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.)
4: @setfilename gforth.info
5: @settitle Gforth Manual
6: @dircategory GNU programming tools
7: @direntry
8: * Gforth: (gforth). A fast interpreter for the Forth language.
9: @end direntry
10: @comment @setchapternewpage odd
11: @macro progstyle {TEXT}
12: @quotation
13: {Programming style note:} \TEXT\
14: @end quotation
15: @end macro
16: @comment %**end of header (This is for running Texinfo on a region.)
17:
18: @ifinfo
19: This file documents Gforth 0.4
20:
21: Copyright @copyright{} 1995-1997 Free Software Foundation, Inc.
22:
23: Permission is granted to make and distribute verbatim copies of
24: this manual provided the copyright notice and this permission notice
25: are preserved on all copies.
26:
27: @ignore
28: Permission is granted to process this file through TeX and print the
29: results, provided the printed document carries a copying permission
30: notice identical to this one except for the removal of this paragraph
31: (this paragraph not being relevant to the printed manual).
32:
33: @end ignore
34: Permission is granted to copy and distribute modified versions of this
35: manual under the conditions for verbatim copying, provided also that the
36: sections entitled "Distribution" and "General Public License" are
37: included exactly as in the original, and provided that the entire
38: resulting derived work is distributed under the terms of a permission
39: notice identical to this one.
40:
41: Permission is granted to copy and distribute translations of this manual
42: into another language, under the above conditions for modified versions,
43: except that the sections entitled "Distribution" and "General Public
44: License" may be included in a translation approved by the author instead
45: of in the original English.
46: @end ifinfo
47:
48: @finalout
49: @titlepage
50: @sp 10
51: @center @titlefont{Gforth Manual}
52: @sp 2
53: @center for version 0.4
54: @sp 2
55: @center Anton Ertl
56: @center Bernd Paysan
57: @center Jens Wilke
58: @sp 3
59: @center This manual is permanently under construction
60:
61: @comment The following two commands start the copyright page.
62: @page
63: @vskip 0pt plus 1filll
64: Copyright @copyright{} 1995--1997 Free Software Foundation, Inc.
65:
66: @comment !! Published by ... or You can get a copy of this manual ...
67:
68: Permission is granted to make and distribute verbatim copies of
69: this manual provided the copyright notice and this permission notice
70: are preserved on all copies.
71:
72: Permission is granted to copy and distribute modified versions of this
73: manual under the conditions for verbatim copying, provided also that the
74: sections entitled "Distribution" and "General Public License" are
75: included exactly as in the original, and provided that the entire
76: resulting derived work is distributed under the terms of a permission
77: notice identical to this one.
78:
79: Permission is granted to copy and distribute translations of this manual
80: into another language, under the above conditions for modified versions,
81: except that the sections entitled "Distribution" and "General Public
82: License" may be included in a translation approved by the author instead
83: of in the original English.
84: @end titlepage
85:
86:
87: @node Top, License, (dir), (dir)
88: @ifinfo
89: Gforth is a free implementation of ANS Forth available on many
90: personal machines. This manual corresponds to version 0.4.
91: @end ifinfo
92:
93: @menu
94: * License::
95: * Goals:: About the Gforth Project
96: * Other Books:: Things you might want to read
97: * Invoking Gforth:: Starting Gforth
98: * Words:: Forth words available in Gforth
99: * Tools:: Programming tools
100: * ANS conformance:: Implementation-defined options etc.
101: * Model:: The abstract machine of Gforth
102: * Integrating Gforth:: Forth as scripting language for applications
103: * Emacs and Gforth:: The Gforth Mode
104: * Image Files:: @code{.fi} files contain compiled code
105: * Engine:: The inner interpreter and the primitives
106: * Bugs:: How to report them
107: * Origin:: Authors and ancestors of Gforth
108: * Word Index:: An item for each Forth word
109: * Concept Index:: A menu covering many topics
110: @end menu
111:
112: @node License, Goals, Top, Top
113: @unnumbered GNU GENERAL PUBLIC LICENSE
114: @center Version 2, June 1991
115:
116: @display
117: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
118: 675 Mass Ave, Cambridge, MA 02139, USA
119:
120: Everyone is permitted to copy and distribute verbatim copies
121: of this license document, but changing it is not allowed.
122: @end display
123:
124: @unnumberedsec Preamble
125:
126: The licenses for most software are designed to take away your
127: freedom to share and change it. By contrast, the GNU General Public
128: License is intended to guarantee your freedom to share and change free
129: software---to make sure the software is free for all its users. This
130: General Public License applies to most of the Free Software
131: Foundation's software and to any other program whose authors commit to
132: using it. (Some other Free Software Foundation software is covered by
133: the GNU Library General Public License instead.) You can apply it to
134: your programs, too.
135:
136: When we speak of free software, we are referring to freedom, not
137: price. Our General Public Licenses are designed to make sure that you
138: have the freedom to distribute copies of free software (and charge for
139: this service if you wish), that you receive source code or can get it
140: if you want it, that you can change the software or use pieces of it
141: in new free programs; and that you know you can do these things.
142:
143: To protect your rights, we need to make restrictions that forbid
144: anyone to deny you these rights or to ask you to surrender the rights.
145: These restrictions translate to certain responsibilities for you if you
146: distribute copies of the software, or if you modify it.
147:
148: For example, if you distribute copies of such a program, whether
149: gratis or for a fee, you must give the recipients all the rights that
150: you have. You must make sure that they, too, receive or can get the
151: source code. And you must show them these terms so they know their
152: rights.
153:
154: We protect your rights with two steps: (1) copyright the software, and
155: (2) offer you this license which gives you legal permission to copy,
156: distribute and/or modify the software.
157:
158: Also, for each author's protection and ours, we want to make certain
159: that everyone understands that there is no warranty for this free
160: software. If the software is modified by someone else and passed on, we
161: want its recipients to know that what they have is not the original, so
162: that any problems introduced by others will not reflect on the original
163: authors' reputations.
164:
165: Finally, any free program is threatened constantly by software
166: patents. We wish to avoid the danger that redistributors of a free
167: program will individually obtain patent licenses, in effect making the
168: program proprietary. To prevent this, we have made it clear that any
169: patent must be licensed for everyone's free use or not licensed at all.
170:
171: The precise terms and conditions for copying, distribution and
172: modification follow.
173:
174: @iftex
175: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
176: @end iftex
177: @ifinfo
178: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
179: @end ifinfo
180:
181: @enumerate 0
182: @item
183: This License applies to any program or other work which contains
184: a notice placed by the copyright holder saying it may be distributed
185: under the terms of this General Public License. The ``Program'', below,
186: refers to any such program or work, and a ``work based on the Program''
187: means either the Program or any derivative work under copyright law:
188: that is to say, a work containing the Program or a portion of it,
189: either verbatim or with modifications and/or translated into another
190: language. (Hereinafter, translation is included without limitation in
191: the term ``modification''.) Each licensee is addressed as ``you''.
192:
193: Activities other than copying, distribution and modification are not
194: covered by this License; they are outside its scope. The act of
195: running the Program is not restricted, and the output from the Program
196: is covered only if its contents constitute a work based on the
197: Program (independent of having been made by running the Program).
198: Whether that is true depends on what the Program does.
199:
200: @item
201: You may copy and distribute verbatim copies of the Program's
202: source code as you receive it, in any medium, provided that you
203: conspicuously and appropriately publish on each copy an appropriate
204: copyright notice and disclaimer of warranty; keep intact all the
205: notices that refer to this License and to the absence of any warranty;
206: and give any other recipients of the Program a copy of this License
207: along with the Program.
208:
209: You may charge a fee for the physical act of transferring a copy, and
210: you may at your option offer warranty protection in exchange for a fee.
211:
212: @item
213: You may modify your copy or copies of the Program or any portion
214: of it, thus forming a work based on the Program, and copy and
215: distribute such modifications or work under the terms of Section 1
216: above, provided that you also meet all of these conditions:
217:
218: @enumerate a
219: @item
220: You must cause the modified files to carry prominent notices
221: stating that you changed the files and the date of any change.
222:
223: @item
224: You must cause any work that you distribute or publish, that in
225: whole or in part contains or is derived from the Program or any
226: part thereof, to be licensed as a whole at no charge to all third
227: parties under the terms of this License.
228:
229: @item
230: If the modified program normally reads commands interactively
231: when run, you must cause it, when started running for such
232: interactive use in the most ordinary way, to print or display an
233: announcement including an appropriate copyright notice and a
234: notice that there is no warranty (or else, saying that you provide
235: a warranty) and that users may redistribute the program under
236: these conditions, and telling the user how to view a copy of this
237: License. (Exception: if the Program itself is interactive but
238: does not normally print such an announcement, your work based on
239: the Program is not required to print an announcement.)
240: @end enumerate
241:
242: These requirements apply to the modified work as a whole. If
243: identifiable sections of that work are not derived from the Program,
244: and can be reasonably considered independent and separate works in
245: themselves, then this License, and its terms, do not apply to those
246: sections when you distribute them as separate works. But when you
247: distribute the same sections as part of a whole which is a work based
248: on the Program, the distribution of the whole must be on the terms of
249: this License, whose permissions for other licensees extend to the
250: entire whole, and thus to each and every part regardless of who wrote it.
251:
252: Thus, it is not the intent of this section to claim rights or contest
253: your rights to work written entirely by you; rather, the intent is to
254: exercise the right to control the distribution of derivative or
255: collective works based on the Program.
256:
257: In addition, mere aggregation of another work not based on the Program
258: with the Program (or with a work based on the Program) on a volume of
259: a storage or distribution medium does not bring the other work under
260: the scope of this License.
261:
262: @item
263: You may copy and distribute the Program (or a work based on it,
264: under Section 2) in object code or executable form under the terms of
265: Sections 1 and 2 above provided that you also do one of the following:
266:
267: @enumerate a
268: @item
269: Accompany it with the complete corresponding machine-readable
270: source code, which must be distributed under the terms of Sections
271: 1 and 2 above on a medium customarily used for software interchange; or,
272:
273: @item
274: Accompany it with a written offer, valid for at least three
275: years, to give any third party, for a charge no more than your
276: cost of physically performing source distribution, a complete
277: machine-readable copy of the corresponding source code, to be
278: distributed under the terms of Sections 1 and 2 above on a medium
279: customarily used for software interchange; or,
280:
281: @item
282: Accompany it with the information you received as to the offer
283: to distribute corresponding source code. (This alternative is
284: allowed only for noncommercial distribution and only if you
285: received the program in object code or executable form with such
286: an offer, in accord with Subsection b above.)
287: @end enumerate
288:
289: The source code for a work means the preferred form of the work for
290: making modifications to it. For an executable work, complete source
291: code means all the source code for all modules it contains, plus any
292: associated interface definition files, plus the scripts used to
293: control compilation and installation of the executable. However, as a
294: special exception, the source code distributed need not include
295: anything that is normally distributed (in either source or binary
296: form) with the major components (compiler, kernel, and so on) of the
297: operating system on which the executable runs, unless that component
298: itself accompanies the executable.
299:
300: If distribution of executable or object code is made by offering
301: access to copy from a designated place, then offering equivalent
302: access to copy the source code from the same place counts as
303: distribution of the source code, even though third parties are not
304: compelled to copy the source along with the object code.
305:
306: @item
307: You may not copy, modify, sublicense, or distribute the Program
308: except as expressly provided under this License. Any attempt
309: otherwise to copy, modify, sublicense or distribute the Program is
310: void, and will automatically terminate your rights under this License.
311: However, parties who have received copies, or rights, from you under
312: this License will not have their licenses terminated so long as such
313: parties remain in full compliance.
314:
315: @item
316: You are not required to accept this License, since you have not
317: signed it. However, nothing else grants you permission to modify or
318: distribute the Program or its derivative works. These actions are
319: prohibited by law if you do not accept this License. Therefore, by
320: modifying or distributing the Program (or any work based on the
321: Program), you indicate your acceptance of this License to do so, and
322: all its terms and conditions for copying, distributing or modifying
323: the Program or works based on it.
324:
325: @item
326: Each time you redistribute the Program (or any work based on the
327: Program), the recipient automatically receives a license from the
328: original licensor to copy, distribute or modify the Program subject to
329: these terms and conditions. You may not impose any further
330: restrictions on the recipients' exercise of the rights granted herein.
331: You are not responsible for enforcing compliance by third parties to
332: this License.
333:
334: @item
335: If, as a consequence of a court judgment or allegation of patent
336: infringement or for any other reason (not limited to patent issues),
337: conditions are imposed on you (whether by court order, agreement or
338: otherwise) that contradict the conditions of this License, they do not
339: excuse you from the conditions of this License. If you cannot
340: distribute so as to satisfy simultaneously your obligations under this
341: License and any other pertinent obligations, then as a consequence you
342: may not distribute the Program at all. For example, if a patent
343: license would not permit royalty-free redistribution of the Program by
344: all those who receive copies directly or indirectly through you, then
345: the only way you could satisfy both it and this License would be to
346: refrain entirely from distribution of the Program.
347:
348: If any portion of this section is held invalid or unenforceable under
349: any particular circumstance, the balance of the section is intended to
350: apply and the section as a whole is intended to apply in other
351: circumstances.
352:
353: It is not the purpose of this section to induce you to infringe any
354: patents or other property right claims or to contest validity of any
355: such claims; this section has the sole purpose of protecting the
356: integrity of the free software distribution system, which is
357: implemented by public license practices. Many people have made
358: generous contributions to the wide range of software distributed
359: through that system in reliance on consistent application of that
360: system; it is up to the author/donor to decide if he or she is willing
361: to distribute software through any other system and a licensee cannot
362: impose that choice.
363:
364: This section is intended to make thoroughly clear what is believed to
365: be a consequence of the rest of this License.
366:
367: @item
368: If the distribution and/or use of the Program is restricted in
369: certain countries either by patents or by copyrighted interfaces, the
370: original copyright holder who places the Program under this License
371: may add an explicit geographical distribution limitation excluding
372: those countries, so that distribution is permitted only in or among
373: countries not thus excluded. In such case, this License incorporates
374: the limitation as if written in the body of this License.
375:
376: @item
377: The Free Software Foundation may publish revised and/or new versions
378: of the General Public License from time to time. Such new versions will
379: be similar in spirit to the present version, but may differ in detail to
380: address new problems or concerns.
381:
382: Each version is given a distinguishing version number. If the Program
383: specifies a version number of this License which applies to it and ``any
384: later version'', you have the option of following the terms and conditions
385: either of that version or of any later version published by the Free
386: Software Foundation. If the Program does not specify a version number of
387: this License, you may choose any version ever published by the Free Software
388: Foundation.
389:
390: @item
391: If you wish to incorporate parts of the Program into other free
392: programs whose distribution conditions are different, write to the author
393: to ask for permission. For software which is copyrighted by the Free
394: Software Foundation, write to the Free Software Foundation; we sometimes
395: make exceptions for this. Our decision will be guided by the two goals
396: of preserving the free status of all derivatives of our free software and
397: of promoting the sharing and reuse of software generally.
398:
399: @iftex
400: @heading NO WARRANTY
401: @end iftex
402: @ifinfo
403: @center NO WARRANTY
404: @end ifinfo
405:
406: @item
407: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
408: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
409: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
410: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
411: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
412: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
413: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
414: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
415: REPAIR OR CORRECTION.
416:
417: @item
418: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
419: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
420: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
421: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
422: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
423: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
424: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
425: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
426: POSSIBILITY OF SUCH DAMAGES.
427: @end enumerate
428:
429: @iftex
430: @heading END OF TERMS AND CONDITIONS
431: @end iftex
432: @ifinfo
433: @center END OF TERMS AND CONDITIONS
434: @end ifinfo
435:
436: @page
437: @unnumberedsec How to Apply These Terms to Your New Programs
438:
439: If you develop a new program, and you want it to be of the greatest
440: possible use to the public, the best way to achieve this is to make it
441: free software which everyone can redistribute and change under these terms.
442:
443: To do so, attach the following notices to the program. It is safest
444: to attach them to the start of each source file to most effectively
445: convey the exclusion of warranty; and each file should have at least
446: the ``copyright'' line and a pointer to where the full notice is found.
447:
448: @smallexample
449: @var{one line to give the program's name and a brief idea of what it does.}
450: Copyright (C) 19@var{yy} @var{name of author}
451:
452: This program is free software; you can redistribute it and/or modify
453: it under the terms of the GNU General Public License as published by
454: the Free Software Foundation; either version 2 of the License, or
455: (at your option) any later version.
456:
457: This program is distributed in the hope that it will be useful,
458: but WITHOUT ANY WARRANTY; without even the implied warranty of
459: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
460: GNU General Public License for more details.
461:
462: You should have received a copy of the GNU General Public License
463: along with this program; if not, write to the Free Software
464: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
465: @end smallexample
466:
467: Also add information on how to contact you by electronic and paper mail.
468:
469: If the program is interactive, make it output a short notice like this
470: when it starts in an interactive mode:
471:
472: @smallexample
473: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
474: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
475: type `show w'.
476: This is free software, and you are welcome to redistribute it
477: under certain conditions; type `show c' for details.
478: @end smallexample
479:
480: The hypothetical commands @samp{show w} and @samp{show c} should show
481: the appropriate parts of the General Public License. Of course, the
482: commands you use may be called something other than @samp{show w} and
483: @samp{show c}; they could even be mouse-clicks or menu items---whatever
484: suits your program.
485:
486: You should also get your employer (if you work as a programmer) or your
487: school, if any, to sign a ``copyright disclaimer'' for the program, if
488: necessary. Here is a sample; alter the names:
489:
490: @smallexample
491: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
492: `Gnomovision' (which makes passes at compilers) written by James Hacker.
493:
494: @var{signature of Ty Coon}, 1 April 1989
495: Ty Coon, President of Vice
496: @end smallexample
497:
498: This General Public License does not permit incorporating your program into
499: proprietary programs. If your program is a subroutine library, you may
500: consider it more useful to permit linking proprietary applications with the
501: library. If this is what you want to do, use the GNU Library General
502: Public License instead of this License.
503:
504: @iftex
505: @unnumbered Preface
506: @cindex Preface
507: This manual documents Gforth. The reader is expected to know
508: Forth. This manual is primarily a reference manual. @xref{Other Books}
509: for introductory material.
510: @end iftex
511:
512: @node Goals, Other Books, License, Top
513: @comment node-name, next, previous, up
514: @chapter Goals of Gforth
515: @cindex Goals
516: The goal of the Gforth Project is to develop a standard model for
517: ANS Forth. This can be split into several subgoals:
518:
519: @itemize @bullet
520: @item
521: Gforth should conform to the Forth standard (ANS Forth).
522: @item
523: It should be a model, i.e. it should define all the
524: implementation-dependent things.
525: @item
526: It should become standard, i.e. widely accepted and used. This goal
527: is the most difficult one.
528: @end itemize
529:
530: To achieve these goals Gforth should be
531: @itemize @bullet
532: @item
533: Similar to previous models (fig-Forth, F83)
534: @item
535: Powerful. It should provide for all the things that are considered
536: necessary today and even some that are not yet considered necessary.
537: @item
538: Efficient. It should not get the reputation of being exceptionally
539: slow.
540: @item
541: Free.
542: @item
543: Available on many machines/easy to port.
544: @end itemize
545:
546: Have we achieved these goals? Gforth conforms to the ANS Forth
547: standard. It may be considered a model, but we have not yet documented
548: which parts of the model are stable and which parts we are likely to
549: change. It certainly has not yet become a de facto standard. It has some
550: similarities and some differences to previous models. It has some
551: powerful features, but not yet everything that we envisioned. We
552: certainly have achieved our execution speed goals (@pxref{Performance}).
553: It is free and available on many machines.
554:
555: @node Other Books, Invoking Gforth, Goals, Top
556: @chapter Other books on ANS Forth
557: @cindex books on Forth
558:
559: As the standard is relatively new, there are not many books out yet. It
560: is not recommended to learn Forth by using Gforth and a book that is
561: not written for ANS Forth, as you will not know your mistakes from the
562: deviations of the book.
563:
564: @cindex standard document for ANS Forth
565: @cindex ANS Forth document
566: There is, of course, the standard, the definite reference if you want to
567: write ANS Forth programs. It is available in printed form from the
568: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
569: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
570: can also get it from Global Engineering Documents (Tel.: USA (800)
571: 854-7179; Fax.: (303) 843-9880) for about $300.
572:
573: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
574: for publication is available electronically and for free in some MS Word
575: format, and it has been converted to HTML. Some pointers to these
576: versions can be found through
577: @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
578:
579: @cindex introductory book
580: @cindex book, introductory
581: @cindex Woehr, Jack: @cite{Forth: The New Model}
582: @cindex @cite{Forth: The new model} (book)
583: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
584: introductory book based on a draft version of the standard. It does not
585: cover the whole standard. It also contains interesting background
586: information (Jack Woehr was in the ANS Forth Technical Committee). It is
587: not appropriate for complete newbies, but programmers experienced in
588: other languages should find it ok.
589:
590: @node Invoking Gforth, Words, Other Books, Top
591: @chapter Invoking Gforth
592: @cindex invoking Gforth
593: @cindex running Gforth
594: @cindex command-line options
595: @cindex options on the command line
596: @cindex flags on the command line
597:
598: You will usually just say @code{gforth}. In many other cases the default
599: Gforth image will be invoked like this:
600:
601: @example
602: gforth [files] [-e forth-code]
603: @end example
604:
605: executing the contents of the files and the Forth code in the order they
606: are given.
607:
608: In general, the command line looks like this:
609:
610: @example
611: gforth [initialization options] [image-specific options]
612: @end example
613:
614: The initialization options must come before the rest of the command
615: line. They are:
616:
617: @table @code
618: @cindex -i, command-line option
619: @cindex --image-file, command-line option
620: @item --image-file @var{file}
621: @itemx -i @var{file}
622: Loads the Forth image @var{file} instead of the default
623: @file{gforth.fi} (@pxref{Image Files}).
624:
625: @cindex --path, command-line option
626: @cindex -p, command-line option
627: @item --path @var{path}
628: @itemx -p @var{path}
629: Uses @var{path} for searching the image file and Forth source code files
630: instead of the default in the environment variable @code{GFORTHPATH} or
631: the path specified at installation time (e.g.,
632: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
633: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
634:
635: @cindex --dictionary-size, command-line option
636: @cindex -m, command-line option
637: @cindex @var{size} parameters for command-line options
638: @cindex size of the dictionary and the stacks
639: @item --dictionary-size @var{size}
640: @itemx -m @var{size}
641: Allocate @var{size} space for the Forth dictionary space instead of
642: using the default specified in the image (typically 256K). The
643: @var{size} specification consists of an integer and a unit (e.g.,
644: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
645: size, in this case Cells), @code{k} (kilobytes), and @code{M}
646: (Megabytes). If no unit is specified, @code{e} is used.
647:
648: @cindex --data-stack-size, command-line option
649: @cindex -d, command-line option
650: @item --data-stack-size @var{size}
651: @itemx -d @var{size}
652: Allocate @var{size} space for the data stack instead of using the
653: default specified in the image (typically 16K).
654:
655: @cindex --return-stack-size, command-line option
656: @cindex -r, command-line option
657: @item --return-stack-size @var{size}
658: @itemx -r @var{size}
659: Allocate @var{size} space for the return stack instead of using the
660: default specified in the image (typically 15K).
661:
662: @cindex --fp-stack-size, command-line option
663: @cindex -f, command-line option
664: @item --fp-stack-size @var{size}
665: @itemx -f @var{size}
666: Allocate @var{size} space for the floating point stack instead of
667: using the default specified in the image (typically 15.5K). In this case
668: the unit specifier @code{e} refers to floating point numbers.
669:
670: @cindex --locals-stack-size, command-line option
671: @cindex -l, command-line option
672: @item --locals-stack-size @var{size}
673: @itemx -l @var{size}
674: Allocate @var{size} space for the locals stack instead of using the
675: default specified in the image (typically 14.5K).
676:
677: @cindex -h, command-line option
678: @cindex --help, command-line option
679: @item --help
680: @itemx -h
681: Print a message about the command-line options
682:
683: @cindex -v, command-line option
684: @cindex --version, command-line option
685: @item --version
686: @itemx -v
687: Print version and exit
688:
689: @cindex --debug, command-line option
690: @item --debug
691: Print some information useful for debugging on startup.
692:
693: @cindex --offset-image, command-line option
694: @item --offset-image
695: Start the dictionary at a slightly different position than would be used
696: otherwise (useful for creating data-relocatable images,
697: @pxref{Data-Relocatable Image Files}).
698:
699: @cindex --no-offset-im, command-line option
700: @item --no-offset-im
701: Start the dictionary at the normal position.
702:
703: @cindex --clear-dictionary, command-line option
704: @item --clear-dictionary
705: Initialize all bytes in the dictionary to 0 before loading the image
706: (@pxref{Data-Relocatable Image Files}).
707:
708: @cindex --die-on-signal, command-line-option
709: @item --die-on-signal
710: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
711: or the segmentation violation SIGSEGV) by translating it into a Forth
712: @code{THROW}. With this option, Gforth exits if it receives such a
713: signal. This option is useful when the engine and/or the image might be
714: severely broken (such that it causes another signal before recovering
715: from the first); this option avoids endless loops in such cases.
716: @end table
717:
718: @cindex loading files at startup
719: @cindex executing code on startup
720: @cindex batch processing with Gforth
721: As explained above, the image-specific command-line arguments for the
722: default image @file{gforth.fi} consist of a sequence of filenames and
723: @code{-e @var{forth-code}} options that are interpreted in the sequence
724: in which they are given. The @code{-e @var{forth-code}} or
725: @code{--evaluate @var{forth-code}} option evaluates the forth
726: code. This option takes only one argument; if you want to evaluate more
727: Forth words, you have to quote them or use several @code{-e}s. To exit
728: after processing the command line (instead of entering interactive mode)
729: append @code{-e bye} to the command line.
730:
731: @cindex versions, invoking other versions of Gforth
732: If you have several versions of Gforth installed, @code{gforth} will
733: invoke the version that was installed last. @code{gforth-@var{version}}
734: invokes a specific version. You may want to use the option
735: @code{--path}, if your environment contains the variable
736: @code{GFORTHPATH}.
737:
738: Not yet implemented:
739: On startup the system first executes the system initialization file
740: (unless the option @code{--no-init-file} is given; note that the system
741: resulting from using this option may not be ANS Forth conformant). Then
742: the user initialization file @file{.gforth.fs} is executed, unless the
743: option @code{--no-rc} is given; this file is first searched in @file{.},
744: then in @file{~}, then in the normal path (see above).
745:
746: @node Words, Tools, Invoking Gforth, Top
747: @chapter Forth Words
748: @cindex Words
749:
750: @menu
751: * Notation::
752: * Arithmetic::
753: * Stack Manipulation::
754: * Memory::
755: * Control Structures::
756: * Locals::
757: * Defining Words::
758: * Structures::
759: * Objects::
760: * Object Oriented Forth::
761: * Tokens for Words::
762: * Wordlists::
763: * Files::
764: * Blocks::
765: * Other I/O::
766: * Programming Tools::
767: * Assembler and Code Words::
768: * Threading Words::
769: * Including Files::
770: @end menu
771:
772: @node Notation, Arithmetic, Words, Words
773: @section Notation
774: @cindex notation of glossary entries
775: @cindex format of glossary entries
776: @cindex glossary notation format
777: @cindex word glossary entry format
778:
779: The Forth words are described in this section in the glossary notation
780: that has become a de-facto standard for Forth texts, i.e.,
781:
782: @format
783: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
784: @end format
785: @var{Description}
786:
787: @table @var
788: @item word
789: @cindex case insensitivity
790: The name of the word. BTW, Gforth is case insensitive, so you can
791: type the words in in lower case (However, @pxref{core-idef}).
792:
793: @item Stack effect
794: @cindex stack effect
795: The stack effect is written in the notation @code{@var{before} --
796: @var{after}}, where @var{before} and @var{after} describe the top of
797: stack entries before and after the execution of the word. The rest of
798: the stack is not touched by the word. The top of stack is rightmost,
799: i.e., a stack sequence is written as it is typed in. Note that Gforth
800: uses a separate floating point stack, but a unified stack
801: notation. Also, return stack effects are not shown in @var{stack
802: effect}, but in @var{Description}. The name of a stack item describes
803: the type and/or the function of the item. See below for a discussion of
804: the types.
805:
806: All words have two stack effects: A compile-time stack effect and a
807: run-time stack effect. The compile-time stack-effect of most words is
808: @var{ -- }. If the compile-time stack-effect of a word deviates from
809: this standard behaviour, or the word does other unusual things at
810: compile time, both stack effects are shown; otherwise only the run-time
811: stack effect is shown.
812:
813: @cindex pronounciation of words
814: @item pronunciation
815: How the word is pronounced.
816:
817: @cindex wordset
818: @item wordset
819: The ANS Forth standard is divided into several wordsets. A standard
820: system need not support all of them. So, the fewer wordsets your program
821: uses the more portable it will be in theory. However, we suspect that
822: most ANS Forth systems on personal machines will feature all
823: wordsets. Words that are not defined in the ANS standard have
824: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
825: describes words that will work in future releases of Gforth;
826: @code{gforth-internal} words are more volatile. Environmental query
827: strings are also displayed like words; you can recognize them by the
828: @code{environment} in the wordset field.
829:
830: @item Description
831: A description of the behaviour of the word.
832: @end table
833:
834: @cindex types of stack items
835: @cindex stack item types
836: The type of a stack item is specified by the character(s) the name
837: starts with:
838:
839: @table @code
840: @item f
841: @cindex @code{f}, stack item type
842: Boolean flags, i.e. @code{false} or @code{true}.
843: @item c
844: @cindex @code{c}, stack item type
845: Char
846: @item w
847: @cindex @code{w}, stack item type
848: Cell, can contain an integer or an address
849: @item n
850: @cindex @code{n}, stack item type
851: signed integer
852: @item u
853: @cindex @code{u}, stack item type
854: unsigned integer
855: @item d
856: @cindex @code{d}, stack item type
857: double sized signed integer
858: @item ud
859: @cindex @code{ud}, stack item type
860: double sized unsigned integer
861: @item r
862: @cindex @code{r}, stack item type
863: Float (on the FP stack)
864: @item a_
865: @cindex @code{a_}, stack item type
866: Cell-aligned address
867: @item c_
868: @cindex @code{c_}, stack item type
869: Char-aligned address (note that a Char may have two bytes in Windows NT)
870: @item f_
871: @cindex @code{f_}, stack item type
872: Float-aligned address
873: @item df_
874: @cindex @code{df_}, stack item type
875: Address aligned for IEEE double precision float
876: @item sf_
877: @cindex @code{sf_}, stack item type
878: Address aligned for IEEE single precision float
879: @item xt
880: @cindex @code{xt}, stack item type
881: Execution token, same size as Cell
882: @item wid
883: @cindex @code{wid}, stack item type
884: Wordlist ID, same size as Cell
885: @item f83name
886: @cindex @code{f83name}, stack item type
887: Pointer to a name structure
888: @item "
889: @cindex @code{"}, stack item type
890: string in the input stream (not the stack). The terminating character is
891: a blank by default. If it is not a blank, it is shown in @code{<>}
892: quotes.
893: @end table
894:
895: @node Arithmetic, Stack Manipulation, Notation, Words
896: @section Arithmetic
897: @cindex arithmetic words
898:
899: @cindex division with potentially negative operands
900: Forth arithmetic is not checked, i.e., you will not hear about integer
901: overflow on addition or multiplication, you may hear about division by
902: zero if you are lucky. The operator is written after the operands, but
903: the operands are still in the original order. I.e., the infix @code{2-1}
904: corresponds to @code{2 1 -}. Forth offers a variety of division
905: operators. If you perform division with potentially negative operands,
906: you do not want to use @code{/} or @code{/mod} with its undefined
907: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
908: former, @pxref{Mixed precision}).
909:
910: @menu
911: * Single precision::
912: * Bitwise operations::
913: * Mixed precision:: operations with single and double-cell integers
914: * Double precision:: Double-cell integer arithmetic
915: * Floating Point::
916: @end menu
917:
918: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
919: @subsection Single precision
920: @cindex single precision arithmetic words
921:
922: doc-+
923: doc--
924: doc-*
925: doc-/
926: doc-mod
927: doc-/mod
928: doc-negate
929: doc-abs
930: doc-min
931: doc-max
932:
933: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
934: @subsection Bitwise operations
935: @cindex bitwise operation words
936:
937: doc-and
938: doc-or
939: doc-xor
940: doc-invert
941: doc-2*
942: doc-2/
943:
944: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
945: @subsection Mixed precision
946: @cindex mixed precision arithmetic words
947:
948: doc-m+
949: doc-*/
950: doc-*/mod
951: doc-m*
952: doc-um*
953: doc-m*/
954: doc-um/mod
955: doc-fm/mod
956: doc-sm/rem
957:
958: @node Double precision, Floating Point, Mixed precision, Arithmetic
959: @subsection Double precision
960: @cindex double precision arithmetic words
961:
962: @cindex double-cell numbers, input format
963: @cindex input format for double-cell numbers
964: The outer (aka text) interpreter converts numbers containing a dot into
965: a double precision number. Note that only numbers with the dot as last
966: character are standard-conforming.
967:
968: doc-d+
969: doc-d-
970: doc-dnegate
971: doc-dabs
972: doc-dmin
973: doc-dmax
974:
975: @node Floating Point, , Double precision, Arithmetic
976: @subsection Floating Point
977: @cindex floating point arithmetic words
978:
979: @cindex floating-point numbers, input format
980: @cindex input format for floating-point numbers
981: The format of floating point numbers recognized by the outer (aka text)
982: interpreter is: a signed decimal number, possibly containing a decimal
983: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
984: by a signed integer (the exponent). E.g., @code{1e} is the same as
985: @code{+1.0e+0}. Note that a number without @code{e}
986: is not interpreted as floating-point number, but as double (if the
987: number contains a @code{.}) or single precision integer. Also,
988: conversions between string and floating point numbers always use base
989: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
990: value greater then 14, the @code{E} may be interpreted as digit and the
991: number will be interpreted as integer, unless it has a signed exponent
992: (both @code{+} and @code{-} are allowed as signs).
993:
994: @cindex angles in trigonometric operations
995: @cindex trigonometric operations
996: Angles in floating point operations are given in radians (a full circle
997: has 2 pi radians). Note, that Gforth has a separate floating point
998: stack, but we use the unified notation.
999:
1000: @cindex floating-point arithmetic, pitfalls
1001: Floating point numbers have a number of unpleasant surprises for the
1002: unwary (e.g., floating point addition is not associative) and even a few
1003: for the wary. You should not use them unless you know what you are doing
1004: or you don't care that the results you get are totally bogus. If you
1005: want to learn about the problems of floating point numbers (and how to
1006: avoid them), you might start with @cite{David Goldberg, What Every
1007: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
1008: Computing Surveys 23(1):5@minus{}48, March 1991}.
1009:
1010: doc-f+
1011: doc-f-
1012: doc-f*
1013: doc-f/
1014: doc-fnegate
1015: doc-fabs
1016: doc-fmax
1017: doc-fmin
1018: doc-floor
1019: doc-fround
1020: doc-f**
1021: doc-fsqrt
1022: doc-fexp
1023: doc-fexpm1
1024: doc-fln
1025: doc-flnp1
1026: doc-flog
1027: doc-falog
1028: doc-fsin
1029: doc-fcos
1030: doc-fsincos
1031: doc-ftan
1032: doc-fasin
1033: doc-facos
1034: doc-fatan
1035: doc-fatan2
1036: doc-fsinh
1037: doc-fcosh
1038: doc-ftanh
1039: doc-fasinh
1040: doc-facosh
1041: doc-fatanh
1042:
1043: @node Stack Manipulation, Memory, Arithmetic, Words
1044: @section Stack Manipulation
1045: @cindex stack manipulation words
1046:
1047: @cindex floating-point stack in the standard
1048: Gforth has a data stack (aka parameter stack) for characters, cells,
1049: addresses, and double cells, a floating point stack for floating point
1050: numbers, a return stack for storing the return addresses of colon
1051: definitions and other data, and a locals stack for storing local
1052: variables. Note that while every sane Forth has a separate floating
1053: point stack, this is not strictly required; an ANS Forth system could
1054: theoretically keep floating point numbers on the data stack. As an
1055: additional difficulty, you don't know how many cells a floating point
1056: number takes. It is reportedly possible to write words in a way that
1057: they work also for a unified stack model, but we do not recommend trying
1058: it. Instead, just say that your program has an environmental dependency
1059: on a separate FP stack.
1060:
1061: @cindex return stack and locals
1062: @cindex locals and return stack
1063: Also, a Forth system is allowed to keep the local variables on the
1064: return stack. This is reasonable, as local variables usually eliminate
1065: the need to use the return stack explicitly. So, if you want to produce
1066: a standard complying program and if you are using local variables in a
1067: word, forget about return stack manipulations in that word (see the
1068: standard document for the exact rules).
1069:
1070: @menu
1071: * Data stack::
1072: * Floating point stack::
1073: * Return stack::
1074: * Locals stack::
1075: * Stack pointer manipulation::
1076: @end menu
1077:
1078: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1079: @subsection Data stack
1080: @cindex data stack manipulation words
1081: @cindex stack manipulations words, data stack
1082:
1083: doc-drop
1084: doc-nip
1085: doc-dup
1086: doc-over
1087: doc-tuck
1088: doc-swap
1089: doc-rot
1090: doc--rot
1091: doc-?dup
1092: doc-pick
1093: doc-roll
1094: doc-2drop
1095: doc-2nip
1096: doc-2dup
1097: doc-2over
1098: doc-2tuck
1099: doc-2swap
1100: doc-2rot
1101:
1102: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1103: @subsection Floating point stack
1104: @cindex floating-point stack manipulation words
1105: @cindex stack manipulation words, floating-point stack
1106:
1107: doc-fdrop
1108: doc-fnip
1109: doc-fdup
1110: doc-fover
1111: doc-ftuck
1112: doc-fswap
1113: doc-frot
1114:
1115: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1116: @subsection Return stack
1117: @cindex return stack manipulation words
1118: @cindex stack manipulation words, return stack
1119:
1120: doc->r
1121: doc-r>
1122: doc-r@
1123: doc-rdrop
1124: doc-2>r
1125: doc-2r>
1126: doc-2r@
1127: doc-2rdrop
1128:
1129: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1130: @subsection Locals stack
1131:
1132: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1133: @subsection Stack pointer manipulation
1134: @cindex stack pointer manipulation words
1135:
1136: doc-sp@
1137: doc-sp!
1138: doc-fp@
1139: doc-fp!
1140: doc-rp@
1141: doc-rp!
1142: doc-lp@
1143: doc-lp!
1144:
1145: @node Memory, Control Structures, Stack Manipulation, Words
1146: @section Memory
1147: @cindex Memory words
1148:
1149: @menu
1150: * Memory Access::
1151: * Address arithmetic::
1152: * Memory Blocks::
1153: @end menu
1154:
1155: @node Memory Access, Address arithmetic, Memory, Memory
1156: @subsection Memory Access
1157: @cindex memory access words
1158:
1159: doc-@
1160: doc-!
1161: doc-+!
1162: doc-c@
1163: doc-c!
1164: doc-2@
1165: doc-2!
1166: doc-f@
1167: doc-f!
1168: doc-sf@
1169: doc-sf!
1170: doc-df@
1171: doc-df!
1172:
1173: @node Address arithmetic, Memory Blocks, Memory Access, Memory
1174: @subsection Address arithmetic
1175: @cindex address arithmetic words
1176:
1177: ANS Forth does not specify the sizes of the data types. Instead, it
1178: offers a number of words for computing sizes and doing address
1179: arithmetic. Basically, address arithmetic is performed in terms of
1180: address units (aus); on most systems the address unit is one byte. Note
1181: that a character may have more than one au, so @code{chars} is no noop
1182: (on systems where it is a noop, it compiles to nothing).
1183:
1184: @cindex alignment of addresses for types
1185: ANS Forth also defines words for aligning addresses for specific
1186: types. Many computers require that accesses to specific data types
1187: must only occur at specific addresses; e.g., that cells may only be
1188: accessed at addresses divisible by 4. Even if a machine allows unaligned
1189: accesses, it can usually perform aligned accesses faster.
1190:
1191: For the performance-conscious: alignment operations are usually only
1192: necessary during the definition of a data structure, not during the
1193: (more frequent) accesses to it.
1194:
1195: ANS Forth defines no words for character-aligning addresses. This is not
1196: an oversight, but reflects the fact that addresses that are not
1197: char-aligned have no use in the standard and therefore will not be
1198: created.
1199:
1200: @cindex @code{CREATE} and alignment
1201: The standard guarantees that addresses returned by @code{CREATE}d words
1202: are cell-aligned; in addition, Gforth guarantees that these addresses
1203: are aligned for all purposes.
1204:
1205: Note that the standard defines a word @code{char}, which has nothing to
1206: do with address arithmetic.
1207:
1208: doc-chars
1209: doc-char+
1210: doc-cells
1211: doc-cell+
1212: doc-cell
1213: doc-align
1214: doc-aligned
1215: doc-floats
1216: doc-float+
1217: doc-float
1218: doc-falign
1219: doc-faligned
1220: doc-sfloats
1221: doc-sfloat+
1222: doc-sfalign
1223: doc-sfaligned
1224: doc-dfloats
1225: doc-dfloat+
1226: doc-dfalign
1227: doc-dfaligned
1228: doc-maxalign
1229: doc-maxaligned
1230: doc-cfalign
1231: doc-cfaligned
1232: doc-address-unit-bits
1233:
1234: @node Memory Blocks, , Address arithmetic, Memory
1235: @subsection Memory Blocks
1236: @cindex memory block words
1237:
1238: doc-move
1239: doc-erase
1240:
1241: While the previous words work on address units, the rest works on
1242: characters.
1243:
1244: doc-cmove
1245: doc-cmove>
1246: doc-fill
1247: doc-blank
1248:
1249: @node Control Structures, Locals, Memory, Words
1250: @section Control Structures
1251: @cindex control structures
1252:
1253: Control structures in Forth cannot be used in interpret state, only in
1254: compile state@footnote{More precisely, they have no interpretation
1255: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
1256: a colon definition. We do not like this limitation, but have not seen a
1257: satisfying way around it yet, although many schemes have been proposed.
1258:
1259: @menu
1260: * Selection::
1261: * Simple Loops::
1262: * Counted Loops::
1263: * Arbitrary control structures::
1264: * Calls and returns::
1265: * Exception Handling::
1266: @end menu
1267:
1268: @node Selection, Simple Loops, Control Structures, Control Structures
1269: @subsection Selection
1270: @cindex selection control structures
1271: @cindex control structures for selection
1272:
1273: @cindex @code{IF} control structure
1274: @example
1275: @var{flag}
1276: IF
1277: @var{code}
1278: ENDIF
1279: @end example
1280: or
1281: @example
1282: @var{flag}
1283: IF
1284: @var{code1}
1285: ELSE
1286: @var{code2}
1287: ENDIF
1288: @end example
1289:
1290: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1291: standard, and @code{ENDIF} is not, although it is quite popular. We
1292: recommend using @code{ENDIF}, because it is less confusing for people
1293: who also know other languages (and is not prone to reinforcing negative
1294: prejudices against Forth in these people). Adding @code{ENDIF} to a
1295: system that only supplies @code{THEN} is simple:
1296: @example
1297: : endif POSTPONE then ; immediate
1298: @end example
1299:
1300: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1301: (adv.)} has the following meanings:
1302: @quotation
1303: ... 2b: following next after in order ... 3d: as a necessary consequence
1304: (if you were there, then you saw them).
1305: @end quotation
1306: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1307: and many other programming languages has the meaning 3d.]
1308:
1309: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1310: you can avoid using @code{?dup}. Using these alternatives is also more
1311: efficient than using @code{?dup}. Definitions in plain standard Forth
1312: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1313: @file{compat/control.fs}.
1314:
1315: @cindex @code{CASE} control structure
1316: @example
1317: @var{n}
1318: CASE
1319: @var{n1} OF @var{code1} ENDOF
1320: @var{n2} OF @var{code2} ENDOF
1321: @dots{}
1322: ENDCASE
1323: @end example
1324:
1325: Executes the first @var{codei}, where the @var{ni} is equal to
1326: @var{n}. A default case can be added by simply writing the code after
1327: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1328: but must not consume it.
1329:
1330: @node Simple Loops, Counted Loops, Selection, Control Structures
1331: @subsection Simple Loops
1332: @cindex simple loops
1333: @cindex loops without count
1334:
1335: @cindex @code{WHILE} loop
1336: @example
1337: BEGIN
1338: @var{code1}
1339: @var{flag}
1340: WHILE
1341: @var{code2}
1342: REPEAT
1343: @end example
1344:
1345: @var{code1} is executed and @var{flag} is computed. If it is true,
1346: @var{code2} is executed and the loop is restarted; If @var{flag} is
1347: false, execution continues after the @code{REPEAT}.
1348:
1349: @cindex @code{UNTIL} loop
1350: @example
1351: BEGIN
1352: @var{code}
1353: @var{flag}
1354: UNTIL
1355: @end example
1356:
1357: @var{code} is executed. The loop is restarted if @code{flag} is false.
1358:
1359: @cindex endless loop
1360: @cindex loops, endless
1361: @example
1362: BEGIN
1363: @var{code}
1364: AGAIN
1365: @end example
1366:
1367: This is an endless loop.
1368:
1369: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1370: @subsection Counted Loops
1371: @cindex counted loops
1372: @cindex loops, counted
1373: @cindex @code{DO} loops
1374:
1375: The basic counted loop is:
1376: @example
1377: @var{limit} @var{start}
1378: ?DO
1379: @var{body}
1380: LOOP
1381: @end example
1382:
1383: This performs one iteration for every integer, starting from @var{start}
1384: and up to, but excluding @var{limit}. The counter, aka index, can be
1385: accessed with @code{i}. E.g., the loop
1386: @example
1387: 10 0 ?DO
1388: i .
1389: LOOP
1390: @end example
1391: prints
1392: @example
1393: 0 1 2 3 4 5 6 7 8 9
1394: @end example
1395: The index of the innermost loop can be accessed with @code{i}, the index
1396: of the next loop with @code{j}, and the index of the third loop with
1397: @code{k}.
1398:
1399: doc-i
1400: doc-j
1401: doc-k
1402:
1403: The loop control data are kept on the return stack, so there are some
1404: restrictions on mixing return stack accesses and counted loop
1405: words. E.g., if you put values on the return stack outside the loop, you
1406: cannot read them inside the loop. If you put values on the return stack
1407: within a loop, you have to remove them before the end of the loop and
1408: before accessing the index of the loop.
1409:
1410: There are several variations on the counted loop:
1411:
1412: @code{LEAVE} leaves the innermost counted loop immediately.
1413:
1414: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1415: (and @code{LOOP} iterates until they become equal by wrap-around
1416: arithmetic). This behaviour is usually not what you want. Therefore,
1417: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1418: @code{?DO}), which do not enter the loop if @var{start} is greater than
1419: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1420: unsigned loop parameters.
1421:
1422: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1423: index by @var{n} instead of by 1. The loop is terminated when the border
1424: between @var{limit-1} and @var{limit} is crossed. E.g.:
1425:
1426: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1427:
1428: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1429:
1430: @cindex negative increment for counted loops
1431: @cindex counted loops with negative increment
1432: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1433:
1434: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1435:
1436: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1437:
1438: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1439: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1440: index by @var{u} each iteration. The loop is terminated when the border
1441: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1442: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1443:
1444: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1445:
1446: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1447:
1448: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1449:
1450: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1451: @code{-LOOP} are not in the ANS Forth standard. However, an
1452: implementation for these words that uses only standard words is provided
1453: in @file{compat/loops.fs}.
1454:
1455: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1456: the loop, independent of the loop parameters. Do not use @code{DO}, even
1457: if you know that the loop is entered in any case. Such knowledge tends
1458: to become invalid during maintenance of a program, and then the
1459: @code{DO} will make trouble.
1460:
1461: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1462: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1463: return stack so @code{EXIT} can get to its return address.
1464:
1465: @cindex @code{FOR} loops
1466: Another counted loop is
1467: @example
1468: @var{n}
1469: FOR
1470: @var{body}
1471: NEXT
1472: @end example
1473: This is the preferred loop of native code compiler writers who are too
1474: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1475: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1476: and ending with 0. Other Forth systems may behave differently, even if
1477: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1478: loops.
1479:
1480: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1481: @subsection Arbitrary control structures
1482: @cindex control structures, user-defined
1483:
1484: @cindex control-flow stack
1485: ANS Forth permits and supports using control structures in a non-nested
1486: way. Information about incomplete control structures is stored on the
1487: control-flow stack. This stack may be implemented on the Forth data
1488: stack, and this is what we have done in Gforth.
1489:
1490: @cindex @code{orig}, control-flow stack item
1491: @cindex @code{dest}, control-flow stack item
1492: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1493: entry represents a backward branch target. A few words are the basis for
1494: building any control structure possible (except control structures that
1495: need storage, like calls, coroutines, and backtracking).
1496:
1497: doc-if
1498: doc-ahead
1499: doc-then
1500: doc-begin
1501: doc-until
1502: doc-again
1503: doc-cs-pick
1504: doc-cs-roll
1505:
1506: On many systems control-flow stack items take one word, in Gforth they
1507: currently take three (this may change in the future). Therefore it is a
1508: really good idea to manipulate the control flow stack with
1509: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1510: words.
1511:
1512: Some standard control structure words are built from these words:
1513:
1514: doc-else
1515: doc-while
1516: doc-repeat
1517:
1518: Gforth adds some more control-structure words:
1519:
1520: doc-endif
1521: doc-?dup-if
1522: doc-?dup-0=-if
1523:
1524: Counted loop words constitute a separate group of words:
1525:
1526: doc-?do
1527: doc-+do
1528: doc-u+do
1529: doc--do
1530: doc-u-do
1531: doc-do
1532: doc-for
1533: doc-loop
1534: doc-+loop
1535: doc--loop
1536: doc-next
1537: doc-leave
1538: doc-?leave
1539: doc-unloop
1540: doc-done
1541:
1542: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1543: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1544: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1545: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1546: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1547: resolved (by using one of the loop-ending words or @code{DONE}).
1548:
1549: Another group of control structure words are
1550:
1551: doc-case
1552: doc-endcase
1553: doc-of
1554: doc-endof
1555:
1556: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1557: @code{cs-roll}.
1558:
1559: @subsubsection Programming Style
1560:
1561: In order to ensure readability we recommend that you do not create
1562: arbitrary control structures directly, but define new control structure
1563: words for the control structure you want and use these words in your
1564: program.
1565:
1566: E.g., instead of writing
1567:
1568: @example
1569: begin
1570: ...
1571: if [ 1 cs-roll ]
1572: ...
1573: again then
1574: @end example
1575:
1576: we recommend defining control structure words, e.g.,
1577:
1578: @example
1579: : while ( dest -- orig dest )
1580: POSTPONE if
1581: 1 cs-roll ; immediate
1582:
1583: : repeat ( orig dest -- )
1584: POSTPONE again
1585: POSTPONE then ; immediate
1586: @end example
1587:
1588: and then using these to create the control structure:
1589:
1590: @example
1591: begin
1592: ...
1593: while
1594: ...
1595: repeat
1596: @end example
1597:
1598: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1599: @code{WHILE} are predefined, so in this example it would not be
1600: necessary to define them.
1601:
1602: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1603: @subsection Calls and returns
1604: @cindex calling a definition
1605: @cindex returning from a definition
1606:
1607: @cindex recursive definitions
1608: A definition can be called simply be writing the name of the definition
1609: to be called. Note that normally a definition is invisible during its
1610: definition. If you want to write a directly recursive definition, you
1611: can use @code{recursive} to make the current definition visible.
1612:
1613: doc-recursive
1614:
1615: Another way to perform a recursive call is
1616:
1617: doc-recurse
1618:
1619: @c @progstyle{
1620: I prefer using @code{recursive} to @code{recurse}, because
1621: calling the definition by name is more descriptive (if the name is
1622: well-chosen) than the somewhat cryptic @code{recurse}. E.g., in a
1623: quicksort implementation, it is much better to read (and think) ``now
1624: sort the partitions'' than to read ``now do a recursive call''.
1625:
1626: For mutual recursion, use @code{defer}red words, like this:
1627:
1628: @example
1629: defer foo
1630:
1631: : bar ( ... -- ... )
1632: ... foo ... ;
1633:
1634: :noname ( ... -- ... )
1635: ... bar ... ;
1636: IS foo
1637: @end example
1638:
1639: When the end of the definition is reached, it returns. An earlier return
1640: can be forced using
1641:
1642: doc-exit
1643:
1644: Don't forget to clean up the return stack and @code{UNLOOP} any
1645: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1646: primitive compiled by @code{EXIT} is
1647:
1648: doc-;s
1649:
1650: @node Exception Handling, , Calls and returns, Control Structures
1651: @subsection Exception Handling
1652: @cindex Exceptions
1653:
1654: doc-catch
1655: doc-throw
1656:
1657: @node Locals, Defining Words, Control Structures, Words
1658: @section Locals
1659: @cindex locals
1660:
1661: Local variables can make Forth programming more enjoyable and Forth
1662: programs easier to read. Unfortunately, the locals of ANS Forth are
1663: laden with restrictions. Therefore, we provide not only the ANS Forth
1664: locals wordset, but also our own, more powerful locals wordset (we
1665: implemented the ANS Forth locals wordset through our locals wordset).
1666:
1667: The ideas in this section have also been published in the paper
1668: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1669: at EuroForth '94; it is available at
1670: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1671:
1672: @menu
1673: * Gforth locals::
1674: * ANS Forth locals::
1675: @end menu
1676:
1677: @node Gforth locals, ANS Forth locals, Locals, Locals
1678: @subsection Gforth locals
1679: @cindex Gforth locals
1680: @cindex locals, Gforth style
1681:
1682: Locals can be defined with
1683:
1684: @example
1685: @{ local1 local2 ... -- comment @}
1686: @end example
1687: or
1688: @example
1689: @{ local1 local2 ... @}
1690: @end example
1691:
1692: E.g.,
1693: @example
1694: : max @{ n1 n2 -- n3 @}
1695: n1 n2 > if
1696: n1
1697: else
1698: n2
1699: endif ;
1700: @end example
1701:
1702: The similarity of locals definitions with stack comments is intended. A
1703: locals definition often replaces the stack comment of a word. The order
1704: of the locals corresponds to the order in a stack comment and everything
1705: after the @code{--} is really a comment.
1706:
1707: This similarity has one disadvantage: It is too easy to confuse locals
1708: declarations with stack comments, causing bugs and making them hard to
1709: find. However, this problem can be avoided by appropriate coding
1710: conventions: Do not use both notations in the same program. If you do,
1711: they should be distinguished using additional means, e.g. by position.
1712:
1713: @cindex types of locals
1714: @cindex locals types
1715: The name of the local may be preceded by a type specifier, e.g.,
1716: @code{F:} for a floating point value:
1717:
1718: @example
1719: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1720: \ complex multiplication
1721: Ar Br f* Ai Bi f* f-
1722: Ar Bi f* Ai Br f* f+ ;
1723: @end example
1724:
1725: @cindex flavours of locals
1726: @cindex locals flavours
1727: @cindex value-flavoured locals
1728: @cindex variable-flavoured locals
1729: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1730: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1731: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1732: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1733: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1734: produces its address (which becomes invalid when the variable's scope is
1735: left). E.g., the standard word @code{emit} can be defined in terms of
1736: @code{type} like this:
1737:
1738: @example
1739: : emit @{ C^ char* -- @}
1740: char* 1 type ;
1741: @end example
1742:
1743: @cindex default type of locals
1744: @cindex locals, default type
1745: A local without type specifier is a @code{W:} local. Both flavours of
1746: locals are initialized with values from the data or FP stack.
1747:
1748: Currently there is no way to define locals with user-defined data
1749: structures, but we are working on it.
1750:
1751: Gforth allows defining locals everywhere in a colon definition. This
1752: poses the following questions:
1753:
1754: @menu
1755: * Where are locals visible by name?::
1756: * How long do locals live?::
1757: * Programming Style::
1758: * Implementation::
1759: @end menu
1760:
1761: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1762: @subsubsection Where are locals visible by name?
1763: @cindex locals visibility
1764: @cindex visibility of locals
1765: @cindex scope of locals
1766:
1767: Basically, the answer is that locals are visible where you would expect
1768: it in block-structured languages, and sometimes a little longer. If you
1769: want to restrict the scope of a local, enclose its definition in
1770: @code{SCOPE}...@code{ENDSCOPE}.
1771:
1772: doc-scope
1773: doc-endscope
1774:
1775: These words behave like control structure words, so you can use them
1776: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1777: arbitrary ways.
1778:
1779: If you want a more exact answer to the visibility question, here's the
1780: basic principle: A local is visible in all places that can only be
1781: reached through the definition of the local@footnote{In compiler
1782: construction terminology, all places dominated by the definition of the
1783: local.}. In other words, it is not visible in places that can be reached
1784: without going through the definition of the local. E.g., locals defined
1785: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1786: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1787: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1788:
1789: The reasoning behind this solution is: We want to have the locals
1790: visible as long as it is meaningful. The user can always make the
1791: visibility shorter by using explicit scoping. In a place that can
1792: only be reached through the definition of a local, the meaning of a
1793: local name is clear. In other places it is not: How is the local
1794: initialized at the control flow path that does not contain the
1795: definition? Which local is meant, if the same name is defined twice in
1796: two independent control flow paths?
1797:
1798: This should be enough detail for nearly all users, so you can skip the
1799: rest of this section. If you really must know all the gory details and
1800: options, read on.
1801:
1802: In order to implement this rule, the compiler has to know which places
1803: are unreachable. It knows this automatically after @code{AHEAD},
1804: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1805: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1806: compiler that the control flow never reaches that place. If
1807: @code{UNREACHABLE} is not used where it could, the only consequence is
1808: that the visibility of some locals is more limited than the rule above
1809: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1810: lie to the compiler), buggy code will be produced.
1811:
1812: doc-unreachable
1813:
1814: Another problem with this rule is that at @code{BEGIN}, the compiler
1815: does not know which locals will be visible on the incoming
1816: back-edge. All problems discussed in the following are due to this
1817: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1818: loops as examples; the discussion also applies to @code{?DO} and other
1819: loops). Perhaps the most insidious example is:
1820: @example
1821: AHEAD
1822: BEGIN
1823: x
1824: [ 1 CS-ROLL ] THEN
1825: @{ x @}
1826: ...
1827: UNTIL
1828: @end example
1829:
1830: This should be legal according to the visibility rule. The use of
1831: @code{x} can only be reached through the definition; but that appears
1832: textually below the use.
1833:
1834: From this example it is clear that the visibility rules cannot be fully
1835: implemented without major headaches. Our implementation treats common
1836: cases as advertised and the exceptions are treated in a safe way: The
1837: compiler makes a reasonable guess about the locals visible after a
1838: @code{BEGIN}; if it is too pessimistic, the
1839: user will get a spurious error about the local not being defined; if the
1840: compiler is too optimistic, it will notice this later and issue a
1841: warning. In the case above the compiler would complain about @code{x}
1842: being undefined at its use. You can see from the obscure examples in
1843: this section that it takes quite unusual control structures to get the
1844: compiler into trouble, and even then it will often do fine.
1845:
1846: If the @code{BEGIN} is reachable from above, the most optimistic guess
1847: is that all locals visible before the @code{BEGIN} will also be
1848: visible after the @code{BEGIN}. This guess is valid for all loops that
1849: are entered only through the @code{BEGIN}, in particular, for normal
1850: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1851: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1852: compiler. When the branch to the @code{BEGIN} is finally generated by
1853: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1854: warns the user if it was too optimistic:
1855: @example
1856: IF
1857: @{ x @}
1858: BEGIN
1859: \ x ?
1860: [ 1 cs-roll ] THEN
1861: ...
1862: UNTIL
1863: @end example
1864:
1865: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1866: optimistically assumes that it lives until the @code{THEN}. It notices
1867: this difference when it compiles the @code{UNTIL} and issues a
1868: warning. The user can avoid the warning, and make sure that @code{x}
1869: is not used in the wrong area by using explicit scoping:
1870: @example
1871: IF
1872: SCOPE
1873: @{ x @}
1874: ENDSCOPE
1875: BEGIN
1876: [ 1 cs-roll ] THEN
1877: ...
1878: UNTIL
1879: @end example
1880:
1881: Since the guess is optimistic, there will be no spurious error messages
1882: about undefined locals.
1883:
1884: If the @code{BEGIN} is not reachable from above (e.g., after
1885: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1886: optimistic guess, as the locals visible after the @code{BEGIN} may be
1887: defined later. Therefore, the compiler assumes that no locals are
1888: visible after the @code{BEGIN}. However, the user can use
1889: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1890: visible at the BEGIN as at the point where the top control-flow stack
1891: item was created.
1892:
1893: doc-assume-live
1894:
1895: E.g.,
1896: @example
1897: @{ x @}
1898: AHEAD
1899: ASSUME-LIVE
1900: BEGIN
1901: x
1902: [ 1 CS-ROLL ] THEN
1903: ...
1904: UNTIL
1905: @end example
1906:
1907: Other cases where the locals are defined before the @code{BEGIN} can be
1908: handled by inserting an appropriate @code{CS-ROLL} before the
1909: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1910: behind the @code{ASSUME-LIVE}).
1911:
1912: Cases where locals are defined after the @code{BEGIN} (but should be
1913: visible immediately after the @code{BEGIN}) can only be handled by
1914: rearranging the loop. E.g., the ``most insidious'' example above can be
1915: arranged into:
1916: @example
1917: BEGIN
1918: @{ x @}
1919: ... 0=
1920: WHILE
1921: x
1922: REPEAT
1923: @end example
1924:
1925: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1926: @subsubsection How long do locals live?
1927: @cindex locals lifetime
1928: @cindex lifetime of locals
1929:
1930: The right answer for the lifetime question would be: A local lives at
1931: least as long as it can be accessed. For a value-flavoured local this
1932: means: until the end of its visibility. However, a variable-flavoured
1933: local could be accessed through its address far beyond its visibility
1934: scope. Ultimately, this would mean that such locals would have to be
1935: garbage collected. Since this entails un-Forth-like implementation
1936: complexities, I adopted the same cowardly solution as some other
1937: languages (e.g., C): The local lives only as long as it is visible;
1938: afterwards its address is invalid (and programs that access it
1939: afterwards are erroneous).
1940:
1941: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1942: @subsubsection Programming Style
1943: @cindex locals programming style
1944: @cindex programming style, locals
1945:
1946: The freedom to define locals anywhere has the potential to change
1947: programming styles dramatically. In particular, the need to use the
1948: return stack for intermediate storage vanishes. Moreover, all stack
1949: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1950: determined arguments) can be eliminated: If the stack items are in the
1951: wrong order, just write a locals definition for all of them; then
1952: write the items in the order you want.
1953:
1954: This seems a little far-fetched and eliminating stack manipulations is
1955: unlikely to become a conscious programming objective. Still, the number
1956: of stack manipulations will be reduced dramatically if local variables
1957: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1958: a traditional implementation of @code{max}).
1959:
1960: This shows one potential benefit of locals: making Forth programs more
1961: readable. Of course, this benefit will only be realized if the
1962: programmers continue to honour the principle of factoring instead of
1963: using the added latitude to make the words longer.
1964:
1965: @cindex single-assignment style for locals
1966: Using @code{TO} can and should be avoided. Without @code{TO},
1967: every value-flavoured local has only a single assignment and many
1968: advantages of functional languages apply to Forth. I.e., programs are
1969: easier to analyse, to optimize and to read: It is clear from the
1970: definition what the local stands for, it does not turn into something
1971: different later.
1972:
1973: E.g., a definition using @code{TO} might look like this:
1974: @example
1975: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1976: u1 u2 min 0
1977: ?do
1978: addr1 c@@ addr2 c@@ -
1979: ?dup-if
1980: unloop exit
1981: then
1982: addr1 char+ TO addr1
1983: addr2 char+ TO addr2
1984: loop
1985: u1 u2 - ;
1986: @end example
1987: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1988: every loop iteration. @code{strcmp} is a typical example of the
1989: readability problems of using @code{TO}. When you start reading
1990: @code{strcmp}, you think that @code{addr1} refers to the start of the
1991: string. Only near the end of the loop you realize that it is something
1992: else.
1993:
1994: This can be avoided by defining two locals at the start of the loop that
1995: are initialized with the right value for the current iteration.
1996: @example
1997: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1998: addr1 addr2
1999: u1 u2 min 0
2000: ?do @{ s1 s2 @}
2001: s1 c@@ s2 c@@ -
2002: ?dup-if
2003: unloop exit
2004: then
2005: s1 char+ s2 char+
2006: loop
2007: 2drop
2008: u1 u2 - ;
2009: @end example
2010: Here it is clear from the start that @code{s1} has a different value
2011: in every loop iteration.
2012:
2013: @node Implementation, , Programming Style, Gforth locals
2014: @subsubsection Implementation
2015: @cindex locals implementation
2016: @cindex implementation of locals
2017:
2018: @cindex locals stack
2019: Gforth uses an extra locals stack. The most compelling reason for
2020: this is that the return stack is not float-aligned; using an extra stack
2021: also eliminates the problems and restrictions of using the return stack
2022: as locals stack. Like the other stacks, the locals stack grows toward
2023: lower addresses. A few primitives allow an efficient implementation:
2024:
2025: doc-@local#
2026: doc-f@local#
2027: doc-laddr#
2028: doc-lp+!#
2029: doc-lp!
2030: doc->l
2031: doc-f>l
2032:
2033: In addition to these primitives, some specializations of these
2034: primitives for commonly occurring inline arguments are provided for
2035: efficiency reasons, e.g., @code{@@local0} as specialization of
2036: @code{@@local#} for the inline argument 0. The following compiling words
2037: compile the right specialized version, or the general version, as
2038: appropriate:
2039:
2040: doc-compile-@local
2041: doc-compile-f@local
2042: doc-compile-lp+!
2043:
2044: Combinations of conditional branches and @code{lp+!#} like
2045: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
2046: is taken) are provided for efficiency and correctness in loops.
2047:
2048: A special area in the dictionary space is reserved for keeping the
2049: local variable names. @code{@{} switches the dictionary pointer to this
2050: area and @code{@}} switches it back and generates the locals
2051: initializing code. @code{W:} etc.@ are normal defining words. This
2052: special area is cleared at the start of every colon definition.
2053:
2054: @cindex wordlist for defining locals
2055: A special feature of Gforth's dictionary is used to implement the
2056: definition of locals without type specifiers: every wordlist (aka
2057: vocabulary) has its own methods for searching
2058: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
2059: with a special search method: When it is searched for a word, it
2060: actually creates that word using @code{W:}. @code{@{} changes the search
2061: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
2062: and then the wordlist for defining locals without type specifiers.
2063:
2064: The lifetime rules support a stack discipline within a colon
2065: definition: The lifetime of a local is either nested with other locals
2066: lifetimes or it does not overlap them.
2067:
2068: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
2069: pointer manipulation is generated. Between control structure words
2070: locals definitions can push locals onto the locals stack. @code{AGAIN}
2071: is the simplest of the other three control flow words. It has to
2072: restore the locals stack depth of the corresponding @code{BEGIN}
2073: before branching. The code looks like this:
2074: @format
2075: @code{lp+!#} current-locals-size @minus{} dest-locals-size
2076: @code{branch} <begin>
2077: @end format
2078:
2079: @code{UNTIL} is a little more complicated: If it branches back, it
2080: must adjust the stack just like @code{AGAIN}. But if it falls through,
2081: the locals stack must not be changed. The compiler generates the
2082: following code:
2083: @format
2084: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
2085: @end format
2086: The locals stack pointer is only adjusted if the branch is taken.
2087:
2088: @code{THEN} can produce somewhat inefficient code:
2089: @format
2090: @code{lp+!#} current-locals-size @minus{} orig-locals-size
2091: <orig target>:
2092: @code{lp+!#} orig-locals-size @minus{} new-locals-size
2093: @end format
2094: The second @code{lp+!#} adjusts the locals stack pointer from the
2095: level at the @var{orig} point to the level after the @code{THEN}. The
2096: first @code{lp+!#} adjusts the locals stack pointer from the current
2097: level to the level at the orig point, so the complete effect is an
2098: adjustment from the current level to the right level after the
2099: @code{THEN}.
2100:
2101: @cindex locals information on the control-flow stack
2102: @cindex control-flow stack items, locals information
2103: In a conventional Forth implementation a dest control-flow stack entry
2104: is just the target address and an orig entry is just the address to be
2105: patched. Our locals implementation adds a wordlist to every orig or dest
2106: item. It is the list of locals visible (or assumed visible) at the point
2107: described by the entry. Our implementation also adds a tag to identify
2108: the kind of entry, in particular to differentiate between live and dead
2109: (reachable and unreachable) orig entries.
2110:
2111: A few unusual operations have to be performed on locals wordlists:
2112:
2113: doc-common-list
2114: doc-sub-list?
2115: doc-list-size
2116:
2117: Several features of our locals wordlist implementation make these
2118: operations easy to implement: The locals wordlists are organised as
2119: linked lists; the tails of these lists are shared, if the lists
2120: contain some of the same locals; and the address of a name is greater
2121: than the address of the names behind it in the list.
2122:
2123: Another important implementation detail is the variable
2124: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
2125: determine if they can be reached directly or only through the branch
2126: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
2127: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
2128: definition, by @code{BEGIN} and usually by @code{THEN}.
2129:
2130: Counted loops are similar to other loops in most respects, but
2131: @code{LEAVE} requires special attention: It performs basically the same
2132: service as @code{AHEAD}, but it does not create a control-flow stack
2133: entry. Therefore the information has to be stored elsewhere;
2134: traditionally, the information was stored in the target fields of the
2135: branches created by the @code{LEAVE}s, by organizing these fields into a
2136: linked list. Unfortunately, this clever trick does not provide enough
2137: space for storing our extended control flow information. Therefore, we
2138: introduce another stack, the leave stack. It contains the control-flow
2139: stack entries for all unresolved @code{LEAVE}s.
2140:
2141: Local names are kept until the end of the colon definition, even if
2142: they are no longer visible in any control-flow path. In a few cases
2143: this may lead to increased space needs for the locals name area, but
2144: usually less than reclaiming this space would cost in code size.
2145:
2146:
2147: @node ANS Forth locals, , Gforth locals, Locals
2148: @subsection ANS Forth locals
2149: @cindex locals, ANS Forth style
2150:
2151: The ANS Forth locals wordset does not define a syntax for locals, but
2152: words that make it possible to define various syntaxes. One of the
2153: possible syntaxes is a subset of the syntax we used in the Gforth locals
2154: wordset, i.e.:
2155:
2156: @example
2157: @{ local1 local2 ... -- comment @}
2158: @end example
2159: or
2160: @example
2161: @{ local1 local2 ... @}
2162: @end example
2163:
2164: The order of the locals corresponds to the order in a stack comment. The
2165: restrictions are:
2166:
2167: @itemize @bullet
2168: @item
2169: Locals can only be cell-sized values (no type specifiers are allowed).
2170: @item
2171: Locals can be defined only outside control structures.
2172: @item
2173: Locals can interfere with explicit usage of the return stack. For the
2174: exact (and long) rules, see the standard. If you don't use return stack
2175: accessing words in a definition using locals, you will be all right. The
2176: purpose of this rule is to make locals implementation on the return
2177: stack easier.
2178: @item
2179: The whole definition must be in one line.
2180: @end itemize
2181:
2182: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
2183: Defining Words}). I.e., they are initialized from the stack. Using their
2184: name produces their value. Their value can be changed using @code{TO}.
2185:
2186: Since this syntax is supported by Gforth directly, you need not do
2187: anything to use it. If you want to port a program using this syntax to
2188: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
2189: syntax on the other system.
2190:
2191: Note that a syntax shown in the standard, section A.13 looks
2192: similar, but is quite different in having the order of locals
2193: reversed. Beware!
2194:
2195: The ANS Forth locals wordset itself consists of the following word
2196:
2197: doc-(local)
2198:
2199: The ANS Forth locals extension wordset defines a syntax, but it is so
2200: awful that we strongly recommend not to use it. We have implemented this
2201: syntax to make porting to Gforth easy, but do not document it here. The
2202: problem with this syntax is that the locals are defined in an order
2203: reversed with respect to the standard stack comment notation, making
2204: programs harder to read, and easier to misread and miswrite. The only
2205: merit of this syntax is that it is easy to implement using the ANS Forth
2206: locals wordset.
2207:
2208: @node Defining Words, Structures, Locals, Words
2209: @section Defining Words
2210: @cindex defining words
2211:
2212: @menu
2213: * Simple Defining Words::
2214: * Colon Definitions::
2215: * User-defined Defining Words::
2216: * Supplying names::
2217: * Interpretation and Compilation Semantics::
2218: @end menu
2219:
2220: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
2221: @subsection Simple Defining Words
2222: @cindex simple defining words
2223: @cindex defining words, simple
2224:
2225: doc-constant
2226: doc-2constant
2227: doc-fconstant
2228: doc-variable
2229: doc-2variable
2230: doc-fvariable
2231: doc-create
2232: doc-user
2233: doc-value
2234: doc-to
2235: doc-defer
2236: doc-is
2237:
2238: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
2239: @subsection Colon Definitions
2240: @cindex colon definitions
2241:
2242: @example
2243: : name ( ... -- ... )
2244: word1 word2 word3 ;
2245: @end example
2246:
2247: creates a word called @code{name}, that, upon execution, executes
2248: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2249:
2250: The explanation above is somewhat superficial. @xref{Interpretation and
2251: Compilation Semantics} for an in-depth discussion of some of the issues
2252: involved.
2253:
2254: doc-:
2255: doc-;
2256:
2257: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2258: @subsection User-defined Defining Words
2259: @cindex user-defined defining words
2260: @cindex defining words, user-defined
2261:
2262: You can create new defining words simply by wrapping defining-time code
2263: around existing defining words and putting the sequence in a colon
2264: definition.
2265:
2266: @cindex @code{CREATE} ... @code{DOES>}
2267: If you want the words defined with your defining words to behave
2268: differently from words defined with standard defining words, you can
2269: write your defining word like this:
2270:
2271: @example
2272: : def-word ( "name" -- )
2273: Create @var{code1}
2274: DOES> ( ... -- ... )
2275: @var{code2} ;
2276:
2277: def-word name
2278: @end example
2279:
2280: Technically, this fragment defines a defining word @code{def-word}, and
2281: a word @code{name}; when you execute @code{name}, the address of the
2282: body of @code{name} is put on the data stack and @var{code2} is executed
2283: (the address of the body of @code{name} is the address @code{HERE}
2284: returns immediately after the @code{CREATE}).
2285:
2286: In other words, if you make the following definitions:
2287:
2288: @example
2289: : def-word1 ( "name" -- )
2290: Create @var{code1} ;
2291:
2292: : action1 ( ... -- ... )
2293: @var{code2} ;
2294:
2295: def-word name1
2296: @end example
2297:
2298: Using @code{name1 action1} is equivalent to using @code{name}.
2299:
2300: E.g., you can implement @code{Constant} in this way:
2301:
2302: @example
2303: : constant ( w "name" -- )
2304: create ,
2305: DOES> ( -- w )
2306: @@ ;
2307: @end example
2308:
2309: When you create a constant with @code{5 constant five}, first a new word
2310: @code{five} is created, then the value 5 is laid down in the body of
2311: @code{five} with @code{,}. When @code{five} is invoked, the address of
2312: the body is put on the stack, and @code{@@} retrieves the value 5.
2313:
2314: @cindex stack effect of @code{DOES>}-parts
2315: @cindex @code{DOES>}-parts, stack effect
2316: In the example above the stack comment after the @code{DOES>} specifies
2317: the stack effect of the defined words, not the stack effect of the
2318: following code (the following code expects the address of the body on
2319: the top of stack, which is not reflected in the stack comment). This is
2320: the convention that I use and recommend (it clashes a bit with using
2321: locals declarations for stack effect specification, though).
2322:
2323: @subsubsection Applications of @code{CREATE..DOES>}
2324: @cindex @code{CREATE} ... @code{DOES>}, applications
2325:
2326: You may wonder how to use this feature. Here are some usage patterns:
2327:
2328: @cindex factoring similar colon definitions
2329: When you see a sequence of code occurring several times, and you can
2330: identify a meaning, you will factor it out as a colon definition. When
2331: you see similar colon definitions, you can factor them using
2332: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
2333: that look very similar:
2334: @example
2335: : ori, ( reg-target reg-source n -- )
2336: 0 asm-reg-reg-imm ;
2337: : andi, ( reg-target reg-source n -- )
2338: 1 asm-reg-reg-imm ;
2339: @end example
2340:
2341: This could be factored with:
2342: @example
2343: : reg-reg-imm ( op-code -- )
2344: create ,
2345: DOES> ( reg-target reg-source n -- )
2346: @@ asm-reg-reg-imm ;
2347:
2348: 0 reg-reg-imm ori,
2349: 1 reg-reg-imm andi,
2350: @end example
2351:
2352: @cindex currying
2353: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2354: supply a part of the parameters for a word (known as @dfn{currying} in
2355: the functional language community). E.g., @code{+} needs two
2356: parameters. Creating versions of @code{+} with one parameter fixed can
2357: be done like this:
2358: @example
2359: : curry+ ( n1 -- )
2360: create ,
2361: DOES> ( n2 -- n1+n2 )
2362: @@ + ;
2363:
2364: 3 curry+ 3+
2365: -2 curry+ 2-
2366: @end example
2367:
2368: @subsubsection The gory details of @code{CREATE..DOES>}
2369: @cindex @code{CREATE} ... @code{DOES>}, details
2370:
2371: doc-does>
2372:
2373: @cindex @code{DOES>} in a separate definition
2374: This means that you need not use @code{CREATE} and @code{DOES>} in the
2375: same definition; E.g., you can put the @code{DOES>}-part in a separate
2376: definition. This allows us to, e.g., select among different DOES>-parts:
2377: @example
2378: : does1
2379: DOES> ( ... -- ... )
2380: ... ;
2381:
2382: : does2
2383: DOES> ( ... -- ... )
2384: ... ;
2385:
2386: : def-word ( ... -- ... )
2387: create ...
2388: IF
2389: does1
2390: ELSE
2391: does2
2392: ENDIF ;
2393: @end example
2394:
2395: @cindex @code{DOES>} in interpretation state
2396: In a standard program you can apply a @code{DOES>}-part only if the last
2397: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2398: will override the behaviour of the last word defined in any case. In a
2399: standard program, you can use @code{DOES>} only in a colon
2400: definition. In Gforth, you can also use it in interpretation state, in a
2401: kind of one-shot mode:
2402: @example
2403: CREATE name ( ... -- ... )
2404: @var{initialization}
2405: DOES>
2406: @var{code} ;
2407: @end example
2408: This is equivalent to the standard
2409: @example
2410: :noname
2411: DOES>
2412: @var{code} ;
2413: CREATE name EXECUTE ( ... -- ... )
2414: @var{initialization}
2415: @end example
2416:
2417: You can get the address of the body of a word with
2418:
2419: doc->body
2420:
2421: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2422: @subsection Supplying names for the defined words
2423: @cindex names for defined words
2424: @cindex defining words, name parameter
2425:
2426: @cindex defining words, name given in a string
2427: By default, defining words take the names for the defined words from the
2428: input stream. Sometimes you want to supply the name from a string. You
2429: can do this with
2430:
2431: doc-nextname
2432:
2433: E.g.,
2434:
2435: @example
2436: s" foo" nextname create
2437: @end example
2438: is equivalent to
2439: @example
2440: create foo
2441: @end example
2442:
2443: @cindex defining words without name
2444: Sometimes you want to define a word without a name. You can do this with
2445:
2446: doc-noname
2447:
2448: @cindex execution token of last defined word
2449: To make any use of the newly defined word, you need its execution
2450: token. You can get it with
2451:
2452: doc-lastxt
2453:
2454: E.g., you can initialize a deferred word with an anonymous colon
2455: definition:
2456: @example
2457: Defer deferred
2458: noname : ( ... -- ... )
2459: ... ;
2460: lastxt IS deferred
2461: @end example
2462:
2463: @code{lastxt} also works when the last word was not defined as
2464: @code{noname}.
2465:
2466: The standard has also recognized the need for anonymous words and
2467: provides
2468:
2469: doc-:noname
2470:
2471: This leaves the execution token for the word on the stack after the
2472: closing @code{;}. You can rewrite the last example with @code{:noname}:
2473: @example
2474: Defer deferred
2475: :noname ( ... -- ... )
2476: ... ;
2477: IS deferred
2478: @end example
2479:
2480: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2481: @subsection Interpretation and Compilation Semantics
2482: @cindex semantics, interpretation and compilation
2483:
2484: @cindex interpretation semantics
2485: The @dfn{interpretation semantics} of a word are what the text
2486: interpreter does when it encounters the word in interpret state. It also
2487: appears in some other contexts, e.g., the execution token returned by
2488: @code{' @var{word}} identifies the interpretation semantics of
2489: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
2490: interpret-state text interpretation of @code{@var{word}}).
2491:
2492: @cindex compilation semantics
2493: The @dfn{compilation semantics} of a word are what the text interpreter
2494: does when it encounters the word in compile state. It also appears in
2495: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2496: standard terminology, ``appends to the current definition''.} the
2497: compilation semantics of @var{word}.
2498:
2499: @cindex execution semantics
2500: The standard also talks about @dfn{execution semantics}. They are used
2501: only for defining the interpretation and compilation semantics of many
2502: words. By default, the interpretation semantics of a word are to
2503: @code{execute} its execution semantics, and the compilation semantics of
2504: a word are to @code{compile,} its execution semantics.@footnote{In
2505: standard terminology: The default interpretation semantics are its
2506: execution semantics; the default compilation semantics are to append its
2507: execution semantics to the execution semantics of the current
2508: definition.}
2509:
2510: @cindex immediate words
2511: You can change the compilation semantics into @code{execute}ing the
2512: execution semantics with
2513:
2514: doc-immediate
2515:
2516: @cindex compile-only words
2517: You can remove the interpretation semantics of a word with
2518:
2519: doc-compile-only
2520: doc-restrict
2521:
2522: Note that ticking (@code{'}) compile-only words gives an error
2523: (``Interpreting a compile-only word'').
2524:
2525: Gforth also allows you to define words with arbitrary combinations of
2526: interpretation and compilation semantics.
2527:
2528: doc-interpret/compile:
2529:
2530: This feature was introduced for implementing @code{TO} and @code{S"}. I
2531: recommend that you do not define such words, as cute as they may be:
2532: they make it hard to get at both parts of the word in some contexts.
2533: E.g., assume you want to get an execution token for the compilation
2534: part. Instead, define two words, one that embodies the interpretation
2535: part, and one that embodies the compilation part.
2536:
2537: There is, however, a potentially useful application of this feature:
2538: Providing differing implementations for the default semantics. While
2539: this introduces redundancy and is therefore usually a bad idea, a
2540: performance improvement may be worth the trouble. E.g., consider the
2541: word @code{foobar}:
2542:
2543: @example
2544: : foobar
2545: foo bar ;
2546: @end example
2547:
2548: Let us assume that @code{foobar} is called so frequently that the
2549: calling overhead would take a significant amount of the run-time. We can
2550: optimize it with @code{interpret/compile:}:
2551:
2552: @example
2553: :noname
2554: foo bar ;
2555: :noname
2556: POSTPONE foo POSTPONE bar ;
2557: interpret/compile: foobar
2558: @end example
2559:
2560: This definition has the same interpretation semantics and essentially
2561: the same compilation semantics as the simple definition of
2562: @code{foobar}, but the implementation of the compilation semantics is
2563: more efficient with respect to run-time.
2564:
2565: @cindex state-smart words are a bad idea
2566: Some people try to use state-smart words to emulate the feature provided
2567: by @code{interpret/compile:} (words are state-smart if they check
2568: @code{STATE} during execution). E.g., they would try to code
2569: @code{foobar} like this:
2570:
2571: @example
2572: : foobar
2573: STATE @@
2574: IF ( compilation state )
2575: POSTPONE foo POSTPONE bar
2576: ELSE
2577: foo bar
2578: ENDIF ; immediate
2579: @end example
2580:
2581: While this works if @code{foobar} is processed only by the text
2582: interpreter, it does not work in other contexts (like @code{'} or
2583: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2584: for a state-smart word, not for the interpretation semantics of the
2585: original @code{foobar}; when you execute this execution token (directly
2586: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2587: state, the result will not be what you expected (i.e., it will not
2588: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2589: write them!
2590:
2591: @cindex defining words with arbitrary semantics combinations
2592: It is also possible to write defining words that define words with
2593: arbitrary combinations of interpretation and compilation semantics (or,
2594: preferably, arbitrary combinations of implementations of the default
2595: semantics). In general, this looks like:
2596:
2597: @example
2598: : def-word
2599: create-interpret/compile
2600: @var{code1}
2601: interpretation>
2602: @var{code2}
2603: <interpretation
2604: compilation>
2605: @var{code3}
2606: <compilation ;
2607: @end example
2608:
2609: For a @var{word} defined with @code{def-word}, the interpretation
2610: semantics are to push the address of the body of @var{word} and perform
2611: @var{code2}, and the compilation semantics are to push the address of
2612: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2613: can also be defined like this:
2614:
2615: @example
2616: : constant ( n "name" -- )
2617: create-interpret/compile
2618: ,
2619: interpretation> ( -- n )
2620: @@
2621: <interpretation
2622: compilation> ( compilation. -- ; run-time. -- n )
2623: @@ postpone literal
2624: <compilation ;
2625: @end example
2626:
2627: doc-create-interpret/compile
2628: doc-interpretation>
2629: doc-<interpretation
2630: doc-compilation>
2631: doc-<compilation
2632:
2633: Note that words defined with @code{interpret/compile:} and
2634: @code{create-interpret/compile} have an extended header structure that
2635: differs from other words; however, unless you try to access them with
2636: plain address arithmetic, you should not notice this. Words for
2637: accessing the header structure usually know how to deal with this; e.g.,
2638: @code{' word >body} also gives you the body of a word created with
2639: @code{create-interpret/compile}.
2640:
2641: @c ----------------------------------------------------------
2642: @node Structures, Objects, Defining Words, Words
2643: @section Structures
2644: @cindex structures
2645: @cindex records
2646:
2647: This section presents the structure package that comes with Gforth. A
2648: version of the package implemented in plain ANS Forth is available in
2649: @file{compat/struct.fs}. This package was inspired by a posting on
2650: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
2651: possibly John Hayes). A version of this section has been published in
2652: ???. Marcel Hendrix provided helpful comments.
2653:
2654: @menu
2655: * Why explicit structure support?::
2656: * Structure Usage::
2657: * Structure Naming Convention::
2658: * Structure Implementation::
2659: * Structure Glossary::
2660: @end menu
2661:
2662: @node Why explicit structure support?, Structure Usage, Structures, Structures
2663: @subsection Why explicit structure support?
2664:
2665: @cindex address arithmetic for structures
2666: @cindex structures using address arithmetic
2667: If we want to use a structure containing several fields, we could simply
2668: reserve memory for it, and access the fields using address arithmetic
2669: (@pxref{Address arithmetic}). As an example, consider a structure with
2670: the following fields
2671:
2672: @table @code
2673: @item a
2674: is a float
2675: @item b
2676: is a cell
2677: @item c
2678: is a float
2679: @end table
2680:
2681: Given the (float-aligned) base address of the structure we get the
2682: address of the field
2683:
2684: @table @code
2685: @item a
2686: without doing anything further.
2687: @item b
2688: with @code{float+}
2689: @item c
2690: with @code{float+ cell+ faligned}
2691: @end table
2692:
2693: It is easy to see that this can become quite tiring.
2694:
2695: Moreover, it is not very readable, because seeing a
2696: @code{cell+} tells us neither which kind of structure is
2697: accessed nor what field is accessed; we have to somehow infer the kind
2698: of structure, and then look up in the documentation, which field of
2699: that structure corresponds to that offset.
2700:
2701: Finally, this kind of address arithmetic also causes maintenance
2702: troubles: If you add or delete a field somewhere in the middle of the
2703: structure, you have to find and change all computations for the fields
2704: afterwards.
2705:
2706: So, instead of using @code{cell+} and friends directly, how
2707: about storing the offsets in constants:
2708:
2709: @example
2710: 0 constant a-offset
2711: 0 float+ constant b-offset
2712: 0 float+ cell+ faligned c-offset
2713: @end example
2714:
2715: Now we can get the address of field @code{x} with @code{x-offset
2716: +}. This is much better in all respects. Of course, you still
2717: have to change all later offset definitions if you add a field. You can
2718: fix this by declaring the offsets in the following way:
2719:
2720: @example
2721: 0 constant a-offset
2722: a-offset float+ constant b-offset
2723: b-offset cell+ faligned constant c-offset
2724: @end example
2725:
2726: Since we always use the offsets with @code{+}, using a defining
2727: word @code{cfield} that includes the @code{+} in the
2728: action of the defined word offers itself:
2729:
2730: @example
2731: : cfield ( n "name" -- )
2732: create ,
2733: does> ( name execution: addr1 -- addr2 )
2734: @@ + ;
2735:
2736: 0 cfield a
2737: 0 a float+ cfield b
2738: 0 b cell+ faligned cfield c
2739: @end example
2740:
2741: Instead of @code{x-offset +}, we now simply write @code{x}.
2742:
2743: The structure field words now can be used quite nicely. However,
2744: their definition is still a bit cumbersome: We have to repeat the
2745: name, the information about size and alignment is distributed before
2746: and after the field definitions etc. The structure package presented
2747: here addresses these problems.
2748:
2749: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
2750: @subsection Structure Usage
2751: @cindex structure usage
2752:
2753: @cindex @code{field} usage
2754: @cindex @code{struct} usage
2755: @cindex @code{end-struct} usage
2756: You can define a structure for a (data-less) linked list with
2757: @example
2758: struct
2759: cell% field list-next
2760: end-struct list%
2761: @end example
2762:
2763: With the address of the list node on the stack, you can compute the
2764: address of the field that contains the address of the next node with
2765: @code{list-next}. E.g., you can determine the length of a list
2766: with:
2767:
2768: @example
2769: : list-length ( list -- n )
2770: \ "list" is a pointer to the first element of a linked list
2771: \ "n" is the length of the list
2772: 0 begin ( list1 n1 )
2773: over
2774: while ( list1 n1 )
2775: 1+ swap list-next @@ swap
2776: repeat
2777: nip ;
2778: @end example
2779:
2780: You can reserve memory for a list node in the dictionary with
2781: @code{list% %allot}, which leaves the address of the list node on the
2782: stack. For the equivalent allocation on the heap you can use @code{list%
2783: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
2784: use @code{list% %allocate}). You can also get the the size of a list
2785: node with @code{list% %size} and it's alignment with @code{list%
2786: %alignment}.
2787:
2788: Note that in ANS Forth the body of a @code{create}d word is
2789: @code{aligned} but not necessarily @code{faligned};
2790: therefore, if you do a
2791: @example
2792: create @emph{name} foo% %allot
2793: @end example
2794:
2795: then the memory alloted for @code{foo%} is
2796: guaranteed to start at the body of @code{@emph{name}} only if
2797: @code{foo%} contains only character, cell and double fields.
2798:
2799: @cindex strcutures containing structures
2800: You can also include a structure @code{foo%} as field of
2801: another structure, with:
2802: @example
2803: struct
2804: ...
2805: foo% field ...
2806: ...
2807: end-struct ...
2808: @end example
2809:
2810: @cindex structure extension
2811: @cindex extended records
2812: Instead of starting with an empty structure, you can also extend an
2813: existing structure. E.g., a plain linked list without data, as defined
2814: above, is hardly useful; You can extend it to a linked list of integers,
2815: like this:@footnote{This feature is also known as @emph{extended
2816: records}. It is the main innovation in the Oberon language; in other
2817: words, adding this feature to Modula-2 led Wirth to create a new
2818: language, write a new compiler etc. Adding this feature to Forth just
2819: requires a few lines of code.}
2820:
2821: @example
2822: list%
2823: cell% field intlist-int
2824: end-struct intlist%
2825: @end example
2826:
2827: @code{intlist%} is a structure with two fields:
2828: @code{list-next} and @code{intlist-int}.
2829:
2830: @cindex structures containing arrays
2831: You can specify an array type containing @emph{n} elements of
2832: type @code{foo%} like this:
2833:
2834: @example
2835: foo% @emph{n} *
2836: @end example
2837:
2838: You can use this array type in any place where you can use a normal
2839: type, e.g., when defining a @code{field}, or with
2840: @code{%allot}.
2841:
2842: @cindex first field optimization
2843: The first field is at the base address of a structure and the word
2844: for this field (e.g., @code{list-next}) actually does not change
2845: the address on the stack. You may be tempted to leave it away in the
2846: interest of run-time and space efficiency. This is not necessary,
2847: because the structure package optimizes this case and compiling such
2848: words does not generate any code. So, in the interest of readability
2849: and maintainability you should include the word for the field when
2850: accessing the field.
2851:
2852: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
2853: @subsection Structure Naming Convention
2854: @cindex structure naming conventions
2855:
2856: The field names that come to (my) mind are often quite generic, and,
2857: if used, would cause frequent name clashes. E.g., many structures
2858: probably contain a @code{counter} field. The structure names
2859: that come to (my) mind are often also the logical choice for the names
2860: of words that create such a structure.
2861:
2862: Therefore, I have adopted the following naming conventions:
2863:
2864: @itemize @bullet
2865: @cindex field naming convention
2866: @item
2867: The names of fields are of the form
2868: @code{@emph{struct}-@emph{field}}, where
2869: @code{@emph{struct}} is the basic name of the structure, and
2870: @code{@emph{field}} is the basic name of the field. You can
2871: think about field words as converting converts the (address of the)
2872: structure into the (address of the) field.
2873:
2874: @cindex structure naming convention
2875: @item
2876: The names of structures are of the form
2877: @code{@emph{struct}%}, where
2878: @code{@emph{struct}} is the basic name of the structure.
2879: @end itemize
2880:
2881: This naming convention does not work that well for fields of extended
2882: structures; e.g., the integer list structure has a field
2883: @code{intlist-int}, but has @code{list-next}, not
2884: @code{intlist-next}.
2885:
2886: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
2887: @subsection Structure Implementation
2888: @cindex structure implementation
2889: @cindex implementation of structures
2890:
2891: The central idea in the implementation is to pass the data about the
2892: structure being built on the stack, not in some global
2893: variable. Everything else falls into place naturally once this design
2894: decision is made.
2895:
2896: The type description on the stack is of the form @emph{align
2897: size}. Keeping the size on the top-of-stack makes dealing with arrays
2898: very simple.
2899:
2900: @code{field} is a defining word that uses @code{create}
2901: and @code{does>}. The body of the field contains the offset
2902: of the field, and the normal @code{does>} action is
2903:
2904: @example
2905: @ +
2906: @end example
2907:
2908: i.e., add the offset to the address, giving the stack effect
2909: @code{addr1 -- addr2} for a field.
2910:
2911: @cindex first field optimization, implementation
2912: This simple structure is slightly complicated by the optimization
2913: for fields with offset 0, which requires a different
2914: @code{does>}-part (because we cannot rely on there being
2915: something on the stack if such a field is invoked during
2916: compilation). Therefore, we put the different @code{does>}-parts
2917: in separate words, and decide which one to invoke based on the
2918: offset. For a zero offset, the field is basically a noop; it is
2919: immediate, and therefore no code is generated when it is compiled.
2920:
2921: @node Structure Glossary, , Structure Implementation, Structures
2922: @subsection Structure Glossary
2923: @cindex structure glossary
2924:
2925: doc-%align
2926: doc-%alignment
2927: doc-%alloc
2928: doc-%allocate
2929: doc-%allot
2930: doc-cell%
2931: doc-char%
2932: doc-dfloat%
2933: doc-double%
2934: doc-end-struct
2935: doc-field
2936: doc-float%
2937: doc-nalign
2938: doc-sfloat%
2939: doc-%size
2940: doc-struct
2941:
2942: @c -------------------------------------------------------------
2943: @node Objects, Object Oriented Forth, Structures, Words
2944: @section Objects
2945: @cindex objects
2946: @cindex object-oriented programming
2947:
2948: @cindex @file{objects.fs}
2949: @cindex @file{oof.fs}
2950: Gforth comes with two packets for object-oriented programming,
2951: @file{objects.fs} and @file{oof.fs}; none of them is preloaded, so you
2952: have to @code{include} them before use. This section describes the
2953: @file{objects.fs} packet. You can find a description (in German) of
2954: @file{oof.fs} in @cite{Object oriented bigFORTH} by Bernd Paysan,
2955: published in @cite{Vierte Dimension} 10(2), 1994. Both packets are
2956: written in ANS Forth and can be used with any other standard Forth.
2957: @c McKewan's and Zsoter's packages
2958: @c this section is a variant of ...
2959:
2960: This section assumes (in some places) that you have read @ref{Structures}.
2961:
2962: @menu
2963: * Properties of the Objects model::
2964: * Why object-oriented programming?::
2965: * Object-Oriented Terminology::
2966: * Basic Objects Usage::
2967: * The class Object::
2968: * Creating objects::
2969: * Object-Oriented Programming Style::
2970: * Class Binding::
2971: * Method conveniences::
2972: * Classes and Scoping::
2973: * Object Interfaces::
2974: * Objects Implementation::
2975: * Comparison with other object models::
2976: * Objects Glossary::
2977: @end menu
2978:
2979: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
2980: and Bernd Paysan helped me with the related works section.
2981:
2982: @node Properties of the Objects model, Why object-oriented programming?, Objects, Objects
2983: @subsection Properties of the @file{objects.fs} model
2984: @cindex @file{objects.fs} properties
2985:
2986: @itemize @bullet
2987: @item
2988: It is straightforward to pass objects on the stack. Passing
2989: selectors on the stack is a little less convenient, but possible.
2990:
2991: @item
2992: Objects are just data structures in memory, and are referenced by
2993: their address. You can create words for objects with normal defining
2994: words like @code{constant}. Likewise, there is no difference
2995: between instance variables that contain objects and those
2996: that contain other data.
2997:
2998: @item
2999: Late binding is efficient and easy to use.
3000:
3001: @item
3002: It avoids parsing, and thus avoids problems with state-smartness
3003: and reduced extensibility; for convenience there are a few parsing
3004: words, but they have non-parsing counterparts. There are also a few
3005: defining words that parse. This is hard to avoid, because all standard
3006: defining words parse (except @code{:noname}); however, such
3007: words are not as bad as many other parsing words, because they are not
3008: state-smart.
3009:
3010: @item
3011: It does not try to incorporate everything. It does a few things
3012: and does them well (IMO). In particular, I did not intend to support
3013: information hiding with this model (although it has features that may
3014: help); you can use a separate package for achieving this.
3015:
3016: @item
3017: It is layered; you don't have to learn and use all features to use this
3018: model. Only a few features are necessary (@xref{Basic Objects Usage},
3019: @xref{The class Object}, @xref{Creating objects}.), the others
3020: are optional and independent of each other.
3021:
3022: @item
3023: An implementation in ANS Forth is available.
3024:
3025: @end itemize
3026:
3027: I have used the technique, on which this model is based, for
3028: implementing the parser generator Gray; we have also used this technique
3029: in Gforth for implementing the various flavours of wordlists (hashed or
3030: not, case-sensitive or not, special-purpose wordlists for locals etc.).
3031:
3032: @node Why object-oriented programming?, Object-Oriented Terminology, Properties of the Objects model, Objects
3033: @subsection Why object-oriented programming?
3034: @cindex object-oriented programming motivation
3035: @cindex motivation for object-oriented programming
3036:
3037: Often we have to deal with several data structures (@emph{objects}),
3038: that have to be treated similarly in some respects, but differ in
3039: others. Graphical objects are the textbook example: circles,
3040: triangles, dinosaurs, icons, and others, and we may want to add more
3041: during program development. We want to apply some operations to any
3042: graphical object, e.g., @code{draw} for displaying it on the
3043: screen. However, @code{draw} has to do something different for
3044: every kind of object.
3045:
3046: We could implement @code{draw} as a big @code{CASE}
3047: control structure that executes the appropriate code depending on the
3048: kind of object to be drawn. This would be not be very elegant, and,
3049: moreover, we would have to change @code{draw} every time we add
3050: a new kind of graphical object (say, a spaceship).
3051:
3052: What we would rather do is: When defining spaceships, we would tell
3053: the system: "Here's how you @code{draw} a spaceship; you figure
3054: out the rest."
3055:
3056: This is the problem that all systems solve that (rightfully) call
3057: themselves object-oriented, and the object-oriented package I present
3058: here also solves this problem (and not much else).
3059:
3060: @node Object-Oriented Terminology, Basic Objects Usage, Why object-oriented programming?, Objects
3061: @subsection Object-Oriented Terminology
3062: @cindex object-oriented terminology
3063: @cindex terminology for object-oriented programming
3064:
3065: This section is mainly for reference, so you don't have to understand
3066: all of it right away. The terminology is mainly Smalltalk-inspired. In
3067: short:
3068:
3069: @table @emph
3070: @cindex class
3071: @item class
3072: a data structure definition with some extras.
3073:
3074: @cindex object
3075: @item object
3076: an instance of the data structure described by the class definition.
3077:
3078: @cindex instance variables
3079: @item instance variables
3080: fields of the data structure.
3081:
3082: @cindex selector
3083: @cindex method selector
3084: @cindex virtual function
3085: @item selector
3086: (or @emph{method selector}) a word (e.g.,
3087: @code{draw}) for performing an operation on a variety of data
3088: structures (classes). A selector describes @emph{what} operation to
3089: perform. In C++ terminology: a (pure) virtual function.
3090:
3091: @cindex method
3092: @item method
3093: the concrete definition that performs the operation
3094: described by the selector for a specific class. A method specifies
3095: @emph{how} the operation is performed for a specific class.
3096:
3097: @cindex selector invocation
3098: @cindex message send
3099: @cindex invoking a selector
3100: @item selector invocation
3101: a call of a selector. One argument of the call (the TOS (top-of-stack))
3102: is used for determining which method is used. In Smalltalk terminology:
3103: a message (consisting of the selector and the other arguments) is sent
3104: to the object.
3105:
3106: @cindex receiving object
3107: @item receiving object
3108: the object used for determining the method executed by a selector
3109: invocation. In our model it is the object that is on the TOS when the
3110: selector is invoked. (@emph{Receiving} comes from Smalltalks
3111: @emph{message} terminology.)
3112:
3113: @cindex child class
3114: @cindex parent class
3115: @cindex inheritance
3116: @item child class
3117: a class that has (@emph{inherits}) all properties (instance variables,
3118: selectors, methods) from a @emph{parent class}. In Smalltalk
3119: terminology: The subclass inherits from the superclass. In C++
3120: terminology: The derived class inherits from the base class.
3121:
3122: @end table
3123:
3124: @c If you wonder about the message sending terminology, it comes from
3125: @c a time when each object had it's own task and objects communicated via
3126: @c message passing; eventually the Smalltalk developers realized that
3127: @c they can do most things through simple (indirect) calls. They kept the
3128: @c terminology.
3129:
3130: @node Basic Objects Usage, The class Object, Object-Oriented Terminology, Objects
3131: @subsection Basic Objects Usage
3132: @cindex basic objects usage
3133: @cindex objects, basic usage
3134:
3135: You can define a class for graphical objects like this:
3136:
3137: @cindex @code{class} usage
3138: @cindex @code{end-class} usage
3139: @cindex @code{selector} usage
3140: @example
3141: object class \ "object" is the parent class
3142: selector draw ( x y graphical -- )
3143: end-class graphical
3144: @end example
3145:
3146: This code defines a class @code{graphical} with an
3147: operation @code{draw}. We can perform the operation
3148: @code{draw} on any @code{graphical} object, e.g.:
3149:
3150: @example
3151: 100 100 t-rex draw
3152: @end example
3153:
3154: where @code{t-rex} is a word (say, a constant) that produces a
3155: graphical object.
3156:
3157: @cindex abstract class
3158: How do we create a graphical object? With the present definitions,
3159: we cannot create a useful graphical object. The class
3160: @code{graphical} describes graphical objects in general, but not
3161: any concrete graphical object type (C++ users would call it an
3162: @emph{abstract class}); e.g., there is no method for the selector
3163: @code{draw} in the class @code{graphical}.
3164:
3165: For concrete graphical objects, we define child classes of the
3166: class @code{graphical}, e.g.:
3167:
3168: @cindex @code{overrides} usage
3169: @cindex @code{field} usage in class definition
3170: @example
3171: graphical class \ "graphical" is the parent class
3172: cell% field circle-radius
3173:
3174: :noname ( x y circle -- )
3175: circle-radius @@ draw-circle ;
3176: overrides draw
3177:
3178: :noname ( n-radius circle -- )
3179: circle-radius ! ;
3180: overrides construct
3181:
3182: end-class circle
3183: @end example
3184:
3185: Here we define a class @code{circle} as a child of @code{graphical},
3186: with a field @code{circle-radius} (which behaves just like a field in
3187: @pxref{Structures}); it defines new methods for the selectors
3188: @code{draw} and @code{construct} (@code{construct} is defined in
3189: @code{object}, the parent class of @code{graphical}).
3190:
3191: Now we can create a circle on the heap (i.e.,
3192: @code{allocate}d memory) with
3193:
3194: @cindex @code{heap-new} usage
3195: @example
3196: 50 circle heap-new constant my-circle
3197: @end example
3198:
3199: @code{heap-new} invokes @code{construct}, thus
3200: initializing the field @code{circle-radius} with 50. We can draw
3201: this new circle at (100,100) with
3202:
3203: @example
3204: 100 100 my-circle draw
3205: @end example
3206:
3207: @cindex selector invocation, restrictions
3208: @cindex class definition, restrictions
3209: Note: You can invoke a selector only if the object on the TOS
3210: (the receiving object) belongs to the class where the selector was
3211: defined or one of its descendents; e.g., you can invoke
3212: @code{draw} only for objects belonging to @code{graphical}
3213: or its descendents (e.g., @code{circle}). Immediately before
3214: @code{end-class}, the search order has to be the same as
3215: immediately after @code{class}.
3216:
3217: @node The class Object, Creating objects, Basic Objects Usage, Objects
3218: @subsection The class @code{object}
3219: @cindex @code{object} class
3220:
3221: When you define a class, you have to specify a parent class. So how do
3222: you start defining classes? There is one class available from the start:
3223: @code{object}. You can use it as ancestor for all classes. It is the
3224: only class that has no parent. It has two selectors: @code{construct}
3225: and @code{print}.
3226:
3227: @node Creating objects, Object-Oriented Programming Style, The class Object, Objects
3228: @subsection Creating objects
3229: @cindex creating objects
3230: @cindex object creation
3231: @cindex object allocation options
3232:
3233: @cindex @code{heap-new} discussion
3234: @cindex @code{dict-new} discussion
3235: @cindex @code{construct} discussion
3236: You can create and initialize an object of a class on the heap with
3237: @code{heap-new} ( ... class -- object ) and in the dictionary
3238: (allocation with @code{allot}) with @code{dict-new} (
3239: ... class -- object ). Both words invoke @code{construct}, which
3240: consumes the stack items indicated by "..." above.
3241:
3242: @cindex @code{init-object} discussion
3243: @cindex @code{class-inst-size} discussion
3244: If you want to allocate memory for an object yourself, you can get its
3245: alignment and size with @code{class-inst-size 2@@} ( class --
3246: align size ). Once you have memory for an object, you can initialize
3247: it with @code{init-object} ( ... class object -- );
3248: @code{construct} does only a part of the necessary work.
3249:
3250: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
3251: @subsection Object-Oriented Programming Style
3252: @cindex object-oriented programming style
3253:
3254: This section is not exhaustive.
3255:
3256: @cindex stack effects of selectors
3257: @cindex selectors and stack effects
3258: In general, it is a good idea to ensure that all methods for the
3259: same selector have the same stack effect: when you invoke a selector,
3260: you often have no idea which method will be invoked, so, unless all
3261: methods have the same stack effect, you will not know the stack effect
3262: of the selector invocation.
3263:
3264: One exception to this rule is methods for the selector
3265: @code{construct}. We know which method is invoked, because we
3266: specify the class to be constructed at the same place. Actually, I
3267: defined @code{construct} as a selector only to give the users a
3268: convenient way to specify initialization. The way it is used, a
3269: mechanism different from selector invocation would be more natural
3270: (but probably would take more code and more space to explain).
3271:
3272: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
3273: @subsection Class Binding
3274: @cindex class binding
3275: @cindex early binding
3276:
3277: @cindex late binding
3278: Normal selector invocations determine the method at run-time depending
3279: on the class of the receiving object (late binding).
3280:
3281: Sometimes we want to invoke a different method. E.g., assume that
3282: you want to use the simple method for @code{print}ing
3283: @code{object}s instead of the possibly long-winded
3284: @code{print} method of the receiver class. You can achieve this
3285: by replacing the invocation of @code{print} with
3286:
3287: @cindex @code{[bind]} usage
3288: @example
3289: [bind] object print
3290: @end example
3291:
3292: in compiled code or
3293:
3294: @cindex @code{bind} usage
3295: @example
3296: bind object print
3297: @end example
3298:
3299: @cindex class binding, alternative to
3300: in interpreted code. Alternatively, you can define the method with a
3301: name (e.g., @code{print-object}), and then invoke it through the
3302: name. Class binding is just a (often more convenient) way to achieve
3303: the same effect; it avoids name clutter and allows you to invoke
3304: methods directly without naming them first.
3305:
3306: @cindex superclass binding
3307: @cindex parent class binding
3308: A frequent use of class binding is this: When we define a method
3309: for a selector, we often want the method to do what the selector does
3310: in the parent class, and a little more. There is a special word for
3311: this purpose: @code{[parent]}; @code{[parent]
3312: @emph{selector}} is equivalent to @code{[bind] @emph{parent
3313: selector}}, where @code{@emph{parent}} is the parent
3314: class of the current class. E.g., a method definition might look like:
3315:
3316: @cindex @code{[parent]} usage
3317: @example
3318: :noname
3319: dup [parent] foo \ do parent's foo on the receiving object
3320: ... \ do some more
3321: ; overrides foo
3322: @end example
3323:
3324: @cindex class binding as optimization
3325: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
3326: March 1997), Andrew McKewan presents class binding as an optimization
3327: technique. I recommend not using it for this purpose unless you are in
3328: an emergency. Late binding is pretty fast with this model anyway, so the
3329: benefit of using class binding is small; the cost of using class binding
3330: where it is not appropriate is reduced maintainability.
3331:
3332: While we are at programming style questions: You should bind
3333: selectors only to ancestor classes of the receiving object. E.g., say,
3334: you know that the receiving object is of class @code{foo} or its
3335: descendents; then you should bind only to @code{foo} and its
3336: ancestors.
3337:
3338: @node Method conveniences, Classes and Scoping, Class Binding, Objects
3339: @subsection Method conveniences
3340: @cindex method conveniences
3341:
3342: In a method you usually access the receiving object pretty often. If
3343: you define the method as a plain colon definition (e.g., with
3344: @code{:noname}), you may have to do a lot of stack
3345: gymnastics. To avoid this, you can define the method with @code{m:
3346: ... ;m}. E.g., you could define the method for
3347: @code{draw}ing a @code{circle} with
3348:
3349: @cindex @code{this} usage
3350: @cindex @code{m:} usage
3351: @cindex @code{;m} usage
3352: @example
3353: m: ( x y circle -- )
3354: ( x y ) this circle-radius @@ draw-circle ;m
3355: @end example
3356:
3357: @cindex @code{exit} in @code{m: ... ;m}
3358: @cindex @code{exitm} discussion
3359: @cindex @code{catch} in @code{m: ... ;m}
3360: When this method is executed, the receiver object is removed from the
3361: stack; you can access it with @code{this} (admittedly, in this
3362: example the use of @code{m: ... ;m} offers no advantage). Note
3363: that I specify the stack effect for the whole method (i.e. including
3364: the receiver object), not just for the code between @code{m:}
3365: and @code{;m}. You cannot use @code{exit} in
3366: @code{m:...;m}; instead, use
3367: @code{exitm}.@footnote{Moreover, for any word that calls
3368: @code{catch} and was defined before loading
3369: @code{objects.fs}, you have to redefine it like I redefined
3370: @code{catch}: @code{: catch this >r catch r> to-this ;}}
3371:
3372: @cindex @code{inst-var} usage
3373: You will frequently use sequences of the form @code{this
3374: @emph{field}} (in the example above: @code{this
3375: circle-radius}). If you use the field only in this way, you can
3376: define it with @code{inst-var} and eliminate the
3377: @code{this} before the field name. E.g., the @code{circle}
3378: class above could also be defined with:
3379:
3380: @example
3381: graphical class
3382: cell% inst-var radius
3383:
3384: m: ( x y circle -- )
3385: radius @@ draw-circle ;m
3386: overrides draw
3387:
3388: m: ( n-radius circle -- )
3389: radius ! ;m
3390: overrides construct
3391:
3392: end-class circle
3393: @end example
3394:
3395: @code{radius} can only be used in @code{circle} and its
3396: descendent classes and inside @code{m:...;m}.
3397:
3398: @cindex @code{inst-value} usage
3399: You can also define fields with @code{inst-value}, which is
3400: to @code{inst-var} what @code{value} is to
3401: @code{variable}. You can change the value of such a field with
3402: @code{[to-inst]}. E.g., we could also define the class
3403: @code{circle} like this:
3404:
3405: @example
3406: graphical class
3407: inst-value radius
3408:
3409: m: ( x y circle -- )
3410: radius draw-circle ;m
3411: overrides draw
3412:
3413: m: ( n-radius circle -- )
3414: [to-inst] radius ;m
3415: overrides construct
3416:
3417: end-class circle
3418: @end example
3419:
3420:
3421: @node Classes and Scoping, Object Interfaces, Method conveniences, Objects
3422: @subsection Classes and Scoping
3423: @cindex classes and scoping
3424: @cindex scoping and classes
3425:
3426: Inheritance is frequent, unlike structure extension. This exacerbates
3427: the problem with the field name convention (@pxref{Structure Naming
3428: Convention}): One always has to remember in which class the field was
3429: originally defined; changing a part of the class structure would require
3430: changes for renaming in otherwise unaffected code.
3431:
3432: @cindex @code{inst-var} visibility
3433: @cindex @code{inst-value} visibility
3434: To solve this problem, I added a scoping mechanism (which was not in my
3435: original charter): A field defined with @code{inst-var} (or
3436: @code{inst-value}) is visible only in the class where it is defined and in
3437: the descendent classes of this class. Using such fields only makes
3438: sense in @code{m:}-defined methods in these classes anyway.
3439:
3440: This scoping mechanism allows us to use the unadorned field name,
3441: because name clashes with unrelated words become much less likely.
3442:
3443: @cindex @code{protected} discussion
3444: @cindex @code{private} discussion
3445: Once we have this mechanism, we can also use it for controlling the
3446: visibility of other words: All words defined after
3447: @code{protected} are visible only in the current class and its
3448: descendents. @code{public} restores the compilation
3449: (i.e. @code{current}) wordlist that was in effect before. If you
3450: have several @code{protected}s without an intervening
3451: @code{public} or @code{set-current}, @code{public}
3452: will restore the compilation wordlist in effect before the first of
3453: these @code{protected}s.
3454:
3455: @node Object Interfaces, Objects Implementation, Classes and Scoping, Objects
3456: @subsection Object Interfaces
3457: @cindex object interfaces
3458: @cindex interfaces for objects
3459:
3460: In this model you can only call selectors defined in the class of the
3461: receiving objects or in one of its ancestors. If you call a selector
3462: with a receiving object that is not in one of these classes, the
3463: result is undefined; if you are lucky, the program crashes
3464: immediately.
3465:
3466: @cindex selectors common to hardly-related classes
3467: Now consider the case when you want to have a selector (or several)
3468: available in two classes: You would have to add the selector to a
3469: common ancestor class, in the worst case to @code{object}. You
3470: may not want to do this, e.g., because someone else is responsible for
3471: this ancestor class.
3472:
3473: The solution for this problem is interfaces. An interface is a
3474: collection of selectors. If a class implements an interface, the
3475: selectors become available to the class and its descendents. A class
3476: can implement an unlimited number of interfaces. For the problem
3477: discussed above, we would define an interface for the selector(s), and
3478: both classes would implement the interface.
3479:
3480: As an example, consider an interface @code{storage} for
3481: writing objects to disk and getting them back, and a class
3482: @code{foo} foo that implements it. The code for this would look
3483: like this:
3484:
3485: @cindex @code{interface} usage
3486: @cindex @code{end-interface} usage
3487: @cindex @code{implementation} usage
3488: @example
3489: interface
3490: selector write ( file object -- )
3491: selector read1 ( file object -- )
3492: end-interface storage
3493:
3494: bar class
3495: storage implementation
3496:
3497: ... overrides write
3498: ... overrides read
3499: ...
3500: end-class foo
3501: @end example
3502:
3503: (I would add a word @code{read} ( file -- object ) that uses
3504: @code{read1} internally, but that's beyond the point illustrated
3505: here.)
3506:
3507: Note that you cannot use @code{protected} in an interface; and
3508: of course you cannot define fields.
3509:
3510: In the Neon model, all selectors are available for all classes;
3511: therefore it does not need interfaces. The price you pay in this model
3512: is slower late binding, and therefore, added complexity to avoid late
3513: binding.
3514:
3515: @node Objects Implementation, Comparison with other object models, Object Interfaces, Objects
3516: @subsection @file{objects.fs} Implementation
3517: @cindex @file{objects.fs} implementation
3518:
3519: @cindex @code{object-map} discussion
3520: An object is a piece of memory, like one of the data structures
3521: described with @code{struct...end-struct}. It has a field
3522: @code{object-map} that points to the method map for the object's
3523: class.
3524:
3525: @cindex method map
3526: @cindex virtual function table
3527: The @emph{method map}@footnote{This is Self terminology; in C++
3528: terminology: virtual function table.} is an array that contains the
3529: execution tokens (XTs) of the methods for the object's class. Each
3530: selector contains an offset into the method maps.
3531:
3532: @cindex @code{selector} implementation, class
3533: @code{selector} is a defining word that uses
3534: @code{create} and @code{does>}. The body of the
3535: selector contains the offset; the @code{does>} action for a
3536: class selector is, basically:
3537:
3538: @example
3539: ( object addr ) @@ over object-map @@ + @@ execute
3540: @end example
3541:
3542: Since @code{object-map} is the first field of the object, it
3543: does not generate any code. As you can see, calling a selector has a
3544: small, constant cost.
3545:
3546: @cindex @code{current-interface} discussion
3547: @cindex class implementation and representation
3548: A class is basically a @code{struct} combined with a method
3549: map. During the class definition the alignment and size of the class
3550: are passed on the stack, just as with @code{struct}s, so
3551: @code{field} can also be used for defining class
3552: fields. However, passing more items on the stack would be
3553: inconvenient, so @code{class} builds a data structure in memory,
3554: which is accessed through the variable
3555: @code{current-interface}. After its definition is complete, the
3556: class is represented on the stack by a pointer (e.g., as parameter for
3557: a child class definition).
3558:
3559: At the start, a new class has the alignment and size of its parent,
3560: and a copy of the parent's method map. Defining new fields extends the
3561: size and alignment; likewise, defining new selectors extends the
3562: method map. @code{overrides} just stores a new XT in the method
3563: map at the offset given by the selector.
3564:
3565: @cindex class binding, implementation
3566: Class binding just gets the XT at the offset given by the selector
3567: from the class's method map and @code{compile,}s (in the case of
3568: @code{[bind]}) it.
3569:
3570: @cindex @code{this} implementation
3571: @cindex @code{catch} and @code{this}
3572: @cindex @code{this} and @code{catch}
3573: I implemented @code{this} as a @code{value}. At the
3574: start of an @code{m:...;m} method the old @code{this} is
3575: stored to the return stack and restored at the end; and the object on
3576: the TOS is stored @code{TO this}. This technique has one
3577: disadvantage: If the user does not leave the method via
3578: @code{;m}, but via @code{throw} or @code{exit},
3579: @code{this} is not restored (and @code{exit} may
3580: crash). To deal with the @code{throw} problem, I have redefined
3581: @code{catch} to save and restore @code{this}; the same
3582: should be done with any word that can catch an exception. As for
3583: @code{exit}, I simply forbid it (as a replacement, there is
3584: @code{exitm}).
3585:
3586: @cindex @code{inst-var} implementation
3587: @code{inst-var} is just the same as @code{field}, with
3588: a different @code{does>} action:
3589: @example
3590: @@ this +
3591: @end example
3592: Similar for @code{inst-value}.
3593:
3594: @cindex class scoping implementation
3595: Each class also has a wordlist that contains the words defined with
3596: @code{inst-var} and @code{inst-value}, and its protected
3597: words. It also has a pointer to its parent. @code{class} pushes
3598: the wordlists of the class an all its ancestors on the search order,
3599: and @code{end-class} drops them.
3600:
3601: @cindex interface implementation
3602: An interface is like a class without fields, parent and protected
3603: words; i.e., it just has a method map. If a class implements an
3604: interface, its method map contains a pointer to the method map of the
3605: interface. The positive offsets in the map are reserved for class
3606: methods, therefore interface map pointers have negative
3607: offsets. Interfaces have offsets that are unique throughout the
3608: system, unlike class selectors, whose offsets are only unique for the
3609: classes where the selector is available (invokable).
3610:
3611: This structure means that interface selectors have to perform one
3612: indirection more than class selectors to find their method. Their body
3613: contains the interface map pointer offset in the class method map, and
3614: the method offset in the interface method map. The
3615: @code{does>} action for an interface selector is, basically:
3616:
3617: @example
3618: ( object selector-body )
3619: 2dup selector-interface @@ ( object selector-body object interface-offset )
3620: swap object-map @@ + @@ ( object selector-body map )
3621: swap selector-offset @@ + @@ execute
3622: @end example
3623:
3624: where @code{object-map} and @code{selector-offset} are
3625: first fields and generate no code.
3626:
3627: As a concrete example, consider the following code:
3628:
3629: @example
3630: interface
3631: selector if1sel1
3632: selector if1sel2
3633: end-interface if1
3634:
3635: object class
3636: if1 implementation
3637: selector cl1sel1
3638: cell% inst-var cl1iv1
3639:
3640: ' m1 overrides construct
3641: ' m2 overrides if1sel1
3642: ' m3 overrides if1sel2
3643: ' m4 overrides cl1sel2
3644: end-class cl1
3645:
3646: create obj1 object dict-new drop
3647: create obj2 cl1 dict-new drop
3648: @end example
3649:
3650: The data structure created by this code (including the data structure
3651: for @code{object}) is shown in the <a
3652: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
3653:
3654: @node Comparison with other object models, Objects Glossary, Objects Implementation, Objects
3655: @subsection Comparison with other object models
3656: @cindex comparison of object models
3657: @cindex object models, comparison
3658:
3659: Many object-oriented Forth extensions have been proposed (@cite{A survey
3660: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
3661: J. Rodriguez and W. F. S. Poehlman lists 17). Here I'll discuss the
3662: relation of @file{objects.fs} to two well-known and two closely-related
3663: (by the use of method maps) models.
3664:
3665: @cindex Neon model
3666: The most popular model currently seems to be the Neon model (see
3667: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
3668: 1997) by Andrew McKewan). The Neon model uses a @code{@emph{selector
3669: object}} syntax, which makes it unnatural to pass objects on the
3670: stack. It also requires that the selector parses the input stream (at
3671: compile time); this leads to reduced extensibility and to bugs that are
3672: hard to find. Finally, it allows using every selector to every object;
3673: this eliminates the need for classes, but makes it harder to create
3674: efficient implementations. A longer version of this critique can be
3675: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
3676: Dimensions, May 1997) by Anton Ertl.
3677:
3678: @cindex Pountain's object-oriented model
3679: Another well-known publication is @cite{Object-Oriented Forth} (Academic
3680: Press, London, 1987) by Dick Pountain. However, it is not really about
3681: object-oriented programming, because it hardly deals with late
3682: binding. Instead, it focuses on features like information hiding and
3683: overloading that are characteristic of modular languages like Ada (83).
3684:
3685: @cindex Zsoter's object-oriented model
3686: In @cite{Does late binding have to be slow?} (Forth Dimensions ??? 1996)
3687: Andras Zsoter describes a model that makes heavy use of an active object
3688: (like @code{this} in @file{objects.fs}): The active object is not only
3689: used for accessing all fields, but also specifies the receiving object
3690: of every selector invocation; you have to change the active object
3691: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
3692: changes more or less implicitly at @code{m: ... ;m}. Such a change at
3693: the method entry point is unnecessary with the Zsoter's model, because
3694: the receiving object is the active object already; OTOH, the explicit
3695: change is absolutely necessary in that model, because otherwise no one
3696: could ever change the active object. An ANS Forth implementation of this
3697: model is available at @url{http://www.forth.org/fig/oopf.html}.
3698:
3699: @cindex @file{oof.fs} object model
3700: The @file{oof.fs} model combines information hiding and overloading
3701: resolution (by keeping names in various wordlists) with object-oriented
3702: programming. It sets the active object implicitly on method entry, but
3703: also allows explicit changing (with @code{>o...o>} or with
3704: @code{with...endwith}). It uses parsing and state-smart objects and
3705: classes for resolving overloading and for early binding: the object or
3706: class parses the selector and determines the method from this. If the
3707: selector is not parsed by an object or class, it performs a call to the
3708: selector for the active object (late binding), like Zsoter's model.
3709: Fields are always accessed through the active object. The big
3710: disadvantage of this model is the parsing and the state-smartness, which
3711: reduces extensibility and increases the opportunities for subtle bugs;
3712: essentially, you are only safe if you never tick or @code{postpone} an
3713: object or class.
3714:
3715: @node Objects Glossary, , Comparison with other object models, Objects
3716: @subsection @file{objects.fs} Glossary
3717: @cindex @file{objects.fs} Glossary
3718:
3719: doc-bind
3720: doc-<bind>
3721: doc-bind'
3722: doc-[bind]
3723: doc-class
3724: doc-class->map
3725: doc-class-inst-size
3726: doc-class-override!
3727: doc-construct
3728: doc-current'
3729: doc-[current]
3730: doc-current-interface
3731: doc-dict-new
3732: doc-drop-order
3733: doc-end-class
3734: doc-end-class-noname
3735: doc-end-interface
3736: doc-end-interface-noname
3737: doc-exitm
3738: doc-heap-new
3739: doc-implementation
3740: doc-init-object
3741: doc-inst-value
3742: doc-inst-var
3743: doc-interface
3744: doc-;m
3745: doc-m:
3746: doc-method
3747: doc-object
3748: doc-overrides
3749: doc-[parent]
3750: doc-print
3751: doc-protected
3752: doc-public
3753: doc-push-order
3754: doc-selector
3755: doc-this
3756: doc-<to-inst>
3757: doc-[to-inst]
3758: doc-to-this
3759: doc-xt-new
3760:
3761: @c -------------------------------------------------------------
3762: @node Object Oriented Forth, Tokens for Words, Objects, Words
3763: @section Object oriented Forth
3764: @cindex oof
3765: @cindex object-oriented programming
3766:
3767: @cindex @file{objects.fs}
3768: @cindex @file{oof.fs}
3769: Gforth comes with two packets for object-oriented programming,
3770: @file{objects.fs} and @file{oof.fs}; none of them is preloaded, so you
3771: have to @code{include} them before use. This section describes the
3772: @file{oof.fs} packet. Both packets are written in ANS Forth and can be
3773: used with any other standard Forth (@pxref{Objects}). This section uses
3774: the same rationale why using object-oriented programming, and the same
3775: terminology.
3776:
3777: The packet described in this section is used in bigFORTH since 1991, and
3778: used for two large applications: a chromatographic system used to
3779: create new medicaments, and a graphic user interface library (MINOS).
3780:
3781: @menu
3782: * Properties of the OOF model::
3783: * Basic OOF Usage::
3784: * The base class object::
3785: * Class Declaration::
3786: * Class Implementation::
3787: @end menu
3788:
3789: @node Properties of the OOF model, Basic OOF Usage, Object Oriented Forth, Object Oriented Forth
3790: @subsection Properties of the OOF model
3791: @cindex @file{oof.fs} properties
3792:
3793: @itemize @bullet
3794: @item
3795: This model combines object oriented programming with information
3796: hiding. It helps you writing large application, where scoping is
3797: necessary, because it provides class-oriented scoping.
3798:
3799: @item
3800: Named objects, object pointers, and object arrays can be created,
3801: selector invocation uses the "object selector" syntax. Selector invocation
3802: to objects and/or selectors on the stack is a bit less convenient, but
3803: possible.
3804:
3805: @item
3806: Selector invocation and instance variable usage of the active object is
3807: straight forward, since both make use of the active object.
3808:
3809: @item
3810: Late binding is efficient and easy to use.
3811:
3812: @item
3813: State-smart objects parse selectors. However, extensibility is provided
3814: using a (parsing) selector @code{postpone} and a selector @code{'}.
3815:
3816: @item
3817: An implementation in ANS Forth is available.
3818:
3819: @end itemize
3820:
3821:
3822: @node Basic OOF Usage, The base class object, Properties of the OOF model, Object Oriented Forth
3823: @subsection Basic OOF Usage
3824: @cindex @file{oof.fs} usage
3825:
3826: Here, I use the same example as for @code{objects} (@pxref{Basic Objects Usage}).
3827:
3828: You can define a class for graphical objects like this:
3829:
3830: @cindex @code{class} usage
3831: @cindex @code{class;} usage
3832: @cindex @code{method} usage
3833: @example
3834: object class graphical \ "object" is the parent class
3835: method draw ( x y graphical -- )
3836: class;
3837: @end example
3838:
3839: This code defines a class @code{graphical} with an
3840: operation @code{draw}. We can perform the operation
3841: @code{draw} on any @code{graphical} object, e.g.:
3842:
3843: @example
3844: 100 100 t-rex draw
3845: @end example
3846:
3847: where @code{t-rex} is an object or object pointer, created with e.g.
3848: @code{graphical : trex}.
3849:
3850: @cindex abstract class
3851: How do we create a graphical object? With the present definitions,
3852: we cannot create a useful graphical object. The class
3853: @code{graphical} describes graphical objects in general, but not
3854: any concrete graphical object type (C++ users would call it an
3855: @emph{abstract class}); e.g., there is no method for the selector
3856: @code{draw} in the class @code{graphical}.
3857:
3858: For concrete graphical objects, we define child classes of the
3859: class @code{graphical}, e.g.:
3860:
3861: @example
3862: graphical class circle \ "graphical" is the parent class
3863: cell var circle-radius
3864: how:
3865: : draw ( x y -- )
3866: circle-radius @@ draw-circle ;
3867:
3868: : init ( n-radius -- (
3869: circle-radius ! ;
3870: class;
3871: @end example
3872:
3873: Here we define a class @code{circle} as a child of @code{graphical},
3874: with a field @code{circle-radius}; it defines new methods for the
3875: selectors @code{draw} and @code{init} (@code{init} is defined in
3876: @code{object}, the parent class of @code{graphical}).
3877:
3878: Now we can create a circle in the dictionary with
3879:
3880: @example
3881: 50 circle : my-circle
3882: @end example
3883:
3884: @code{:} invokes @code{init}, thus initializing the field
3885: @code{circle-radius} with 50. We can draw this new circle at (100,100)
3886: with
3887:
3888: @example
3889: 100 100 my-circle draw
3890: @end example
3891:
3892: @cindex selector invocation, restrictions
3893: @cindex class definition, restrictions
3894: Note: You can invoke a selector only if the receiving object belongs to
3895: the class where the selector was defined or one of its descendents;
3896: e.g., you can invoke @code{draw} only for objects belonging to
3897: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
3898: mechanism will check if you try to invoke a selector that is not
3899: defined in this class hierarchy, so you'll get an error at compilation
3900: time.
3901:
3902:
3903: @node The base class object, Class Declaration, Basic OOF Usage, Object Oriented Forth
3904: @subsection The base class @file{object}
3905: @cindex @file{oof.fs} base class
3906:
3907: When you define a class, you have to specify a parent class. So how do
3908: you start defining classes? There is one class available from the start:
3909: @code{object}. You have to use it as ancestor for all classes. It is the
3910: only class that has no parent. Classes are also objects, except that
3911: they don't have instance variables; class manipulation such as
3912: inheritance or changing definitions of a class is handled through
3913: selectors of the class @code{object}.
3914:
3915: @code{object} provides a number of selectors:
3916:
3917: @itemize @bullet
3918: @item
3919: @code{class} for subclassing, @code{definitions} to add definitions
3920: later on, and @code{class?} to get type informations (is the class a
3921: subclass of the class passed on the stack?).
3922: doc---object-class
3923: doc---object-definitions
3924: doc---object-class?
3925:
3926: @item
3927: @code{init} and @code{dispose} as constructor and destroctor of the
3928: object. @code{init} is invocated after the object's memory is allocated,
3929: while @code{dispose} also handles deallocation. Thus if you redefine
3930: @code{dispose}, you have to call the parent's dispose with @code{super
3931: dispose}, too.
3932: doc---object-init
3933: doc---object-dispose
3934:
3935: @item
3936: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
3937: @code{[]} to create named and unnamed objects and object arrays or
3938: object pointers.
3939: doc---object-new
3940: doc---object-new[]
3941: doc---object-:
3942: doc---object-ptr
3943: doc---object-asptr
3944: doc---object-[]
3945:
3946: @item
3947: @code{::} and @code{super} for expicit scoping. You should use expicit
3948: scoping only for super classes or classes with the same set of instance
3949: variables. Explicit scoped selectors use early binding.
3950: doc---object-::
3951: doc---object-super
3952:
3953: @item
3954: @code{self} to get the address of the object
3955: doc---object-self
3956:
3957: @item
3958: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
3959: pointers and instance defers.
3960: doc---object-bind
3961: doc---object-bound
3962: doc---object-link
3963: doc---object-is
3964:
3965: @item
3966: @code{'} to obtain selector tokens, @code{send} to invocate selectors
3967: form the stack, and @code{postpone} to generate selector invocation code.
3968: doc---object-'
3969: doc---object-postpone
3970:
3971: @item
3972: @code{with} and @code{endwith} to select the active object from the
3973: stack, and enabling it's scope. Using @code{with} and @code{endwith}
3974: also allows to create code using selector @code{postpone} without being
3975: trapped bye the state-smart objects.
3976: doc---object-with
3977: doc---object-endwith
3978:
3979: @end itemize
3980:
3981: @node Class Declaration, Class Implementation, The base class object, Object Oriented Forth
3982: @subsection Class Declaration
3983: @cindex class declaration
3984:
3985: @itemize @bullet
3986: @item
3987: Instance variables
3988: doc---oof-var
3989:
3990: @item
3991: Object pointers
3992: doc---oof-ptr
3993: doc---oof-asptr
3994:
3995: @item
3996: Instance defers
3997: doc---oof-defer
3998:
3999: @item
4000: Method selectors
4001: doc---oof-early
4002: doc---oof-method
4003:
4004: @item
4005: Class wide variables
4006: doc---oof-static
4007:
4008: @item
4009: End declaration
4010: doc---oof-how:
4011: doc---oof-class;
4012:
4013: @end itemize
4014:
4015: @node Class Implementation, , Class Declaration, Object Oriented Forth
4016: @subsection Class Implementation
4017: @cindex class implementation
4018:
4019: @c -------------------------------------------------------------
4020: @node Tokens for Words, Wordlists, Object Oriented Forth, Words
4021: @section Tokens for Words
4022: @cindex tokens for words
4023:
4024: This chapter describes the creation and use of tokens that represent
4025: words on the stack (and in data space).
4026:
4027: Named words have interpretation and compilation semantics. Unnamed words
4028: just have execution semantics.
4029:
4030: @cindex execution token
4031: An @dfn{execution token} represents the execution semantics of an
4032: unnamed word. An execution token occupies one cell. As explained in
4033: section @ref{Supplying names}, the execution token of the last words
4034: defined can be produced with
4035:
4036: short-lastxt
4037:
4038: You can perform the semantics represented by an execution token with
4039: doc-execute
4040: You can compile the word with
4041: doc-compile,
4042:
4043: @cindex code field address
4044: @cindex CFA
4045: In Gforth, the abstract data type @emph{execution token} is implemented
4046: as CFA (code field address).
4047:
4048: The interpretation semantics of a named word are also represented by an
4049: execution token. You can get it with
4050:
4051: doc-[']
4052: doc-'
4053:
4054: For literals, you use @code{'} in interpreted code and @code{[']} in
4055: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
4056: by complaining about compile-only words. To get an execution token for a
4057: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
4058: @var{X} drop}.
4059:
4060: @cindex compilation token
4061: The compilation semantics are represented by a @dfn{compilation token}
4062: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
4063: execution token. The compilation semantics represented by the
4064: compilation token can be performed with @code{execute}, which consumes
4065: the whole compilation token, with an additional stack effect determined
4066: by the represented compilation semantics.
4067:
4068: doc-[comp']
4069: doc-comp'
4070:
4071: You can compile the compilation semantics with @code{postpone,}. I.e.,
4072: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
4073: @var{word}}.
4074:
4075: doc-postpone,
4076:
4077: At present, the @var{w} part of a compilation token is an execution
4078: token, and the @var{xt} part represents either @code{execute} or
4079: @code{compile,}. However, don't rely on that knowledge, unless necessary;
4080: we may introduce unusual compilation tokens in the future (e.g.,
4081: compilation tokens representing the compilation semantics of literals).
4082:
4083: @cindex name token
4084: @cindex name field address
4085: @cindex NFA
4086: Named words are also represented by the @dfn{name token}. The abstract
4087: data type @emph{name token} is implemented as NFA (name field address).
4088:
4089: doc-find-name
4090: doc-name>int
4091: doc-name?int
4092: doc-name>comp
4093: doc-name>string
4094:
4095: @node Wordlists, Files, Tokens for Words, Words
4096: @section Wordlists
4097:
4098: @node Files, Blocks, Wordlists, Words
4099: @section Files
4100:
4101: @node Blocks, Other I/O, Files, Words
4102: @section Blocks
4103:
4104: @node Other I/O, Programming Tools, Blocks, Words
4105: @section Other I/O
4106:
4107: @node Programming Tools, Assembler and Code Words, Other I/O, Words
4108: @section Programming Tools
4109: @cindex programming tools
4110:
4111: @menu
4112: * Debugging:: Simple and quick.
4113: * Assertions:: Making your programs self-checking.
4114: * Singlestep Debugger:: Executing your program word by word.
4115: @end menu
4116:
4117: @node Debugging, Assertions, Programming Tools, Programming Tools
4118: @subsection Debugging
4119: @cindex debugging
4120:
4121: The simple debugging aids provided in @file{debugs.fs}
4122: are meant to support a different style of debugging than the
4123: tracing/stepping debuggers used in languages with long turn-around
4124: times.
4125:
4126: A much better (faster) way in fast-compiling languages is to add
4127: printing code at well-selected places, let the program run, look at
4128: the output, see where things went wrong, add more printing code, etc.,
4129: until the bug is found.
4130:
4131: The word @code{~~} is easy to insert. It just prints debugging
4132: information (by default the source location and the stack contents). It
4133: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
4134: query-replace them with nothing). The deferred words
4135: @code{printdebugdata} and @code{printdebugline} control the output of
4136: @code{~~}. The default source location output format works well with
4137: Emacs' compilation mode, so you can step through the program at the
4138: source level using @kbd{C-x `} (the advantage over a stepping debugger
4139: is that you can step in any direction and you know where the crash has
4140: happened or where the strange data has occurred).
4141:
4142: Note that the default actions clobber the contents of the pictured
4143: numeric output string, so you should not use @code{~~}, e.g., between
4144: @code{<#} and @code{#>}.
4145:
4146: doc-~~
4147: doc-printdebugdata
4148: doc-printdebugline
4149:
4150: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
4151: @subsection Assertions
4152: @cindex assertions
4153:
4154: It is a good idea to make your programs self-checking, in particular, if
4155: you use an assumption (e.g., that a certain field of a data structure is
4156: never zero) that may become wrong during maintenance. Gforth supports
4157: assertions for this purpose. They are used like this:
4158:
4159: @example
4160: assert( @var{flag} )
4161: @end example
4162:
4163: The code between @code{assert(} and @code{)} should compute a flag, that
4164: should be true if everything is alright and false otherwise. It should
4165: not change anything else on the stack. The overall stack effect of the
4166: assertion is @code{( -- )}. E.g.
4167:
4168: @example
4169: assert( 1 1 + 2 = ) \ what we learn in school
4170: assert( dup 0<> ) \ assert that the top of stack is not zero
4171: assert( false ) \ this code should not be reached
4172: @end example
4173:
4174: The need for assertions is different at different times. During
4175: debugging, we want more checking, in production we sometimes care more
4176: for speed. Therefore, assertions can be turned off, i.e., the assertion
4177: becomes a comment. Depending on the importance of an assertion and the
4178: time it takes to check it, you may want to turn off some assertions and
4179: keep others turned on. Gforth provides several levels of assertions for
4180: this purpose:
4181:
4182: doc-assert0(
4183: doc-assert1(
4184: doc-assert2(
4185: doc-assert3(
4186: doc-assert(
4187: doc-)
4188:
4189: @code{Assert(} is the same as @code{assert1(}. The variable
4190: @code{assert-level} specifies the highest assertions that are turned
4191: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
4192: @code{assert1(} assertions perform checking, while @code{assert2(} and
4193: @code{assert3(} assertions are treated as comments.
4194:
4195: Note that the @code{assert-level} is evaluated at compile-time, not at
4196: run-time. I.e., you cannot turn assertions on or off at run-time, you
4197: have to set the @code{assert-level} appropriately before compiling a
4198: piece of code. You can compile several pieces of code at several
4199: @code{assert-level}s (e.g., a trusted library at level 1 and newly
4200: written code at level 3).
4201:
4202: doc-assert-level
4203:
4204: If an assertion fails, a message compatible with Emacs' compilation mode
4205: is produced and the execution is aborted (currently with @code{ABORT"}.
4206: If there is interest, we will introduce a special throw code. But if you
4207: intend to @code{catch} a specific condition, using @code{throw} is
4208: probably more appropriate than an assertion).
4209:
4210: @node Singlestep Debugger, , Assertions, Programming Tools
4211: @subsection Singlestep Debugger
4212: @cindex singlestep Debugger
4213: @cindex debugging Singlestep
4214: @cindex @code{dbg}
4215: @cindex @code{BREAK:}
4216: @cindex @code{BREAK"}
4217:
4218: When a new word is created there's often the need to check whether it behaves
4219: correctly or not. You can do this by typing @code{dbg badword}. This might
4220: look like:
4221: @example
4222: : badword 0 DO i . LOOP ; ok
4223: 2 dbg badword
4224: : badword
4225: Scanning code...
4226:
4227: Nesting debugger ready!
4228:
4229: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
4230: 400D4740 8049F68 DO -> [ 0 ]
4231: 400D4744 804A0C8 i -> [ 1 ] 00000
4232: 400D4748 400C5E60 . -> 0 [ 0 ]
4233: 400D474C 8049D0C LOOP -> [ 0 ]
4234: 400D4744 804A0C8 i -> [ 1 ] 00001
4235: 400D4748 400C5E60 . -> 1 [ 0 ]
4236: 400D474C 8049D0C LOOP -> [ 0 ]
4237: 400D4758 804B384 ; -> ok
4238: @end example
4239:
4240: Each line displayed is one step. You always have to hit return to
4241: execute the next word that is displayed. If you don't want to execute
4242: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
4243: an overview what keys are available:
4244:
4245: @table @i
4246:
4247: @item <return>
4248: Next; Execute the next word.
4249:
4250: @item n
4251: Nest; Single step through next word.
4252:
4253: @item u
4254: Unnest; Stop debugging and execute rest of word. If we got to this word
4255: with nest, continue debugging with the calling word.
4256:
4257: @item d
4258: Done; Stop debugging and execute rest.
4259:
4260: @item s
4261: Stopp; Abort immediately.
4262:
4263: @end table
4264:
4265: Debugging large application with this mechanism is very difficult, because
4266: you have to nest very deep into the program before the interesting part
4267: begins. This takes a lot of time.
4268:
4269: To do it more directly put a @code{BREAK:} command into your source code.
4270: When program execution reaches @code{BREAK:} the single step debugger is
4271: invoked and you have all the features described above.
4272:
4273: If you have more than one part to debug it is useful to know where the
4274: program has stopped at the moment. You can do this by the
4275: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
4276: string is typed out when the ``breakpoint'' is reached.
4277:
4278: @node Assembler and Code Words, Threading Words, Programming Tools, Words
4279: @section Assembler and Code Words
4280: @cindex assembler
4281: @cindex code words
4282:
4283: Gforth provides some words for defining primitives (words written in
4284: machine code), and for defining the the machine-code equivalent of
4285: @code{DOES>}-based defining words. However, the machine-independent
4286: nature of Gforth poses a few problems: First of all, Gforth runs on
4287: several architectures, so it can provide no standard assembler. What's
4288: worse is that the register allocation not only depends on the processor,
4289: but also on the @code{gcc} version and options used.
4290:
4291: The words that Gforth offers encapsulate some system dependences (e.g., the
4292: header structure), so a system-independent assembler may be used in
4293: Gforth. If you do not have an assembler, you can compile machine code
4294: directly with @code{,} and @code{c,}.
4295:
4296: doc-assembler
4297: doc-code
4298: doc-end-code
4299: doc-;code
4300: doc-flush-icache
4301:
4302: If @code{flush-icache} does not work correctly, @code{code} words
4303: etc. will not work (reliably), either.
4304:
4305: These words are rarely used. Therefore they reside in @code{code.fs},
4306: which is usually not loaded (except @code{flush-icache}, which is always
4307: present). You can load them with @code{require code.fs}.
4308:
4309: @cindex registers of the inner interpreter
4310: In the assembly code you will want to refer to the inner interpreter's
4311: registers (e.g., the data stack pointer) and you may want to use other
4312: registers for temporary storage. Unfortunately, the register allocation
4313: is installation-dependent.
4314:
4315: The easiest solution is to use explicit register declarations
4316: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
4317: GNU C Manual}) for all of the inner interpreter's registers: You have to
4318: compile Gforth with @code{-DFORCE_REG} (configure option
4319: @code{--enable-force-reg}) and the appropriate declarations must be
4320: present in the @code{machine.h} file (see @code{mips.h} for an example;
4321: you can find a full list of all declarable register symbols with
4322: @code{grep register engine.c}). If you give explicit registers to all
4323: variables that are declared at the beginning of @code{engine()}, you
4324: should be able to use the other caller-saved registers for temporary
4325: storage. Alternatively, you can use the @code{gcc} option
4326: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
4327: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
4328: (however, this restriction on register allocation may slow Gforth
4329: significantly).
4330:
4331: If this solution is not viable (e.g., because @code{gcc} does not allow
4332: you to explicitly declare all the registers you need), you have to find
4333: out by looking at the code where the inner interpreter's registers
4334: reside and which registers can be used for temporary storage. You can
4335: get an assembly listing of the engine's code with @code{make engine.s}.
4336:
4337: In any case, it is good practice to abstract your assembly code from the
4338: actual register allocation. E.g., if the data stack pointer resides in
4339: register @code{$17}, create an alias for this register called @code{sp},
4340: and use that in your assembly code.
4341:
4342: @cindex code words, portable
4343: Another option for implementing normal and defining words efficiently
4344: is: adding the wanted functionality to the source of Gforth. For normal
4345: words you just have to edit @file{primitives} (@pxref{Automatic
4346: Generation}), defining words (equivalent to @code{;CODE} words, for fast
4347: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
4348: @file{prims2x.fs}, and possibly @file{cross.fs}.
4349:
4350:
4351: @node Threading Words, Including Files, Assembler and Code Words, Words
4352: @section Threading Words
4353: @cindex threading words
4354:
4355: @cindex code address
4356: These words provide access to code addresses and other threading stuff
4357: in Gforth (and, possibly, other interpretive Forths). It more or less
4358: abstracts away the differences between direct and indirect threading
4359: (and, for direct threading, the machine dependences). However, at
4360: present this wordset is still incomplete. It is also pretty low-level;
4361: some day it will hopefully be made unnecessary by an internals wordset
4362: that abstracts implementation details away completely.
4363:
4364: doc->code-address
4365: doc->does-code
4366: doc-code-address!
4367: doc-does-code!
4368: doc-does-handler!
4369: doc-/does-handler
4370:
4371: The code addresses produced by various defining words are produced by
4372: the following words:
4373:
4374: doc-docol:
4375: doc-docon:
4376: doc-dovar:
4377: doc-douser:
4378: doc-dodefer:
4379: doc-dofield:
4380:
4381: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
4382: with @code{>DOES-CODE}. If the word was defined in that way, the value
4383: returned is different from 0 and identifies the @code{DOES>} used by the
4384: defining word.
4385:
4386: @node Including Files, Include and Require, Threading Words, Words
4387: @section Including Files
4388: @cindex including files
4389:
4390: @menu
4391: * Include and Require::
4392: * Path Handling::
4393: @end menu
4394:
4395: @node Include and Require, Path Handling, Including Files, Including Files
4396: @subsection Include and Requrie
4397:
4398: There a two words to include the source files more intelligently.
4399:
4400: doc-include
4401: doc-require
4402:
4403: @node Path Handling, , Include and Require, Including Files
4404: @subsection Path Handling
4405: @cindex path handling
4406:
4407: In larger program projects it is often neccassary to build up a structured
4408: directory tree. Standard Forth programs are somewhere more central because
4409: they must be accessed from some more other programs. To achieve this it is
4410: possible to manipulate the search path in which Gforth tries to find the
4411: source file.
4412:
4413: doc-fpath+
4414: doc-fpath=
4415: doc-.fpath
4416:
4417: Using fpath and require would look like:
4418:
4419: @example
4420:
4421: fpath= /usr/lib/forth/|./
4422:
4423: require timer.fs
4424:
4425: ...
4426:
4427: @end example
4428:
4429: @cindex ~+
4430: There is another nice feature which is similar to C's @code{include <...>}
4431: and @code{include "..."}. For example: You have a program seperated into
4432: several files in a subdirectory and you want to include some other files
4433: in this subdirectory from within the program. You have to tell Gforth that
4434: you are now looking relative to the directory the current file comes from.
4435: You can tell this Gforth by using the prefix @code{~+/} in front of the
4436: filename. It is also possible to add it to the search path.
4437:
4438: If you have the need to look for a file in the Forth search path, you could
4439: use this Gforth feature in your application.
4440:
4441: doc-open-fpath-file
4442:
4443: It is even possible to use your own search paths. Create a search path like
4444: this:
4445:
4446: @example
4447:
4448: Make a buffer for the path:
4449: create mypath 100 chars , \ maximum length (is checked)
4450: 0 , \ real len
4451: 100 chars allot \ space for path
4452:
4453: @end example
4454:
4455: You have the same functions for the forth search path in an generic version
4456: for different pathes.
4457:
4458: doc-path+
4459: doc-path=
4460: doc-.path
4461: doc-open-path-file
4462:
4463: @c ******************************************************************
4464: @node Tools, ANS conformance, Words, Top
4465: @chapter Tools
4466:
4467: @menu
4468: * ANS Report:: Report the words used, sorted by wordset.
4469: @end menu
4470:
4471: See also @ref{Emacs and Gforth}.
4472:
4473: @node ANS Report, , Tools, Tools
4474: @section @file{ans-report.fs}: Report the words used, sorted by wordset
4475: @cindex @file{ans-report.fs}
4476: @cindex report the words used in your program
4477: @cindex words used in your program
4478:
4479: If you want to label a Forth program as ANS Forth Program, you must
4480: document which wordsets the program uses; for extension wordsets, it is
4481: helpful to list the words the program requires from these wordsets
4482: (because Forth systems are allowed to provide only some words of them).
4483:
4484: The @file{ans-report.fs} tool makes it easy for you to determine which
4485: words from which wordset and which non-ANS words your application
4486: uses. You simply have to include @file{ans-report.fs} before loading the
4487: program you want to check. After loading your program, you can get the
4488: report with @code{print-ans-report}. A typical use is to run this as
4489: batch job like this:
4490: @example
4491: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
4492: @end example
4493:
4494: The output looks like this (for @file{compat/control.fs}):
4495: @example
4496: The program uses the following words
4497: from CORE :
4498: : POSTPONE THEN ; immediate ?dup IF 0=
4499: from BLOCK-EXT :
4500: \
4501: from FILE :
4502: (
4503: @end example
4504:
4505: @subsection Caveats
4506:
4507: Note that @file{ans-report.fs} just checks which words are used, not whether
4508: they are used in an ANS Forth conforming way!
4509:
4510: Some words are defined in several wordsets in the
4511: standard. @file{ans-report.fs} reports them for only one of the
4512: wordsets, and not necessarily the one you expect. It depends on usage
4513: which wordset is the right one to specify. E.g., if you only use the
4514: compilation semantics of @code{S"}, it is a Core word; if you also use
4515: its interpretation semantics, it is a File word.
4516:
4517: @c ******************************************************************
4518: @node ANS conformance, Model, Tools, Top
4519: @chapter ANS conformance
4520: @cindex ANS conformance of Gforth
4521:
4522: To the best of our knowledge, Gforth is an
4523:
4524: ANS Forth System
4525: @itemize @bullet
4526: @item providing the Core Extensions word set
4527: @item providing the Block word set
4528: @item providing the Block Extensions word set
4529: @item providing the Double-Number word set
4530: @item providing the Double-Number Extensions word set
4531: @item providing the Exception word set
4532: @item providing the Exception Extensions word set
4533: @item providing the Facility word set
4534: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
4535: @item providing the File Access word set
4536: @item providing the File Access Extensions word set
4537: @item providing the Floating-Point word set
4538: @item providing the Floating-Point Extensions word set
4539: @item providing the Locals word set
4540: @item providing the Locals Extensions word set
4541: @item providing the Memory-Allocation word set
4542: @item providing the Memory-Allocation Extensions word set (that one's easy)
4543: @item providing the Programming-Tools word set
4544: @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
4545: @item providing the Search-Order word set
4546: @item providing the Search-Order Extensions word set
4547: @item providing the String word set
4548: @item providing the String Extensions word set (another easy one)
4549: @end itemize
4550:
4551: @cindex system documentation
4552: In addition, ANS Forth systems are required to document certain
4553: implementation choices. This chapter tries to meet these
4554: requirements. In many cases it gives a way to ask the system for the
4555: information instead of providing the information directly, in
4556: particular, if the information depends on the processor, the operating
4557: system or the installation options chosen, or if they are likely to
4558: change during the maintenance of Gforth.
4559:
4560: @comment The framework for the rest has been taken from pfe.
4561:
4562: @menu
4563: * The Core Words::
4564: * The optional Block word set::
4565: * The optional Double Number word set::
4566: * The optional Exception word set::
4567: * The optional Facility word set::
4568: * The optional File-Access word set::
4569: * The optional Floating-Point word set::
4570: * The optional Locals word set::
4571: * The optional Memory-Allocation word set::
4572: * The optional Programming-Tools word set::
4573: * The optional Search-Order word set::
4574: @end menu
4575:
4576:
4577: @c =====================================================================
4578: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
4579: @comment node-name, next, previous, up
4580: @section The Core Words
4581: @c =====================================================================
4582: @cindex core words, system documentation
4583: @cindex system documentation, core words
4584:
4585: @menu
4586: * core-idef:: Implementation Defined Options
4587: * core-ambcond:: Ambiguous Conditions
4588: * core-other:: Other System Documentation
4589: @end menu
4590:
4591: @c ---------------------------------------------------------------------
4592: @node core-idef, core-ambcond, The Core Words, The Core Words
4593: @subsection Implementation Defined Options
4594: @c ---------------------------------------------------------------------
4595: @cindex core words, implementation-defined options
4596: @cindex implementation-defined options, core words
4597:
4598:
4599: @table @i
4600: @item (Cell) aligned addresses:
4601: @cindex cell-aligned addresses
4602: @cindex aligned addresses
4603: processor-dependent. Gforth's alignment words perform natural alignment
4604: (e.g., an address aligned for a datum of size 8 is divisible by
4605: 8). Unaligned accesses usually result in a @code{-23 THROW}.
4606:
4607: @item @code{EMIT} and non-graphic characters:
4608: @cindex @code{EMIT} and non-graphic characters
4609: @cindex non-graphic characters and @code{EMIT}
4610: The character is output using the C library function (actually, macro)
4611: @code{putc}.
4612:
4613: @item character editing of @code{ACCEPT} and @code{EXPECT}:
4614: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
4615: @cindex editing in @code{ACCEPT} and @code{EXPECT}
4616: @cindex @code{ACCEPT}, editing
4617: @cindex @code{EXPECT}, editing
4618: This is modeled on the GNU readline library (@pxref{Readline
4619: Interaction, , Command Line Editing, readline, The GNU Readline
4620: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
4621: producing a full word completion every time you type it (instead of
4622: producing the common prefix of all completions).
4623:
4624: @item character set:
4625: @cindex character set
4626: The character set of your computer and display device. Gforth is
4627: 8-bit-clean (but some other component in your system may make trouble).
4628:
4629: @item Character-aligned address requirements:
4630: @cindex character-aligned address requirements
4631: installation-dependent. Currently a character is represented by a C
4632: @code{unsigned char}; in the future we might switch to @code{wchar_t}
4633: (Comments on that requested).
4634:
4635: @item character-set extensions and matching of names:
4636: @cindex character-set extensions and matching of names
4637: @cindex case sensitivity for name lookup
4638: @cindex name lookup, case sensitivity
4639: @cindex locale and case sensitivity
4640: Any character except the ASCII NUL charcter can be used in a
4641: name. Matching is case-insensitive (except in @code{TABLE}s). The
4642: matching is performed using the C function @code{strncasecmp}, whose
4643: function is probably influenced by the locale. E.g., the @code{C} locale
4644: does not know about accents and umlauts, so they are matched
4645: case-sensitively in that locale. For portability reasons it is best to
4646: write programs such that they work in the @code{C} locale. Then one can
4647: use libraries written by a Polish programmer (who might use words
4648: containing ISO Latin-2 encoded characters) and by a French programmer
4649: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
4650: funny results for some of the words (which ones, depends on the font you
4651: are using)). Also, the locale you prefer may not be available in other
4652: operating systems. Hopefully, Unicode will solve these problems one day.
4653:
4654: @item conditions under which control characters match a space delimiter:
4655: @cindex space delimiters
4656: @cindex control characters as delimiters
4657: If @code{WORD} is called with the space character as a delimiter, all
4658: white-space characters (as identified by the C macro @code{isspace()})
4659: are delimiters. @code{PARSE}, on the other hand, treats space like other
4660: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
4661: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
4662: interpreter (aka text interpreter) by default, treats all white-space
4663: characters as delimiters.
4664:
4665: @item format of the control flow stack:
4666: @cindex control flow stack, format
4667: The data stack is used as control flow stack. The size of a control flow
4668: stack item in cells is given by the constant @code{cs-item-size}. At the
4669: time of this writing, an item consists of a (pointer to a) locals list
4670: (third), an address in the code (second), and a tag for identifying the
4671: item (TOS). The following tags are used: @code{defstart},
4672: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
4673: @code{scopestart}.
4674:
4675: @item conversion of digits > 35
4676: @cindex digits > 35
4677: The characters @code{[\]^_'} are the digits with the decimal value
4678: 36@minus{}41. There is no way to input many of the larger digits.
4679:
4680: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
4681: @cindex @code{EXPECT}, display after end of input
4682: @cindex @code{ACCEPT}, display after end of input
4683: The cursor is moved to the end of the entered string. If the input is
4684: terminated using the @kbd{Return} key, a space is typed.
4685:
4686: @item exception abort sequence of @code{ABORT"}:
4687: @cindex exception abort sequence of @code{ABORT"}
4688: @cindex @code{ABORT"}, exception abort sequence
4689: The error string is stored into the variable @code{"error} and a
4690: @code{-2 throw} is performed.
4691:
4692: @item input line terminator:
4693: @cindex input line terminator
4694: @cindex line terminator on input
4695: @cindex newline charcter on input
4696: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
4697: lines. One of these characters is typically produced when you type the
4698: @kbd{Enter} or @kbd{Return} key.
4699:
4700: @item maximum size of a counted string:
4701: @cindex maximum size of a counted string
4702: @cindex counted string, maximum size
4703: @code{s" /counted-string" environment? drop .}. Currently 255 characters
4704: on all ports, but this may change.
4705:
4706: @item maximum size of a parsed string:
4707: @cindex maximum size of a parsed string
4708: @cindex parsed string, maximum size
4709: Given by the constant @code{/line}. Currently 255 characters.
4710:
4711: @item maximum size of a definition name, in characters:
4712: @cindex maximum size of a definition name, in characters
4713: @cindex name, maximum length
4714: 31
4715:
4716: @item maximum string length for @code{ENVIRONMENT?}, in characters:
4717: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
4718: @cindex @code{ENVIRONMENT?} string length, maximum
4719: 31
4720:
4721: @item method of selecting the user input device:
4722: @cindex user input device, method of selecting
4723: The user input device is the standard input. There is currently no way to
4724: change it from within Gforth. However, the input can typically be
4725: redirected in the command line that starts Gforth.
4726:
4727: @item method of selecting the user output device:
4728: @cindex user output device, method of selecting
4729: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
4730: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
4731: output, so output on a terminal does not become visible before the next
4732: newline or buffer overflow. Output on non-terminals is invisible until
4733: the buffer overflows.
4734:
4735: @item methods of dictionary compilation:
4736: What are we expected to document here?
4737:
4738: @item number of bits in one address unit:
4739: @cindex number of bits in one address unit
4740: @cindex address unit, size in bits
4741: @code{s" address-units-bits" environment? drop .}. 8 in all current
4742: ports.
4743:
4744: @item number representation and arithmetic:
4745: @cindex number representation and arithmetic
4746: Processor-dependent. Binary two's complement on all current ports.
4747:
4748: @item ranges for integer types:
4749: @cindex ranges for integer types
4750: @cindex integer types, ranges
4751: Installation-dependent. Make environmental queries for @code{MAX-N},
4752: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
4753: unsigned (and positive) types is 0. The lower bound for signed types on
4754: two's complement and one's complement machines machines can be computed
4755: by adding 1 to the upper bound.
4756:
4757: @item read-only data space regions:
4758: @cindex read-only data space regions
4759: @cindex data-space, read-only regions
4760: The whole Forth data space is writable.
4761:
4762: @item size of buffer at @code{WORD}:
4763: @cindex size of buffer at @code{WORD}
4764: @cindex @code{WORD} buffer size
4765: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
4766: shared with the pictured numeric output string. If overwriting
4767: @code{PAD} is acceptable, it is as large as the remaining dictionary
4768: space, although only as much can be sensibly used as fits in a counted
4769: string.
4770:
4771: @item size of one cell in address units:
4772: @cindex cell size
4773: @code{1 cells .}.
4774:
4775: @item size of one character in address units:
4776: @cindex char size
4777: @code{1 chars .}. 1 on all current ports.
4778:
4779: @item size of the keyboard terminal buffer:
4780: @cindex size of the keyboard terminal buffer
4781: @cindex terminal buffer, size
4782: Varies. You can determine the size at a specific time using @code{lp@@
4783: tib - .}. It is shared with the locals stack and TIBs of files that
4784: include the current file. You can change the amount of space for TIBs
4785: and locals stack at Gforth startup with the command line option
4786: @code{-l}.
4787:
4788: @item size of the pictured numeric output buffer:
4789: @cindex size of the pictured numeric output buffer
4790: @cindex pictured numeric output buffer, size
4791: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
4792: shared with @code{WORD}.
4793:
4794: @item size of the scratch area returned by @code{PAD}:
4795: @cindex size of the scratch area returned by @code{PAD}
4796: @cindex @code{PAD} size
4797: The remainder of dictionary space. @code{unused pad here - - .}.
4798:
4799: @item system case-sensitivity characteristics:
4800: @cindex case-sensitivity characteristics
4801: Dictionary searches are case insensitive (except in
4802: @code{TABLE}s). However, as explained above under @i{character-set
4803: extensions}, the matching for non-ASCII characters is determined by the
4804: locale you are using. In the default @code{C} locale all non-ASCII
4805: characters are matched case-sensitively.
4806:
4807: @item system prompt:
4808: @cindex system prompt
4809: @cindex prompt
4810: @code{ ok} in interpret state, @code{ compiled} in compile state.
4811:
4812: @item division rounding:
4813: @cindex division rounding
4814: installation dependent. @code{s" floored" environment? drop .}. We leave
4815: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
4816: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
4817:
4818: @item values of @code{STATE} when true:
4819: @cindex @code{STATE} values
4820: -1.
4821:
4822: @item values returned after arithmetic overflow:
4823: On two's complement machines, arithmetic is performed modulo
4824: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
4825: arithmetic (with appropriate mapping for signed types). Division by zero
4826: typically results in a @code{-55 throw} (Floating-point unidentified
4827: fault), although a @code{-10 throw} (divide by zero) would be more
4828: appropriate.
4829:
4830: @item whether the current definition can be found after @t{DOES>}:
4831: @cindex @t{DOES>}, visibility of current definition
4832: No.
4833:
4834: @end table
4835:
4836: @c ---------------------------------------------------------------------
4837: @node core-ambcond, core-other, core-idef, The Core Words
4838: @subsection Ambiguous conditions
4839: @c ---------------------------------------------------------------------
4840: @cindex core words, ambiguous conditions
4841: @cindex ambiguous conditions, core words
4842:
4843: @table @i
4844:
4845: @item a name is neither a word nor a number:
4846: @cindex name not found
4847: @cindex Undefined word
4848: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
4849: preserves the data and FP stack, so you don't lose more work than
4850: necessary.
4851:
4852: @item a definition name exceeds the maximum length allowed:
4853: @cindex Word name too long
4854: @code{-19 throw} (Word name too long)
4855:
4856: @item addressing a region not inside the various data spaces of the forth system:
4857: @cindex Invalid memory address
4858: The stacks, code space and name space are accessible. Machine code space is
4859: typically readable. Accessing other addresses gives results dependent on
4860: the operating system. On decent systems: @code{-9 throw} (Invalid memory
4861: address).
4862:
4863: @item argument type incompatible with parameter:
4864: @cindex Argument type mismatch
4865: This is usually not caught. Some words perform checks, e.g., the control
4866: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
4867: mismatch).
4868:
4869: @item attempting to obtain the execution token of a word with undefined execution semantics:
4870: @cindex Interpreting a compile-only word, for @code{'} etc.
4871: @cindex execution token of words with undefined execution semantics
4872: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
4873: get an execution token for @code{compile-only-error} (which performs a
4874: @code{-14 throw} when executed).
4875:
4876: @item dividing by zero:
4877: @cindex dividing by zero
4878: @cindex floating point unidentified fault, integer division
4879: @cindex divide by zero
4880: typically results in a @code{-55 throw} (floating point unidentified
4881: fault), although a @code{-10 throw} (divide by zero) would be more
4882: appropriate.
4883:
4884: @item insufficient data stack or return stack space:
4885: @cindex insufficient data stack or return stack space
4886: @cindex stack overflow
4887: @cindex Address alignment exception, stack overflow
4888: @cindex Invalid memory address, stack overflow
4889: Depending on the operating system, the installation, and the invocation
4890: of Gforth, this is either checked by the memory management hardware, or
4891: it is not checked. If it is checked, you typically get a @code{-9 throw}
4892: (Invalid memory address) as soon as the overflow happens. If it is not
4893: check, overflows typically result in mysterious illegal memory accesses,
4894: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
4895: (Address alignment exception); they might also destroy the internal data
4896: structure of @code{ALLOCATE} and friends, resulting in various errors in
4897: these words.
4898:
4899: @item insufficient space for loop control parameters:
4900: @cindex insufficient space for loop control parameters
4901: like other return stack overflows.
4902:
4903: @item insufficient space in the dictionary:
4904: @cindex insufficient space in the dictionary
4905: @cindex dictionary overflow
4906: Depending on the operating system, the installation, and the invocation
4907: of Gforth, this is either checked by the memory management hardware, or
4908: it is not checked. Similar results as stack overflows. However,
4909: typically the error appears at a different place when one inserts or
4910: removes code. Also, the @code{THROW} does not relieve the situation (it
4911: does for stack overflows).
4912:
4913: @item interpreting a word with undefined interpretation semantics:
4914: @cindex interpreting a word with undefined interpretation semantics
4915: @cindex Interpreting a compile-only word
4916: For some words, we have defined interpretation semantics. For the
4917: others: @code{-14 throw} (Interpreting a compile-only word).
4918:
4919: @item modifying the contents of the input buffer or a string literal:
4920: @cindex modifying the contents of the input buffer or a string literal
4921: These are located in writable memory and can be modified.
4922:
4923: @item overflow of the pictured numeric output string:
4924: @cindex overflow of the pictured numeric output string
4925: @cindex pictured numeric output string, overflow
4926: Not checked. Runs into the dictionary and destroys it (at least,
4927: partially).
4928:
4929: @item parsed string overflow:
4930: @cindex parsed string overflow
4931: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
4932:
4933: @item producing a result out of range:
4934: @cindex result out of range
4935: On two's complement machines, arithmetic is performed modulo
4936: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
4937: arithmetic (with appropriate mapping for signed types). Division by zero
4938: typically results in a @code{-55 throw} (floatingpoint unidentified
4939: fault), although a @code{-10 throw} (divide by zero) would be more
4940: appropriate. @code{convert} and @code{>number} currently overflow
4941: silently.
4942:
4943: @item reading from an empty data or return stack:
4944: @cindex stack empty
4945: @cindex stack underflow
4946: The data stack is checked by the outer (aka text) interpreter after
4947: every word executed. If it has underflowed, a @code{-4 throw} (Stack
4948: underflow) is performed. Apart from that, stacks may be checked or not,
4949: depending on operating system, installation, and invocation. The
4950: consequences of stack underflows are similar to the consequences of
4951: stack overflows. Note that even if the system uses checking (through the
4952: MMU), your program may have to underflow by a significant number of
4953: stack items to trigger the reaction (the reason for this is that the
4954: MMU, and therefore the checking, works with a page-size granularity).
4955:
4956: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
4957: @cindex unexpected end of the input buffer
4958: @cindex zero-length string as a name
4959: @cindex Attempt to use zero-length string as a name
4960: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
4961: use zero-length string as a name). Words like @code{'} probably will not
4962: find what they search. Note that it is possible to create zero-length
4963: names with @code{nextname} (should it not?).
4964:
4965: @item @code{>IN} greater than input buffer:
4966: @cindex @code{>IN} greater than input buffer
4967: The next invocation of a parsing word returns a string with length 0.
4968:
4969: @item @code{RECURSE} appears after @code{DOES>}:
4970: @cindex @code{RECURSE} appears after @code{DOES>}
4971: Compiles a recursive call to the defining word, not to the defined word.
4972:
4973: @item argument input source different than current input source for @code{RESTORE-INPUT}:
4974: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
4975: @cindex Argument type mismatch, @code{RESTORE-INPUT}
4976: @cindex @code{RESTORE-INPUT}, Argument type mismatch
4977: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
4978: the end of the file was reached), its source-id may be
4979: reused. Therefore, restoring an input source specification referencing a
4980: closed file may lead to unpredictable results instead of a @code{-12
4981: THROW}.
4982:
4983: In the future, Gforth may be able to restore input source specifications
4984: from other than the current input source.
4985:
4986: @item data space containing definitions gets de-allocated:
4987: @cindex data space containing definitions gets de-allocated
4988: Deallocation with @code{allot} is not checked. This typically results in
4989: memory access faults or execution of illegal instructions.
4990:
4991: @item data space read/write with incorrect alignment:
4992: @cindex data space read/write with incorrect alignment
4993: @cindex alignment faults
4994: @cindex Address alignment exception
4995: Processor-dependent. Typically results in a @code{-23 throw} (Address
4996: alignment exception). Under Linux on a 486 or later processor with
4997: alignment turned on, incorrect alignment results in a @code{-9 throw}
4998: (Invalid memory address). There are reportedly some processors with
4999: alignment restrictions that do not report them.
5000:
5001: @item data space pointer not properly aligned, @code{,}, @code{C,}:
5002: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
5003: Like other alignment errors.
5004:
5005: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
5006: Like other stack underflows.
5007:
5008: @item loop control parameters not available:
5009: @cindex loop control parameters not available
5010: Not checked. The counted loop words simply assume that the top of return
5011: stack items are loop control parameters and behave accordingly.
5012:
5013: @item most recent definition does not have a name (@code{IMMEDIATE}):
5014: @cindex most recent definition does not have a name (@code{IMMEDIATE})
5015: @cindex last word was headerless
5016: @code{abort" last word was headerless"}.
5017:
5018: @item name not defined by @code{VALUE} used by @code{TO}:
5019: @cindex name not defined by @code{VALUE} used by @code{TO}
5020: @cindex @code{TO} on non-@code{VALUE}s
5021: @cindex Invalid name argument, @code{TO}
5022: @code{-32 throw} (Invalid name argument) (unless name is a local or was
5023: defined by @code{CONSTANT}; in the latter case it just changes the constant).
5024:
5025: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
5026: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
5027: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
5028: @code{-13 throw} (Undefined word)
5029:
5030: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
5031: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
5032: Gforth behaves as if they were of the same type. I.e., you can predict
5033: the behaviour by interpreting all parameters as, e.g., signed.
5034:
5035: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
5036: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
5037: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
5038: compilation semantics of @code{TO}.
5039:
5040: @item String longer than a counted string returned by @code{WORD}:
5041: @cindex String longer than a counted string returned by @code{WORD}
5042: @cindex @code{WORD}, string overflow
5043: Not checked. The string will be ok, but the count will, of course,
5044: contain only the least significant bits of the length.
5045:
5046: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
5047: @cindex @code{LSHIFT}, large shift counts
5048: @cindex @code{RSHIFT}, large shift counts
5049: Processor-dependent. Typical behaviours are returning 0 and using only
5050: the low bits of the shift count.
5051:
5052: @item word not defined via @code{CREATE}:
5053: @cindex @code{>BODY} of non-@code{CREATE}d words
5054: @code{>BODY} produces the PFA of the word no matter how it was defined.
5055:
5056: @cindex @code{DOES>} of non-@code{CREATE}d words
5057: @code{DOES>} changes the execution semantics of the last defined word no
5058: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
5059: @code{CREATE , DOES>}.
5060:
5061: @item words improperly used outside @code{<#} and @code{#>}:
5062: Not checked. As usual, you can expect memory faults.
5063:
5064: @end table
5065:
5066:
5067: @c ---------------------------------------------------------------------
5068: @node core-other, , core-ambcond, The Core Words
5069: @subsection Other system documentation
5070: @c ---------------------------------------------------------------------
5071: @cindex other system documentation, core words
5072: @cindex core words, other system documentation
5073:
5074: @table @i
5075: @item nonstandard words using @code{PAD}:
5076: @cindex @code{PAD} use by nonstandard words
5077: None.
5078:
5079: @item operator's terminal facilities available:
5080: @cindex operator's terminal facilities available
5081: After processing the command line, Gforth goes into interactive mode,
5082: and you can give commands to Gforth interactively. The actual facilities
5083: available depend on how you invoke Gforth.
5084:
5085: @item program data space available:
5086: @cindex program data space available
5087: @cindex data space available
5088: @code{UNUSED .} gives the remaining dictionary space. The total
5089: dictionary space can be specified with the @code{-m} switch
5090: (@pxref{Invoking Gforth}) when Gforth starts up.
5091:
5092: @item return stack space available:
5093: @cindex return stack space available
5094: You can compute the total return stack space in cells with
5095: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
5096: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
5097:
5098: @item stack space available:
5099: @cindex stack space available
5100: You can compute the total data stack space in cells with
5101: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
5102: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
5103:
5104: @item system dictionary space required, in address units:
5105: @cindex system dictionary space required, in address units
5106: Type @code{here forthstart - .} after startup. At the time of this
5107: writing, this gives 80080 (bytes) on a 32-bit system.
5108: @end table
5109:
5110:
5111: @c =====================================================================
5112: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
5113: @section The optional Block word set
5114: @c =====================================================================
5115: @cindex system documentation, block words
5116: @cindex block words, system documentation
5117:
5118: @menu
5119: * block-idef:: Implementation Defined Options
5120: * block-ambcond:: Ambiguous Conditions
5121: * block-other:: Other System Documentation
5122: @end menu
5123:
5124:
5125: @c ---------------------------------------------------------------------
5126: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
5127: @subsection Implementation Defined Options
5128: @c ---------------------------------------------------------------------
5129: @cindex implementation-defined options, block words
5130: @cindex block words, implementation-defined options
5131:
5132: @table @i
5133: @item the format for display by @code{LIST}:
5134: @cindex @code{LIST} display format
5135: First the screen number is displayed, then 16 lines of 64 characters,
5136: each line preceded by the line number.
5137:
5138: @item the length of a line affected by @code{\}:
5139: @cindex length of a line affected by @code{\}
5140: @cindex @code{\}, line length in blocks
5141: 64 characters.
5142: @end table
5143:
5144:
5145: @c ---------------------------------------------------------------------
5146: @node block-ambcond, block-other, block-idef, The optional Block word set
5147: @subsection Ambiguous conditions
5148: @c ---------------------------------------------------------------------
5149: @cindex block words, ambiguous conditions
5150: @cindex ambiguous conditions, block words
5151:
5152: @table @i
5153: @item correct block read was not possible:
5154: @cindex block read not possible
5155: Typically results in a @code{throw} of some OS-derived value (between
5156: -512 and -2048). If the blocks file was just not long enough, blanks are
5157: supplied for the missing portion.
5158:
5159: @item I/O exception in block transfer:
5160: @cindex I/O exception in block transfer
5161: @cindex block transfer, I/O exception
5162: Typically results in a @code{throw} of some OS-derived value (between
5163: -512 and -2048).
5164:
5165: @item invalid block number:
5166: @cindex invalid block number
5167: @cindex block number invalid
5168: @code{-35 throw} (Invalid block number)
5169:
5170: @item a program directly alters the contents of @code{BLK}:
5171: @cindex @code{BLK}, altering @code{BLK}
5172: The input stream is switched to that other block, at the same
5173: position. If the storing to @code{BLK} happens when interpreting
5174: non-block input, the system will get quite confused when the block ends.
5175:
5176: @item no current block buffer for @code{UPDATE}:
5177: @cindex @code{UPDATE}, no current block buffer
5178: @code{UPDATE} has no effect.
5179:
5180: @end table
5181:
5182: @c ---------------------------------------------------------------------
5183: @node block-other, , block-ambcond, The optional Block word set
5184: @subsection Other system documentation
5185: @c ---------------------------------------------------------------------
5186: @cindex other system documentation, block words
5187: @cindex block words, other system documentation
5188:
5189: @table @i
5190: @item any restrictions a multiprogramming system places on the use of buffer addresses:
5191: No restrictions (yet).
5192:
5193: @item the number of blocks available for source and data:
5194: depends on your disk space.
5195:
5196: @end table
5197:
5198:
5199: @c =====================================================================
5200: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
5201: @section The optional Double Number word set
5202: @c =====================================================================
5203: @cindex system documentation, double words
5204: @cindex double words, system documentation
5205:
5206: @menu
5207: * double-ambcond:: Ambiguous Conditions
5208: @end menu
5209:
5210:
5211: @c ---------------------------------------------------------------------
5212: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
5213: @subsection Ambiguous conditions
5214: @c ---------------------------------------------------------------------
5215: @cindex double words, ambiguous conditions
5216: @cindex ambiguous conditions, double words
5217:
5218: @table @i
5219: @item @var{d} outside of range of @var{n} in @code{D>S}:
5220: @cindex @code{D>S}, @var{d} out of range of @var{n}
5221: The least significant cell of @var{d} is produced.
5222:
5223: @end table
5224:
5225:
5226: @c =====================================================================
5227: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
5228: @section The optional Exception word set
5229: @c =====================================================================
5230: @cindex system documentation, exception words
5231: @cindex exception words, system documentation
5232:
5233: @menu
5234: * exception-idef:: Implementation Defined Options
5235: @end menu
5236:
5237:
5238: @c ---------------------------------------------------------------------
5239: @node exception-idef, , The optional Exception word set, The optional Exception word set
5240: @subsection Implementation Defined Options
5241: @c ---------------------------------------------------------------------
5242: @cindex implementation-defined options, exception words
5243: @cindex exception words, implementation-defined options
5244:
5245: @table @i
5246: @item @code{THROW}-codes used in the system:
5247: @cindex @code{THROW}-codes used in the system
5248: The codes -256@minus{}-511 are used for reporting signals. The mapping
5249: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
5250: codes -512@minus{}-2047 are used for OS errors (for file and memory
5251: allocation operations). The mapping from OS error numbers to throw codes
5252: is -512@minus{}@code{errno}. One side effect of this mapping is that
5253: undefined OS errors produce a message with a strange number; e.g.,
5254: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
5255: @end table
5256:
5257: @c =====================================================================
5258: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
5259: @section The optional Facility word set
5260: @c =====================================================================
5261: @cindex system documentation, facility words
5262: @cindex facility words, system documentation
5263:
5264: @menu
5265: * facility-idef:: Implementation Defined Options
5266: * facility-ambcond:: Ambiguous Conditions
5267: @end menu
5268:
5269:
5270: @c ---------------------------------------------------------------------
5271: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
5272: @subsection Implementation Defined Options
5273: @c ---------------------------------------------------------------------
5274: @cindex implementation-defined options, facility words
5275: @cindex facility words, implementation-defined options
5276:
5277: @table @i
5278: @item encoding of keyboard events (@code{EKEY}):
5279: @cindex keyboard events, encoding in @code{EKEY}
5280: @cindex @code{EKEY}, encoding of keyboard events
5281: Not yet implemented.
5282:
5283: @item duration of a system clock tick:
5284: @cindex duration of a system clock tick
5285: @cindex clock tick duration
5286: System dependent. With respect to @code{MS}, the time is specified in
5287: microseconds. How well the OS and the hardware implement this, is
5288: another question.
5289:
5290: @item repeatability to be expected from the execution of @code{MS}:
5291: @cindex repeatability to be expected from the execution of @code{MS}
5292: @cindex @code{MS}, repeatability to be expected
5293: System dependent. On Unix, a lot depends on load. If the system is
5294: lightly loaded, and the delay is short enough that Gforth does not get
5295: swapped out, the performance should be acceptable. Under MS-DOS and
5296: other single-tasking systems, it should be good.
5297:
5298: @end table
5299:
5300:
5301: @c ---------------------------------------------------------------------
5302: @node facility-ambcond, , facility-idef, The optional Facility word set
5303: @subsection Ambiguous conditions
5304: @c ---------------------------------------------------------------------
5305: @cindex facility words, ambiguous conditions
5306: @cindex ambiguous conditions, facility words
5307:
5308: @table @i
5309: @item @code{AT-XY} can't be performed on user output device:
5310: @cindex @code{AT-XY} can't be performed on user output device
5311: Largely terminal dependent. No range checks are done on the arguments.
5312: No errors are reported. You may see some garbage appearing, you may see
5313: simply nothing happen.
5314:
5315: @end table
5316:
5317:
5318: @c =====================================================================
5319: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
5320: @section The optional File-Access word set
5321: @c =====================================================================
5322: @cindex system documentation, file words
5323: @cindex file words, system documentation
5324:
5325: @menu
5326: * file-idef:: Implementation Defined Options
5327: * file-ambcond:: Ambiguous Conditions
5328: @end menu
5329:
5330: @c ---------------------------------------------------------------------
5331: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
5332: @subsection Implementation Defined Options
5333: @c ---------------------------------------------------------------------
5334: @cindex implementation-defined options, file words
5335: @cindex file words, implementation-defined options
5336:
5337: @table @i
5338: @item file access methods used:
5339: @cindex file access methods used
5340: @code{R/O}, @code{R/W} and @code{BIN} work as you would
5341: expect. @code{W/O} translates into the C file opening mode @code{w} (or
5342: @code{wb}): The file is cleared, if it exists, and created, if it does
5343: not (with both @code{open-file} and @code{create-file}). Under Unix
5344: @code{create-file} creates a file with 666 permissions modified by your
5345: umask.
5346:
5347: @item file exceptions:
5348: @cindex file exceptions
5349: The file words do not raise exceptions (except, perhaps, memory access
5350: faults when you pass illegal addresses or file-ids).
5351:
5352: @item file line terminator:
5353: @cindex file line terminator
5354: System-dependent. Gforth uses C's newline character as line
5355: terminator. What the actual character code(s) of this are is
5356: system-dependent.
5357:
5358: @item file name format:
5359: @cindex file name format
5360: System dependent. Gforth just uses the file name format of your OS.
5361:
5362: @item information returned by @code{FILE-STATUS}:
5363: @cindex @code{FILE-STATUS}, returned information
5364: @code{FILE-STATUS} returns the most powerful file access mode allowed
5365: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
5366: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
5367: along with the returned mode.
5368:
5369: @item input file state after an exception when including source:
5370: @cindex exception when including source
5371: All files that are left via the exception are closed.
5372:
5373: @item @var{ior} values and meaning:
5374: @cindex @var{ior} values and meaning
5375: The @var{ior}s returned by the file and memory allocation words are
5376: intended as throw codes. They typically are in the range
5377: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
5378: @var{ior}s is -512@minus{}@var{errno}.
5379:
5380: @item maximum depth of file input nesting:
5381: @cindex maximum depth of file input nesting
5382: @cindex file input nesting, maximum depth
5383: limited by the amount of return stack, locals/TIB stack, and the number
5384: of open files available. This should not give you troubles.
5385:
5386: @item maximum size of input line:
5387: @cindex maximum size of input line
5388: @cindex input line size, maximum
5389: @code{/line}. Currently 255.
5390:
5391: @item methods of mapping block ranges to files:
5392: @cindex mapping block ranges to files
5393: @cindex files containing blocks
5394: @cindex blocks in files
5395: By default, blocks are accessed in the file @file{blocks.fb} in the
5396: current working directory. The file can be switched with @code{USE}.
5397:
5398: @item number of string buffers provided by @code{S"}:
5399: @cindex @code{S"}, number of string buffers
5400: 1
5401:
5402: @item size of string buffer used by @code{S"}:
5403: @cindex @code{S"}, size of string buffer
5404: @code{/line}. currently 255.
5405:
5406: @end table
5407:
5408: @c ---------------------------------------------------------------------
5409: @node file-ambcond, , file-idef, The optional File-Access word set
5410: @subsection Ambiguous conditions
5411: @c ---------------------------------------------------------------------
5412: @cindex file words, ambiguous conditions
5413: @cindex ambiguous conditions, file words
5414:
5415: @table @i
5416: @item attempting to position a file outside its boundaries:
5417: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
5418: @code{REPOSITION-FILE} is performed as usual: Afterwards,
5419: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
5420:
5421: @item attempting to read from file positions not yet written:
5422: @cindex reading from file positions not yet written
5423: End-of-file, i.e., zero characters are read and no error is reported.
5424:
5425: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
5426: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid
5427: An appropriate exception may be thrown, but a memory fault or other
5428: problem is more probable.
5429:
5430: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
5431: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
5432: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
5433: The @var{ior} produced by the operation, that discovered the problem, is
5434: thrown.
5435:
5436: @item named file cannot be opened (@code{INCLUDED}):
5437: @cindex @code{INCLUDED}, named file cannot be opened
5438: The @var{ior} produced by @code{open-file} is thrown.
5439:
5440: @item requesting an unmapped block number:
5441: @cindex unmapped block numbers
5442: There are no unmapped legal block numbers. On some operating systems,
5443: writing a block with a large number may overflow the file system and
5444: have an error message as consequence.
5445:
5446: @item using @code{source-id} when @code{blk} is non-zero:
5447: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
5448: @code{source-id} performs its function. Typically it will give the id of
5449: the source which loaded the block. (Better ideas?)
5450:
5451: @end table
5452:
5453:
5454: @c =====================================================================
5455: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
5456: @section The optional Floating-Point word set
5457: @c =====================================================================
5458: @cindex system documentation, floating-point words
5459: @cindex floating-point words, system documentation
5460:
5461: @menu
5462: * floating-idef:: Implementation Defined Options
5463: * floating-ambcond:: Ambiguous Conditions
5464: @end menu
5465:
5466:
5467: @c ---------------------------------------------------------------------
5468: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
5469: @subsection Implementation Defined Options
5470: @c ---------------------------------------------------------------------
5471: @cindex implementation-defined options, floating-point words
5472: @cindex floating-point words, implementation-defined options
5473:
5474: @table @i
5475: @item format and range of floating point numbers:
5476: @cindex format and range of floating point numbers
5477: @cindex floating point numbers, format and range
5478: System-dependent; the @code{double} type of C.
5479:
5480: @item results of @code{REPRESENT} when @var{float} is out of range:
5481: @cindex @code{REPRESENT}, results when @var{float} is out of range
5482: System dependent; @code{REPRESENT} is implemented using the C library
5483: function @code{ecvt()} and inherits its behaviour in this respect.
5484:
5485: @item rounding or truncation of floating-point numbers:
5486: @cindex rounding of floating-point numbers
5487: @cindex truncation of floating-point numbers
5488: @cindex floating-point numbers, rounding or truncation
5489: System dependent; the rounding behaviour is inherited from the hosting C
5490: compiler. IEEE-FP-based (i.e., most) systems by default round to
5491: nearest, and break ties by rounding to even (i.e., such that the last
5492: bit of the mantissa is 0).
5493:
5494: @item size of floating-point stack:
5495: @cindex floating-point stack size
5496: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
5497: the floating-point stack (in floats). You can specify this on startup
5498: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
5499:
5500: @item width of floating-point stack:
5501: @cindex floating-point stack width
5502: @code{1 floats}.
5503:
5504: @end table
5505:
5506:
5507: @c ---------------------------------------------------------------------
5508: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
5509: @subsection Ambiguous conditions
5510: @c ---------------------------------------------------------------------
5511: @cindex floating-point words, ambiguous conditions
5512: @cindex ambiguous conditions, floating-point words
5513:
5514: @table @i
5515: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
5516: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
5517: System-dependent. Typically results in a @code{-23 THROW} like other
5518: alignment violations.
5519:
5520: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
5521: @cindex @code{f@@} used with an address that is not float aligned
5522: @cindex @code{f!} used with an address that is not float aligned
5523: System-dependent. Typically results in a @code{-23 THROW} like other
5524: alignment violations.
5525:
5526: @item floating-point result out of range:
5527: @cindex floating-point result out of range
5528: System-dependent. Can result in a @code{-55 THROW} (Floating-point
5529: unidentified fault), or can produce a special value representing, e.g.,
5530: Infinity.
5531:
5532: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
5533: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
5534: System-dependent. Typically results in an alignment fault like other
5535: alignment violations.
5536:
5537: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
5538: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
5539: The floating-point number is converted into decimal nonetheless.
5540:
5541: @item Both arguments are equal to zero (@code{FATAN2}):
5542: @cindex @code{FATAN2}, both arguments are equal to zero
5543: System-dependent. @code{FATAN2} is implemented using the C library
5544: function @code{atan2()}.
5545:
5546: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
5547: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
5548: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
5549: because of small errors and the tan will be a very large (or very small)
5550: but finite number.
5551:
5552: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
5553: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
5554: The result is rounded to the nearest float.
5555:
5556: @item dividing by zero:
5557: @cindex dividing by zero, floating-point
5558: @cindex floating-point dividing by zero
5559: @cindex floating-point unidentified fault, FP divide-by-zero
5560: @code{-55 throw} (Floating-point unidentified fault)
5561:
5562: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
5563: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
5564: System dependent. On IEEE-FP based systems the number is converted into
5565: an infinity.
5566:
5567: @item @var{float}<1 (@code{FACOSH}):
5568: @cindex @code{FACOSH}, @var{float}<1
5569: @cindex floating-point unidentified fault, @code{FACOSH}
5570: @code{-55 throw} (Floating-point unidentified fault)
5571:
5572: @item @var{float}=<-1 (@code{FLNP1}):
5573: @cindex @code{FLNP1}, @var{float}=<-1
5574: @cindex floating-point unidentified fault, @code{FLNP1}
5575: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
5576: negative infinity is typically produced for @var{float}=-1.
5577:
5578: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
5579: @cindex @code{FLN}, @var{float}=<0
5580: @cindex @code{FLOG}, @var{float}=<0
5581: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
5582: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
5583: negative infinity is typically produced for @var{float}=0.
5584:
5585: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
5586: @cindex @code{FASINH}, @var{float}<0
5587: @cindex @code{FSQRT}, @var{float}<0
5588: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
5589: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
5590: produces values for these inputs on my Linux box (Bug in the C library?)
5591:
5592: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
5593: @cindex @code{FACOS}, |@var{float}|>1
5594: @cindex @code{FASIN}, |@var{float}|>1
5595: @cindex @code{FATANH}, |@var{float}|>1
5596: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
5597: @code{-55 throw} (Floating-point unidentified fault).
5598:
5599: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
5600: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
5601: @cindex floating-point unidentified fault, @code{F>D}
5602: @code{-55 throw} (Floating-point unidentified fault).
5603:
5604: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
5605: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
5606: This does not happen.
5607: @end table
5608:
5609: @c =====================================================================
5610: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
5611: @section The optional Locals word set
5612: @c =====================================================================
5613: @cindex system documentation, locals words
5614: @cindex locals words, system documentation
5615:
5616: @menu
5617: * locals-idef:: Implementation Defined Options
5618: * locals-ambcond:: Ambiguous Conditions
5619: @end menu
5620:
5621:
5622: @c ---------------------------------------------------------------------
5623: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
5624: @subsection Implementation Defined Options
5625: @c ---------------------------------------------------------------------
5626: @cindex implementation-defined options, locals words
5627: @cindex locals words, implementation-defined options
5628:
5629: @table @i
5630: @item maximum number of locals in a definition:
5631: @cindex maximum number of locals in a definition
5632: @cindex locals, maximum number in a definition
5633: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
5634: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
5635: characters. The number of locals in a definition is bounded by the size
5636: of locals-buffer, which contains the names of the locals.
5637:
5638: @end table
5639:
5640:
5641: @c ---------------------------------------------------------------------
5642: @node locals-ambcond, , locals-idef, The optional Locals word set
5643: @subsection Ambiguous conditions
5644: @c ---------------------------------------------------------------------
5645: @cindex locals words, ambiguous conditions
5646: @cindex ambiguous conditions, locals words
5647:
5648: @table @i
5649: @item executing a named local in interpretation state:
5650: @cindex local in interpretation state
5651: @cindex Interpreting a compile-only word, for a local
5652: Locals have no interpretation semantics. If you try to perform the
5653: interpretation semantics, you will get a @code{-14 throw} somewhere
5654: (Interpreting a compile-only word). If you perform the compilation
5655: semantics, the locals access will be compiled (irrespective of state).
5656:
5657: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
5658: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
5659: @cindex @code{TO} on non-@code{VALUE}s and non-locals
5660: @cindex Invalid name argument, @code{TO}
5661: @code{-32 throw} (Invalid name argument)
5662:
5663: @end table
5664:
5665:
5666: @c =====================================================================
5667: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
5668: @section The optional Memory-Allocation word set
5669: @c =====================================================================
5670: @cindex system documentation, memory-allocation words
5671: @cindex memory-allocation words, system documentation
5672:
5673: @menu
5674: * memory-idef:: Implementation Defined Options
5675: @end menu
5676:
5677:
5678: @c ---------------------------------------------------------------------
5679: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
5680: @subsection Implementation Defined Options
5681: @c ---------------------------------------------------------------------
5682: @cindex implementation-defined options, memory-allocation words
5683: @cindex memory-allocation words, implementation-defined options
5684:
5685: @table @i
5686: @item values and meaning of @var{ior}:
5687: @cindex @var{ior} values and meaning
5688: The @var{ior}s returned by the file and memory allocation words are
5689: intended as throw codes. They typically are in the range
5690: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
5691: @var{ior}s is -512@minus{}@var{errno}.
5692:
5693: @end table
5694:
5695: @c =====================================================================
5696: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
5697: @section The optional Programming-Tools word set
5698: @c =====================================================================
5699: @cindex system documentation, programming-tools words
5700: @cindex programming-tools words, system documentation
5701:
5702: @menu
5703: * programming-idef:: Implementation Defined Options
5704: * programming-ambcond:: Ambiguous Conditions
5705: @end menu
5706:
5707:
5708: @c ---------------------------------------------------------------------
5709: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
5710: @subsection Implementation Defined Options
5711: @c ---------------------------------------------------------------------
5712: @cindex implementation-defined options, programming-tools words
5713: @cindex programming-tools words, implementation-defined options
5714:
5715: @table @i
5716: @item ending sequence for input following @code{;CODE} and @code{CODE}:
5717: @cindex @code{;CODE} ending sequence
5718: @cindex @code{CODE} ending sequence
5719: @code{END-CODE}
5720:
5721: @item manner of processing input following @code{;CODE} and @code{CODE}:
5722: @cindex @code{;CODE}, processing input
5723: @cindex @code{CODE}, processing input
5724: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
5725: the input is processed by the text interpreter, (starting) in interpret
5726: state.
5727:
5728: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
5729: @cindex @code{ASSEMBLER}, search order capability
5730: The ANS Forth search order word set.
5731:
5732: @item source and format of display by @code{SEE}:
5733: @cindex @code{SEE}, source and format of output
5734: The source for @code{see} is the intermediate code used by the inner
5735: interpreter. The current @code{see} tries to output Forth source code
5736: as well as possible.
5737:
5738: @end table
5739:
5740: @c ---------------------------------------------------------------------
5741: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
5742: @subsection Ambiguous conditions
5743: @c ---------------------------------------------------------------------
5744: @cindex programming-tools words, ambiguous conditions
5745: @cindex ambiguous conditions, programming-tools words
5746:
5747: @table @i
5748:
5749: @item deleting the compilation wordlist (@code{FORGET}):
5750: @cindex @code{FORGET}, deleting the compilation wordlist
5751: Not implemented (yet).
5752:
5753: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
5754: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
5755: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
5756: @cindex control-flow stack underflow
5757: This typically results in an @code{abort"} with a descriptive error
5758: message (may change into a @code{-22 throw} (Control structure mismatch)
5759: in the future). You may also get a memory access error. If you are
5760: unlucky, this ambiguous condition is not caught.
5761:
5762: @item @var{name} can't be found (@code{FORGET}):
5763: @cindex @code{FORGET}, @var{name} can't be found
5764: Not implemented (yet).
5765:
5766: @item @var{name} not defined via @code{CREATE}:
5767: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
5768: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
5769: the execution semantics of the last defined word no matter how it was
5770: defined.
5771:
5772: @item @code{POSTPONE} applied to @code{[IF]}:
5773: @cindex @code{POSTPONE} applied to @code{[IF]}
5774: @cindex @code{[IF]} and @code{POSTPONE}
5775: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
5776: equivalent to @code{[IF]}.
5777:
5778: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
5779: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
5780: Continue in the same state of conditional compilation in the next outer
5781: input source. Currently there is no warning to the user about this.
5782:
5783: @item removing a needed definition (@code{FORGET}):
5784: @cindex @code{FORGET}, removing a needed definition
5785: Not implemented (yet).
5786:
5787: @end table
5788:
5789:
5790: @c =====================================================================
5791: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
5792: @section The optional Search-Order word set
5793: @c =====================================================================
5794: @cindex system documentation, search-order words
5795: @cindex search-order words, system documentation
5796:
5797: @menu
5798: * search-idef:: Implementation Defined Options
5799: * search-ambcond:: Ambiguous Conditions
5800: @end menu
5801:
5802:
5803: @c ---------------------------------------------------------------------
5804: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
5805: @subsection Implementation Defined Options
5806: @c ---------------------------------------------------------------------
5807: @cindex implementation-defined options, search-order words
5808: @cindex search-order words, implementation-defined options
5809:
5810: @table @i
5811: @item maximum number of word lists in search order:
5812: @cindex maximum number of word lists in search order
5813: @cindex search order, maximum depth
5814: @code{s" wordlists" environment? drop .}. Currently 16.
5815:
5816: @item minimum search order:
5817: @cindex minimum search order
5818: @cindex search order, minimum
5819: @code{root root}.
5820:
5821: @end table
5822:
5823: @c ---------------------------------------------------------------------
5824: @node search-ambcond, , search-idef, The optional Search-Order word set
5825: @subsection Ambiguous conditions
5826: @c ---------------------------------------------------------------------
5827: @cindex search-order words, ambiguous conditions
5828: @cindex ambiguous conditions, search-order words
5829:
5830: @table @i
5831: @item changing the compilation wordlist (during compilation):
5832: @cindex changing the compilation wordlist (during compilation)
5833: @cindex compilation wordlist, change before definition ends
5834: The word is entered into the wordlist that was the compilation wordlist
5835: at the start of the definition. Any changes to the name field (e.g.,
5836: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
5837: are applied to the latest defined word (as reported by @code{last} or
5838: @code{lastxt}), if possible, irrespective of the compilation wordlist.
5839:
5840: @item search order empty (@code{previous}):
5841: @cindex @code{previous}, search order empty
5842: @cindex Vocstack empty, @code{previous}
5843: @code{abort" Vocstack empty"}.
5844:
5845: @item too many word lists in search order (@code{also}):
5846: @cindex @code{also}, too many word lists in search order
5847: @cindex Vocstack full, @code{also}
5848: @code{abort" Vocstack full"}.
5849:
5850: @end table
5851:
5852: @c ***************************************************************
5853: @node Model, Integrating Gforth, ANS conformance, Top
5854: @chapter Model
5855:
5856: This chapter has yet to be written. It will contain information, on
5857: which internal structures you can rely.
5858:
5859: @c ***************************************************************
5860: @node Integrating Gforth, Emacs and Gforth, Model, Top
5861: @chapter Integrating Gforth into C programs
5862:
5863: This is not yet implemented.
5864:
5865: Several people like to use Forth as scripting language for applications
5866: that are otherwise written in C, C++, or some other language.
5867:
5868: The Forth system ATLAST provides facilities for embedding it into
5869: applications; unfortunately it has several disadvantages: most
5870: importantly, it is not based on ANS Forth, and it is apparently dead
5871: (i.e., not developed further and not supported). The facilities
5872: provided by Gforth in this area are inspired by ATLASTs facilities, so
5873: making the switch should not be hard.
5874:
5875: We also tried to design the interface such that it can easily be
5876: implemented by other Forth systems, so that we may one day arrive at a
5877: standardized interface. Such a standard interface would allow you to
5878: replace the Forth system without having to rewrite C code.
5879:
5880: You embed the Gforth interpreter by linking with the library
5881: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
5882: global symbols in this library that belong to the interface, have the
5883: prefix @code{forth_}. (Global symbols that are used internally have the
5884: prefix @code{gforth_}).
5885:
5886: You can include the declarations of Forth types and the functions and
5887: variables of the interface with @code{#include <forth.h>}.
5888:
5889: Types.
5890:
5891: Variables.
5892:
5893: Data and FP Stack pointer. Area sizes.
5894:
5895: functions.
5896:
5897: forth_init(imagefile)
5898: forth_evaluate(string) exceptions?
5899: forth_goto(address) (or forth_execute(xt)?)
5900: forth_continue() (a corountining mechanism)
5901:
5902: Adding primitives.
5903:
5904: No checking.
5905:
5906: Signals?
5907:
5908: Accessing the Stacks
5909:
5910: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
5911: @chapter Emacs and Gforth
5912: @cindex Emacs and Gforth
5913:
5914: @cindex @file{gforth.el}
5915: @cindex @file{forth.el}
5916: @cindex Rydqvist, Goran
5917: @cindex comment editing commands
5918: @cindex @code{\}, editing with Emacs
5919: @cindex debug tracer editing commands
5920: @cindex @code{~~}, removal with Emacs
5921: @cindex Forth mode in Emacs
5922: Gforth comes with @file{gforth.el}, an improved version of
5923: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
5924: improvements are a better (but still not perfect) handling of
5925: indentation. I have also added comment paragraph filling (@kbd{M-q}),
5926: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
5927: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
5928: stuff I do not use alone, even though some of it only makes sense for
5929: TILE. To get a description of these features, enter Forth mode and type
5930: @kbd{C-h m}.
5931:
5932: @cindex source location of error or debugging output in Emacs
5933: @cindex error output, finding the source location in Emacs
5934: @cindex debugging output, finding the source location in Emacs
5935: In addition, Gforth supports Emacs quite well: The source code locations
5936: given in error messages, debugging output (from @code{~~}) and failed
5937: assertion messages are in the right format for Emacs' compilation mode
5938: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
5939: Manual}) so the source location corresponding to an error or other
5940: message is only a few keystrokes away (@kbd{C-x `} for the next error,
5941: @kbd{C-c C-c} for the error under the cursor).
5942:
5943: @cindex @file{TAGS} file
5944: @cindex @file{etags.fs}
5945: @cindex viewing the source of a word in Emacs
5946: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
5947: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
5948: contains the definitions of all words defined afterwards. You can then
5949: find the source for a word using @kbd{M-.}. Note that emacs can use
5950: several tags files at the same time (e.g., one for the Gforth sources
5951: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
5952: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
5953: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
5954: @file{/usr/local/share/gforth/0.2.0/TAGS}).
5955:
5956: @cindex @file{.emacs}
5957: To get all these benefits, add the following lines to your @file{.emacs}
5958: file:
5959:
5960: @example
5961: (autoload 'forth-mode "gforth.el")
5962: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
5963: @end example
5964:
5965: @node Image Files, Engine, Emacs and Gforth, Top
5966: @chapter Image Files
5967: @cindex image files
5968: @cindex @code{.fi} files
5969: @cindex precompiled Forth code
5970: @cindex dictionary in persistent form
5971: @cindex persistent form of dictionary
5972:
5973: An image file is a file containing an image of the Forth dictionary,
5974: i.e., compiled Forth code and data residing in the dictionary. By
5975: convention, we use the extension @code{.fi} for image files.
5976:
5977: @menu
5978: * Image File Background:: Why have image files?
5979: * Non-Relocatable Image Files:: don't always work.
5980: * Data-Relocatable Image Files:: are better.
5981: * Fully Relocatable Image Files:: better yet.
5982: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
5983: * Running Image Files:: @code{gforth -i @var{file}} or @var{file}.
5984: * Modifying the Startup Sequence:: and turnkey applications.
5985: @end menu
5986:
5987: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
5988: @section Image File Background
5989: @cindex image file background
5990:
5991: Our Forth system consists not only of primitives, but also of
5992: definitions written in Forth. Since the Forth compiler itself belongs to
5993: those definitions, it is not possible to start the system with the
5994: primitives and the Forth source alone. Therefore we provide the Forth
5995: code as an image file in nearly executable form. At the start of the
5996: system a C routine loads the image file into memory, optionally
5997: relocates the addresses, then sets up the memory (stacks etc.) according
5998: to information in the image file, and starts executing Forth code.
5999:
6000: The image file variants represent different compromises between the
6001: goals of making it easy to generate image files and making them
6002: portable.
6003:
6004: @cindex relocation at run-time
6005: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
6006: run-time. This avoids many of the complications discussed below (image
6007: files are data relocatable without further ado), but costs performance
6008: (one addition per memory access).
6009:
6010: @cindex relocation at load-time
6011: By contrast, our loader performs relocation at image load time. The
6012: loader also has to replace tokens standing for primitive calls with the
6013: appropriate code-field addresses (or code addresses in the case of
6014: direct threading).
6015:
6016: There are three kinds of image files, with different degrees of
6017: relocatability: non-relocatable, data-relocatable, and fully relocatable
6018: image files.
6019:
6020: @cindex image file loader
6021: @cindex relocating loader
6022: @cindex loader for image files
6023: These image file variants have several restrictions in common; they are
6024: caused by the design of the image file loader:
6025:
6026: @itemize @bullet
6027: @item
6028: There is only one segment; in particular, this means, that an image file
6029: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
6030: them). And the contents of the stacks are not represented, either.
6031:
6032: @item
6033: The only kinds of relocation supported are: adding the same offset to
6034: all cells that represent data addresses; and replacing special tokens
6035: with code addresses or with pieces of machine code.
6036:
6037: If any complex computations involving addresses are performed, the
6038: results cannot be represented in the image file. Several applications that
6039: use such computations come to mind:
6040: @itemize @minus
6041: @item
6042: Hashing addresses (or data structures which contain addresses) for table
6043: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
6044: purpose, you will have no problem, because the hash tables are
6045: recomputed automatically when the system is started. If you use your own
6046: hash tables, you will have to do something similar.
6047:
6048: @item
6049: There's a cute implementation of doubly-linked lists that uses
6050: @code{XOR}ed addresses. You could represent such lists as singly-linked
6051: in the image file, and restore the doubly-linked representation on
6052: startup.@footnote{In my opinion, though, you should think thrice before
6053: using a doubly-linked list (whatever implementation).}
6054:
6055: @item
6056: The code addresses of run-time routines like @code{docol:} cannot be
6057: represented in the image file (because their tokens would be replaced by
6058: machine code in direct threaded implementations). As a workaround,
6059: compute these addresses at run-time with @code{>code-address} from the
6060: executions tokens of appropriate words (see the definitions of
6061: @code{docol:} and friends in @file{kernel.fs}).
6062:
6063: @item
6064: On many architectures addresses are represented in machine code in some
6065: shifted or mangled form. You cannot put @code{CODE} words that contain
6066: absolute addresses in this form in a relocatable image file. Workarounds
6067: are representing the address in some relative form (e.g., relative to
6068: the CFA, which is present in some register), or loading the address from
6069: a place where it is stored in a non-mangled form.
6070: @end itemize
6071: @end itemize
6072:
6073: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
6074: @section Non-Relocatable Image Files
6075: @cindex non-relocatable image files
6076: @cindex image files, non-relocatable
6077:
6078: These files are simple memory dumps of the dictionary. They are specific
6079: to the executable (i.e., @file{gforth} file) they were created
6080: with. What's worse, they are specific to the place on which the
6081: dictionary resided when the image was created. Now, there is no
6082: guarantee that the dictionary will reside at the same place the next
6083: time you start Gforth, so there's no guarantee that a non-relocatable
6084: image will work the next time (Gforth will complain instead of crashing,
6085: though).
6086:
6087: You can create a non-relocatable image file with
6088:
6089: doc-savesystem
6090:
6091: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
6092: @section Data-Relocatable Image Files
6093: @cindex data-relocatable image files
6094: @cindex image files, data-relocatable
6095:
6096: These files contain relocatable data addresses, but fixed code addresses
6097: (instead of tokens). They are specific to the executable (i.e.,
6098: @file{gforth} file) they were created with. For direct threading on some
6099: architectures (e.g., the i386), data-relocatable images do not work. You
6100: get a data-relocatable image, if you use @file{gforthmi} with a
6101: Gforth binary that is not doubly indirect threaded (@pxref{Fully
6102: Relocatable Image Files}).
6103:
6104: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
6105: @section Fully Relocatable Image Files
6106: @cindex fully relocatable image files
6107: @cindex image files, fully relocatable
6108:
6109: @cindex @file{kern*.fi}, relocatability
6110: @cindex @file{gforth.fi}, relocatability
6111: These image files have relocatable data addresses, and tokens for code
6112: addresses. They can be used with different binaries (e.g., with and
6113: without debugging) on the same machine, and even across machines with
6114: the same data formats (byte order, cell size, floating point
6115: format). However, they are usually specific to the version of Gforth
6116: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
6117: are fully relocatable.
6118:
6119: There are two ways to create a fully relocatable image file:
6120:
6121: @menu
6122: * gforthmi:: The normal way
6123: * cross.fs:: The hard way
6124: @end menu
6125:
6126: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
6127: @subsection @file{gforthmi}
6128: @cindex @file{comp-i.fs}
6129: @cindex @file{gforthmi}
6130:
6131: You will usually use @file{gforthmi}. If you want to create an
6132: image @var{file} that contains everything you would load by invoking
6133: Gforth with @code{gforth @var{options}}, you simply say
6134: @example
6135: gforthmi @var{file} @var{options}
6136: @end example
6137:
6138: E.g., if you want to create an image @file{asm.fi} that has the file
6139: @file{asm.fs} loaded in addition to the usual stuff, you could do it
6140: like this:
6141:
6142: @example
6143: gforthmi asm.fi asm.fs
6144: @end example
6145:
6146: @file{gforthmi} works like this: It produces two non-relocatable
6147: images for different addresses and then compares them. Its output
6148: reflects this: first you see the output (if any) of the two Gforth
6149: invocations that produce the nonrelocatable image files, then you see
6150: the output of the comparing program: It displays the offset used for
6151: data addresses and the offset used for code addresses;
6152: moreover, for each cell that cannot be represented correctly in the
6153: image files, it displays a line like the following one:
6154:
6155: @example
6156: 78DC BFFFFA50 BFFFFA40
6157: @end example
6158:
6159: This means that at offset $78dc from @code{forthstart}, one input image
6160: contains $bffffa50, and the other contains $bffffa40. Since these cells
6161: cannot be represented correctly in the output image, you should examine
6162: these places in the dictionary and verify that these cells are dead
6163: (i.e., not read before they are written).
6164:
6165: @cindex @code{savesystem} during @file{gforthmi}
6166: @cindex @code{bye} during @file{gforthmi}
6167: @cindex doubly indirect threaded code
6168: @cindex environment variable @code{GFORTHD}
6169: @cindex @code{GFORTHD} environment variable
6170: @cindex @code{gforth-ditc}
6171: There are a few wrinkles: After processing the passed @var{options}, the
6172: words @code{savesystem} and @code{bye} must be visible. A special doubly
6173: indirect threaded version of the @file{gforth} executable is used for
6174: creating the nonrelocatable images; you can pass the exact filename of
6175: this executable through the environment variable @code{GFORTHD}
6176: (default: @file{gforth-ditc}); if you pass a version that is not doubly
6177: indirect threaded, you will not get a fully relocatable image, but a
6178: data-relocatable image (because there is no code address offset).
6179:
6180: @node cross.fs, , gforthmi, Fully Relocatable Image Files
6181: @subsection @file{cross.fs}
6182: @cindex @file{cross.fs}
6183: @cindex cross-compiler
6184: @cindex metacompiler
6185:
6186: You can also use @code{cross}, a batch compiler that accepts a Forth-like
6187: programming language. This @code{cross} language has to be documented
6188: yet.
6189:
6190: @cindex target compiler
6191: @code{cross} also allows you to create image files for machines with
6192: different data sizes and data formats than the one used for generating
6193: the image file. You can also use it to create an application image that
6194: does not contain a Forth compiler. These features are bought with
6195: restrictions and inconveniences in programming. E.g., addresses have to
6196: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
6197: order to make the code relocatable.
6198:
6199:
6200: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
6201: @section Stack and Dictionary Sizes
6202: @cindex image file, stack and dictionary sizes
6203: @cindex dictionary size default
6204: @cindex stack size default
6205:
6206: If you invoke Gforth with a command line flag for the size
6207: (@pxref{Invoking Gforth}), the size you specify is stored in the
6208: dictionary. If you save the dictionary with @code{savesystem} or create
6209: an image with @file{gforthmi}, this size will become the default
6210: for the resulting image file. E.g., the following will create a
6211: fully relocatable version of gforth.fi with a 1MB dictionary:
6212:
6213: @example
6214: gforthmi gforth.fi -m 1M
6215: @end example
6216:
6217: In other words, if you want to set the default size for the dictionary
6218: and the stacks of an image, just invoke @file{gforthmi} with the
6219: appropriate options when creating the image.
6220:
6221: @cindex stack size, cache-friendly
6222: Note: For cache-friendly behaviour (i.e., good performance), you should
6223: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
6224: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
6225: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
6226:
6227: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
6228: @section Running Image Files
6229: @cindex running image files
6230: @cindex invoking image files
6231: @cindex image file invocation
6232:
6233: @cindex -i, invoke image file
6234: @cindex --image file, invoke image file
6235: You can invoke Gforth with an image file @var{image} instead of the
6236: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
6237: @example
6238: gforth -i @var{image}
6239: @end example
6240:
6241: @cindex executable image file
6242: @cindex image files, executable
6243: If your operating system supports starting scripts with a line of the
6244: form @code{#! ...}, you just have to type the image file name to start
6245: Gforth with this image file (note that the file extension @code{.fi} is
6246: just a convention). I.e., to run Gforth with the image file @var{image},
6247: you can just type @var{image} instead of @code{gforth -i @var{image}}.
6248:
6249: doc-#!
6250:
6251: @node Modifying the Startup Sequence, , Running Image Files, Image Files
6252: @section Modifying the Startup Sequence
6253: @cindex startup sequence for image file
6254: @cindex image file initialization sequence
6255: @cindex initialization sequence of image file
6256:
6257: You can add your own initialization to the startup sequence through the
6258: deferred word
6259:
6260: doc-'cold
6261:
6262: @code{'cold} is invoked just before the image-specific command line
6263: processing (by default, loading files and evaluating (@code{-e}) strings)
6264: starts.
6265:
6266: A sequence for adding your initialization usually looks like this:
6267:
6268: @example
6269: :noname
6270: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
6271: ... \ your stuff
6272: ; IS 'cold
6273: @end example
6274:
6275: @cindex turnkey image files
6276: @cindex image files, turnkey applications
6277: You can make a turnkey image by letting @code{'cold} execute a word
6278: (your turnkey application) that never returns; instead, it exits Gforth
6279: via @code{bye} or @code{throw}.
6280:
6281: @cindex command-line arguments, access
6282: @cindex arguments on the command line, access
6283: You can access the (image-specific) command-line arguments through the
6284: variables @code{argc} and @code{argv}. @code{arg} provides conventient
6285: access to @code{argv}.
6286:
6287: doc-argc
6288: doc-argv
6289: doc-arg
6290:
6291: If @code{'cold} exits normally, Gforth processes the command-line
6292: arguments as files to be loaded and strings to be evaluated. Therefore,
6293: @code{'cold} should remove the arguments it has used in this case.
6294:
6295: @c ******************************************************************
6296: @node Engine, Bugs, Image Files, Top
6297: @chapter Engine
6298: @cindex engine
6299: @cindex virtual machine
6300:
6301: Reading this section is not necessary for programming with Gforth. It
6302: may be helpful for finding your way in the Gforth sources.
6303:
6304: The ideas in this section have also been published in the papers
6305: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
6306: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
6307: Ertl, presented at EuroForth '93; the latter is available at
6308: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
6309:
6310: @menu
6311: * Portability::
6312: * Threading::
6313: * Primitives::
6314: * Performance::
6315: @end menu
6316:
6317: @node Portability, Threading, Engine, Engine
6318: @section Portability
6319: @cindex engine portability
6320:
6321: One of the main goals of the effort is availability across a wide range
6322: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
6323: this goal by manually coding the engine in assembly language for several
6324: then-popular processors. This approach is very labor-intensive and the
6325: results are short-lived due to progress in computer architecture.
6326:
6327: @cindex C, using C for the engine
6328: Others have avoided this problem by coding in C, e.g., Mitch Bradley
6329: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
6330: particularly popular for UNIX-based Forths due to the large variety of
6331: architectures of UNIX machines. Unfortunately an implementation in C
6332: does not mix well with the goals of efficiency and with using
6333: traditional techniques: Indirect or direct threading cannot be expressed
6334: in C, and switch threading, the fastest technique available in C, is
6335: significantly slower. Another problem with C is that it is very
6336: cumbersome to express double integer arithmetic.
6337:
6338: @cindex GNU C for the engine
6339: @cindex long long
6340: Fortunately, there is a portable language that does not have these
6341: limitations: GNU C, the version of C processed by the GNU C compiler
6342: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
6343: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
6344: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
6345: threading possible, its @code{long long} type (@pxref{Long Long, ,
6346: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
6347: double numbers@footnote{Unfortunately, long longs are not implemented
6348: properly on all machines (e.g., on alpha-osf1, long longs are only 64
6349: bits, the same size as longs (and pointers), but they should be twice as
6350: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
6351: C Manual}). So, we had to implement doubles in C after all. Still, on
6352: most machines we can use long longs and achieve better performance than
6353: with the emulation package.}. GNU C is available for free on all
6354: important (and many unimportant) UNIX machines, VMS, 80386s running
6355: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
6356: on all these machines.
6357:
6358: Writing in a portable language has the reputation of producing code that
6359: is slower than assembly. For our Forth engine we repeatedly looked at
6360: the code produced by the compiler and eliminated most compiler-induced
6361: inefficiencies by appropriate changes in the source code.
6362:
6363: @cindex explicit register declarations
6364: @cindex --enable-force-reg, configuration flag
6365: @cindex -DFORCE_REG
6366: However, register allocation cannot be portably influenced by the
6367: programmer, leading to some inefficiencies on register-starved
6368: machines. We use explicit register declarations (@pxref{Explicit Reg
6369: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
6370: improve the speed on some machines. They are turned on by using the
6371: configuration flag @code{--enable-force-reg} (@code{gcc} switch
6372: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
6373: machine, but also on the compiler version: On some machines some
6374: compiler versions produce incorrect code when certain explicit register
6375: declarations are used. So by default @code{-DFORCE_REG} is not used.
6376:
6377: @node Threading, Primitives, Portability, Engine
6378: @section Threading
6379: @cindex inner interpreter implementation
6380: @cindex threaded code implementation
6381:
6382: @cindex labels as values
6383: GNU C's labels as values extension (available since @code{gcc-2.0},
6384: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
6385: makes it possible to take the address of @var{label} by writing
6386: @code{&&@var{label}}. This address can then be used in a statement like
6387: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
6388: @code{goto x}.
6389:
6390: @cindex NEXT, indirect threaded
6391: @cindex indirect threaded inner interpreter
6392: @cindex inner interpreter, indirect threaded
6393: With this feature an indirect threaded NEXT looks like:
6394: @example
6395: cfa = *ip++;
6396: ca = *cfa;
6397: goto *ca;
6398: @end example
6399: @cindex instruction pointer
6400: For those unfamiliar with the names: @code{ip} is the Forth instruction
6401: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
6402: execution token and points to the code field of the next word to be
6403: executed; The @code{ca} (code address) fetched from there points to some
6404: executable code, e.g., a primitive or the colon definition handler
6405: @code{docol}.
6406:
6407: @cindex NEXT, direct threaded
6408: @cindex direct threaded inner interpreter
6409: @cindex inner interpreter, direct threaded
6410: Direct threading is even simpler:
6411: @example
6412: ca = *ip++;
6413: goto *ca;
6414: @end example
6415:
6416: Of course we have packaged the whole thing neatly in macros called
6417: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
6418:
6419: @menu
6420: * Scheduling::
6421: * Direct or Indirect Threaded?::
6422: * DOES>::
6423: @end menu
6424:
6425: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
6426: @subsection Scheduling
6427: @cindex inner interpreter optimization
6428:
6429: There is a little complication: Pipelined and superscalar processors,
6430: i.e., RISC and some modern CISC machines can process independent
6431: instructions while waiting for the results of an instruction. The
6432: compiler usually reorders (schedules) the instructions in a way that
6433: achieves good usage of these delay slots. However, on our first tries
6434: the compiler did not do well on scheduling primitives. E.g., for
6435: @code{+} implemented as
6436: @example
6437: n=sp[0]+sp[1];
6438: sp++;
6439: sp[0]=n;
6440: NEXT;
6441: @end example
6442: the NEXT comes strictly after the other code, i.e., there is nearly no
6443: scheduling. After a little thought the problem becomes clear: The
6444: compiler cannot know that sp and ip point to different addresses (and
6445: the version of @code{gcc} we used would not know it even if it was
6446: possible), so it could not move the load of the cfa above the store to
6447: the TOS. Indeed the pointers could be the same, if code on or very near
6448: the top of stack were executed. In the interest of speed we chose to
6449: forbid this probably unused ``feature'' and helped the compiler in
6450: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
6451: the goto part (@code{NEXT_P2}). @code{+} now looks like:
6452: @example
6453: n=sp[0]+sp[1];
6454: sp++;
6455: NEXT_P1;
6456: sp[0]=n;
6457: NEXT_P2;
6458: @end example
6459: This can be scheduled optimally by the compiler.
6460:
6461: This division can be turned off with the switch @code{-DCISC_NEXT}. This
6462: switch is on by default on machines that do not profit from scheduling
6463: (e.g., the 80386), in order to preserve registers.
6464:
6465: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
6466: @subsection Direct or Indirect Threaded?
6467: @cindex threading, direct or indirect?
6468:
6469: @cindex -DDIRECT_THREADED
6470: Both! After packaging the nasty details in macro definitions we
6471: realized that we could switch between direct and indirect threading by
6472: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
6473: defining a few machine-specific macros for the direct-threading case.
6474: On the Forth level we also offer access words that hide the
6475: differences between the threading methods (@pxref{Threading Words}).
6476:
6477: Indirect threading is implemented completely machine-independently.
6478: Direct threading needs routines for creating jumps to the executable
6479: code (e.g. to docol or dodoes). These routines are inherently
6480: machine-dependent, but they do not amount to many source lines. I.e.,
6481: even porting direct threading to a new machine is a small effort.
6482:
6483: @cindex --enable-indirect-threaded, configuration flag
6484: @cindex --enable-direct-threaded, configuration flag
6485: The default threading method is machine-dependent. You can enforce a
6486: specific threading method when building Gforth with the configuration
6487: flag @code{--enable-direct-threaded} or
6488: @code{--enable-indirect-threaded}. Note that direct threading is not
6489: supported on all machines.
6490:
6491: @node DOES>, , Direct or Indirect Threaded?, Threading
6492: @subsection DOES>
6493: @cindex @code{DOES>} implementation
6494:
6495: @cindex dodoes routine
6496: @cindex DOES-code
6497: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
6498: the chunk of code executed by every word defined by a
6499: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
6500: the Forth code to be executed, i.e. the code after the
6501: @code{DOES>} (the DOES-code)? There are two solutions:
6502:
6503: In fig-Forth the code field points directly to the dodoes and the
6504: DOES-code address is stored in the cell after the code address (i.e. at
6505: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
6506: the Forth-79 and all later standards, because in fig-Forth this address
6507: lies in the body (which is illegal in these standards). However, by
6508: making the code field larger for all words this solution becomes legal
6509: again. We use this approach for the indirect threaded version and for
6510: direct threading on some machines. Leaving a cell unused in most words
6511: is a bit wasteful, but on the machines we are targeting this is hardly a
6512: problem. The other reason for having a code field size of two cells is
6513: to avoid having different image files for direct and indirect threaded
6514: systems (direct threaded systems require two-cell code fields on many
6515: machines).
6516:
6517: @cindex DOES-handler
6518: The other approach is that the code field points or jumps to the cell
6519: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
6520: this address (the DOES-handler). @code{dodoes} can then get the
6521: DOES-code address by computing the code address, i.e., the address of
6522: the jump to dodoes, and add the length of that jump field. A variant of
6523: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
6524: return address (which can be found in the return register on RISCs) is
6525: the DOES-code address. Since the two cells available in the code field
6526: are used up by the jump to the code address in direct threading on many
6527: architectures, we use this approach for direct threading on these
6528: architectures. We did not want to add another cell to the code field.
6529:
6530: @node Primitives, Performance, Threading, Engine
6531: @section Primitives
6532: @cindex primitives, implementation
6533: @cindex virtual machine instructions, implementation
6534:
6535: @menu
6536: * Automatic Generation::
6537: * TOS Optimization::
6538: * Produced code::
6539: @end menu
6540:
6541: @node Automatic Generation, TOS Optimization, Primitives, Primitives
6542: @subsection Automatic Generation
6543: @cindex primitives, automatic generation
6544:
6545: @cindex @file{prims2x.fs}
6546: Since the primitives are implemented in a portable language, there is no
6547: longer any need to minimize the number of primitives. On the contrary,
6548: having many primitives has an advantage: speed. In order to reduce the
6549: number of errors in primitives and to make programming them easier, we
6550: provide a tool, the primitive generator (@file{prims2x.fs}), that
6551: automatically generates most (and sometimes all) of the C code for a
6552: primitive from the stack effect notation. The source for a primitive
6553: has the following form:
6554:
6555: @cindex primitive source format
6556: @format
6557: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
6558: [@code{""}@var{glossary entry}@code{""}]
6559: @var{C code}
6560: [@code{:}
6561: @var{Forth code}]
6562: @end format
6563:
6564: The items in brackets are optional. The category and glossary fields
6565: are there for generating the documentation, the Forth code is there
6566: for manual implementations on machines without GNU C. E.g., the source
6567: for the primitive @code{+} is:
6568: @example
6569: + n1 n2 -- n core plus
6570: n = n1+n2;
6571: @end example
6572:
6573: This looks like a specification, but in fact @code{n = n1+n2} is C
6574: code. Our primitive generation tool extracts a lot of information from
6575: the stack effect notations@footnote{We use a one-stack notation, even
6576: though we have separate data and floating-point stacks; The separate
6577: notation can be generated easily from the unified notation.}: The number
6578: of items popped from and pushed on the stack, their type, and by what
6579: name they are referred to in the C code. It then generates a C code
6580: prelude and postlude for each primitive. The final C code for @code{+}
6581: looks like this:
6582:
6583: @example
6584: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
6585: /* */ /* documentation */
6586: @{
6587: DEF_CA /* definition of variable ca (indirect threading) */
6588: Cell n1; /* definitions of variables */
6589: Cell n2;
6590: Cell n;
6591: n1 = (Cell) sp[1]; /* input */
6592: n2 = (Cell) TOS;
6593: sp += 1; /* stack adjustment */
6594: NAME("+") /* debugging output (with -DDEBUG) */
6595: @{
6596: n = n1+n2; /* C code taken from the source */
6597: @}
6598: NEXT_P1; /* NEXT part 1 */
6599: TOS = (Cell)n; /* output */
6600: NEXT_P2; /* NEXT part 2 */
6601: @}
6602: @end example
6603:
6604: This looks long and inefficient, but the GNU C compiler optimizes quite
6605: well and produces optimal code for @code{+} on, e.g., the R3000 and the
6606: HP RISC machines: Defining the @code{n}s does not produce any code, and
6607: using them as intermediate storage also adds no cost.
6608:
6609: There are also other optimizations, that are not illustrated by this
6610: example: Assignments between simple variables are usually for free (copy
6611: propagation). If one of the stack items is not used by the primitive
6612: (e.g. in @code{drop}), the compiler eliminates the load from the stack
6613: (dead code elimination). On the other hand, there are some things that
6614: the compiler does not do, therefore they are performed by
6615: @file{prims2x.fs}: The compiler does not optimize code away that stores
6616: a stack item to the place where it just came from (e.g., @code{over}).
6617:
6618: While programming a primitive is usually easy, there are a few cases
6619: where the programmer has to take the actions of the generator into
6620: account, most notably @code{?dup}, but also words that do not (always)
6621: fall through to NEXT.
6622:
6623: @node TOS Optimization, Produced code, Automatic Generation, Primitives
6624: @subsection TOS Optimization
6625: @cindex TOS optimization for primitives
6626: @cindex primitives, keeping the TOS in a register
6627:
6628: An important optimization for stack machine emulators, e.g., Forth
6629: engines, is keeping one or more of the top stack items in
6630: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
6631: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
6632: @itemize @bullet
6633: @item
6634: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
6635: due to fewer loads from and stores to the stack.
6636: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
6637: @var{y<n}, due to additional moves between registers.
6638: @end itemize
6639:
6640: @cindex -DUSE_TOS
6641: @cindex -DUSE_NO_TOS
6642: In particular, keeping one item in a register is never a disadvantage,
6643: if there are enough registers. Keeping two items in registers is a
6644: disadvantage for frequent words like @code{?branch}, constants,
6645: variables, literals and @code{i}. Therefore our generator only produces
6646: code that keeps zero or one items in registers. The generated C code
6647: covers both cases; the selection between these alternatives is made at
6648: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
6649: code for @code{+} is just a simple variable name in the one-item case,
6650: otherwise it is a macro that expands into @code{sp[0]}. Note that the
6651: GNU C compiler tries to keep simple variables like @code{TOS} in
6652: registers, and it usually succeeds, if there are enough registers.
6653:
6654: @cindex -DUSE_FTOS
6655: @cindex -DUSE_NO_FTOS
6656: The primitive generator performs the TOS optimization for the
6657: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
6658: operations the benefit of this optimization is even larger:
6659: floating-point operations take quite long on most processors, but can be
6660: performed in parallel with other operations as long as their results are
6661: not used. If the FP-TOS is kept in a register, this works. If
6662: it is kept on the stack, i.e., in memory, the store into memory has to
6663: wait for the result of the floating-point operation, lengthening the
6664: execution time of the primitive considerably.
6665:
6666: The TOS optimization makes the automatic generation of primitives a
6667: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
6668: @code{TOS} is not sufficient. There are some special cases to
6669: consider:
6670: @itemize @bullet
6671: @item In the case of @code{dup ( w -- w w )} the generator must not
6672: eliminate the store to the original location of the item on the stack,
6673: if the TOS optimization is turned on.
6674: @item Primitives with stack effects of the form @code{--}
6675: @var{out1}...@var{outy} must store the TOS to the stack at the start.
6676: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
6677: must load the TOS from the stack at the end. But for the null stack
6678: effect @code{--} no stores or loads should be generated.
6679: @end itemize
6680:
6681: @node Produced code, , TOS Optimization, Primitives
6682: @subsection Produced code
6683: @cindex primitives, assembly code listing
6684:
6685: @cindex @file{engine.s}
6686: To see what assembly code is produced for the primitives on your machine
6687: with your compiler and your flag settings, type @code{make engine.s} and
6688: look at the resulting file @file{engine.s}.
6689:
6690: @node Performance, , Primitives, Engine
6691: @section Performance
6692: @cindex performance of some Forth interpreters
6693: @cindex engine performance
6694: @cindex benchmarking Forth systems
6695: @cindex Gforth performance
6696:
6697: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
6698: impossible to write a significantly faster engine.
6699:
6700: On register-starved machines like the 386 architecture processors
6701: improvements are possible, because @code{gcc} does not utilize the
6702: registers as well as a human, even with explicit register declarations;
6703: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
6704: and hand-tuned it for the 486; this system is 1.19 times faster on the
6705: Sieve benchmark on a 486DX2/66 than Gforth compiled with
6706: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
6707:
6708: @cindex Win32Forth performance
6709: @cindex NT Forth performance
6710: @cindex eforth performance
6711: @cindex ThisForth performance
6712: @cindex PFE performance
6713: @cindex TILE performance
6714: However, this potential advantage of assembly language implementations
6715: is not necessarily realized in complete Forth systems: We compared
6716: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
6717: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
6718: 1994) and Eforth (with and without peephole (aka pinhole) optimization
6719: of the threaded code); all these systems were written in assembly
6720: language. We also compared Gforth with three systems written in C:
6721: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
6722: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
6723: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
6724: -fomit-frame-pointer; ThisForth employs peephole optimization of the
6725: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
6726: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
6727: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
6728: 486DX2/66 with similar memory performance under Windows NT. Marcel
6729: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
6730: added the peephole optimizer, ran the benchmarks and reported the
6731: results.
6732:
6733: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
6734: matrix multiplication come from the Stanford integer benchmarks and have
6735: been translated into Forth by Martin Fraeman; we used the versions
6736: included in the TILE Forth package, but with bigger data set sizes; and
6737: a recursive Fibonacci number computation for benchmarking calling
6738: performance. The following table shows the time taken for the benchmarks
6739: scaled by the time taken by Gforth (in other words, it shows the speedup
6740: factor that Gforth achieved over the other systems).
6741:
6742: @example
6743: relative Win32- NT eforth This-
6744: time Gforth Forth Forth eforth +opt PFE Forth TILE
6745: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
6746: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
6747: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
6748: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
6749: @end example
6750:
6751: You may find the good performance of Gforth compared with the systems
6752: written in assembly language quite surprising. One important reason for
6753: the disappointing performance of these systems is probably that they are
6754: not written optimally for the 486 (e.g., they use the @code{lods}
6755: instruction). In addition, Win32Forth uses a comfortable, but costly
6756: method for relocating the Forth image: like @code{cforth}, it computes
6757: the actual addresses at run time, resulting in two address computations
6758: per NEXT (@pxref{Image File Background}).
6759:
6760: Only Eforth with the peephole optimizer performs comparable to
6761: Gforth. The speedups achieved with peephole optimization of threaded
6762: code are quite remarkable. Adding a peephole optimizer to Gforth should
6763: cause similar speedups.
6764:
6765: The speedup of Gforth over PFE, ThisForth and TILE can be easily
6766: explained with the self-imposed restriction of the latter systems to
6767: standard C, which makes efficient threading impossible (however, the
6768: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
6769: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
6770: Moreover, current C compilers have a hard time optimizing other aspects
6771: of the ThisForth and the TILE source.
6772:
6773: Note that the performance of Gforth on 386 architecture processors
6774: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
6775: failed to allocate any of the virtual machine registers into real
6776: machine registers by itself and would not work correctly with explicit
6777: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
6778: running the Sieve) than the one measured above.
6779:
6780: Note also that there have been several releases of Win32Forth since the
6781: release presented here, so the results presented here may have little
6782: predictive value for the performance of Win32Forth today.
6783:
6784: @cindex @file{Benchres}
6785: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
6786: Maierhofer (presented at EuroForth '95), an indirect threaded version of
6787: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
6788: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
6789: threaded version used here. The paper available at
6790: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
6791: it also contains numbers for some native code systems. You can find a
6792: newer version of these measurements at
6793: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
6794: find numbers for Gforth on various machines in @file{Benchres}.
6795:
6796: @node Bugs, Origin, Engine, Top
6797: @chapter Bugs
6798: @cindex bug reporting
6799:
6800: Known bugs are described in the file BUGS in the Gforth distribution.
6801:
6802: If you find a bug, please send a bug report to
6803: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should
6804: describe the Gforth version used (it is announced at the start of an
6805: interactive Gforth session), the machine and operating system (on Unix
6806: systems you can use @code{uname -a} to produce this information), the
6807: installation options (send the @file{config.status} file), and a
6808: complete list of changes you (or your installer) have made to the Gforth
6809: sources (if any); it should contain a program (or a sequence of keyboard
6810: commands) that reproduces the bug and a description of what you think
6811: constitutes the buggy behaviour.
6812:
6813: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
6814: to Report Bugs, gcc.info, GNU C Manual}.
6815:
6816:
6817: @node Origin, Word Index, Bugs, Top
6818: @chapter Authors and Ancestors of Gforth
6819:
6820: @section Authors and Contributors
6821: @cindex authors of Gforth
6822: @cindex contributors to Gforth
6823:
6824: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
6825: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
6826: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
6827: with their continuous feedback. Lennart Benshop contributed
6828: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
6829: support for calling C libraries. Helpful comments also came from Paul
6830: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
6831: Wavrik, Barrie Stott and Marc de Groot.
6832:
6833: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
6834: and autoconf, among others), and to the creators of the Internet: Gforth
6835: was developed across the Internet, and its authors have not met
6836: physically yet.
6837:
6838: @section Pedigree
6839: @cindex Pedigree of Gforth
6840:
6841: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
6842: Dirk Zoller) will cross-fertilize each other. Of course, a significant
6843: part of the design of Gforth was prescribed by ANS Forth.
6844:
6845: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
6846: 32 bit native code version of VolksForth for the Atari ST, written
6847: mostly by Dietrich Weineck.
6848:
6849: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
6850: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
6851: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
6852:
6853: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
6854: Forth-83 standard. !! Pedigree? When?
6855:
6856: A team led by Bill Ragsdale implemented fig-Forth on many processors in
6857: 1979. Robert Selzer and Bill Ragsdale developed the original
6858: implementation of fig-Forth for the 6502 based on microForth.
6859:
6860: The principal architect of microForth was Dean Sanderson. microForth was
6861: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
6862: the 1802, and subsequently implemented on the 8080, the 6800 and the
6863: Z80.
6864:
6865: All earlier Forth systems were custom-made, usually by Charles Moore,
6866: who discovered (as he puts it) Forth during the late 60s. The first full
6867: Forth existed in 1971.
6868:
6869: A part of the information in this section comes from @cite{The Evolution
6870: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
6871: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
6872: Notices 28(3), 1993. You can find more historical and genealogical
6873: information about Forth there.
6874:
6875: @node Word Index, Concept Index, Origin, Top
6876: @unnumbered Word Index
6877:
6878: This index is as incomplete as the manual. Each word is listed with
6879: stack effect and wordset.
6880:
6881: @printindex fn
6882:
6883: @node Concept Index, , Word Index, Top
6884: @unnumbered Concept and Word Index
6885:
6886: This index is as incomplete as the manual. Not all entries listed are
6887: present verbatim in the text. Only the names are listed for the words
6888: here.
6889:
6890: @printindex cp
6891:
6892: @contents
6893: @bye
6894:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gforth.ds CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth / doc