREATE}:
1.43    ! anton    4181: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
        !          4182: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
1.37      anton    4183: the execution semantics of the last defined word no matter how it was
                   4184: defined.
1.14      anton    4185: 
1.15      anton    4186: @item @code{POSTPONE} applied to @code{[IF]}:
1.43    ! anton    4187: @cindex @code{POSTPONE} applied to @code{[IF]}
        !          4188: @cindex @code{[IF]} and @code{POSTPONE}
1.15      anton    4189: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
                   4190: equivalent to @code{[IF]}.
1.14      anton    4191: 
1.15      anton    4192: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
1.43    ! anton    4193: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
1.15      anton    4194: Continue in the same state of conditional compilation in the next outer
                   4195: input source. Currently there is no warning to the user about this.
1.14      anton    4196: 
1.15      anton    4197: @item removing a needed definition (@code{FORGET}):
1.43    ! anton    4198: @cindex @code{FORGET}, removing a needed definition
1.15      anton    4199: Not implemented (yet).
1.14      anton    4200: 
                   4201: @end table
                   4202: 
                   4203: 
                   4204: @c =====================================================================
1.15      anton    4205: @node  The optional Search-Order word set,  , The optional Programming-Tools word set, ANS conformance
                   4206: @section The optional Search-Order word set
1.14      anton    4207: @c =====================================================================
1.43    ! anton    4208: @cindex system documentation, search-order words
        !          4209: @cindex search-order words, system documentation
1.14      anton    4210: 
                   4211: @menu
1.15      anton    4212: * search-idef::                 Implementation Defined Options                 
                   4213: * search-ambcond::              Ambiguous Conditions              
1.14      anton    4214: @end menu
                   4215: 
                   4216: 
                   4217: @c ---------------------------------------------------------------------
1.15      anton    4218: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14      anton    4219: @subsection Implementation Defined Options
                   4220: @c ---------------------------------------------------------------------
1.43    ! anton    4221: @cindex implementation-defined options, search-order words
        !          4222: @cindex search-order words, implementation-defined options
1.14      anton    4223: 
                   4224: @table @i
1.15      anton    4225: @item maximum number of word lists in search order:
1.43    ! anton    4226: @cindex maximum number of word lists in search order
        !          4227: @cindex search order, maximum depth
1.15      anton    4228: @code{s" wordlists" environment? drop .}. Currently 16.
                   4229: 
                   4230: @item minimum search order:
1.43    ! anton    4231: @cindex minimum search order
        !          4232: @cindex search order, minimum
1.15      anton    4233: @code{root root}.
1.14      anton    4234: 
                   4235: @end table
                   4236: 
                   4237: @c ---------------------------------------------------------------------
1.15      anton    4238: @node search-ambcond,  , search-idef, The optional Search-Order word set
1.14      anton    4239: @subsection Ambiguous conditions
                   4240: @c ---------------------------------------------------------------------
1.43    ! anton    4241: @cindex search-order words, ambiguous conditions
        !          4242: @cindex ambiguous conditions, search-order words
1.14      anton    4243: 
                   4244: @table @i
1.15      anton    4245: @item changing the compilation wordlist (during compilation):
1.43    ! anton    4246: @cindex changing the compilation wordlist (during compilation)
        !          4247: @cindex compilation wordlist, change before definition ends
1.33      anton    4248: The word is entered into the wordlist that was the compilation wordlist
                   4249: at the start of the definition. Any changes to the name field (e.g.,
                   4250: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
                   4251: are applied to the latest defined word (as reported by @code{last} or
                   4252: @code{lastxt}), if possible, irrespective of the compilation wordlist.
1.14      anton    4253: 
1.15      anton    4254: @item search order empty (@code{previous}):
1.43    ! anton    4255: @cindex @code{previous}, search order empty
        !          4256: @cindex Vocstack empty, @code{previous}
1.15      anton    4257: @code{abort" Vocstack empty"}.
1.14      anton    4258: 
1.15      anton    4259: @item too many word lists in search order (@code{also}):
1.43    ! anton    4260: @cindex @code{also}, too many word lists in search order
        !          4261: @cindex Vocstack full, @code{also}
1.15      anton    4262: @code{abort" Vocstack full"}.
1.14      anton    4263: 
                   4264: @end table
1.13      anton    4265: 
1.43    ! anton    4266: @c ***************************************************************
1.34      anton    4267: @node Model, Integrating Gforth, ANS conformance, Top
                   4268: @chapter Model
                   4269: 
                   4270: This chapter has yet to be written. It will contain information, on
                   4271: which internal structures you can rely.
                   4272: 
1.43    ! anton    4273: @c ***************************************************************
1.34      anton    4274: @node Integrating Gforth, Emacs and Gforth, Model, Top
                   4275: @chapter Integrating Gforth into C programs
                   4276: 
                   4277: This is not yet implemented.
                   4278: 
                   4279: Several people like to use Forth as scripting language for applications
                   4280: that are otherwise written in C, C++, or some other language.
                   4281: 
                   4282: The Forth system ATLAST provides facilities for embedding it into
                   4283: applications; unfortunately it has several disadvantages: most
1.36      anton    4284: importantly, it is not based on ANS Forth, and it is apparently dead
1.34      anton    4285: (i.e., not developed further and not supported). The facilities
                   4286: provided by Gforth in this area are inspired by ATLASTs facilities, so
                   4287: making the switch should not be hard.
                   4288: 
                   4289: We also tried to design the interface such that it can easily be
                   4290: implemented by other Forth systems, so that we may one day arrive at a
                   4291: standardized interface. Such a standard interface would allow you to
                   4292: replace the Forth system without having to rewrite C code.
                   4293: 
                   4294: You embed the Gforth interpreter by linking with the library
                   4295: @code{libgforth.a} (give the compiler the option @code{-lgforth}).  All
                   4296: global symbols in this library that belong to the interface, have the
                   4297: prefix @code{forth_}. (Global symbols that are used internally have the
                   4298: prefix @code{gforth_}).
                   4299: 
                   4300: You can include the declarations of Forth types and the functions and
1.36      anton    4301: variables of the interface with @code{#include <forth.h>}.
1.34      anton    4302: 
                   4303: Types.
1.13      anton    4304: 
1.34      anton    4305: Variables.
                   4306: 
                   4307: Data and FP Stack pointer. Area sizes.
                   4308: 
                   4309: functions.
                   4310: 
                   4311: forth_init(imagefile)
                   4312: forth_evaluate(string) exceptions?
                   4313: forth_goto(address) (or forth_execute(xt)?)
                   4314: forth_continue() (a corountining mechanism)
                   4315: 
                   4316: Adding primitives.
                   4317: 
                   4318: No checking.
                   4319: 
                   4320: Signals?
                   4321: 
                   4322: Accessing the Stacks
1.4       anton    4323: 
1.43    ! anton    4324: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
1.17      anton    4325: @chapter Emacs and Gforth
1.43    ! anton    4326: @cindex Emacs and Gforth
1.4       anton    4327: 
1.43    ! anton    4328: @cindex @file{gforth.el}
        !          4329: @cindex @file{forth.el}
        !          4330: @cindex Rydqvist, Goran
        !          4331: @cindex comment editing commands
        !          4332: @cindex @code{\}, editing with Emacs
        !          4333: @cindex debug tracer editing commands
        !          4334: @cindex @code{~~}, removal with Emacs
        !          4335: @cindex Forth mode in Emacs
1.17      anton    4336: Gforth comes with @file{gforth.el}, an improved version of
1.33      anton    4337: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.4       anton    4338: improvements are a better (but still not perfect) handling of
                   4339: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8       anton    4340: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
                   4341: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
                   4342: stuff I do not use alone, even though some of it only makes sense for
                   4343: TILE. To get a description of these features, enter Forth mode and type
                   4344: @kbd{C-h m}.
1.4       anton    4345: 
1.43    ! anton    4346: @cindex source location of error or debugging output in Emacs
        !          4347: @cindex error output, finding the source location in Emacs
        !          4348: @cindex debugging output, finding the source location in Emacs
1.17      anton    4349: In addition, Gforth supports Emacs quite well: The source code locations
1.4       anton    4350: given in error messages, debugging output (from @code{~~}) and failed
                   4351: assertion messages are in the right format for Emacs' compilation mode
                   4352: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
                   4353: Manual}) so the source location corresponding to an error or other
                   4354: message is only a few keystrokes away (@kbd{C-x `} for the next error,
                   4355: @kbd{C-c C-c} for the error under the cursor).
                   4356: 
1.43    ! anton    4357: @cindex @file{TAGS} file
        !          4358: @cindex @file{etags.fs}
        !          4359: @cindex viewing the source of a word in Emacs
1.4       anton    4360: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
                   4361: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
                   4362: contains the definitions of all words defined afterwards. You can then
                   4363: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17      anton    4364: several tags files at the same time (e.g., one for the Gforth sources
1.28      anton    4365: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
                   4366: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
                   4367: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
1.33      anton    4368: @file{/usr/local/share/gforth/0.2.0/TAGS}).
1.4       anton    4369: 
1.43    ! anton    4370: @cindex @file{.emacs}
1.4       anton    4371: To get all these benefits, add the following lines to your @file{.emacs}
                   4372: file:
                   4373: 
                   4374: @example
                   4375: (autoload 'forth-mode "gforth.el")
                   4376: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
                   4377: @end example
                   4378: 
1.43    ! anton    4379: @node Image Files, Engine, Emacs and Gforth, Top
        !          4380: @chapter Image Files
        !          4381: @cindex image files
        !          4382: @cindex @code{.fi} files
        !          4383: @cindex precompiled Forth code
        !          4384: @cindex dictionary in persistent form
        !          4385: @cindex persistent form of dictionary
        !          4386: 
        !          4387: An image file is a file containing an image of the Forth dictionary,
        !          4388: i.e., compiled Forth code and data residing in the dictionary.  By
        !          4389: convention, we use the extension @code{.fi} for image files.
        !          4390: 
        !          4391: @menu
        !          4392: * Image File Background::          Why have image files?
        !          4393: * Non-Relocatable Image Files::    don't always work.
        !          4394: * Data-Relocatable Image Files::   are better.
        !          4395: * Fully Relocatable Image Files::  are hard to create.
        !          4396: * Stack and Dictionary Sizes::     Setting the default sizes for an image.
        !          4397: * Running Image Files::            @code{gforth -i @var{file}} or @var{file}.
        !          4398: * Modifying the Startup Sequence:: and turnkey applications.
        !          4399: @end menu
        !          4400: 
        !          4401: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
        !          4402: @section Image File Background
        !          4403: @cindex image file background
        !          4404: 
        !          4405: Our Forth system consists not only of primitives, but also of
        !          4406: definitions written in Forth. Since the Forth compiler itself belongs
        !          4407: to those definitions, it is not possible to start the system with the
        !          4408: primitives and the Forth source alone. Therefore we provide the Forth
        !          4409: code as an image file in nearly executable form. At the start of the
        !          4410: system a C routine loads the image file into memory, sets up the
        !          4411: memory (stacks etc.) according to information in the image file, and
        !          4412: starts executing Forth code.
        !          4413: 
        !          4414: The image file variants represent different compromises between the
        !          4415: goals of making it easy to generate image files and making them
        !          4416: portable.
        !          4417: 
        !          4418: @cindex relocation at run-time
        !          4419: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
        !          4420: run-time. This avoids many of the complications discussed below (image
        !          4421: files are data relocatable without further ado), but costs performance
        !          4422: (one addition per memory access).
        !          4423: 
        !          4424: @cindex relocation at load-time
        !          4425: Our loader performs relocation at image load time. The loader also has
        !          4426: to replace tokens standing for primitive calls with the appropriate
        !          4427: code-field addresses (or code addresses in the case of direct
        !          4428: threading).
        !          4429: 
        !          4430: There are three kinds of image files, with different degrees of
        !          4431: relocatability: non-relocatable, data-relocatable, and fully relocatable
        !          4432: image files.
        !          4433: 
        !          4434: @cindex image file loader
        !          4435: @cindex relocating loader
        !          4436: @cindex loader for image files
        !          4437: These image file variants have several restrictions in common; they are
        !          4438: caused by the design of the image file loader:
        !          4439: 
        !          4440: @itemize @bullet
        !          4441: @item
        !          4442: There is only one segment; in particular, this means, that an image file
        !          4443: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
        !          4444: them). And the contents of the stacks are not represented, either.
        !          4445: 
        !          4446: @item
        !          4447: The only kinds of relocation supported are: the addition of the same
        !          4448: offset to all cells that represent data addresses; and the replacement of
        !          4449: special tokens with code addresses.
        !          4450: 
        !          4451: If any complex computations involving addresses are performed, the
        !          4452: results cannot be represented in the image file. Several applications that
        !          4453: use such computations come to mind:
        !          4454: @itemize @minus
        !          4455: @item
        !          4456: Hashing addresses (or data structures which contain addresses) for table
        !          4457: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
        !          4458: purpose, you will have no problem, because the hash tables are
        !          4459: recomputed automatically when the system is started. If you use your own
        !          4460: hash tables, you will have to do something similar.
        !          4461: 
        !          4462: @item
        !          4463: There's a cute implementation of doubly-linked lists that uses
        !          4464: @code{XOR}ed addresses. You could represent such lists as singly-linked
        !          4465: in the image file, and restore the doubly-linked representation on
        !          4466: startup.@footnote{In my opinion, though, you should think thrice before
        !          4467: using a doubly-linked list (whatever implementation).}
        !          4468: 
        !          4469: @item
        !          4470: On many architectures addresses are represented in machine code in some
        !          4471: shifted or mangled form. You cannot put @code{CODE} words that contain
        !          4472: absolute addresses in this form in a relocatable image file. Workarounds
        !          4473: are representing the address in some relative form (e.g., relative to
        !          4474: the CFA, which is present in some register), or loading the address from
        !          4475: a place where it is stored in a non-mangled form.
        !          4476: @end itemize
        !          4477: @end itemize
        !          4478: 
        !          4479: @node  Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
        !          4480: @section Non-Relocatable Image Files
        !          4481: @cindex non-relocatable image files
        !          4482: @cindex image files, non-relocatable
        !          4483: 
        !          4484: These files are simple memory dumps of the dictionary. They are specific
        !          4485: to the executable (i.e., @file{gforth} file) they were created
        !          4486: with. What's worse, they are specific to the place on which the
        !          4487: dictionary resided when the image was created. Now, there is no
        !          4488: guarantee that the dictionary will reside at the same place the next
        !          4489: time you start Gforth, so there's no guarantee that a non-relocatable
        !          4490: image will work the next time (Gforth will complain instead of crashing,
        !          4491: though).
        !          4492: 
        !          4493: You can create a non-relocatable image file with
        !          4494: 
        !          4495: doc-savesystem
        !          4496: 
        !          4497: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
        !          4498: @section Data-Relocatable Image Files
        !          4499: @cindex data-relocatable image files
        !          4500: @cindex image files, data-relocatable
        !          4501: 
        !          4502: @cindex @file{gforth.fi}, relocatability
        !          4503: These files contain relocatable data addresses, but fixed code addresses
        !          4504: (instead of tokens). They are specific to the executable (i.e.,
        !          4505: @file{gforth} file) they were created with. E.g., the image
        !          4506: @code{gforth.fi} is a data-relocatable image file.
        !          4507: 
        !          4508: @cindex @file{comp-image.fs}
        !          4509: You can create a data-relocatable image file by creating two
        !          4510: non-relocatable image files for different base addresses and processing
        !          4511: them with @file{comp-image.fs}, which generates the relocation
        !          4512: information by comparing the images.
        !          4513: 
        !          4514: After loading @file{comp-image.fs}, you can invoke the word
        !          4515: @code{comp-image}, which reads the names of the two input files and the
        !          4516: name of the output file from the input stream. This program also
        !          4517: produces some output on the standard output: It displays the offset
        !          4518: (i.e., the difference between the base addresses of the images);
        !          4519: moreover, for each cell that cannot be represented correctly in the
        !          4520: image files, it displays a line like the following one:
        !          4521: 
        !          4522: @example
        !          4523:      78DC         BFFFFA50         BFFFFA40
        !          4524: @end example
        !          4525: 
        !          4526: This means that at offset $78dc from @code{forthstart}, one input image
        !          4527: contains $bffffa50, and the other contains $bffffa40. Since these cells
        !          4528: cannot be represented correctly in the output image, you should examine
        !          4529: these places in the dictionary and verify that these cells are dead
        !          4530: (i.e., not read before they are written).
        !          4531: 
        !          4532: E.g., if you want to create an image @file{asm.fi} that has the file
        !          4533: @file{asm.fs} loaded in addition to the usual stuff, you could do it
        !          4534: like this:
        !          4535: 
        !          4536: @example
        !          4537: gforth --clear-dictionary asm.fs "savesystem asm.fi1 bye"
        !          4538: gforth --clear-dictionary --offset-image asm.fs "savesystem asm.fi2 bye"
        !          4539: gforth -m 100000 comp-image.fs -e "comp-image asm.fi1 asm.fi2 asm.fi bye"
        !          4540: @end example
        !          4541: 
        !          4542: @cindex --clear-dictionary, creating image files
        !          4543: @cindex --offset-image, creating image files
        !          4544: The flag @code{--clear-dictionary} ensures that the dictionary memory is
        !          4545: cleared at the start (so you won't see @code{comp-image} messages about
        !          4546: spurious differences). The flag @code{--offset-image} loads the image at
        !          4547: a small offset from its normal position, ensuring that the resulting
        !          4548: nonrelocatable images have different bases (otherwise @code{comp-image}
        !          4549: cannot work).
        !          4550: 
        !          4551: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
        !          4552: @section Fully Relocatable Image Files
        !          4553: @cindex fully relocatable image files
        !          4554: @cindex image files, fully relocatable
        !          4555: 
        !          4556: @cindex @file{kern*.fi}, relocatability
        !          4557: These image files have relocatable data addresses, and tokens for code
        !          4558: addresses. They are still a bit machine dependent with respect to the
        !          4559: size and format (byte order, floating point format) of their data. The
        !          4560: @file{kernlxxx.fi} files are fully relocatable.
        !          4561: 
        !          4562: @cindex cross-compiler
        !          4563: @cindex metacompiler
        !          4564: At present a fully relocatable image file can only be produced by
        !          4565: @code{cross}, a batch compiler that accepts a Forth-like programming
        !          4566: language. This @code{cross} language has to be documented
        !          4567: yet.
        !          4568: 
        !          4569: @cindex target compiler
        !          4570: @code{cross} not only allows the programmer to create fully relocatable
        !          4571: image files, but also to create image files for machines with different
        !          4572: data sizes and data formats than the one used for generating the image
        !          4573: file. This convenience is bought with restrictions and inconveniences in
        !          4574: programming. E.g., addresses have to be stored in memory with special
        !          4575: words (@code{A!}, @code{A,}, etc.) in order to make the code
        !          4576: relocatable.
        !          4577: 
        !          4578: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
        !          4579: @section Stack and Dictionary Sizes
        !          4580: @cindex image file, stack and dictionary sizes
        !          4581: @cindex dictionary size default
        !          4582: @cindex stack size default
        !          4583: 
        !          4584: If you invoke Gforth with a command line flag for the size
        !          4585: (@pxref{Invoking Gforth}), the size you specify is stored in the
        !          4586: dictionary. If you save the dictionary with @code{savesystem}, this size
        !          4587: will become the default for the resulting image file. E.g., the
        !          4588: following will create a non-relocatable version of gforth.fi with a 1MB
        !          4589: dictionary:
        !          4590: 
        !          4591: @example
        !          4592: gforth -m 1M -e "savesystem gforth1M.fi bye"
        !          4593: @end example
        !          4594: 
        !          4595: In other words, if you want to set the default size for the dictionary
        !          4596: and the stacks of an image, just invoke Gforth with the appropriate
        !          4597: options when creating the image. When creating a data-relocatable image,
        !          4598: you have to use the same size options for both @code{savesystem}s.
        !          4599: 
        !          4600: @cindex stack size, cache-friendly
        !          4601: Note: For cache-friendly behaviour (i.e., good performance), you should
        !          4602: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
        !          4603: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
        !          4604: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
        !          4605: 
        !          4606: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
        !          4607: @section Running Image Files
        !          4608: @cindex running image files
        !          4609: @cindex invoking image files
        !          4610: @cindex image file invocation
        !          4611: 
        !          4612: @cindex -i, invoke image file
        !          4613: @cindex --image file, invoke image file
        !          4614: You can invoke Gforth with an image file @var{image} instead of the
        !          4615: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
        !          4616: @example
        !          4617: gforth -i @var{image}
        !          4618: @end example
        !          4619: 
        !          4620: @cindex executable image file
        !          4621: @cindex image files, executable
        !          4622: If your operating system supports starting scripts with a line of the
        !          4623: form @code{#! ...}, you can make your image file executable, and you'll
        !          4624: just have to type the image file name to start Gforth with this image
        !          4625: file (note that the file extension @code{.fi} is just a convention).
        !          4626: 
        !          4627: I.e., in most Unix systems, you just have to make the image file
        !          4628: @var{image} executable with
        !          4629: 
        !          4630: @example
        !          4631: chmod +x @var{image}
        !          4632: @end example
        !          4633: 
        !          4634: and then you can invoke it by simply typing @var{image} instead of
        !          4635: @code{gforth -i @var{image}}.
        !          4636: 
        !          4637: @node Modifying the Startup Sequence,  , Running Image Files, Image Files
        !          4638: @section Modifying the Startup Sequence
        !          4639: @cindex startup sequence for image file
        !          4640: @cindex image file initialization sequence
        !          4641: @cindex initialization sequence of image file
        !          4642: 
        !          4643: You can add your own initialization to the startup sequence through the
        !          4644: deferred word
        !          4645: 
        !          4646: doc-'cold
        !          4647: 
        !          4648: @code{'cold} is invoked just before the image-specific command line
        !          4649: processing (by default, loading files and evaluating (@code{-e}) strings
        !          4650: starts.
        !          4651: 
        !          4652: A sequence for adding your initialization usually looks like this:
        !          4653: 
        !          4654: @example
        !          4655: :noname
        !          4656:     Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
        !          4657:     ... \ your stuff
        !          4658: ; IS 'cold
        !          4659: @end example
        !          4660: 
        !          4661: @cindex turnkey image files
        !          4662: @cindex image files, turnkey applications
        !          4663: You can make a turnkey image by letting @code{'cold} execute a word
        !          4664: (your turnkey application) that never returns; instead, it exits Gforth
        !          4665: via @code{bye} or @code{throw}.
        !          4666: 
        !          4667: @c ******************************************************************
        !          4668: @node Engine, Bugs, Image Files, Top
        !          4669: @chapter Engine
        !          4670: @cindex engine
        !          4671: @cindex virtual machine
1.3       anton    4672: 
1.17      anton    4673: Reading this section is not necessary for programming with Gforth. It
1.43    ! anton    4674: may be helpful for finding your way in the Gforth sources.
1.3       anton    4675: 
1.24      anton    4676: The ideas in this section have also been published in the papers
                   4677: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
                   4678: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
                   4679: Ertl, presented at EuroForth '93; the latter is available at
                   4680: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
                   4681: 
1.4       anton    4682: @menu
                   4683: * Portability::                 
                   4684: * Threading::                   
                   4685: * Primitives::                  
1.17      anton    4686: * Performance::                 
1.4       anton    4687: @end menu
                   4688: 
1.43    ! anton    4689: @node Portability, Threading, Engine, Engine
1.3       anton    4690: @section Portability
1.43    ! anton    4691: @cindex engine portability
1.3       anton    4692: 
                   4693: One of the main goals of the effort is availability across a wide range
                   4694: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
                   4695: this goal by manually coding the engine in assembly language for several
                   4696: then-popular processors. This approach is very labor-intensive and the
                   4697: results are short-lived due to progress in computer architecture.
                   4698: 
1.43    ! anton    4699: @cindex C, using C for the engine
1.3       anton    4700: Others have avoided this problem by coding in C, e.g., Mitch Bradley
                   4701: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
                   4702: particularly popular for UNIX-based Forths due to the large variety of
                   4703: architectures of UNIX machines. Unfortunately an implementation in C
                   4704: does not mix well with the goals of efficiency and with using
                   4705: traditional techniques: Indirect or direct threading cannot be expressed
                   4706: in C, and switch threading, the fastest technique available in C, is
1.43    ! anton    4707: significantly slower. Another problem with C is that it is very
1.3       anton    4708: cumbersome to express double integer arithmetic.
                   4709: 
1.43    ! anton    4710: @cindex GNU C for the engine
        !          4711: @cindex long long
1.3       anton    4712: Fortunately, there is a portable language that does not have these
                   4713: limitations: GNU C, the version of C processed by the GNU C compiler
                   4714: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
                   4715: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
                   4716: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
                   4717: threading possible, its @code{long long} type (@pxref{Long Long, ,
1.33      anton    4718: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
1.32      anton    4719: double numbers@footnote{Unfortunately, long longs are not implemented
                   4720: properly on all machines (e.g., on alpha-osf1, long longs are only 64
                   4721: bits, the same size as longs (and pointers), but they should be twice as
                   4722: long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
                   4723: C Manual}). So, we had to implement doubles in C after all. Still, on
                   4724: most machines we can use long longs and achieve better performance than
                   4725: with the emulation package.}. GNU C is available for free on all
                   4726: important (and many unimportant) UNIX machines, VMS, 80386s running
                   4727: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
                   4728: on all these machines.
1.3       anton    4729: 
                   4730: Writing in a portable language has the reputation of producing code that
                   4731: is slower than assembly. For our Forth engine we repeatedly looked at
                   4732: the code produced by the compiler and eliminated most compiler-induced
1.43    ! anton    4733: inefficiencies by appropriate changes in the source code.
1.3       anton    4734: 
1.43    ! anton    4735: @cindex explicit register declarations
        !          4736: @cindex --enable-force-reg, configuration flag
        !          4737: @cindex -DFORCE_REG
1.3       anton    4738: However, register allocation cannot be portably influenced by the
                   4739: programmer, leading to some inefficiencies on register-starved
                   4740: machines. We use explicit register declarations (@pxref{Explicit Reg
                   4741: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
                   4742: improve the speed on some machines. They are turned on by using the
1.43    ! anton    4743: configuration flag @code{--enable-force-reg} (@code{gcc} switch
        !          4744: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
        !          4745: machine, but also on the compiler version: On some machines some
        !          4746: compiler versions produce incorrect code when certain explicit register
        !          4747: declarations are used. So by default @code{-DFORCE_REG} is not used.
1.3       anton    4748: 
1.43    ! anton    4749: @node Threading, Primitives, Portability, Engine
1.3       anton    4750: @section Threading
1.43    ! anton    4751: @cindex inner interpreter implementation
        !          4752: @cindex threaded code implementation
1.3       anton    4753: 
1.43    ! anton    4754: @cindex labels as values
1.3       anton    4755: GNU C's labels as values extension (available since @code{gcc-2.0},
                   4756: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
                   4757: makes it possible to take the address of @var{label} by writing
                   4758: @code{&&@var{label}}.  This address can then be used in a statement like
                   4759: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
                   4760: @code{goto x}.
                   4761: 
1.43    ! anton    4762: @cindex NEXT, indirect threaded
        !          4763: @cindex indirect threaded inner interpreter
        !          4764: @cindex inner interpreter, indirect threaded
1.3       anton    4765: With this feature an indirect threaded NEXT looks like:
                   4766: @example
                   4767: cfa = *ip++;
                   4768: ca = *cfa;
                   4769: goto *ca;
                   4770: @end example
1.43    ! anton    4771: @cindex instruction pointer
1.3       anton    4772: For those unfamiliar with the names: @code{ip} is the Forth instruction
                   4773: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
                   4774: execution token and points to the code field of the next word to be
                   4775: executed; The @code{ca} (code address) fetched from there points to some
                   4776: executable code, e.g., a primitive or the colon definition handler
                   4777: @code{docol}.
                   4778: 
1.43    ! anton    4779: @cindex NEXT, direct threaded
        !          4780: @cindex direct threaded inner interpreter
        !          4781: @cindex inner interpreter, direct threaded
1.3       anton    4782: Direct threading is even simpler:
                   4783: @example
                   4784: ca = *ip++;
                   4785: goto *ca;
                   4786: @end example
                   4787: 
                   4788: Of course we have packaged the whole thing neatly in macros called
                   4789: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
                   4790: 
1.4       anton    4791: @menu
                   4792: * Scheduling::                  
                   4793: * Direct or Indirect Threaded?::  
                   4794: * DOES>::                       
                   4795: @end menu
                   4796: 
                   4797: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3       anton    4798: @subsection Scheduling
1.43    ! anton    4799: @cindex inner interpreter optimization
1.3       anton    4800: 
                   4801: There is a little complication: Pipelined and superscalar processors,
                   4802: i.e., RISC and some modern CISC machines can process independent
                   4803: instructions while waiting for the results of an instruction. The
                   4804: compiler usually reorders (schedules) the instructions in a way that
                   4805: achieves good usage of these delay slots. However, on our first tries
                   4806: the compiler did not do well on scheduling primitives. E.g., for
                   4807: @code{+} implemented as
                   4808: @example
                   4809: n=sp[0]+sp[1];
                   4810: sp++;
                   4811: sp[0]=n;
                   4812: NEXT;
                   4813: @end example
                   4814: the NEXT comes strictly after the other code, i.e., there is nearly no
                   4815: scheduling. After a little thought the problem becomes clear: The
                   4816: compiler cannot know that sp and ip point to different addresses (and
1.4       anton    4817: the version of @code{gcc} we used would not know it even if it was
                   4818: possible), so it could not move the load of the cfa above the store to
                   4819: the TOS. Indeed the pointers could be the same, if code on or very near
                   4820: the top of stack were executed. In the interest of speed we chose to
                   4821: forbid this probably unused ``feature'' and helped the compiler in
                   4822: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
                   4823: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3       anton    4824: @example
                   4825: n=sp[0]+sp[1];
                   4826: sp++;
                   4827: NEXT_P1;
                   4828: sp[0]=n;
                   4829: NEXT_P2;
                   4830: @end example
1.4       anton    4831: This can be scheduled optimally by the compiler.
1.3       anton    4832: 
                   4833: This division can be turned off with the switch @code{-DCISC_NEXT}. This
                   4834: switch is on by default on machines that do not profit from scheduling
                   4835: (e.g., the 80386), in order to preserve registers.
                   4836: 
1.4       anton    4837: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3       anton    4838: @subsection Direct or Indirect Threaded?
1.43    ! anton    4839: @cindex threading, direct or indirect?
1.3       anton    4840: 
1.43    ! anton    4841: @cindex -DDIRECT_THREADED
1.3       anton    4842: Both! After packaging the nasty details in macro definitions we
                   4843: realized that we could switch between direct and indirect threading by
                   4844: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
                   4845: defining a few machine-specific macros for the direct-threading case.
                   4846: On the Forth level we also offer access words that hide the
                   4847: differences between the threading methods (@pxref{Threading Words}).
                   4848: 
1.43    ! anton    4849: Indirect threading is implemented completely machine-independently.
        !          4850: Direct threading needs routines for creating jumps to the executable
        !          4851: code (e.g. to docol or dodoes). These routines are inherently
        !          4852: machine-dependent, but they do not amount to many source lines. I.e.,
        !          4853: even porting direct threading to a new machine is a small effort.
        !          4854: 
        !          4855: @cindex --enable-indirect-threaded, configuration flag
        !          4856: @cindex --enable-direct-threaded, configuration flag
        !          4857: The default threading method is machine-dependent. You can enforce a
        !          4858: specific threading method when building Gforth with the configuration
        !          4859: flag @code{--enable-direct-threaded} or
        !          4860: @code{--enable-indirect-threaded}. Note that direct threading is not
        !          4861: supported on all machines.
1.3       anton    4862: 
1.4       anton    4863: @node DOES>,  , Direct or Indirect Threaded?, Threading
1.3       anton    4864: @subsection DOES>
1.43    ! anton    4865: @cindex @code{DOES>} implementation
        !          4866: 
        !          4867: @cindex dodoes routine
        !          4868: @cindex DOES-code
1.3       anton    4869: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
                   4870: the chunk of code executed by every word defined by a
                   4871: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
1.43    ! anton    4872: the Forth code to be executed, i.e. the code after the
        !          4873: @code{DOES>} (the DOES-code)? There are two solutions:
1.3       anton    4874: 
                   4875: In fig-Forth the code field points directly to the dodoes and the
1.43    ! anton    4876: DOES-code address is stored in the cell after the code address (i.e. at
        !          4877: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
        !          4878: the Forth-79 and all later standards, because in fig-Forth this address
1.3       anton    4879: lies in the body (which is illegal in these standards). However, by
                   4880: making the code field larger for all words this solution becomes legal
1.43    ! anton    4881: again. We use this approach for the indirect threaded version and for
        !          4882: direct threading on some machines. Leaving a cell unused in most words
        !          4883: is a bit wasteful, but on the machines we are targeting this is hardly a
        !          4884: problem. The other reason for having a code field size of two cells is
        !          4885: to avoid having different image files for direct and indirect threaded
        !          4886: systems (direct threaded systems require two-cell code fields on many
        !          4887: machines).
1.3       anton    4888: 
1.43    ! anton    4889: @cindex DOES-handler
1.3       anton    4890: The other approach is that the code field points or jumps to the cell
                   4891: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
1.43    ! anton    4892: this address (the DOES-handler). @code{dodoes} can then get the
        !          4893: DOES-code address by computing the code address, i.e., the address of
        !          4894: the jump to dodoes, and add the length of that jump field. A variant of
        !          4895: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
        !          4896: return address (which can be found in the return register on RISCs) is
        !          4897: the DOES-code address. Since the two cells available in the code field
        !          4898: are used up by the jump to the code address in direct threading on many
        !          4899: architectures, we use this approach for direct threading on these
        !          4900: architectures. We did not want to add another cell to the code field.
1.3       anton    4901: 
1.43    ! anton    4902: @node Primitives, Performance, Threading, Engine
1.3       anton    4903: @section Primitives
1.43    ! anton    4904: @cindex primitives, implementation
        !          4905: @cindex virtual machine instructions, implementation
1.3       anton    4906: 
1.4       anton    4907: @menu
                   4908: * Automatic Generation::        
                   4909: * TOS Optimization::            
                   4910: * Produced code::               
                   4911: @end menu
                   4912: 
                   4913: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3       anton    4914: @subsection Automatic Generation
1.43    ! anton    4915: @cindex primitives, automatic generation
1.3       anton    4916: 
1.43    ! anton    4917: @cindex @file{prims2x.fs}
1.3       anton    4918: Since the primitives are implemented in a portable language, there is no
                   4919: longer any need to minimize the number of primitives. On the contrary,
1.43    ! anton    4920: having many primitives has an advantage: speed. In order to reduce the
1.3       anton    4921: number of errors in primitives and to make programming them easier, we
                   4922: provide a tool, the primitive generator (@file{prims2x.fs}), that
                   4923: automatically generates most (and sometimes all) of the C code for a
                   4924: primitive from the stack effect notation.  The source for a primitive
                   4925: has the following form:
                   4926: 
1.43    ! anton    4927: @cindex primitive source format
1.3       anton    4928: @format
                   4929: @var{Forth-name}       @var{stack-effect}      @var{category}  [@var{pronounc.}]
                   4930: [@code{""}@var{glossary entry}@code{""}]
                   4931: @var{C code}
                   4932: [@code{:}
                   4933: @var{Forth code}]
                   4934: @end format
                   4935: 
                   4936: The items in brackets are optional. The category and glossary fields
                   4937: are there for generating the documentation, the Forth code is there
                   4938: for manual implementations on machines without GNU C. E.g., the source
                   4939: for the primitive @code{+} is:
                   4940: @example
                   4941: +    n1 n2 -- n    core    plus
                   4942: n = n1+n2;
                   4943: @end example
                   4944: 
                   4945: This looks like a specification, but in fact @code{n = n1+n2} is C
                   4946: code. Our primitive generation tool extracts a lot of information from
                   4947: the stack effect notations@footnote{We use a one-stack notation, even
                   4948: though we have separate data and floating-point stacks; The separate
                   4949: notation can be generated easily from the unified notation.}: The number
                   4950: of items popped from and pushed on the stack, their type, and by what
                   4951: name they are referred to in the C code. It then generates a C code
                   4952: prelude and postlude for each primitive. The final C code for @code{+}
                   4953: looks like this:
                   4954: 
                   4955: @example
                   4956: I_plus:        /* + ( n1 n2 -- n ) */  /* label, stack effect */
                   4957: /*  */                          /* documentation */
1.4       anton    4958: @{
1.3       anton    4959: DEF_CA                          /* definition of variable ca (indirect threading) */
                   4960: Cell n1;                        /* definitions of variables */
                   4961: Cell n2;
                   4962: Cell n;
                   4963: n1 = (Cell) sp[1];              /* input */
                   4964: n2 = (Cell) TOS;
                   4965: sp += 1;                        /* stack adjustment */
                   4966: NAME("+")                       /* debugging output (with -DDEBUG) */
1.4       anton    4967: @{
1.3       anton    4968: n = n1+n2;                      /* C code taken from the source */
1.4       anton    4969: @}
1.3       anton    4970: NEXT_P1;                        /* NEXT part 1 */
                   4971: TOS = (Cell)n;                  /* output */
                   4972: NEXT_P2;                        /* NEXT part 2 */
1.4       anton    4973: @}
1.3       anton    4974: @end example
                   4975: 
                   4976: This looks long and inefficient, but the GNU C compiler optimizes quite
                   4977: well and produces optimal code for @code{+} on, e.g., the R3000 and the
                   4978: HP RISC machines: Defining the @code{n}s does not produce any code, and
                   4979: using them as intermediate storage also adds no cost.
                   4980: 
                   4981: There are also other optimizations, that are not illustrated by this
                   4982: example: Assignments between simple variables are usually for free (copy
                   4983: propagation). If one of the stack items is not used by the primitive
                   4984: (e.g.  in @code{drop}), the compiler eliminates the load from the stack
                   4985: (dead code elimination). On the other hand, there are some things that
                   4986: the compiler does not do, therefore they are performed by
                   4987: @file{prims2x.fs}: The compiler does not optimize code away that stores
                   4988: a stack item to the place where it just came from (e.g., @code{over}).
                   4989: 
                   4990: While programming a primitive is usually easy, there are a few cases
                   4991: where the programmer has to take the actions of the generator into
                   4992: account, most notably @code{?dup}, but also words that do not (always)
                   4993: fall through to NEXT.
                   4994: 
1.4       anton    4995: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3       anton    4996: @subsection TOS Optimization
1.43    ! anton    4997: @cindex TOS optimization for primitives
        !          4998: @cindex primitives, keeping the TOS in a register
1.3       anton    4999: 
                   5000: An important optimization for stack machine emulators, e.g., Forth
                   5001: engines, is keeping  one or more of the top stack items in
1.4       anton    5002: registers.  If a word has the stack effect @var{in1}...@var{inx} @code{--}
                   5003: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.34      anton    5004: @itemize @bullet
1.3       anton    5005: @item
                   5006: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
                   5007: due to fewer loads from and stores to the stack.
                   5008: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
                   5009: @var{y<n}, due to additional moves between registers.
                   5010: @end itemize
                   5011: 
1.43    ! anton    5012: @cindex -DUSE_TOS
        !          5013: @cindex -DUSE_NO_TOS
1.3       anton    5014: In particular, keeping one item in a register is never a disadvantage,
                   5015: if there are enough registers. Keeping two items in registers is a
                   5016: disadvantage for frequent words like @code{?branch}, constants,
                   5017: variables, literals and @code{i}. Therefore our generator only produces
                   5018: code that keeps zero or one items in registers. The generated C code
                   5019: covers both cases; the selection between these alternatives is made at
                   5020: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
                   5021: code for @code{+} is just a simple variable name in the one-item case,
                   5022: otherwise it is a macro that expands into @code{sp[0]}. Note that the
                   5023: GNU C compiler tries to keep simple variables like @code{TOS} in
                   5024: registers, and it usually succeeds, if there are enough registers.
                   5025: 
1.43    ! anton    5026: @cindex -DUSE_FTOS
        !          5027: @cindex -DUSE_NO_FTOS
1.3       anton    5028: The primitive generator performs the TOS optimization for the
                   5029: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
                   5030: operations the benefit of this optimization is even larger:
                   5031: floating-point operations take quite long on most processors, but can be
                   5032: performed in parallel with other operations as long as their results are
                   5033: not used. If the FP-TOS is kept in a register, this works. If
                   5034: it is kept on the stack, i.e., in memory, the store into memory has to
                   5035: wait for the result of the floating-point operation, lengthening the
                   5036: execution time of the primitive considerably.
                   5037: 
                   5038: The TOS optimization makes the automatic generation of primitives a
                   5039: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
                   5040: @code{TOS} is not sufficient. There are some special cases to
                   5041: consider:
1.34      anton    5042: @itemize @bullet
1.3       anton    5043: @item In the case of @code{dup ( w -- w w )} the generator must not
                   5044: eliminate the store to the original location of the item on the stack,
                   5045: if the TOS optimization is turned on.
1.4       anton    5046: @item Primitives with stack effects of the form @code{--}
                   5047: @var{out1}...@var{outy} must store the TOS to the stack at the start.
                   5048: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3       anton    5049: must load the TOS from the stack at the end. But for the null stack
                   5050: effect @code{--} no stores or loads should be generated.
                   5051: @end itemize
                   5052: 
1.4       anton    5053: @node Produced code,  , TOS Optimization, Primitives
1.3       anton    5054: @subsection Produced code
1.43    ! anton    5055: @cindex primitives, assembly code listing
1.3       anton    5056: 
1.43    ! anton    5057: @cindex @file{engine.s}
1.3       anton    5058: To see what assembly code is produced for the primitives on your machine
                   5059: with your compiler and your flag settings, type @code{make engine.s} and
1.4       anton    5060: look at the resulting file @file{engine.s}.
1.3       anton    5061: 
1.43    ! anton    5062: @node  Performance,  , Primitives, Engine
1.17      anton    5063: @section Performance
1.43    ! anton    5064: @cindex performance of some Forth interpreters
        !          5065: @cindex engine performance
        !          5066: @cindex benchmarking Forth systems
        !          5067: @cindex Gforth performance
1.17      anton    5068: 
                   5069: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
                   5070: impossible to write a significantly faster engine.
                   5071: 
                   5072: On register-starved machines like the 386 architecture processors
                   5073: improvements are possible, because @code{gcc} does not utilize the
                   5074: registers as well as a human, even with explicit register declarations;
                   5075: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
                   5076: and hand-tuned it for the 486; this system is 1.19 times faster on the
                   5077: Sieve benchmark on a 486DX2/66 than Gforth compiled with
                   5078: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
                   5079: 
1.43    ! anton    5080: @cindex Win32Forth performance
        !          5081: @cindex NT Forth performance
        !          5082: @cindex eforth performance
        !          5083: @cindex ThisForth performance
        !          5084: @cindex PFE performance
        !          5085: @cindex TILE performance
1.17      anton    5086: However, this potential advantage of assembly language implementations
                   5087: is not necessarily realized in complete Forth systems: We compared
1.26      anton    5088: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
                   5089: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
                   5090: 1994) and Eforth (with and without peephole (aka pinhole) optimization
                   5091: of the threaded code); all these systems were written in assembly
1.30      anton    5092: language. We also compared Gforth with three systems written in C:
1.32      anton    5093: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
                   5094: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
                   5095: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
                   5096: -fomit-frame-pointer; ThisForth employs peephole optimization of the
                   5097: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
                   5098: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
                   5099: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
                   5100: 486DX2/66 with similar memory performance under Windows NT. Marcel
                   5101: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
                   5102: added the peephole optimizer, ran the benchmarks and reported the
                   5103: results.
1.17      anton    5104:  
                   5105: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
                   5106: matrix multiplication come from the Stanford integer benchmarks and have
                   5107: been translated into Forth by Martin Fraeman; we used the versions
1.30      anton    5108: included in the TILE Forth package, but with bigger data set sizes; and
                   5109: a recursive Fibonacci number computation for benchmarking calling
                   5110: performance. The following table shows the time taken for the benchmarks
                   5111: scaled by the time taken by Gforth (in other words, it shows the speedup
                   5112: factor that Gforth achieved over the other systems).
1.17      anton    5113: 
                   5114: @example
1.30      anton    5115: relative      Win32-    NT       eforth       This-
                   5116:   time  Gforth Forth Forth eforth  +opt   PFE Forth  TILE
1.32      anton    5117: sieve     1.00  1.39  1.14   1.39  0.85  1.58  3.18  8.58
                   5118: bubble    1.00  1.31  1.41   1.48  0.88  1.50        3.88
1.38      anton    5119: matmul    1.00  1.47  1.35   1.46  0.74  1.58        4.09
                   5120: fib       1.00  1.52  1.34   1.22  0.86  1.74  2.99  4.30
1.17      anton    5121: @end example
                   5122: 
                   5123: You may find the good performance of Gforth compared with the systems
                   5124: written in assembly language quite surprising. One important reason for
                   5125: the disappointing performance of these systems is probably that they are
                   5126: not written optimally for the 486 (e.g., they use the @code{lods}
                   5127: instruction). In addition, Win32Forth uses a comfortable, but costly
                   5128: method for relocating the Forth image: like @code{cforth}, it computes
                   5129: the actual addresses at run time, resulting in two address computations
1.43    ! anton    5130: per NEXT (@pxref{Image File Background}).
1.17      anton    5131: 
1.26      anton    5132: Only Eforth with the peephole optimizer performs comparable to
                   5133: Gforth. The speedups achieved with peephole optimization of threaded
                   5134: code are quite remarkable. Adding a peephole optimizer to Gforth should
                   5135: cause similar speedups.
                   5136: 
1.30      anton    5137: The speedup of Gforth over PFE, ThisForth and TILE can be easily
1.43    ! anton    5138: explained with the self-imposed restriction of the latter systems to
        !          5139: standard C, which makes efficient threading impossible (however, the
        !          5140: measured implementation of PFE uses a GNU C extension: @ref{Global Reg
        !          5141: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
        !          5142: Moreover, current C compilers have a hard time optimizing other aspects
        !          5143: of the ThisForth and the TILE source.
1.17      anton    5144: 
                   5145: Note that the performance of Gforth on 386 architecture processors
                   5146: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
                   5147: failed to allocate any of the virtual machine registers into real
                   5148: machine registers by itself and would not work correctly with explicit
                   5149: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
                   5150: running the Sieve) than the one measured above.
                   5151: 
1.43    ! anton    5152: Note also that there have been several releases of Win32Forth since the
        !          5153: release presented here, so the results presented here may have little
        !          5154: predictive value for the performance of Win32Forth today.
        !          5155: 
        !          5156: @cindex @file{Benchres}
1.26      anton    5157: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
                   5158: Maierhofer (presented at EuroForth '95), an indirect threaded version of
                   5159: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
                   5160: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
                   5161: used here. The paper available at
1.24      anton    5162: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
1.43    ! anton    5163: it also contains numbers for some native code systems. You can find a
        !          5164: newer version of these measurements at
        !          5165: @file{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
        !          5166: find numbers for Gforth on various machines in @file{Benchres}.
1.24      anton    5167: 
1.43    ! anton    5168: @node Bugs, Origin, Engine, Top
1.4       anton    5169: @chapter Bugs
1.43    ! anton    5170: @cindex bug reporting
1.4       anton    5171: 
1.17      anton    5172: Known bugs are described in the file BUGS in the Gforth distribution.
                   5173: 
1.24      anton    5174: If you find a bug, please send a bug report to
1.32      anton    5175: @code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
1.17      anton    5176: describe the Gforth version used (it is announced at the start of an
                   5177: interactive Gforth session), the machine and operating system (on Unix
                   5178: systems you can use @code{uname -a} to produce this information), the
1.43    ! anton    5179: installation options (send the @file{config.status} file), and a
1.24      anton    5180: complete list of changes you (or your installer) have made to the Gforth
                   5181: sources (if any); it should contain a program (or a sequence of keyboard
                   5182: commands) that reproduces the bug and a description of what you think
                   5183: constitutes the buggy behaviour.
1.17      anton    5184: 
                   5185: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
                   5186: to Report Bugs, gcc.info, GNU C Manual}.
                   5187: 
                   5188: 
1.29      anton    5189: @node Origin, Word Index, Bugs, Top
                   5190: @chapter Authors and Ancestors of Gforth
                   5191: 
                   5192: @section Authors and Contributors
1.43    ! anton    5193: @cindex authors of Gforth
        !          5194: @cindex contributors to Gforth
1.29      anton    5195: 
                   5196: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
1.30      anton    5197: Ertl. The third major author was Jens Wilke.  Lennart Benschop (who was
                   5198: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
                   5199: with their continuous feedback. Lennart Benshop contributed
1.29      anton    5200: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
                   5201: support for calling C libraries. Helpful comments also came from Paul
1.37      anton    5202: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.39      anton    5203: Wavrik, Barrie Stott and Marc de Groot.
1.29      anton    5204: 
1.30      anton    5205: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
                   5206: and autoconf, among others), and to the creators of the Internet: Gforth
                   5207: was developed across the Internet, and its authors have not met
                   5208: physically yet.
                   5209: 
1.29      anton    5210: @section Pedigree
1.43    ! anton    5211: @cindex Pedigree of Gforth
1.4       anton    5212: 
1.17      anton    5213: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24      anton    5214: Dirk Zoller) will cross-fertilize each other. Of course, a significant
                   5215: part of the design of Gforth was prescribed by ANS Forth.
1.17      anton    5216: 
1.23      pazsan   5217: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
                   5218: 32 bit native code version of VolksForth for the Atari ST, written
                   5219: mostly by Dietrich Weineck.
                   5220: 
                   5221: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
                   5222: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24      anton    5223: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17      anton    5224: 
1.34      anton    5225: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17      anton    5226: Forth-83 standard. !! Pedigree? When?
                   5227: 
                   5228: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24      anton    5229: 1979. Robert Selzer and Bill Ragsdale developed the original
                   5230: implementation of fig-Forth for the 6502 based on microForth.
                   5231: 
                   5232: The principal architect of microForth was Dean Sanderson. microForth was
1.41      anton    5233: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
1.24      anton    5234: the 1802, and subsequently implemented on the 8080, the 6800 and the
                   5235: Z80.
1.17      anton    5236: 
1.24      anton    5237: All earlier Forth systems were custom-made, usually by Charles Moore,
1.30      anton    5238: who discovered (as he puts it) Forth during the late 60s. The first full
                   5239: Forth existed in 1971.
1.17      anton    5240: 
                   5241: A part of the information in this section comes from @cite{The Evolution
                   5242: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
                   5243: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
                   5244: Notices 28(3), 1993.  You can find more historical and genealogical
                   5245: information about Forth there.
                   5246: 
1.43    ! anton    5247: @node Word Index, Concept Index, Origin, Top
        !          5248: @unnumbered Word Index
1.4       anton    5249: 
1.18      anton    5250: This index is as incomplete as the manual. Each word is listed with
                   5251: stack effect and wordset.
1.17      anton    5252: 
                   5253: @printindex fn
                   5254: 
1.43    ! anton    5255: @node Concept Index,  , Word Index, Top
        !          5256: @unnumbered Concept and Word Index
        !          5257: 
        !          5258: This index is as incomplete as the manual. Not all entries listed are
        !          5259: present verbatim in the text. Only the names are listed for the words
        !          5260: here.
1.17      anton    5261: 
1.43    ! anton    5262: @printindex cp
1.1       anton    5263: 
                   5264: @contents
                   5265: @bye
                   5266: 

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