Annotation of gforth/doc/gforth.ds, revision 1.8
1.1 anton 1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3: @comment %**start of header (This is for running Texinfo on a region.)
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
1.3 anton 11: @macro progstyle {TEXT}
12: @quotation
13: {Programming style note:} \TEXT\
14: @end quotation
15: @end macro
1.1 anton 16: @comment %**end of header (This is for running Texinfo on a region.)
17:
18: @ifinfo
1.5 anton 19: This file documents Gforth 0.4
1.1 anton 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
1.5 anton 53: @center for version 0.4
1.1 anton 54: @sp 2
55: @center Anton Ertl
1.6 pazsan 56: @center Bernd Paysan
1.5 anton 57: @center Jens Wilke
1.1 anton 58: @sp 3
1.6 pazsan 59: @center This manual is permanently under construction
1.1 anton 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
1.5 anton 90: personal machines. This manual corresponds to version 0.4.
1.1 anton 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:
1.5 anton 699: @cindex --no-offset-im, command-line option
700: @item --no-offset-im
701: Start the dictionary at the normal position.
702:
1.1 anton 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}).
1.5 anton 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.
1.1 anton 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::
1.5 anton 754: * Memory::
1.1 anton 755: * Control Structures::
756: * Locals::
757: * Defining Words::
1.5 anton 758: * Structures::
1.6 pazsan 759: * Objects::
760: * Object Oriented Forth::
1.1 anton 761: * Tokens for Words::
762: * Wordlists::
763: * Files::
764: * Blocks::
765: * Other I/O::
766: * Programming Tools::
1.7 pazsan 767: * Assembler and Code Words::
1.1 anton 768: * Threading Words::
1.5 anton 769: * Including Files::
1.1 anton 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:
1.3 anton 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
1.1 anton 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:
1.5 anton 2208: @node Defining Words, Structures, Locals, Words
1.1 anton 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:
1.5 anton 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 -------------------------------------------------------------
1.6 pazsan 2943: @node Objects, Object Oriented Forth, Structures, Words
1.5 anton 2944: @section Objects
2945: @cindex objects
2946: @cindex object-oriented programming
2947:
2948: @cindex @file{objects.fs}
2949: @cindex @file{oof.fs}
1.8 ! pazsan 2950: Gforth comes with three packets for object-oriented programming,
! 2951: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
! 2952: is preloaded, so you have to @code{include} them before use. This
! 2953: section describes the @file{objects.fs} packet. You can find a
! 2954: description (in German) of @file{oof.fs} in @cite{Object oriented
! 2955: bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension} 10(2),
! 2956: 1994. Both packets are written in ANS Forth and can be used with any
! 2957: other standard Forth.
1.5 anton 2958: @c McKewan's and Zsoter's packages
2959: @c this section is a variant of ...
2960:
2961: This section assumes (in some places) that you have read @ref{Structures}.
2962:
2963: @menu
2964: * Properties of the Objects model::
2965: * Why object-oriented programming?::
2966: * Object-Oriented Terminology::
2967: * Basic Objects Usage::
2968: * The class Object::
2969: * Creating objects::
2970: * Object-Oriented Programming Style::
2971: * Class Binding::
2972: * Method conveniences::
2973: * Classes and Scoping::
2974: * Object Interfaces::
2975: * Objects Implementation::
2976: * Comparison with other object models::
2977: * Objects Glossary::
2978: @end menu
2979:
2980: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
2981: and Bernd Paysan helped me with the related works section.
2982:
2983: @node Properties of the Objects model, Why object-oriented programming?, Objects, Objects
2984: @subsection Properties of the @file{objects.fs} model
2985: @cindex @file{objects.fs} properties
2986:
2987: @itemize @bullet
2988: @item
2989: It is straightforward to pass objects on the stack. Passing
2990: selectors on the stack is a little less convenient, but possible.
2991:
2992: @item
2993: Objects are just data structures in memory, and are referenced by
2994: their address. You can create words for objects with normal defining
2995: words like @code{constant}. Likewise, there is no difference
2996: between instance variables that contain objects and those
2997: that contain other data.
2998:
2999: @item
3000: Late binding is efficient and easy to use.
3001:
3002: @item
3003: It avoids parsing, and thus avoids problems with state-smartness
3004: and reduced extensibility; for convenience there are a few parsing
3005: words, but they have non-parsing counterparts. There are also a few
3006: defining words that parse. This is hard to avoid, because all standard
3007: defining words parse (except @code{:noname}); however, such
3008: words are not as bad as many other parsing words, because they are not
3009: state-smart.
3010:
3011: @item
3012: It does not try to incorporate everything. It does a few things
3013: and does them well (IMO). In particular, I did not intend to support
3014: information hiding with this model (although it has features that may
3015: help); you can use a separate package for achieving this.
3016:
3017: @item
3018: It is layered; you don't have to learn and use all features to use this
3019: model. Only a few features are necessary (@xref{Basic Objects Usage},
3020: @xref{The class Object}, @xref{Creating objects}.), the others
3021: are optional and independent of each other.
3022:
3023: @item
3024: An implementation in ANS Forth is available.
3025:
3026: @end itemize
3027:
3028: I have used the technique, on which this model is based, for
3029: implementing the parser generator Gray; we have also used this technique
3030: in Gforth for implementing the various flavours of wordlists (hashed or
3031: not, case-sensitive or not, special-purpose wordlists for locals etc.).
3032:
3033: @node Why object-oriented programming?, Object-Oriented Terminology, Properties of the Objects model, Objects
3034: @subsection Why object-oriented programming?
3035: @cindex object-oriented programming motivation
3036: @cindex motivation for object-oriented programming
3037:
3038: Often we have to deal with several data structures (@emph{objects}),
3039: that have to be treated similarly in some respects, but differ in
3040: others. Graphical objects are the textbook example: circles,
3041: triangles, dinosaurs, icons, and others, and we may want to add more
3042: during program development. We want to apply some operations to any
3043: graphical object, e.g., @code{draw} for displaying it on the
3044: screen. However, @code{draw} has to do something different for
3045: every kind of object.
3046:
3047: We could implement @code{draw} as a big @code{CASE}
3048: control structure that executes the appropriate code depending on the
3049: kind of object to be drawn. This would be not be very elegant, and,
3050: moreover, we would have to change @code{draw} every time we add
3051: a new kind of graphical object (say, a spaceship).
3052:
3053: What we would rather do is: When defining spaceships, we would tell
3054: the system: "Here's how you @code{draw} a spaceship; you figure
3055: out the rest."
3056:
3057: This is the problem that all systems solve that (rightfully) call
3058: themselves object-oriented, and the object-oriented package I present
3059: here also solves this problem (and not much else).
3060:
3061: @node Object-Oriented Terminology, Basic Objects Usage, Why object-oriented programming?, Objects
3062: @subsection Object-Oriented Terminology
3063: @cindex object-oriented terminology
3064: @cindex terminology for object-oriented programming
3065:
3066: This section is mainly for reference, so you don't have to understand
3067: all of it right away. The terminology is mainly Smalltalk-inspired. In
3068: short:
3069:
3070: @table @emph
3071: @cindex class
3072: @item class
3073: a data structure definition with some extras.
3074:
3075: @cindex object
3076: @item object
3077: an instance of the data structure described by the class definition.
3078:
3079: @cindex instance variables
3080: @item instance variables
3081: fields of the data structure.
3082:
3083: @cindex selector
3084: @cindex method selector
3085: @cindex virtual function
3086: @item selector
3087: (or @emph{method selector}) a word (e.g.,
3088: @code{draw}) for performing an operation on a variety of data
3089: structures (classes). A selector describes @emph{what} operation to
3090: perform. In C++ terminology: a (pure) virtual function.
3091:
3092: @cindex method
3093: @item method
3094: the concrete definition that performs the operation
3095: described by the selector for a specific class. A method specifies
3096: @emph{how} the operation is performed for a specific class.
3097:
3098: @cindex selector invocation
3099: @cindex message send
3100: @cindex invoking a selector
3101: @item selector invocation
3102: a call of a selector. One argument of the call (the TOS (top-of-stack))
3103: is used for determining which method is used. In Smalltalk terminology:
3104: a message (consisting of the selector and the other arguments) is sent
3105: to the object.
3106:
3107: @cindex receiving object
3108: @item receiving object
3109: the object used for determining the method executed by a selector
3110: invocation. In our model it is the object that is on the TOS when the
3111: selector is invoked. (@emph{Receiving} comes from Smalltalks
3112: @emph{message} terminology.)
3113:
3114: @cindex child class
3115: @cindex parent class
3116: @cindex inheritance
3117: @item child class
3118: a class that has (@emph{inherits}) all properties (instance variables,
3119: selectors, methods) from a @emph{parent class}. In Smalltalk
3120: terminology: The subclass inherits from the superclass. In C++
3121: terminology: The derived class inherits from the base class.
3122:
3123: @end table
3124:
3125: @c If you wonder about the message sending terminology, it comes from
3126: @c a time when each object had it's own task and objects communicated via
3127: @c message passing; eventually the Smalltalk developers realized that
3128: @c they can do most things through simple (indirect) calls. They kept the
3129: @c terminology.
3130:
3131: @node Basic Objects Usage, The class Object, Object-Oriented Terminology, Objects
3132: @subsection Basic Objects Usage
3133: @cindex basic objects usage
3134: @cindex objects, basic usage
3135:
3136: You can define a class for graphical objects like this:
3137:
3138: @cindex @code{class} usage
3139: @cindex @code{end-class} usage
3140: @cindex @code{selector} usage
3141: @example
3142: object class \ "object" is the parent class
3143: selector draw ( x y graphical -- )
3144: end-class graphical
3145: @end example
3146:
3147: This code defines a class @code{graphical} with an
3148: operation @code{draw}. We can perform the operation
3149: @code{draw} on any @code{graphical} object, e.g.:
3150:
3151: @example
3152: 100 100 t-rex draw
3153: @end example
3154:
3155: where @code{t-rex} is a word (say, a constant) that produces a
3156: graphical object.
3157:
3158: @cindex abstract class
3159: How do we create a graphical object? With the present definitions,
3160: we cannot create a useful graphical object. The class
3161: @code{graphical} describes graphical objects in general, but not
3162: any concrete graphical object type (C++ users would call it an
3163: @emph{abstract class}); e.g., there is no method for the selector
3164: @code{draw} in the class @code{graphical}.
3165:
3166: For concrete graphical objects, we define child classes of the
3167: class @code{graphical}, e.g.:
3168:
3169: @cindex @code{overrides} usage
3170: @cindex @code{field} usage in class definition
3171: @example
3172: graphical class \ "graphical" is the parent class
3173: cell% field circle-radius
3174:
3175: :noname ( x y circle -- )
3176: circle-radius @@ draw-circle ;
3177: overrides draw
3178:
3179: :noname ( n-radius circle -- )
3180: circle-radius ! ;
3181: overrides construct
3182:
3183: end-class circle
3184: @end example
3185:
3186: Here we define a class @code{circle} as a child of @code{graphical},
3187: with a field @code{circle-radius} (which behaves just like a field in
3188: @pxref{Structures}); it defines new methods for the selectors
3189: @code{draw} and @code{construct} (@code{construct} is defined in
3190: @code{object}, the parent class of @code{graphical}).
3191:
3192: Now we can create a circle on the heap (i.e.,
3193: @code{allocate}d memory) with
3194:
3195: @cindex @code{heap-new} usage
3196: @example
3197: 50 circle heap-new constant my-circle
3198: @end example
3199:
3200: @code{heap-new} invokes @code{construct}, thus
3201: initializing the field @code{circle-radius} with 50. We can draw
3202: this new circle at (100,100) with
3203:
3204: @example
3205: 100 100 my-circle draw
3206: @end example
3207:
3208: @cindex selector invocation, restrictions
3209: @cindex class definition, restrictions
3210: Note: You can invoke a selector only if the object on the TOS
3211: (the receiving object) belongs to the class where the selector was
3212: defined or one of its descendents; e.g., you can invoke
3213: @code{draw} only for objects belonging to @code{graphical}
3214: or its descendents (e.g., @code{circle}). Immediately before
3215: @code{end-class}, the search order has to be the same as
3216: immediately after @code{class}.
3217:
3218: @node The class Object, Creating objects, Basic Objects Usage, Objects
3219: @subsection The class @code{object}
3220: @cindex @code{object} class
3221:
3222: When you define a class, you have to specify a parent class. So how do
3223: you start defining classes? There is one class available from the start:
3224: @code{object}. You can use it as ancestor for all classes. It is the
3225: only class that has no parent. It has two selectors: @code{construct}
3226: and @code{print}.
3227:
3228: @node Creating objects, Object-Oriented Programming Style, The class Object, Objects
3229: @subsection Creating objects
3230: @cindex creating objects
3231: @cindex object creation
3232: @cindex object allocation options
3233:
3234: @cindex @code{heap-new} discussion
3235: @cindex @code{dict-new} discussion
3236: @cindex @code{construct} discussion
3237: You can create and initialize an object of a class on the heap with
3238: @code{heap-new} ( ... class -- object ) and in the dictionary
3239: (allocation with @code{allot}) with @code{dict-new} (
3240: ... class -- object ). Both words invoke @code{construct}, which
3241: consumes the stack items indicated by "..." above.
3242:
3243: @cindex @code{init-object} discussion
3244: @cindex @code{class-inst-size} discussion
3245: If you want to allocate memory for an object yourself, you can get its
3246: alignment and size with @code{class-inst-size 2@@} ( class --
3247: align size ). Once you have memory for an object, you can initialize
3248: it with @code{init-object} ( ... class object -- );
3249: @code{construct} does only a part of the necessary work.
3250:
3251: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
3252: @subsection Object-Oriented Programming Style
3253: @cindex object-oriented programming style
3254:
3255: This section is not exhaustive.
3256:
3257: @cindex stack effects of selectors
3258: @cindex selectors and stack effects
3259: In general, it is a good idea to ensure that all methods for the
3260: same selector have the same stack effect: when you invoke a selector,
3261: you often have no idea which method will be invoked, so, unless all
3262: methods have the same stack effect, you will not know the stack effect
3263: of the selector invocation.
3264:
3265: One exception to this rule is methods for the selector
3266: @code{construct}. We know which method is invoked, because we
3267: specify the class to be constructed at the same place. Actually, I
3268: defined @code{construct} as a selector only to give the users a
3269: convenient way to specify initialization. The way it is used, a
3270: mechanism different from selector invocation would be more natural
3271: (but probably would take more code and more space to explain).
3272:
3273: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
3274: @subsection Class Binding
3275: @cindex class binding
3276: @cindex early binding
3277:
3278: @cindex late binding
3279: Normal selector invocations determine the method at run-time depending
3280: on the class of the receiving object (late binding).
3281:
3282: Sometimes we want to invoke a different method. E.g., assume that
3283: you want to use the simple method for @code{print}ing
3284: @code{object}s instead of the possibly long-winded
3285: @code{print} method of the receiver class. You can achieve this
3286: by replacing the invocation of @code{print} with
3287:
3288: @cindex @code{[bind]} usage
3289: @example
3290: [bind] object print
3291: @end example
3292:
3293: in compiled code or
3294:
3295: @cindex @code{bind} usage
3296: @example
3297: bind object print
3298: @end example
3299:
3300: @cindex class binding, alternative to
3301: in interpreted code. Alternatively, you can define the method with a
3302: name (e.g., @code{print-object}), and then invoke it through the
3303: name. Class binding is just a (often more convenient) way to achieve
3304: the same effect; it avoids name clutter and allows you to invoke
3305: methods directly without naming them first.
3306:
3307: @cindex superclass binding
3308: @cindex parent class binding
3309: A frequent use of class binding is this: When we define a method
3310: for a selector, we often want the method to do what the selector does
3311: in the parent class, and a little more. There is a special word for
3312: this purpose: @code{[parent]}; @code{[parent]
3313: @emph{selector}} is equivalent to @code{[bind] @emph{parent
3314: selector}}, where @code{@emph{parent}} is the parent
3315: class of the current class. E.g., a method definition might look like:
3316:
3317: @cindex @code{[parent]} usage
3318: @example
3319: :noname
3320: dup [parent] foo \ do parent's foo on the receiving object
3321: ... \ do some more
3322: ; overrides foo
3323: @end example
3324:
3325: @cindex class binding as optimization
3326: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
3327: March 1997), Andrew McKewan presents class binding as an optimization
3328: technique. I recommend not using it for this purpose unless you are in
3329: an emergency. Late binding is pretty fast with this model anyway, so the
3330: benefit of using class binding is small; the cost of using class binding
3331: where it is not appropriate is reduced maintainability.
3332:
3333: While we are at programming style questions: You should bind
3334: selectors only to ancestor classes of the receiving object. E.g., say,
3335: you know that the receiving object is of class @code{foo} or its
3336: descendents; then you should bind only to @code{foo} and its
3337: ancestors.
3338:
3339: @node Method conveniences, Classes and Scoping, Class Binding, Objects
3340: @subsection Method conveniences
3341: @cindex method conveniences
3342:
3343: In a method you usually access the receiving object pretty often. If
3344: you define the method as a plain colon definition (e.g., with
3345: @code{:noname}), you may have to do a lot of stack
3346: gymnastics. To avoid this, you can define the method with @code{m:
3347: ... ;m}. E.g., you could define the method for
3348: @code{draw}ing a @code{circle} with
3349:
3350: @cindex @code{this} usage
3351: @cindex @code{m:} usage
3352: @cindex @code{;m} usage
3353: @example
3354: m: ( x y circle -- )
3355: ( x y ) this circle-radius @@ draw-circle ;m
3356: @end example
3357:
3358: @cindex @code{exit} in @code{m: ... ;m}
3359: @cindex @code{exitm} discussion
3360: @cindex @code{catch} in @code{m: ... ;m}
3361: When this method is executed, the receiver object is removed from the
3362: stack; you can access it with @code{this} (admittedly, in this
3363: example the use of @code{m: ... ;m} offers no advantage). Note
3364: that I specify the stack effect for the whole method (i.e. including
3365: the receiver object), not just for the code between @code{m:}
3366: and @code{;m}. You cannot use @code{exit} in
3367: @code{m:...;m}; instead, use
3368: @code{exitm}.@footnote{Moreover, for any word that calls
3369: @code{catch} and was defined before loading
3370: @code{objects.fs}, you have to redefine it like I redefined
3371: @code{catch}: @code{: catch this >r catch r> to-this ;}}
3372:
3373: @cindex @code{inst-var} usage
3374: You will frequently use sequences of the form @code{this
3375: @emph{field}} (in the example above: @code{this
3376: circle-radius}). If you use the field only in this way, you can
3377: define it with @code{inst-var} and eliminate the
3378: @code{this} before the field name. E.g., the @code{circle}
3379: class above could also be defined with:
3380:
3381: @example
3382: graphical class
3383: cell% inst-var radius
3384:
3385: m: ( x y circle -- )
3386: radius @@ draw-circle ;m
3387: overrides draw
3388:
3389: m: ( n-radius circle -- )
3390: radius ! ;m
3391: overrides construct
3392:
3393: end-class circle
3394: @end example
3395:
3396: @code{radius} can only be used in @code{circle} and its
3397: descendent classes and inside @code{m:...;m}.
3398:
3399: @cindex @code{inst-value} usage
3400: You can also define fields with @code{inst-value}, which is
3401: to @code{inst-var} what @code{value} is to
3402: @code{variable}. You can change the value of such a field with
3403: @code{[to-inst]}. E.g., we could also define the class
3404: @code{circle} like this:
3405:
3406: @example
3407: graphical class
3408: inst-value radius
3409:
3410: m: ( x y circle -- )
3411: radius draw-circle ;m
3412: overrides draw
3413:
3414: m: ( n-radius circle -- )
3415: [to-inst] radius ;m
3416: overrides construct
3417:
3418: end-class circle
3419: @end example
3420:
3421:
3422: @node Classes and Scoping, Object Interfaces, Method conveniences, Objects
3423: @subsection Classes and Scoping
3424: @cindex classes and scoping
3425: @cindex scoping and classes
3426:
3427: Inheritance is frequent, unlike structure extension. This exacerbates
3428: the problem with the field name convention (@pxref{Structure Naming
3429: Convention}): One always has to remember in which class the field was
3430: originally defined; changing a part of the class structure would require
3431: changes for renaming in otherwise unaffected code.
3432:
3433: @cindex @code{inst-var} visibility
3434: @cindex @code{inst-value} visibility
3435: To solve this problem, I added a scoping mechanism (which was not in my
3436: original charter): A field defined with @code{inst-var} (or
3437: @code{inst-value}) is visible only in the class where it is defined and in
3438: the descendent classes of this class. Using such fields only makes
3439: sense in @code{m:}-defined methods in these classes anyway.
3440:
3441: This scoping mechanism allows us to use the unadorned field name,
3442: because name clashes with unrelated words become much less likely.
3443:
3444: @cindex @code{protected} discussion
3445: @cindex @code{private} discussion
3446: Once we have this mechanism, we can also use it for controlling the
3447: visibility of other words: All words defined after
3448: @code{protected} are visible only in the current class and its
3449: descendents. @code{public} restores the compilation
3450: (i.e. @code{current}) wordlist that was in effect before. If you
3451: have several @code{protected}s without an intervening
3452: @code{public} or @code{set-current}, @code{public}
3453: will restore the compilation wordlist in effect before the first of
3454: these @code{protected}s.
3455:
3456: @node Object Interfaces, Objects Implementation, Classes and Scoping, Objects
3457: @subsection Object Interfaces
3458: @cindex object interfaces
3459: @cindex interfaces for objects
3460:
3461: In this model you can only call selectors defined in the class of the
3462: receiving objects or in one of its ancestors. If you call a selector
3463: with a receiving object that is not in one of these classes, the
3464: result is undefined; if you are lucky, the program crashes
3465: immediately.
3466:
3467: @cindex selectors common to hardly-related classes
3468: Now consider the case when you want to have a selector (or several)
3469: available in two classes: You would have to add the selector to a
3470: common ancestor class, in the worst case to @code{object}. You
3471: may not want to do this, e.g., because someone else is responsible for
3472: this ancestor class.
3473:
3474: The solution for this problem is interfaces. An interface is a
3475: collection of selectors. If a class implements an interface, the
3476: selectors become available to the class and its descendents. A class
3477: can implement an unlimited number of interfaces. For the problem
3478: discussed above, we would define an interface for the selector(s), and
3479: both classes would implement the interface.
3480:
3481: As an example, consider an interface @code{storage} for
3482: writing objects to disk and getting them back, and a class
3483: @code{foo} foo that implements it. The code for this would look
3484: like this:
3485:
3486: @cindex @code{interface} usage
3487: @cindex @code{end-interface} usage
3488: @cindex @code{implementation} usage
3489: @example
3490: interface
3491: selector write ( file object -- )
3492: selector read1 ( file object -- )
3493: end-interface storage
3494:
3495: bar class
3496: storage implementation
3497:
3498: ... overrides write
3499: ... overrides read
3500: ...
3501: end-class foo
3502: @end example
3503:
3504: (I would add a word @code{read} ( file -- object ) that uses
3505: @code{read1} internally, but that's beyond the point illustrated
3506: here.)
3507:
3508: Note that you cannot use @code{protected} in an interface; and
3509: of course you cannot define fields.
3510:
3511: In the Neon model, all selectors are available for all classes;
3512: therefore it does not need interfaces. The price you pay in this model
3513: is slower late binding, and therefore, added complexity to avoid late
3514: binding.
3515:
3516: @node Objects Implementation, Comparison with other object models, Object Interfaces, Objects
3517: @subsection @file{objects.fs} Implementation
3518: @cindex @file{objects.fs} implementation
3519:
3520: @cindex @code{object-map} discussion
3521: An object is a piece of memory, like one of the data structures
3522: described with @code{struct...end-struct}. It has a field
3523: @code{object-map} that points to the method map for the object's
3524: class.
3525:
3526: @cindex method map
3527: @cindex virtual function table
3528: The @emph{method map}@footnote{This is Self terminology; in C++
3529: terminology: virtual function table.} is an array that contains the
3530: execution tokens (XTs) of the methods for the object's class. Each
3531: selector contains an offset into the method maps.
3532:
3533: @cindex @code{selector} implementation, class
3534: @code{selector} is a defining word that uses
3535: @code{create} and @code{does>}. The body of the
3536: selector contains the offset; the @code{does>} action for a
3537: class selector is, basically:
3538:
3539: @example
3540: ( object addr ) @@ over object-map @@ + @@ execute
3541: @end example
3542:
3543: Since @code{object-map} is the first field of the object, it
3544: does not generate any code. As you can see, calling a selector has a
3545: small, constant cost.
3546:
3547: @cindex @code{current-interface} discussion
3548: @cindex class implementation and representation
3549: A class is basically a @code{struct} combined with a method
3550: map. During the class definition the alignment and size of the class
3551: are passed on the stack, just as with @code{struct}s, so
3552: @code{field} can also be used for defining class
3553: fields. However, passing more items on the stack would be
3554: inconvenient, so @code{class} builds a data structure in memory,
3555: which is accessed through the variable
3556: @code{current-interface}. After its definition is complete, the
3557: class is represented on the stack by a pointer (e.g., as parameter for
3558: a child class definition).
3559:
3560: At the start, a new class has the alignment and size of its parent,
3561: and a copy of the parent's method map. Defining new fields extends the
3562: size and alignment; likewise, defining new selectors extends the
3563: method map. @code{overrides} just stores a new XT in the method
3564: map at the offset given by the selector.
3565:
3566: @cindex class binding, implementation
3567: Class binding just gets the XT at the offset given by the selector
3568: from the class's method map and @code{compile,}s (in the case of
3569: @code{[bind]}) it.
3570:
3571: @cindex @code{this} implementation
3572: @cindex @code{catch} and @code{this}
3573: @cindex @code{this} and @code{catch}
3574: I implemented @code{this} as a @code{value}. At the
3575: start of an @code{m:...;m} method the old @code{this} is
3576: stored to the return stack and restored at the end; and the object on
3577: the TOS is stored @code{TO this}. This technique has one
3578: disadvantage: If the user does not leave the method via
3579: @code{;m}, but via @code{throw} or @code{exit},
3580: @code{this} is not restored (and @code{exit} may
3581: crash). To deal with the @code{throw} problem, I have redefined
3582: @code{catch} to save and restore @code{this}; the same
3583: should be done with any word that can catch an exception. As for
3584: @code{exit}, I simply forbid it (as a replacement, there is
3585: @code{exitm}).
3586:
3587: @cindex @code{inst-var} implementation
3588: @code{inst-var} is just the same as @code{field}, with
3589: a different @code{does>} action:
3590: @example
3591: @@ this +
3592: @end example
3593: Similar for @code{inst-value}.
3594:
3595: @cindex class scoping implementation
3596: Each class also has a wordlist that contains the words defined with
3597: @code{inst-var} and @code{inst-value}, and its protected
3598: words. It also has a pointer to its parent. @code{class} pushes
3599: the wordlists of the class an all its ancestors on the search order,
3600: and @code{end-class} drops them.
3601:
3602: @cindex interface implementation
3603: An interface is like a class without fields, parent and protected
3604: words; i.e., it just has a method map. If a class implements an
3605: interface, its method map contains a pointer to the method map of the
3606: interface. The positive offsets in the map are reserved for class
3607: methods, therefore interface map pointers have negative
3608: offsets. Interfaces have offsets that are unique throughout the
3609: system, unlike class selectors, whose offsets are only unique for the
3610: classes where the selector is available (invokable).
3611:
3612: This structure means that interface selectors have to perform one
3613: indirection more than class selectors to find their method. Their body
3614: contains the interface map pointer offset in the class method map, and
3615: the method offset in the interface method map. The
3616: @code{does>} action for an interface selector is, basically:
3617:
3618: @example
3619: ( object selector-body )
3620: 2dup selector-interface @@ ( object selector-body object interface-offset )
3621: swap object-map @@ + @@ ( object selector-body map )
3622: swap selector-offset @@ + @@ execute
3623: @end example
3624:
3625: where @code{object-map} and @code{selector-offset} are
3626: first fields and generate no code.
3627:
3628: As a concrete example, consider the following code:
3629:
3630: @example
3631: interface
3632: selector if1sel1
3633: selector if1sel2
3634: end-interface if1
3635:
3636: object class
3637: if1 implementation
3638: selector cl1sel1
3639: cell% inst-var cl1iv1
3640:
3641: ' m1 overrides construct
3642: ' m2 overrides if1sel1
3643: ' m3 overrides if1sel2
3644: ' m4 overrides cl1sel2
3645: end-class cl1
3646:
3647: create obj1 object dict-new drop
3648: create obj2 cl1 dict-new drop
3649: @end example
3650:
3651: The data structure created by this code (including the data structure
3652: for @code{object}) is shown in the <a
3653: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
3654:
3655: @node Comparison with other object models, Objects Glossary, Objects Implementation, Objects
3656: @subsection Comparison with other object models
3657: @cindex comparison of object models
3658: @cindex object models, comparison
3659:
3660: Many object-oriented Forth extensions have been proposed (@cite{A survey
3661: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
3662: J. Rodriguez and W. F. S. Poehlman lists 17). Here I'll discuss the
3663: relation of @file{objects.fs} to two well-known and two closely-related
3664: (by the use of method maps) models.
3665:
3666: @cindex Neon model
3667: The most popular model currently seems to be the Neon model (see
3668: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
3669: 1997) by Andrew McKewan). The Neon model uses a @code{@emph{selector
3670: object}} syntax, which makes it unnatural to pass objects on the
3671: stack. It also requires that the selector parses the input stream (at
3672: compile time); this leads to reduced extensibility and to bugs that are
3673: hard to find. Finally, it allows using every selector to every object;
3674: this eliminates the need for classes, but makes it harder to create
3675: efficient implementations. A longer version of this critique can be
3676: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
3677: Dimensions, May 1997) by Anton Ertl.
3678:
3679: @cindex Pountain's object-oriented model
3680: Another well-known publication is @cite{Object-Oriented Forth} (Academic
3681: Press, London, 1987) by Dick Pountain. However, it is not really about
3682: object-oriented programming, because it hardly deals with late
3683: binding. Instead, it focuses on features like information hiding and
3684: overloading that are characteristic of modular languages like Ada (83).
3685:
3686: @cindex Zsoter's object-oriented model
3687: In @cite{Does late binding have to be slow?} (Forth Dimensions ??? 1996)
3688: Andras Zsoter describes a model that makes heavy use of an active object
3689: (like @code{this} in @file{objects.fs}): The active object is not only
3690: used for accessing all fields, but also specifies the receiving object
3691: of every selector invocation; you have to change the active object
3692: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
3693: changes more or less implicitly at @code{m: ... ;m}. Such a change at
3694: the method entry point is unnecessary with the Zsoter's model, because
3695: the receiving object is the active object already; OTOH, the explicit
3696: change is absolutely necessary in that model, because otherwise no one
3697: could ever change the active object. An ANS Forth implementation of this
3698: model is available at @url{http://www.forth.org/fig/oopf.html}.
3699:
3700: @cindex @file{oof.fs} object model
3701: The @file{oof.fs} model combines information hiding and overloading
3702: resolution (by keeping names in various wordlists) with object-oriented
3703: programming. It sets the active object implicitly on method entry, but
3704: also allows explicit changing (with @code{>o...o>} or with
3705: @code{with...endwith}). It uses parsing and state-smart objects and
3706: classes for resolving overloading and for early binding: the object or
3707: class parses the selector and determines the method from this. If the
3708: selector is not parsed by an object or class, it performs a call to the
3709: selector for the active object (late binding), like Zsoter's model.
3710: Fields are always accessed through the active object. The big
3711: disadvantage of this model is the parsing and the state-smartness, which
3712: reduces extensibility and increases the opportunities for subtle bugs;
3713: essentially, you are only safe if you never tick or @code{postpone} an
3714: object or class.
3715:
3716: @node Objects Glossary, , Comparison with other object models, Objects
3717: @subsection @file{objects.fs} Glossary
3718: @cindex @file{objects.fs} Glossary
3719:
3720: doc-bind
3721: doc-<bind>
3722: doc-bind'
3723: doc-[bind]
3724: doc-class
3725: doc-class->map
3726: doc-class-inst-size
3727: doc-class-override!
3728: doc-construct
3729: doc-current'
3730: doc-[current]
3731: doc-current-interface
3732: doc-dict-new
3733: doc-drop-order
3734: doc-end-class
3735: doc-end-class-noname
3736: doc-end-interface
3737: doc-end-interface-noname
3738: doc-exitm
3739: doc-heap-new
3740: doc-implementation
3741: doc-init-object
3742: doc-inst-value
3743: doc-inst-var
3744: doc-interface
3745: doc-;m
3746: doc-m:
3747: doc-method
3748: doc-object
3749: doc-overrides
3750: doc-[parent]
3751: doc-print
3752: doc-protected
3753: doc-public
3754: doc-push-order
3755: doc-selector
3756: doc-this
3757: doc-<to-inst>
3758: doc-[to-inst]
3759: doc-to-this
3760: doc-xt-new
3761:
3762: @c -------------------------------------------------------------
1.8 ! pazsan 3763: @node Object Oriented Forth, Mini-OOF, Objects, Words
1.6 pazsan 3764: @section Object oriented Forth
3765: @cindex oof
3766: @cindex object-oriented programming
3767:
3768: @cindex @file{objects.fs}
3769: @cindex @file{oof.fs}
3770: Gforth comes with two packets for object-oriented programming,
3771: @file{objects.fs} and @file{oof.fs}; none of them is preloaded, so you
3772: have to @code{include} them before use. This section describes the
3773: @file{oof.fs} packet. Both packets are written in ANS Forth and can be
1.7 pazsan 3774: used with any other standard Forth (@pxref{Objects}). This section uses
1.6 pazsan 3775: the same rationale why using object-oriented programming, and the same
3776: terminology.
3777:
3778: The packet described in this section is used in bigFORTH since 1991, and
3779: used for two large applications: a chromatographic system used to
3780: create new medicaments, and a graphic user interface library (MINOS).
3781:
3782: @menu
3783: * Properties of the OOF model::
3784: * Basic OOF Usage::
3785: * The base class object::
1.7 pazsan 3786: * Class Declaration::
3787: * Class Implementation::
1.6 pazsan 3788: @end menu
3789:
1.7 pazsan 3790: @node Properties of the OOF model, Basic OOF Usage, Object Oriented Forth, Object Oriented Forth
1.6 pazsan 3791: @subsection Properties of the OOF model
3792: @cindex @file{oof.fs} properties
3793:
3794: @itemize @bullet
3795: @item
3796: This model combines object oriented programming with information
3797: hiding. It helps you writing large application, where scoping is
3798: necessary, because it provides class-oriented scoping.
3799:
3800: @item
3801: Named objects, object pointers, and object arrays can be created,
3802: selector invocation uses the "object selector" syntax. Selector invocation
3803: to objects and/or selectors on the stack is a bit less convenient, but
3804: possible.
3805:
3806: @item
3807: Selector invocation and instance variable usage of the active object is
3808: straight forward, since both make use of the active object.
3809:
3810: @item
3811: Late binding is efficient and easy to use.
3812:
3813: @item
3814: State-smart objects parse selectors. However, extensibility is provided
3815: using a (parsing) selector @code{postpone} and a selector @code{'}.
3816:
3817: @item
3818: An implementation in ANS Forth is available.
3819:
3820: @end itemize
3821:
3822:
1.7 pazsan 3823: @node Basic OOF Usage, The base class object, Properties of the OOF model, Object Oriented Forth
1.6 pazsan 3824: @subsection Basic OOF Usage
3825: @cindex @file{oof.fs} usage
3826:
3827: Here, I use the same example as for @code{objects} (@pxref{Basic Objects Usage}).
3828:
3829: You can define a class for graphical objects like this:
3830:
3831: @cindex @code{class} usage
3832: @cindex @code{class;} usage
3833: @cindex @code{method} usage
3834: @example
3835: object class graphical \ "object" is the parent class
3836: method draw ( x y graphical -- )
3837: class;
3838: @end example
3839:
3840: This code defines a class @code{graphical} with an
3841: operation @code{draw}. We can perform the operation
3842: @code{draw} on any @code{graphical} object, e.g.:
3843:
3844: @example
3845: 100 100 t-rex draw
3846: @end example
3847:
3848: where @code{t-rex} is an object or object pointer, created with e.g.
3849: @code{graphical : trex}.
3850:
3851: @cindex abstract class
3852: How do we create a graphical object? With the present definitions,
3853: we cannot create a useful graphical object. The class
3854: @code{graphical} describes graphical objects in general, but not
3855: any concrete graphical object type (C++ users would call it an
3856: @emph{abstract class}); e.g., there is no method for the selector
3857: @code{draw} in the class @code{graphical}.
3858:
3859: For concrete graphical objects, we define child classes of the
3860: class @code{graphical}, e.g.:
3861:
3862: @example
3863: graphical class circle \ "graphical" is the parent class
3864: cell var circle-radius
3865: how:
3866: : draw ( x y -- )
3867: circle-radius @@ draw-circle ;
3868:
3869: : init ( n-radius -- (
3870: circle-radius ! ;
3871: class;
3872: @end example
3873:
3874: Here we define a class @code{circle} as a child of @code{graphical},
3875: with a field @code{circle-radius}; it defines new methods for the
3876: selectors @code{draw} and @code{init} (@code{init} is defined in
3877: @code{object}, the parent class of @code{graphical}).
3878:
3879: Now we can create a circle in the dictionary with
3880:
3881: @example
3882: 50 circle : my-circle
3883: @end example
3884:
3885: @code{:} invokes @code{init}, thus initializing the field
3886: @code{circle-radius} with 50. We can draw this new circle at (100,100)
3887: with
3888:
3889: @example
3890: 100 100 my-circle draw
3891: @end example
3892:
3893: @cindex selector invocation, restrictions
3894: @cindex class definition, restrictions
3895: Note: You can invoke a selector only if the receiving object belongs to
3896: the class where the selector was defined or one of its descendents;
3897: e.g., you can invoke @code{draw} only for objects belonging to
3898: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
1.7 pazsan 3899: mechanism will check if you try to invoke a selector that is not
1.6 pazsan 3900: defined in this class hierarchy, so you'll get an error at compilation
3901: time.
3902:
3903:
1.7 pazsan 3904: @node The base class object, Class Declaration, Basic OOF Usage, Object Oriented Forth
1.6 pazsan 3905: @subsection The base class @file{object}
3906: @cindex @file{oof.fs} base class
3907:
3908: When you define a class, you have to specify a parent class. So how do
3909: you start defining classes? There is one class available from the start:
3910: @code{object}. You have to use it as ancestor for all classes. It is the
3911: only class that has no parent. Classes are also objects, except that
3912: they don't have instance variables; class manipulation such as
3913: inheritance or changing definitions of a class is handled through
3914: selectors of the class @code{object}.
3915:
3916: @code{object} provides a number of selectors:
3917:
3918: @itemize @bullet
3919: @item
3920: @code{class} for subclassing, @code{definitions} to add definitions
3921: later on, and @code{class?} to get type informations (is the class a
3922: subclass of the class passed on the stack?).
1.7 pazsan 3923: doc---object-class
3924: doc---object-definitions
3925: doc---object-class?
1.6 pazsan 3926:
3927: @item
3928: @code{init} and @code{dispose} as constructor and destroctor of the
3929: object. @code{init} is invocated after the object's memory is allocated,
3930: while @code{dispose} also handles deallocation. Thus if you redefine
3931: @code{dispose}, you have to call the parent's dispose with @code{super
3932: dispose}, too.
1.7 pazsan 3933: doc---object-init
3934: doc---object-dispose
1.6 pazsan 3935:
3936: @item
1.7 pazsan 3937: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
3938: @code{[]} to create named and unnamed objects and object arrays or
3939: object pointers.
3940: doc---object-new
3941: doc---object-new[]
3942: doc---object-:
3943: doc---object-ptr
3944: doc---object-asptr
3945: doc---object-[]
1.6 pazsan 3946:
3947: @item
3948: @code{::} and @code{super} for expicit scoping. You should use expicit
3949: scoping only for super classes or classes with the same set of instance
3950: variables. Explicit scoped selectors use early binding.
1.7 pazsan 3951: doc---object-::
3952: doc---object-super
1.6 pazsan 3953:
3954: @item
3955: @code{self} to get the address of the object
1.7 pazsan 3956: doc---object-self
1.6 pazsan 3957:
3958: @item
3959: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
3960: pointers and instance defers.
1.7 pazsan 3961: doc---object-bind
3962: doc---object-bound
3963: doc---object-link
3964: doc---object-is
1.6 pazsan 3965:
3966: @item
3967: @code{'} to obtain selector tokens, @code{send} to invocate selectors
3968: form the stack, and @code{postpone} to generate selector invocation code.
1.7 pazsan 3969: doc---object-'
3970: doc---object-postpone
1.6 pazsan 3971:
3972: @item
3973: @code{with} and @code{endwith} to select the active object from the
3974: stack, and enabling it's scope. Using @code{with} and @code{endwith}
3975: also allows to create code using selector @code{postpone} without being
3976: trapped bye the state-smart objects.
1.7 pazsan 3977: doc---object-with
3978: doc---object-endwith
1.6 pazsan 3979:
3980: @end itemize
3981:
1.7 pazsan 3982: @node Class Declaration, Class Implementation, The base class object, Object Oriented Forth
3983: @subsection Class Declaration
3984: @cindex class declaration
3985:
3986: @itemize @bullet
3987: @item
3988: Instance variables
3989: doc---oof-var
3990:
3991: @item
3992: Object pointers
3993: doc---oof-ptr
3994: doc---oof-asptr
3995:
3996: @item
3997: Instance defers
3998: doc---oof-defer
3999:
4000: @item
4001: Method selectors
4002: doc---oof-early
4003: doc---oof-method
4004:
4005: @item
4006: Class wide variables
4007: doc---oof-static
4008:
4009: @item
4010: End declaration
4011: doc---oof-how:
4012: doc---oof-class;
4013:
4014: @end itemize
4015:
4016: @node Class Implementation, , Class Declaration, Object Oriented Forth
4017: @subsection Class Implementation
4018: @cindex class implementation
4019:
1.8 ! pazsan 4020: @node Mini-OOF, Tokens for Words, Object Oriented Forth, Words
! 4021: @section Mini-OOF
! 4022: @cindex mini-oof
! 4023:
! 4024: Gforth's third object oriented Forth package is a 12-liner. It uses a
! 4025: bit of a mixture of the @file{object.fs} and the @file{oof.fs} syntax,
! 4026: and reduces to the bare minimum of features.
! 4027:
! 4028: @example
! 4029: : method ( m v -- m' v ) Create over , swap cell+ swap
! 4030: DOES> ( ... o -- ... ) @ over @ + @ execute ;
! 4031: : var ( m v size -- m v' ) Create over , +
! 4032: DOES> ( o -- addr ) @ + ;
! 4033: : class ( class -- class methods vars ) dup 2@ ;
! 4034: : end-class ( class methods vars -- )
! 4035: Create here >r , dup , 2 cells ?DO ['] noop , cell +LOOP
! 4036: cell+ dup cell+ swap @ 2 - cells r> 2 cells + swap move ;
! 4037: : defines ( xt class -- ) ' >body @ + ! ;
! 4038: : new ( class -- o ) here over @ allot swap over ! ;
! 4039: : :: ( class "name" -- ) ' >body @ + @ compile, ;
! 4040: Create object 1 cells , 2 cells ,
! 4041: @end example
! 4042:
1.6 pazsan 4043: @c -------------------------------------------------------------
1.8 ! pazsan 4044: @node Tokens for Words, Wordlists, Mini-OOF, Words
1.1 anton 4045: @section Tokens for Words
4046: @cindex tokens for words
4047:
4048: This chapter describes the creation and use of tokens that represent
4049: words on the stack (and in data space).
4050:
4051: Named words have interpretation and compilation semantics. Unnamed words
4052: just have execution semantics.
4053:
4054: @cindex execution token
4055: An @dfn{execution token} represents the execution semantics of an
4056: unnamed word. An execution token occupies one cell. As explained in
4057: section @ref{Supplying names}, the execution token of the last words
4058: defined can be produced with
4059:
4060: short-lastxt
4061:
4062: You can perform the semantics represented by an execution token with
4063: doc-execute
4064: You can compile the word with
4065: doc-compile,
4066:
4067: @cindex code field address
4068: @cindex CFA
4069: In Gforth, the abstract data type @emph{execution token} is implemented
4070: as CFA (code field address).
4071:
4072: The interpretation semantics of a named word are also represented by an
4073: execution token. You can get it with
4074:
4075: doc-[']
4076: doc-'
4077:
4078: For literals, you use @code{'} in interpreted code and @code{[']} in
4079: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
4080: by complaining about compile-only words. To get an execution token for a
4081: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
4082: @var{X} drop}.
4083:
4084: @cindex compilation token
4085: The compilation semantics are represented by a @dfn{compilation token}
4086: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
4087: execution token. The compilation semantics represented by the
4088: compilation token can be performed with @code{execute}, which consumes
4089: the whole compilation token, with an additional stack effect determined
4090: by the represented compilation semantics.
4091:
4092: doc-[comp']
4093: doc-comp'
4094:
4095: You can compile the compilation semantics with @code{postpone,}. I.e.,
4096: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
4097: @var{word}}.
4098:
4099: doc-postpone,
4100:
4101: At present, the @var{w} part of a compilation token is an execution
4102: token, and the @var{xt} part represents either @code{execute} or
4103: @code{compile,}. However, don't rely on that knowledge, unless necessary;
4104: we may introduce unusual compilation tokens in the future (e.g.,
4105: compilation tokens representing the compilation semantics of literals).
4106:
4107: @cindex name token
4108: @cindex name field address
4109: @cindex NFA
4110: Named words are also represented by the @dfn{name token}. The abstract
4111: data type @emph{name token} is implemented as NFA (name field address).
4112:
4113: doc-find-name
4114: doc-name>int
4115: doc-name?int
4116: doc-name>comp
4117: doc-name>string
4118:
4119: @node Wordlists, Files, Tokens for Words, Words
4120: @section Wordlists
4121:
4122: @node Files, Blocks, Wordlists, Words
4123: @section Files
4124:
4125: @node Blocks, Other I/O, Files, Words
4126: @section Blocks
4127:
4128: @node Other I/O, Programming Tools, Blocks, Words
4129: @section Other I/O
4130:
1.7 pazsan 4131: @node Programming Tools, Assembler and Code Words, Other I/O, Words
1.1 anton 4132: @section Programming Tools
4133: @cindex programming tools
4134:
4135: @menu
4136: * Debugging:: Simple and quick.
4137: * Assertions:: Making your programs self-checking.
1.6 pazsan 4138: * Singlestep Debugger:: Executing your program word by word.
1.1 anton 4139: @end menu
4140:
4141: @node Debugging, Assertions, Programming Tools, Programming Tools
4142: @subsection Debugging
4143: @cindex debugging
4144:
1.2 jwilke 4145: The simple debugging aids provided in @file{debugs.fs}
1.1 anton 4146: are meant to support a different style of debugging than the
4147: tracing/stepping debuggers used in languages with long turn-around
4148: times.
4149:
4150: A much better (faster) way in fast-compiling languages is to add
4151: printing code at well-selected places, let the program run, look at
4152: the output, see where things went wrong, add more printing code, etc.,
4153: until the bug is found.
4154:
4155: The word @code{~~} is easy to insert. It just prints debugging
4156: information (by default the source location and the stack contents). It
4157: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
4158: query-replace them with nothing). The deferred words
4159: @code{printdebugdata} and @code{printdebugline} control the output of
4160: @code{~~}. The default source location output format works well with
4161: Emacs' compilation mode, so you can step through the program at the
4162: source level using @kbd{C-x `} (the advantage over a stepping debugger
4163: is that you can step in any direction and you know where the crash has
4164: happened or where the strange data has occurred).
4165:
4166: Note that the default actions clobber the contents of the pictured
4167: numeric output string, so you should not use @code{~~}, e.g., between
4168: @code{<#} and @code{#>}.
4169:
4170: doc-~~
4171: doc-printdebugdata
4172: doc-printdebugline
4173:
1.2 jwilke 4174: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
1.1 anton 4175: @subsection Assertions
4176: @cindex assertions
4177:
4178: It is a good idea to make your programs self-checking, in particular, if
4179: you use an assumption (e.g., that a certain field of a data structure is
4180: never zero) that may become wrong during maintenance. Gforth supports
4181: assertions for this purpose. They are used like this:
4182:
4183: @example
4184: assert( @var{flag} )
4185: @end example
4186:
4187: The code between @code{assert(} and @code{)} should compute a flag, that
4188: should be true if everything is alright and false otherwise. It should
4189: not change anything else on the stack. The overall stack effect of the
4190: assertion is @code{( -- )}. E.g.
4191:
4192: @example
4193: assert( 1 1 + 2 = ) \ what we learn in school
4194: assert( dup 0<> ) \ assert that the top of stack is not zero
4195: assert( false ) \ this code should not be reached
4196: @end example
4197:
4198: The need for assertions is different at different times. During
4199: debugging, we want more checking, in production we sometimes care more
4200: for speed. Therefore, assertions can be turned off, i.e., the assertion
4201: becomes a comment. Depending on the importance of an assertion and the
4202: time it takes to check it, you may want to turn off some assertions and
4203: keep others turned on. Gforth provides several levels of assertions for
4204: this purpose:
4205:
4206: doc-assert0(
4207: doc-assert1(
4208: doc-assert2(
4209: doc-assert3(
4210: doc-assert(
4211: doc-)
4212:
4213: @code{Assert(} is the same as @code{assert1(}. The variable
4214: @code{assert-level} specifies the highest assertions that are turned
4215: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
4216: @code{assert1(} assertions perform checking, while @code{assert2(} and
4217: @code{assert3(} assertions are treated as comments.
4218:
4219: Note that the @code{assert-level} is evaluated at compile-time, not at
4220: run-time. I.e., you cannot turn assertions on or off at run-time, you
4221: have to set the @code{assert-level} appropriately before compiling a
4222: piece of code. You can compile several pieces of code at several
4223: @code{assert-level}s (e.g., a trusted library at level 1 and newly
4224: written code at level 3).
4225:
4226: doc-assert-level
4227:
4228: If an assertion fails, a message compatible with Emacs' compilation mode
4229: is produced and the execution is aborted (currently with @code{ABORT"}.
4230: If there is interest, we will introduce a special throw code. But if you
4231: intend to @code{catch} a specific condition, using @code{throw} is
4232: probably more appropriate than an assertion).
4233:
1.2 jwilke 4234: @node Singlestep Debugger, , Assertions, Programming Tools
4235: @subsection Singlestep Debugger
4236: @cindex singlestep Debugger
4237: @cindex debugging Singlestep
4238: @cindex @code{dbg}
4239: @cindex @code{BREAK:}
4240: @cindex @code{BREAK"}
4241:
4242: When a new word is created there's often the need to check whether it behaves
1.5 anton 4243: correctly or not. You can do this by typing @code{dbg badword}. This might
1.2 jwilke 4244: look like:
4245: @example
4246: : badword 0 DO i . LOOP ; ok
4247: 2 dbg badword
4248: : badword
4249: Scanning code...
4250:
4251: Nesting debugger ready!
4252:
4253: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
4254: 400D4740 8049F68 DO -> [ 0 ]
4255: 400D4744 804A0C8 i -> [ 1 ] 00000
4256: 400D4748 400C5E60 . -> 0 [ 0 ]
4257: 400D474C 8049D0C LOOP -> [ 0 ]
4258: 400D4744 804A0C8 i -> [ 1 ] 00001
4259: 400D4748 400C5E60 . -> 1 [ 0 ]
4260: 400D474C 8049D0C LOOP -> [ 0 ]
4261: 400D4758 804B384 ; -> ok
4262: @end example
4263:
1.5 anton 4264: Each line displayed is one step. You always have to hit return to
4265: execute the next word that is displayed. If you don't want to execute
4266: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
4267: an overview what keys are available:
1.2 jwilke 4268:
4269: @table @i
4270:
1.4 anton 4271: @item <return>
1.5 anton 4272: Next; Execute the next word.
1.2 jwilke 4273:
4274: @item n
1.5 anton 4275: Nest; Single step through next word.
1.2 jwilke 4276:
4277: @item u
1.5 anton 4278: Unnest; Stop debugging and execute rest of word. If we got to this word
4279: with nest, continue debugging with the calling word.
1.2 jwilke 4280:
4281: @item d
1.5 anton 4282: Done; Stop debugging and execute rest.
1.2 jwilke 4283:
4284: @item s
1.5 anton 4285: Stopp; Abort immediately.
1.2 jwilke 4286:
4287: @end table
4288:
4289: Debugging large application with this mechanism is very difficult, because
4290: you have to nest very deep into the program before the interesting part
4291: begins. This takes a lot of time.
4292:
4293: To do it more directly put a @code{BREAK:} command into your source code.
4294: When program execution reaches @code{BREAK:} the single step debugger is
4295: invoked and you have all the features described above.
4296:
4297: If you have more than one part to debug it is useful to know where the
4298: program has stopped at the moment. You can do this by the
4299: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
4300: string is typed out when the ``breakpoint'' is reached.
4301:
1.7 pazsan 4302: @node Assembler and Code Words, Threading Words, Programming Tools, Words
4303: @section Assembler and Code Words
1.1 anton 4304: @cindex assembler
4305: @cindex code words
4306:
4307: Gforth provides some words for defining primitives (words written in
4308: machine code), and for defining the the machine-code equivalent of
4309: @code{DOES>}-based defining words. However, the machine-independent
4310: nature of Gforth poses a few problems: First of all, Gforth runs on
4311: several architectures, so it can provide no standard assembler. What's
4312: worse is that the register allocation not only depends on the processor,
4313: but also on the @code{gcc} version and options used.
4314:
4315: The words that Gforth offers encapsulate some system dependences (e.g., the
4316: header structure), so a system-independent assembler may be used in
4317: Gforth. If you do not have an assembler, you can compile machine code
4318: directly with @code{,} and @code{c,}.
4319:
4320: doc-assembler
4321: doc-code
4322: doc-end-code
4323: doc-;code
4324: doc-flush-icache
4325:
4326: If @code{flush-icache} does not work correctly, @code{code} words
4327: etc. will not work (reliably), either.
4328:
4329: These words are rarely used. Therefore they reside in @code{code.fs},
4330: which is usually not loaded (except @code{flush-icache}, which is always
4331: present). You can load them with @code{require code.fs}.
4332:
4333: @cindex registers of the inner interpreter
4334: In the assembly code you will want to refer to the inner interpreter's
4335: registers (e.g., the data stack pointer) and you may want to use other
4336: registers for temporary storage. Unfortunately, the register allocation
4337: is installation-dependent.
4338:
4339: The easiest solution is to use explicit register declarations
4340: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
4341: GNU C Manual}) for all of the inner interpreter's registers: You have to
4342: compile Gforth with @code{-DFORCE_REG} (configure option
4343: @code{--enable-force-reg}) and the appropriate declarations must be
4344: present in the @code{machine.h} file (see @code{mips.h} for an example;
4345: you can find a full list of all declarable register symbols with
4346: @code{grep register engine.c}). If you give explicit registers to all
4347: variables that are declared at the beginning of @code{engine()}, you
4348: should be able to use the other caller-saved registers for temporary
4349: storage. Alternatively, you can use the @code{gcc} option
4350: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
4351: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
4352: (however, this restriction on register allocation may slow Gforth
4353: significantly).
4354:
4355: If this solution is not viable (e.g., because @code{gcc} does not allow
4356: you to explicitly declare all the registers you need), you have to find
4357: out by looking at the code where the inner interpreter's registers
4358: reside and which registers can be used for temporary storage. You can
4359: get an assembly listing of the engine's code with @code{make engine.s}.
4360:
4361: In any case, it is good practice to abstract your assembly code from the
4362: actual register allocation. E.g., if the data stack pointer resides in
4363: register @code{$17}, create an alias for this register called @code{sp},
4364: and use that in your assembly code.
4365:
4366: @cindex code words, portable
4367: Another option for implementing normal and defining words efficiently
4368: is: adding the wanted functionality to the source of Gforth. For normal
4369: words you just have to edit @file{primitives} (@pxref{Automatic
4370: Generation}), defining words (equivalent to @code{;CODE} words, for fast
4371: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
4372: @file{prims2x.fs}, and possibly @file{cross.fs}.
4373:
4374:
1.2 jwilke 4375: @node Threading Words, Including Files, Assembler and Code Words, Words
1.1 anton 4376: @section Threading Words
4377: @cindex threading words
4378:
4379: @cindex code address
4380: These words provide access to code addresses and other threading stuff
4381: in Gforth (and, possibly, other interpretive Forths). It more or less
4382: abstracts away the differences between direct and indirect threading
4383: (and, for direct threading, the machine dependences). However, at
4384: present this wordset is still incomplete. It is also pretty low-level;
4385: some day it will hopefully be made unnecessary by an internals wordset
4386: that abstracts implementation details away completely.
4387:
4388: doc->code-address
4389: doc->does-code
4390: doc-code-address!
4391: doc-does-code!
4392: doc-does-handler!
4393: doc-/does-handler
4394:
4395: The code addresses produced by various defining words are produced by
4396: the following words:
4397:
4398: doc-docol:
4399: doc-docon:
4400: doc-dovar:
4401: doc-douser:
4402: doc-dodefer:
4403: doc-dofield:
4404:
4405: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
4406: with @code{>DOES-CODE}. If the word was defined in that way, the value
4407: returned is different from 0 and identifies the @code{DOES>} used by the
4408: defining word.
1.2 jwilke 4409:
1.7 pazsan 4410: @node Including Files, Include and Require, Threading Words, Words
1.5 anton 4411: @section Including Files
1.2 jwilke 4412: @cindex including files
4413:
1.7 pazsan 4414: @menu
4415: * Include and Require::
4416: * Path Handling::
4417: @end menu
4418:
4419: @node Include and Require, Path Handling, Including Files, Including Files
1.2 jwilke 4420: @subsection Include and Requrie
4421:
4422: There a two words to include the source files more intelligently.
4423:
4424: doc-include
4425: doc-require
4426:
1.7 pazsan 4427: @node Path Handling, , Include and Require, Including Files
4428: @subsection Path Handling
1.2 jwilke 4429: @cindex path handling
4430:
4431: In larger program projects it is often neccassary to build up a structured
1.5 anton 4432: directory tree. Standard Forth programs are somewhere more central because
1.2 jwilke 4433: they must be accessed from some more other programs. To achieve this it is
1.5 anton 4434: possible to manipulate the search path in which Gforth tries to find the
1.2 jwilke 4435: source file.
4436:
4437: doc-fpath+
4438: doc-fpath=
4439: doc-.fpath
4440:
4441: Using fpath and require would look like:
4442:
4443: @example
4444:
4445: fpath= /usr/lib/forth/|./
4446:
4447: require timer.fs
4448:
4449: ...
4450:
4451: @end example
4452:
4453: @cindex ~+
4454: There is another nice feature which is similar to C's @code{include <...>}
4455: and @code{include "..."}. For example: You have a program seperated into
1.5 anton 4456: several files in a subdirectory and you want to include some other files
4457: in this subdirectory from within the program. You have to tell Gforth that
4458: you are now looking relative to the directory the current file comes from.
4459: You can tell this Gforth by using the prefix @code{~+/} in front of the
1.2 jwilke 4460: filename. It is also possible to add it to the search path.
4461:
1.5 anton 4462: If you have the need to look for a file in the Forth search path, you could
4463: use this Gforth feature in your application.
1.2 jwilke 4464:
4465: doc-open-fpath-file
4466:
4467: It is even possible to use your own search paths. Create a search path like
4468: this:
4469:
4470: @example
4471:
4472: Make a buffer for the path:
4473: create mypath 100 chars , \ maximum length (is checked)
4474: 0 , \ real len
4475: 100 chars allot \ space for path
4476:
4477: @end example
4478:
4479: You have the same functions for the forth search path in an generic version
4480: for different pathes.
4481:
4482: doc-path+
4483: doc-path=
4484: doc-.path
4485: doc-open-path-file
1.1 anton 4486:
1.5 anton 4487: @c ******************************************************************
1.1 anton 4488: @node Tools, ANS conformance, Words, Top
4489: @chapter Tools
4490:
4491: @menu
4492: * ANS Report:: Report the words used, sorted by wordset.
4493: @end menu
4494:
4495: See also @ref{Emacs and Gforth}.
4496:
4497: @node ANS Report, , Tools, Tools
4498: @section @file{ans-report.fs}: Report the words used, sorted by wordset
4499: @cindex @file{ans-report.fs}
4500: @cindex report the words used in your program
4501: @cindex words used in your program
4502:
4503: If you want to label a Forth program as ANS Forth Program, you must
4504: document which wordsets the program uses; for extension wordsets, it is
4505: helpful to list the words the program requires from these wordsets
4506: (because Forth systems are allowed to provide only some words of them).
4507:
4508: The @file{ans-report.fs} tool makes it easy for you to determine which
4509: words from which wordset and which non-ANS words your application
4510: uses. You simply have to include @file{ans-report.fs} before loading the
4511: program you want to check. After loading your program, you can get the
4512: report with @code{print-ans-report}. A typical use is to run this as
4513: batch job like this:
4514: @example
4515: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
4516: @end example
4517:
4518: The output looks like this (for @file{compat/control.fs}):
4519: @example
4520: The program uses the following words
4521: from CORE :
4522: : POSTPONE THEN ; immediate ?dup IF 0=
4523: from BLOCK-EXT :
4524: \
4525: from FILE :
4526: (
4527: @end example
4528:
4529: @subsection Caveats
4530:
4531: Note that @file{ans-report.fs} just checks which words are used, not whether
4532: they are used in an ANS Forth conforming way!
4533:
4534: Some words are defined in several wordsets in the
4535: standard. @file{ans-report.fs} reports them for only one of the
4536: wordsets, and not necessarily the one you expect. It depends on usage
4537: which wordset is the right one to specify. E.g., if you only use the
4538: compilation semantics of @code{S"}, it is a Core word; if you also use
4539: its interpretation semantics, it is a File word.
4540:
4541: @c ******************************************************************
4542: @node ANS conformance, Model, Tools, Top
4543: @chapter ANS conformance
4544: @cindex ANS conformance of Gforth
4545:
4546: To the best of our knowledge, Gforth is an
4547:
4548: ANS Forth System
4549: @itemize @bullet
4550: @item providing the Core Extensions word set
4551: @item providing the Block word set
4552: @item providing the Block Extensions word set
4553: @item providing the Double-Number word set
4554: @item providing the Double-Number Extensions word set
4555: @item providing the Exception word set
4556: @item providing the Exception Extensions word set
4557: @item providing the Facility word set
4558: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
4559: @item providing the File Access word set
4560: @item providing the File Access Extensions word set
4561: @item providing the Floating-Point word set
4562: @item providing the Floating-Point Extensions word set
4563: @item providing the Locals word set
4564: @item providing the Locals Extensions word set
4565: @item providing the Memory-Allocation word set
4566: @item providing the Memory-Allocation Extensions word set (that one's easy)
4567: @item providing the Programming-Tools word set
4568: @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
4569: @item providing the Search-Order word set
4570: @item providing the Search-Order Extensions word set
4571: @item providing the String word set
4572: @item providing the String Extensions word set (another easy one)
4573: @end itemize
4574:
4575: @cindex system documentation
4576: In addition, ANS Forth systems are required to document certain
4577: implementation choices. This chapter tries to meet these
4578: requirements. In many cases it gives a way to ask the system for the
4579: information instead of providing the information directly, in
4580: particular, if the information depends on the processor, the operating
4581: system or the installation options chosen, or if they are likely to
4582: change during the maintenance of Gforth.
4583:
4584: @comment The framework for the rest has been taken from pfe.
4585:
4586: @menu
4587: * The Core Words::
4588: * The optional Block word set::
4589: * The optional Double Number word set::
4590: * The optional Exception word set::
4591: * The optional Facility word set::
4592: * The optional File-Access word set::
4593: * The optional Floating-Point word set::
4594: * The optional Locals word set::
4595: * The optional Memory-Allocation word set::
4596: * The optional Programming-Tools word set::
4597: * The optional Search-Order word set::
4598: @end menu
4599:
4600:
4601: @c =====================================================================
4602: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
4603: @comment node-name, next, previous, up
4604: @section The Core Words
4605: @c =====================================================================
4606: @cindex core words, system documentation
4607: @cindex system documentation, core words
4608:
4609: @menu
4610: * core-idef:: Implementation Defined Options
4611: * core-ambcond:: Ambiguous Conditions
4612: * core-other:: Other System Documentation
4613: @end menu
4614:
4615: @c ---------------------------------------------------------------------
4616: @node core-idef, core-ambcond, The Core Words, The Core Words
4617: @subsection Implementation Defined Options
4618: @c ---------------------------------------------------------------------
4619: @cindex core words, implementation-defined options
4620: @cindex implementation-defined options, core words
4621:
4622:
4623: @table @i
4624: @item (Cell) aligned addresses:
4625: @cindex cell-aligned addresses
4626: @cindex aligned addresses
4627: processor-dependent. Gforth's alignment words perform natural alignment
4628: (e.g., an address aligned for a datum of size 8 is divisible by
4629: 8). Unaligned accesses usually result in a @code{-23 THROW}.
4630:
4631: @item @code{EMIT} and non-graphic characters:
4632: @cindex @code{EMIT} and non-graphic characters
4633: @cindex non-graphic characters and @code{EMIT}
4634: The character is output using the C library function (actually, macro)
4635: @code{putc}.
4636:
4637: @item character editing of @code{ACCEPT} and @code{EXPECT}:
4638: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
4639: @cindex editing in @code{ACCEPT} and @code{EXPECT}
4640: @cindex @code{ACCEPT}, editing
4641: @cindex @code{EXPECT}, editing
4642: This is modeled on the GNU readline library (@pxref{Readline
4643: Interaction, , Command Line Editing, readline, The GNU Readline
4644: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
4645: producing a full word completion every time you type it (instead of
4646: producing the common prefix of all completions).
4647:
4648: @item character set:
4649: @cindex character set
4650: The character set of your computer and display device. Gforth is
4651: 8-bit-clean (but some other component in your system may make trouble).
4652:
4653: @item Character-aligned address requirements:
4654: @cindex character-aligned address requirements
4655: installation-dependent. Currently a character is represented by a C
4656: @code{unsigned char}; in the future we might switch to @code{wchar_t}
4657: (Comments on that requested).
4658:
4659: @item character-set extensions and matching of names:
4660: @cindex character-set extensions and matching of names
4661: @cindex case sensitivity for name lookup
4662: @cindex name lookup, case sensitivity
4663: @cindex locale and case sensitivity
4664: Any character except the ASCII NUL charcter can be used in a
4665: name. Matching is case-insensitive (except in @code{TABLE}s). The
4666: matching is performed using the C function @code{strncasecmp}, whose
4667: function is probably influenced by the locale. E.g., the @code{C} locale
4668: does not know about accents and umlauts, so they are matched
4669: case-sensitively in that locale. For portability reasons it is best to
4670: write programs such that they work in the @code{C} locale. Then one can
4671: use libraries written by a Polish programmer (who might use words
4672: containing ISO Latin-2 encoded characters) and by a French programmer
4673: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
4674: funny results for some of the words (which ones, depends on the font you
4675: are using)). Also, the locale you prefer may not be available in other
4676: operating systems. Hopefully, Unicode will solve these problems one day.
4677:
4678: @item conditions under which control characters match a space delimiter:
4679: @cindex space delimiters
4680: @cindex control characters as delimiters
4681: If @code{WORD} is called with the space character as a delimiter, all
4682: white-space characters (as identified by the C macro @code{isspace()})
4683: are delimiters. @code{PARSE}, on the other hand, treats space like other
4684: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
4685: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
4686: interpreter (aka text interpreter) by default, treats all white-space
4687: characters as delimiters.
4688:
4689: @item format of the control flow stack:
4690: @cindex control flow stack, format
4691: The data stack is used as control flow stack. The size of a control flow
4692: stack item in cells is given by the constant @code{cs-item-size}. At the
4693: time of this writing, an item consists of a (pointer to a) locals list
4694: (third), an address in the code (second), and a tag for identifying the
4695: item (TOS). The following tags are used: @code{defstart},
4696: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
4697: @code{scopestart}.
4698:
4699: @item conversion of digits > 35
4700: @cindex digits > 35
4701: The characters @code{[\]^_'} are the digits with the decimal value
4702: 36@minus{}41. There is no way to input many of the larger digits.
4703:
4704: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
4705: @cindex @code{EXPECT}, display after end of input
4706: @cindex @code{ACCEPT}, display after end of input
4707: The cursor is moved to the end of the entered string. If the input is
4708: terminated using the @kbd{Return} key, a space is typed.
4709:
4710: @item exception abort sequence of @code{ABORT"}:
4711: @cindex exception abort sequence of @code{ABORT"}
4712: @cindex @code{ABORT"}, exception abort sequence
4713: The error string is stored into the variable @code{"error} and a
4714: @code{-2 throw} is performed.
4715:
4716: @item input line terminator:
4717: @cindex input line terminator
4718: @cindex line terminator on input
4719: @cindex newline charcter on input
4720: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
4721: lines. One of these characters is typically produced when you type the
4722: @kbd{Enter} or @kbd{Return} key.
4723:
4724: @item maximum size of a counted string:
4725: @cindex maximum size of a counted string
4726: @cindex counted string, maximum size
4727: @code{s" /counted-string" environment? drop .}. Currently 255 characters
4728: on all ports, but this may change.
4729:
4730: @item maximum size of a parsed string:
4731: @cindex maximum size of a parsed string
4732: @cindex parsed string, maximum size
4733: Given by the constant @code{/line}. Currently 255 characters.
4734:
4735: @item maximum size of a definition name, in characters:
4736: @cindex maximum size of a definition name, in characters
4737: @cindex name, maximum length
4738: 31
4739:
4740: @item maximum string length for @code{ENVIRONMENT?}, in characters:
4741: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
4742: @cindex @code{ENVIRONMENT?} string length, maximum
4743: 31
4744:
4745: @item method of selecting the user input device:
4746: @cindex user input device, method of selecting
4747: The user input device is the standard input. There is currently no way to
4748: change it from within Gforth. However, the input can typically be
4749: redirected in the command line that starts Gforth.
4750:
4751: @item method of selecting the user output device:
4752: @cindex user output device, method of selecting
4753: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
4754: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
4755: output, so output on a terminal does not become visible before the next
4756: newline or buffer overflow. Output on non-terminals is invisible until
4757: the buffer overflows.
4758:
4759: @item methods of dictionary compilation:
4760: What are we expected to document here?
4761:
4762: @item number of bits in one address unit:
4763: @cindex number of bits in one address unit
4764: @cindex address unit, size in bits
4765: @code{s" address-units-bits" environment? drop .}. 8 in all current
4766: ports.
4767:
4768: @item number representation and arithmetic:
4769: @cindex number representation and arithmetic
4770: Processor-dependent. Binary two's complement on all current ports.
4771:
4772: @item ranges for integer types:
4773: @cindex ranges for integer types
4774: @cindex integer types, ranges
4775: Installation-dependent. Make environmental queries for @code{MAX-N},
4776: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
4777: unsigned (and positive) types is 0. The lower bound for signed types on
4778: two's complement and one's complement machines machines can be computed
4779: by adding 1 to the upper bound.
4780:
4781: @item read-only data space regions:
4782: @cindex read-only data space regions
4783: @cindex data-space, read-only regions
4784: The whole Forth data space is writable.
4785:
4786: @item size of buffer at @code{WORD}:
4787: @cindex size of buffer at @code{WORD}
4788: @cindex @code{WORD} buffer size
4789: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
4790: shared with the pictured numeric output string. If overwriting
4791: @code{PAD} is acceptable, it is as large as the remaining dictionary
4792: space, although only as much can be sensibly used as fits in a counted
4793: string.
4794:
4795: @item size of one cell in address units:
4796: @cindex cell size
4797: @code{1 cells .}.
4798:
4799: @item size of one character in address units:
4800: @cindex char size
4801: @code{1 chars .}. 1 on all current ports.
4802:
4803: @item size of the keyboard terminal buffer:
4804: @cindex size of the keyboard terminal buffer
4805: @cindex terminal buffer, size
4806: Varies. You can determine the size at a specific time using @code{lp@@
4807: tib - .}. It is shared with the locals stack and TIBs of files that
4808: include the current file. You can change the amount of space for TIBs
4809: and locals stack at Gforth startup with the command line option
4810: @code{-l}.
4811:
4812: @item size of the pictured numeric output buffer:
4813: @cindex size of the pictured numeric output buffer
4814: @cindex pictured numeric output buffer, size
4815: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
4816: shared with @code{WORD}.
4817:
4818: @item size of the scratch area returned by @code{PAD}:
4819: @cindex size of the scratch area returned by @code{PAD}
4820: @cindex @code{PAD} size
4821: The remainder of dictionary space. @code{unused pad here - - .}.
4822:
4823: @item system case-sensitivity characteristics:
4824: @cindex case-sensitivity characteristics
4825: Dictionary searches are case insensitive (except in
4826: @code{TABLE}s). However, as explained above under @i{character-set
4827: extensions}, the matching for non-ASCII characters is determined by the
4828: locale you are using. In the default @code{C} locale all non-ASCII
4829: characters are matched case-sensitively.
4830:
4831: @item system prompt:
4832: @cindex system prompt
4833: @cindex prompt
4834: @code{ ok} in interpret state, @code{ compiled} in compile state.
4835:
4836: @item division rounding:
4837: @cindex division rounding
4838: installation dependent. @code{s" floored" environment? drop .}. We leave
4839: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
4840: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
4841:
4842: @item values of @code{STATE} when true:
4843: @cindex @code{STATE} values
4844: -1.
4845:
4846: @item values returned after arithmetic overflow:
4847: On two's complement machines, arithmetic is performed modulo
4848: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
4849: arithmetic (with appropriate mapping for signed types). Division by zero
4850: typically results in a @code{-55 throw} (Floating-point unidentified
4851: fault), although a @code{-10 throw} (divide by zero) would be more
4852: appropriate.
4853:
4854: @item whether the current definition can be found after @t{DOES>}:
4855: @cindex @t{DOES>}, visibility of current definition
4856: No.
4857:
4858: @end table
4859:
4860: @c ---------------------------------------------------------------------
4861: @node core-ambcond, core-other, core-idef, The Core Words
4862: @subsection Ambiguous conditions
4863: @c ---------------------------------------------------------------------
4864: @cindex core words, ambiguous conditions
4865: @cindex ambiguous conditions, core words
4866:
4867: @table @i
4868:
4869: @item a name is neither a word nor a number:
4870: @cindex name not found
4871: @cindex Undefined word
4872: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
4873: preserves the data and FP stack, so you don't lose more work than
4874: necessary.
4875:
4876: @item a definition name exceeds the maximum length allowed:
4877: @cindex Word name too long
4878: @code{-19 throw} (Word name too long)
4879:
4880: @item addressing a region not inside the various data spaces of the forth system:
4881: @cindex Invalid memory address
4882: The stacks, code space and name space are accessible. Machine code space is
4883: typically readable. Accessing other addresses gives results dependent on
4884: the operating system. On decent systems: @code{-9 throw} (Invalid memory
4885: address).
4886:
4887: @item argument type incompatible with parameter:
4888: @cindex Argument type mismatch
4889: This is usually not caught. Some words perform checks, e.g., the control
4890: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
4891: mismatch).
4892:
4893: @item attempting to obtain the execution token of a word with undefined execution semantics:
4894: @cindex Interpreting a compile-only word, for @code{'} etc.
4895: @cindex execution token of words with undefined execution semantics
4896: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
4897: get an execution token for @code{compile-only-error} (which performs a
4898: @code{-14 throw} when executed).
4899:
4900: @item dividing by zero:
4901: @cindex dividing by zero
4902: @cindex floating point unidentified fault, integer division
4903: @cindex divide by zero
4904: typically results in a @code{-55 throw} (floating point unidentified
4905: fault), although a @code{-10 throw} (divide by zero) would be more
4906: appropriate.
4907:
4908: @item insufficient data stack or return stack space:
4909: @cindex insufficient data stack or return stack space
4910: @cindex stack overflow
4911: @cindex Address alignment exception, stack overflow
4912: @cindex Invalid memory address, stack overflow
4913: Depending on the operating system, the installation, and the invocation
4914: of Gforth, this is either checked by the memory management hardware, or
4915: it is not checked. If it is checked, you typically get a @code{-9 throw}
4916: (Invalid memory address) as soon as the overflow happens. If it is not
4917: check, overflows typically result in mysterious illegal memory accesses,
4918: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
4919: (Address alignment exception); they might also destroy the internal data
4920: structure of @code{ALLOCATE} and friends, resulting in various errors in
4921: these words.
4922:
4923: @item insufficient space for loop control parameters:
4924: @cindex insufficient space for loop control parameters
4925: like other return stack overflows.
4926:
4927: @item insufficient space in the dictionary:
4928: @cindex insufficient space in the dictionary
4929: @cindex dictionary overflow
4930: Depending on the operating system, the installation, and the invocation
4931: of Gforth, this is either checked by the memory management hardware, or
4932: it is not checked. Similar results as stack overflows. However,
4933: typically the error appears at a different place when one inserts or
4934: removes code. Also, the @code{THROW} does not relieve the situation (it
4935: does for stack overflows).
4936:
4937: @item interpreting a word with undefined interpretation semantics:
4938: @cindex interpreting a word with undefined interpretation semantics
4939: @cindex Interpreting a compile-only word
4940: For some words, we have defined interpretation semantics. For the
4941: others: @code{-14 throw} (Interpreting a compile-only word).
4942:
4943: @item modifying the contents of the input buffer or a string literal:
4944: @cindex modifying the contents of the input buffer or a string literal
4945: These are located in writable memory and can be modified.
4946:
4947: @item overflow of the pictured numeric output string:
4948: @cindex overflow of the pictured numeric output string
4949: @cindex pictured numeric output string, overflow
4950: Not checked. Runs into the dictionary and destroys it (at least,
4951: partially).
4952:
4953: @item parsed string overflow:
4954: @cindex parsed string overflow
4955: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
4956:
4957: @item producing a result out of range:
4958: @cindex result out of range
4959: On two's complement machines, arithmetic is performed modulo
4960: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
4961: arithmetic (with appropriate mapping for signed types). Division by zero
4962: typically results in a @code{-55 throw} (floatingpoint unidentified
4963: fault), although a @code{-10 throw} (divide by zero) would be more
4964: appropriate. @code{convert} and @code{>number} currently overflow
4965: silently.
4966:
4967: @item reading from an empty data or return stack:
4968: @cindex stack empty
4969: @cindex stack underflow
4970: The data stack is checked by the outer (aka text) interpreter after
4971: every word executed. If it has underflowed, a @code{-4 throw} (Stack
4972: underflow) is performed. Apart from that, stacks may be checked or not,
4973: depending on operating system, installation, and invocation. The
4974: consequences of stack underflows are similar to the consequences of
4975: stack overflows. Note that even if the system uses checking (through the
4976: MMU), your program may have to underflow by a significant number of
4977: stack items to trigger the reaction (the reason for this is that the
4978: MMU, and therefore the checking, works with a page-size granularity).
4979:
4980: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
4981: @cindex unexpected end of the input buffer
4982: @cindex zero-length string as a name
4983: @cindex Attempt to use zero-length string as a name
4984: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
4985: use zero-length string as a name). Words like @code{'} probably will not
4986: find what they search. Note that it is possible to create zero-length
4987: names with @code{nextname} (should it not?).
4988:
4989: @item @code{>IN} greater than input buffer:
4990: @cindex @code{>IN} greater than input buffer
4991: The next invocation of a parsing word returns a string with length 0.
4992:
4993: @item @code{RECURSE} appears after @code{DOES>}:
4994: @cindex @code{RECURSE} appears after @code{DOES>}
4995: Compiles a recursive call to the defining word, not to the defined word.
4996:
4997: @item argument input source different than current input source for @code{RESTORE-INPUT}:
4998: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
4999: @cindex Argument type mismatch, @code{RESTORE-INPUT}
5000: @cindex @code{RESTORE-INPUT}, Argument type mismatch
5001: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
5002: the end of the file was reached), its source-id may be
5003: reused. Therefore, restoring an input source specification referencing a
5004: closed file may lead to unpredictable results instead of a @code{-12
5005: THROW}.
5006:
5007: In the future, Gforth may be able to restore input source specifications
5008: from other than the current input source.
5009:
5010: @item data space containing definitions gets de-allocated:
5011: @cindex data space containing definitions gets de-allocated
5012: Deallocation with @code{allot} is not checked. This typically results in
5013: memory access faults or execution of illegal instructions.
5014:
5015: @item data space read/write with incorrect alignment:
5016: @cindex data space read/write with incorrect alignment
5017: @cindex alignment faults
5018: @cindex Address alignment exception
5019: Processor-dependent. Typically results in a @code{-23 throw} (Address
5020: alignment exception). Under Linux on a 486 or later processor with
5021: alignment turned on, incorrect alignment results in a @code{-9 throw}
5022: (Invalid memory address). There are reportedly some processors with
5023: alignment restrictions that do not report them.
5024:
5025: @item data space pointer not properly aligned, @code{,}, @code{C,}:
5026: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
5027: Like other alignment errors.
5028:
5029: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
5030: Like other stack underflows.
5031:
5032: @item loop control parameters not available:
5033: @cindex loop control parameters not available
5034: Not checked. The counted loop words simply assume that the top of return
5035: stack items are loop control parameters and behave accordingly.
5036:
5037: @item most recent definition does not have a name (@code{IMMEDIATE}):
5038: @cindex most recent definition does not have a name (@code{IMMEDIATE})
5039: @cindex last word was headerless
5040: @code{abort" last word was headerless"}.
5041:
5042: @item name not defined by @code{VALUE} used by @code{TO}:
5043: @cindex name not defined by @code{VALUE} used by @code{TO}
5044: @cindex @code{TO} on non-@code{VALUE}s
5045: @cindex Invalid name argument, @code{TO}
5046: @code{-32 throw} (Invalid name argument) (unless name is a local or was
5047: defined by @code{CONSTANT}; in the latter case it just changes the constant).
5048:
5049: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
5050: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
5051: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
5052: @code{-13 throw} (Undefined word)
5053:
5054: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
5055: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
5056: Gforth behaves as if they were of the same type. I.e., you can predict
5057: the behaviour by interpreting all parameters as, e.g., signed.
5058:
5059: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
5060: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
5061: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
5062: compilation semantics of @code{TO}.
5063:
5064: @item String longer than a counted string returned by @code{WORD}:
5065: @cindex String longer than a counted string returned by @code{WORD}
5066: @cindex @code{WORD}, string overflow
5067: Not checked. The string will be ok, but the count will, of course,
5068: contain only the least significant bits of the length.
5069:
5070: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
5071: @cindex @code{LSHIFT}, large shift counts
5072: @cindex @code{RSHIFT}, large shift counts
5073: Processor-dependent. Typical behaviours are returning 0 and using only
5074: the low bits of the shift count.
5075:
5076: @item word not defined via @code{CREATE}:
5077: @cindex @code{>BODY} of non-@code{CREATE}d words
5078: @code{>BODY} produces the PFA of the word no matter how it was defined.
5079:
5080: @cindex @code{DOES>} of non-@code{CREATE}d words
5081: @code{DOES>} changes the execution semantics of the last defined word no
5082: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
5083: @code{CREATE , DOES>}.
5084:
5085: @item words improperly used outside @code{<#} and @code{#>}:
5086: Not checked. As usual, you can expect memory faults.
5087:
5088: @end table
5089:
5090:
5091: @c ---------------------------------------------------------------------
5092: @node core-other, , core-ambcond, The Core Words
5093: @subsection Other system documentation
5094: @c ---------------------------------------------------------------------
5095: @cindex other system documentation, core words
5096: @cindex core words, other system documentation
5097:
5098: @table @i
5099: @item nonstandard words using @code{PAD}:
5100: @cindex @code{PAD} use by nonstandard words
5101: None.
5102:
5103: @item operator's terminal facilities available:
5104: @cindex operator's terminal facilities available
5105: After processing the command line, Gforth goes into interactive mode,
5106: and you can give commands to Gforth interactively. The actual facilities
5107: available depend on how you invoke Gforth.
5108:
5109: @item program data space available:
5110: @cindex program data space available
5111: @cindex data space available
5112: @code{UNUSED .} gives the remaining dictionary space. The total
5113: dictionary space can be specified with the @code{-m} switch
5114: (@pxref{Invoking Gforth}) when Gforth starts up.
5115:
5116: @item return stack space available:
5117: @cindex return stack space available
5118: You can compute the total return stack space in cells with
5119: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
5120: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
5121:
5122: @item stack space available:
5123: @cindex stack space available
5124: You can compute the total data stack space in cells with
5125: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
5126: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
5127:
5128: @item system dictionary space required, in address units:
5129: @cindex system dictionary space required, in address units
5130: Type @code{here forthstart - .} after startup. At the time of this
5131: writing, this gives 80080 (bytes) on a 32-bit system.
5132: @end table
5133:
5134:
5135: @c =====================================================================
5136: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
5137: @section The optional Block word set
5138: @c =====================================================================
5139: @cindex system documentation, block words
5140: @cindex block words, system documentation
5141:
5142: @menu
5143: * block-idef:: Implementation Defined Options
5144: * block-ambcond:: Ambiguous Conditions
5145: * block-other:: Other System Documentation
5146: @end menu
5147:
5148:
5149: @c ---------------------------------------------------------------------
5150: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
5151: @subsection Implementation Defined Options
5152: @c ---------------------------------------------------------------------
5153: @cindex implementation-defined options, block words
5154: @cindex block words, implementation-defined options
5155:
5156: @table @i
5157: @item the format for display by @code{LIST}:
5158: @cindex @code{LIST} display format
5159: First the screen number is displayed, then 16 lines of 64 characters,
5160: each line preceded by the line number.
5161:
5162: @item the length of a line affected by @code{\}:
5163: @cindex length of a line affected by @code{\}
5164: @cindex @code{\}, line length in blocks
5165: 64 characters.
5166: @end table
5167:
5168:
5169: @c ---------------------------------------------------------------------
5170: @node block-ambcond, block-other, block-idef, The optional Block word set
5171: @subsection Ambiguous conditions
5172: @c ---------------------------------------------------------------------
5173: @cindex block words, ambiguous conditions
5174: @cindex ambiguous conditions, block words
5175:
5176: @table @i
5177: @item correct block read was not possible:
5178: @cindex block read not possible
5179: Typically results in a @code{throw} of some OS-derived value (between
5180: -512 and -2048). If the blocks file was just not long enough, blanks are
5181: supplied for the missing portion.
5182:
5183: @item I/O exception in block transfer:
5184: @cindex I/O exception in block transfer
5185: @cindex block transfer, I/O exception
5186: Typically results in a @code{throw} of some OS-derived value (between
5187: -512 and -2048).
5188:
5189: @item invalid block number:
5190: @cindex invalid block number
5191: @cindex block number invalid
5192: @code{-35 throw} (Invalid block number)
5193:
5194: @item a program directly alters the contents of @code{BLK}:
5195: @cindex @code{BLK}, altering @code{BLK}
5196: The input stream is switched to that other block, at the same
5197: position. If the storing to @code{BLK} happens when interpreting
5198: non-block input, the system will get quite confused when the block ends.
5199:
5200: @item no current block buffer for @code{UPDATE}:
5201: @cindex @code{UPDATE}, no current block buffer
5202: @code{UPDATE} has no effect.
5203:
5204: @end table
5205:
5206: @c ---------------------------------------------------------------------
5207: @node block-other, , block-ambcond, The optional Block word set
5208: @subsection Other system documentation
5209: @c ---------------------------------------------------------------------
5210: @cindex other system documentation, block words
5211: @cindex block words, other system documentation
5212:
5213: @table @i
5214: @item any restrictions a multiprogramming system places on the use of buffer addresses:
5215: No restrictions (yet).
5216:
5217: @item the number of blocks available for source and data:
5218: depends on your disk space.
5219:
5220: @end table
5221:
5222:
5223: @c =====================================================================
5224: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
5225: @section The optional Double Number word set
5226: @c =====================================================================
5227: @cindex system documentation, double words
5228: @cindex double words, system documentation
5229:
5230: @menu
5231: * double-ambcond:: Ambiguous Conditions
5232: @end menu
5233:
5234:
5235: @c ---------------------------------------------------------------------
5236: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
5237: @subsection Ambiguous conditions
5238: @c ---------------------------------------------------------------------
5239: @cindex double words, ambiguous conditions
5240: @cindex ambiguous conditions, double words
5241:
5242: @table @i
5243: @item @var{d} outside of range of @var{n} in @code{D>S}:
5244: @cindex @code{D>S}, @var{d} out of range of @var{n}
5245: The least significant cell of @var{d} is produced.
5246:
5247: @end table
5248:
5249:
5250: @c =====================================================================
5251: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
5252: @section The optional Exception word set
5253: @c =====================================================================
5254: @cindex system documentation, exception words
5255: @cindex exception words, system documentation
5256:
5257: @menu
5258: * exception-idef:: Implementation Defined Options
5259: @end menu
5260:
5261:
5262: @c ---------------------------------------------------------------------
5263: @node exception-idef, , The optional Exception word set, The optional Exception word set
5264: @subsection Implementation Defined Options
5265: @c ---------------------------------------------------------------------
5266: @cindex implementation-defined options, exception words
5267: @cindex exception words, implementation-defined options
5268:
5269: @table @i
5270: @item @code{THROW}-codes used in the system:
5271: @cindex @code{THROW}-codes used in the system
5272: The codes -256@minus{}-511 are used for reporting signals. The mapping
5273: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
5274: codes -512@minus{}-2047 are used for OS errors (for file and memory
5275: allocation operations). The mapping from OS error numbers to throw codes
5276: is -512@minus{}@code{errno}. One side effect of this mapping is that
5277: undefined OS errors produce a message with a strange number; e.g.,
5278: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
5279: @end table
5280:
5281: @c =====================================================================
5282: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
5283: @section The optional Facility word set
5284: @c =====================================================================
5285: @cindex system documentation, facility words
5286: @cindex facility words, system documentation
5287:
5288: @menu
5289: * facility-idef:: Implementation Defined Options
5290: * facility-ambcond:: Ambiguous Conditions
5291: @end menu
5292:
5293:
5294: @c ---------------------------------------------------------------------
5295: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
5296: @subsection Implementation Defined Options
5297: @c ---------------------------------------------------------------------
5298: @cindex implementation-defined options, facility words
5299: @cindex facility words, implementation-defined options
5300:
5301: @table @i
5302: @item encoding of keyboard events (@code{EKEY}):
5303: @cindex keyboard events, encoding in @code{EKEY}
5304: @cindex @code{EKEY}, encoding of keyboard events
5305: Not yet implemented.
5306:
5307: @item duration of a system clock tick:
5308: @cindex duration of a system clock tick
5309: @cindex clock tick duration
5310: System dependent. With respect to @code{MS}, the time is specified in
5311: microseconds. How well the OS and the hardware implement this, is
5312: another question.
5313:
5314: @item repeatability to be expected from the execution of @code{MS}:
5315: @cindex repeatability to be expected from the execution of @code{MS}
5316: @cindex @code{MS}, repeatability to be expected
5317: System dependent. On Unix, a lot depends on load. If the system is
5318: lightly loaded, and the delay is short enough that Gforth does not get
5319: swapped out, the performance should be acceptable. Under MS-DOS and
5320: other single-tasking systems, it should be good.
5321:
5322: @end table
5323:
5324:
5325: @c ---------------------------------------------------------------------
5326: @node facility-ambcond, , facility-idef, The optional Facility word set
5327: @subsection Ambiguous conditions
5328: @c ---------------------------------------------------------------------
5329: @cindex facility words, ambiguous conditions
5330: @cindex ambiguous conditions, facility words
5331:
5332: @table @i
5333: @item @code{AT-XY} can't be performed on user output device:
5334: @cindex @code{AT-XY} can't be performed on user output device
5335: Largely terminal dependent. No range checks are done on the arguments.
5336: No errors are reported. You may see some garbage appearing, you may see
5337: simply nothing happen.
5338:
5339: @end table
5340:
5341:
5342: @c =====================================================================
5343: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
5344: @section The optional File-Access word set
5345: @c =====================================================================
5346: @cindex system documentation, file words
5347: @cindex file words, system documentation
5348:
5349: @menu
5350: * file-idef:: Implementation Defined Options
5351: * file-ambcond:: Ambiguous Conditions
5352: @end menu
5353:
5354: @c ---------------------------------------------------------------------
5355: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
5356: @subsection Implementation Defined Options
5357: @c ---------------------------------------------------------------------
5358: @cindex implementation-defined options, file words
5359: @cindex file words, implementation-defined options
5360:
5361: @table @i
5362: @item file access methods used:
5363: @cindex file access methods used
5364: @code{R/O}, @code{R/W} and @code{BIN} work as you would
5365: expect. @code{W/O} translates into the C file opening mode @code{w} (or
5366: @code{wb}): The file is cleared, if it exists, and created, if it does
5367: not (with both @code{open-file} and @code{create-file}). Under Unix
5368: @code{create-file} creates a file with 666 permissions modified by your
5369: umask.
5370:
5371: @item file exceptions:
5372: @cindex file exceptions
5373: The file words do not raise exceptions (except, perhaps, memory access
5374: faults when you pass illegal addresses or file-ids).
5375:
5376: @item file line terminator:
5377: @cindex file line terminator
5378: System-dependent. Gforth uses C's newline character as line
5379: terminator. What the actual character code(s) of this are is
5380: system-dependent.
5381:
5382: @item file name format:
5383: @cindex file name format
5384: System dependent. Gforth just uses the file name format of your OS.
5385:
5386: @item information returned by @code{FILE-STATUS}:
5387: @cindex @code{FILE-STATUS}, returned information
5388: @code{FILE-STATUS} returns the most powerful file access mode allowed
5389: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
5390: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
5391: along with the returned mode.
5392:
5393: @item input file state after an exception when including source:
5394: @cindex exception when including source
5395: All files that are left via the exception are closed.
5396:
5397: @item @var{ior} values and meaning:
5398: @cindex @var{ior} values and meaning
5399: The @var{ior}s returned by the file and memory allocation words are
5400: intended as throw codes. They typically are in the range
5401: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
5402: @var{ior}s is -512@minus{}@var{errno}.
5403:
5404: @item maximum depth of file input nesting:
5405: @cindex maximum depth of file input nesting
5406: @cindex file input nesting, maximum depth
5407: limited by the amount of return stack, locals/TIB stack, and the number
5408: of open files available. This should not give you troubles.
5409:
5410: @item maximum size of input line:
5411: @cindex maximum size of input line
5412: @cindex input line size, maximum
5413: @code{/line}. Currently 255.
5414:
5415: @item methods of mapping block ranges to files:
5416: @cindex mapping block ranges to files
5417: @cindex files containing blocks
5418: @cindex blocks in files
5419: By default, blocks are accessed in the file @file{blocks.fb} in the
5420: current working directory. The file can be switched with @code{USE}.
5421:
5422: @item number of string buffers provided by @code{S"}:
5423: @cindex @code{S"}, number of string buffers
5424: 1
5425:
5426: @item size of string buffer used by @code{S"}:
5427: @cindex @code{S"}, size of string buffer
5428: @code{/line}. currently 255.
5429:
5430: @end table
5431:
5432: @c ---------------------------------------------------------------------
5433: @node file-ambcond, , file-idef, The optional File-Access word set
5434: @subsection Ambiguous conditions
5435: @c ---------------------------------------------------------------------
5436: @cindex file words, ambiguous conditions
5437: @cindex ambiguous conditions, file words
5438:
5439: @table @i
5440: @item attempting to position a file outside its boundaries:
5441: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
5442: @code{REPOSITION-FILE} is performed as usual: Afterwards,
5443: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
5444:
5445: @item attempting to read from file positions not yet written:
5446: @cindex reading from file positions not yet written
5447: End-of-file, i.e., zero characters are read and no error is reported.
5448:
5449: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
5450: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid
5451: An appropriate exception may be thrown, but a memory fault or other
5452: problem is more probable.
5453:
5454: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
5455: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
5456: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
5457: The @var{ior} produced by the operation, that discovered the problem, is
5458: thrown.
5459:
5460: @item named file cannot be opened (@code{INCLUDED}):
5461: @cindex @code{INCLUDED}, named file cannot be opened
5462: The @var{ior} produced by @code{open-file} is thrown.
5463:
5464: @item requesting an unmapped block number:
5465: @cindex unmapped block numbers
5466: There are no unmapped legal block numbers. On some operating systems,
5467: writing a block with a large number may overflow the file system and
5468: have an error message as consequence.
5469:
5470: @item using @code{source-id} when @code{blk} is non-zero:
5471: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
5472: @code{source-id} performs its function. Typically it will give the id of
5473: the source which loaded the block. (Better ideas?)
5474:
5475: @end table
5476:
5477:
5478: @c =====================================================================
5479: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
5480: @section The optional Floating-Point word set
5481: @c =====================================================================
5482: @cindex system documentation, floating-point words
5483: @cindex floating-point words, system documentation
5484:
5485: @menu
5486: * floating-idef:: Implementation Defined Options
5487: * floating-ambcond:: Ambiguous Conditions
5488: @end menu
5489:
5490:
5491: @c ---------------------------------------------------------------------
5492: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
5493: @subsection Implementation Defined Options
5494: @c ---------------------------------------------------------------------
5495: @cindex implementation-defined options, floating-point words
5496: @cindex floating-point words, implementation-defined options
5497:
5498: @table @i
5499: @item format and range of floating point numbers:
5500: @cindex format and range of floating point numbers
5501: @cindex floating point numbers, format and range
5502: System-dependent; the @code{double} type of C.
5503:
5504: @item results of @code{REPRESENT} when @var{float} is out of range:
5505: @cindex @code{REPRESENT}, results when @var{float} is out of range
5506: System dependent; @code{REPRESENT} is implemented using the C library
5507: function @code{ecvt()} and inherits its behaviour in this respect.
5508:
5509: @item rounding or truncation of floating-point numbers:
5510: @cindex rounding of floating-point numbers
5511: @cindex truncation of floating-point numbers
5512: @cindex floating-point numbers, rounding or truncation
5513: System dependent; the rounding behaviour is inherited from the hosting C
5514: compiler. IEEE-FP-based (i.e., most) systems by default round to
5515: nearest, and break ties by rounding to even (i.e., such that the last
5516: bit of the mantissa is 0).
5517:
5518: @item size of floating-point stack:
5519: @cindex floating-point stack size
5520: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
5521: the floating-point stack (in floats). You can specify this on startup
5522: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
5523:
5524: @item width of floating-point stack:
5525: @cindex floating-point stack width
5526: @code{1 floats}.
5527:
5528: @end table
5529:
5530:
5531: @c ---------------------------------------------------------------------
5532: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
5533: @subsection Ambiguous conditions
5534: @c ---------------------------------------------------------------------
5535: @cindex floating-point words, ambiguous conditions
5536: @cindex ambiguous conditions, floating-point words
5537:
5538: @table @i
5539: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
5540: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
5541: System-dependent. Typically results in a @code{-23 THROW} like other
5542: alignment violations.
5543:
5544: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
5545: @cindex @code{f@@} used with an address that is not float aligned
5546: @cindex @code{f!} used with an address that is not float aligned
5547: System-dependent. Typically results in a @code{-23 THROW} like other
5548: alignment violations.
5549:
5550: @item floating-point result out of range:
5551: @cindex floating-point result out of range
5552: System-dependent. Can result in a @code{-55 THROW} (Floating-point
5553: unidentified fault), or can produce a special value representing, e.g.,
5554: Infinity.
5555:
5556: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
5557: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
5558: System-dependent. Typically results in an alignment fault like other
5559: alignment violations.
5560:
5561: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
5562: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
5563: The floating-point number is converted into decimal nonetheless.
5564:
5565: @item Both arguments are equal to zero (@code{FATAN2}):
5566: @cindex @code{FATAN2}, both arguments are equal to zero
5567: System-dependent. @code{FATAN2} is implemented using the C library
5568: function @code{atan2()}.
5569:
5570: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
5571: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
5572: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
5573: because of small errors and the tan will be a very large (or very small)
5574: but finite number.
5575:
5576: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
5577: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
5578: The result is rounded to the nearest float.
5579:
5580: @item dividing by zero:
5581: @cindex dividing by zero, floating-point
5582: @cindex floating-point dividing by zero
5583: @cindex floating-point unidentified fault, FP divide-by-zero
5584: @code{-55 throw} (Floating-point unidentified fault)
5585:
5586: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
5587: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
5588: System dependent. On IEEE-FP based systems the number is converted into
5589: an infinity.
5590:
5591: @item @var{float}<1 (@code{FACOSH}):
5592: @cindex @code{FACOSH}, @var{float}<1
5593: @cindex floating-point unidentified fault, @code{FACOSH}
5594: @code{-55 throw} (Floating-point unidentified fault)
5595:
5596: @item @var{float}=<-1 (@code{FLNP1}):
5597: @cindex @code{FLNP1}, @var{float}=<-1
5598: @cindex floating-point unidentified fault, @code{FLNP1}
5599: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
5600: negative infinity is typically produced for @var{float}=-1.
5601:
5602: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
5603: @cindex @code{FLN}, @var{float}=<0
5604: @cindex @code{FLOG}, @var{float}=<0
5605: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
5606: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
5607: negative infinity is typically produced for @var{float}=0.
5608:
5609: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
5610: @cindex @code{FASINH}, @var{float}<0
5611: @cindex @code{FSQRT}, @var{float}<0
5612: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
5613: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
5614: produces values for these inputs on my Linux box (Bug in the C library?)
5615:
5616: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
5617: @cindex @code{FACOS}, |@var{float}|>1
5618: @cindex @code{FASIN}, |@var{float}|>1
5619: @cindex @code{FATANH}, |@var{float}|>1
5620: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
5621: @code{-55 throw} (Floating-point unidentified fault).
5622:
5623: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
5624: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
5625: @cindex floating-point unidentified fault, @code{F>D}
5626: @code{-55 throw} (Floating-point unidentified fault).
5627:
5628: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
5629: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
5630: This does not happen.
5631: @end table
5632:
5633: @c =====================================================================
5634: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
5635: @section The optional Locals word set
5636: @c =====================================================================
5637: @cindex system documentation, locals words
5638: @cindex locals words, system documentation
5639:
5640: @menu
5641: * locals-idef:: Implementation Defined Options
5642: * locals-ambcond:: Ambiguous Conditions
5643: @end menu
5644:
5645:
5646: @c ---------------------------------------------------------------------
5647: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
5648: @subsection Implementation Defined Options
5649: @c ---------------------------------------------------------------------
5650: @cindex implementation-defined options, locals words
5651: @cindex locals words, implementation-defined options
5652:
5653: @table @i
5654: @item maximum number of locals in a definition:
5655: @cindex maximum number of locals in a definition
5656: @cindex locals, maximum number in a definition
5657: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
5658: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
5659: characters. The number of locals in a definition is bounded by the size
5660: of locals-buffer, which contains the names of the locals.
5661:
5662: @end table
5663:
5664:
5665: @c ---------------------------------------------------------------------
5666: @node locals-ambcond, , locals-idef, The optional Locals word set
5667: @subsection Ambiguous conditions
5668: @c ---------------------------------------------------------------------
5669: @cindex locals words, ambiguous conditions
5670: @cindex ambiguous conditions, locals words
5671:
5672: @table @i
5673: @item executing a named local in interpretation state:
5674: @cindex local in interpretation state
5675: @cindex Interpreting a compile-only word, for a local
5676: Locals have no interpretation semantics. If you try to perform the
5677: interpretation semantics, you will get a @code{-14 throw} somewhere
5678: (Interpreting a compile-only word). If you perform the compilation
5679: semantics, the locals access will be compiled (irrespective of state).
5680:
5681: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
5682: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
5683: @cindex @code{TO} on non-@code{VALUE}s and non-locals
5684: @cindex Invalid name argument, @code{TO}
5685: @code{-32 throw} (Invalid name argument)
5686:
5687: @end table
5688:
5689:
5690: @c =====================================================================
5691: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
5692: @section The optional Memory-Allocation word set
5693: @c =====================================================================
5694: @cindex system documentation, memory-allocation words
5695: @cindex memory-allocation words, system documentation
5696:
5697: @menu
5698: * memory-idef:: Implementation Defined Options
5699: @end menu
5700:
5701:
5702: @c ---------------------------------------------------------------------
5703: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
5704: @subsection Implementation Defined Options
5705: @c ---------------------------------------------------------------------
5706: @cindex implementation-defined options, memory-allocation words
5707: @cindex memory-allocation words, implementation-defined options
5708:
5709: @table @i
5710: @item values and meaning of @var{ior}:
5711: @cindex @var{ior} values and meaning
5712: The @var{ior}s returned by the file and memory allocation words are
5713: intended as throw codes. They typically are in the range
5714: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
5715: @var{ior}s is -512@minus{}@var{errno}.
5716:
5717: @end table
5718:
5719: @c =====================================================================
5720: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
5721: @section The optional Programming-Tools word set
5722: @c =====================================================================
5723: @cindex system documentation, programming-tools words
5724: @cindex programming-tools words, system documentation
5725:
5726: @menu
5727: * programming-idef:: Implementation Defined Options
5728: * programming-ambcond:: Ambiguous Conditions
5729: @end menu
5730:
5731:
5732: @c ---------------------------------------------------------------------
5733: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
5734: @subsection Implementation Defined Options
5735: @c ---------------------------------------------------------------------
5736: @cindex implementation-defined options, programming-tools words
5737: @cindex programming-tools words, implementation-defined options
5738:
5739: @table @i
5740: @item ending sequence for input following @code{;CODE} and @code{CODE}:
5741: @cindex @code{;CODE} ending sequence
5742: @cindex @code{CODE} ending sequence
5743: @code{END-CODE}
5744:
5745: @item manner of processing input following @code{;CODE} and @code{CODE}:
5746: @cindex @code{;CODE}, processing input
5747: @cindex @code{CODE}, processing input
5748: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
5749: the input is processed by the text interpreter, (starting) in interpret
5750: state.
5751:
5752: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
5753: @cindex @code{ASSEMBLER}, search order capability
5754: The ANS Forth search order word set.
5755:
5756: @item source and format of display by @code{SEE}:
5757: @cindex @code{SEE}, source and format of output
5758: The source for @code{see} is the intermediate code used by the inner
5759: interpreter. The current @code{see} tries to output Forth source code
5760: as well as possible.
5761:
5762: @end table
5763:
5764: @c ---------------------------------------------------------------------
5765: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
5766: @subsection Ambiguous conditions
5767: @c ---------------------------------------------------------------------
5768: @cindex programming-tools words, ambiguous conditions
5769: @cindex ambiguous conditions, programming-tools words
5770:
5771: @table @i
5772:
5773: @item deleting the compilation wordlist (@code{FORGET}):
5774: @cindex @code{FORGET}, deleting the compilation wordlist
5775: Not implemented (yet).
5776:
5777: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
5778: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
5779: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
5780: @cindex control-flow stack underflow
5781: This typically results in an @code{abort"} with a descriptive error
5782: message (may change into a @code{-22 throw} (Control structure mismatch)
5783: in the future). You may also get a memory access error. If you are
5784: unlucky, this ambiguous condition is not caught.
5785:
5786: @item @var{name} can't be found (@code{FORGET}):
5787: @cindex @code{FORGET}, @var{name} can't be found
5788: Not implemented (yet).
5789:
5790: @item @var{name} not defined via @code{CREATE}:
5791: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
5792: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
5793: the execution semantics of the last defined word no matter how it was
5794: defined.
5795:
5796: @item @code{POSTPONE} applied to @code{[IF]}:
5797: @cindex @code{POSTPONE} applied to @code{[IF]}
5798: @cindex @code{[IF]} and @code{POSTPONE}
5799: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
5800: equivalent to @code{[IF]}.
5801:
5802: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
5803: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
5804: Continue in the same state of conditional compilation in the next outer
5805: input source. Currently there is no warning to the user about this.
5806:
5807: @item removing a needed definition (@code{FORGET}):
5808: @cindex @code{FORGET}, removing a needed definition
5809: Not implemented (yet).
5810:
5811: @end table
5812:
5813:
5814: @c =====================================================================
5815: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
5816: @section The optional Search-Order word set
5817: @c =====================================================================
5818: @cindex system documentation, search-order words
5819: @cindex search-order words, system documentation
5820:
5821: @menu
5822: * search-idef:: Implementation Defined Options
5823: * search-ambcond:: Ambiguous Conditions
5824: @end menu
5825:
5826:
5827: @c ---------------------------------------------------------------------
5828: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
5829: @subsection Implementation Defined Options
5830: @c ---------------------------------------------------------------------
5831: @cindex implementation-defined options, search-order words
5832: @cindex search-order words, implementation-defined options
5833:
5834: @table @i
5835: @item maximum number of word lists in search order:
5836: @cindex maximum number of word lists in search order
5837: @cindex search order, maximum depth
5838: @code{s" wordlists" environment? drop .}. Currently 16.
5839:
5840: @item minimum search order:
5841: @cindex minimum search order
5842: @cindex search order, minimum
5843: @code{root root}.
5844:
5845: @end table
5846:
5847: @c ---------------------------------------------------------------------
5848: @node search-ambcond, , search-idef, The optional Search-Order word set
5849: @subsection Ambiguous conditions
5850: @c ---------------------------------------------------------------------
5851: @cindex search-order words, ambiguous conditions
5852: @cindex ambiguous conditions, search-order words
5853:
5854: @table @i
5855: @item changing the compilation wordlist (during compilation):
5856: @cindex changing the compilation wordlist (during compilation)
5857: @cindex compilation wordlist, change before definition ends
5858: The word is entered into the wordlist that was the compilation wordlist
5859: at the start of the definition. Any changes to the name field (e.g.,
5860: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
5861: are applied to the latest defined word (as reported by @code{last} or
5862: @code{lastxt}), if possible, irrespective of the compilation wordlist.
5863:
5864: @item search order empty (@code{previous}):
5865: @cindex @code{previous}, search order empty
5866: @cindex Vocstack empty, @code{previous}
5867: @code{abort" Vocstack empty"}.
5868:
5869: @item too many word lists in search order (@code{also}):
5870: @cindex @code{also}, too many word lists in search order
5871: @cindex Vocstack full, @code{also}
5872: @code{abort" Vocstack full"}.
5873:
5874: @end table
5875:
5876: @c ***************************************************************
5877: @node Model, Integrating Gforth, ANS conformance, Top
5878: @chapter Model
5879:
5880: This chapter has yet to be written. It will contain information, on
5881: which internal structures you can rely.
5882:
5883: @c ***************************************************************
5884: @node Integrating Gforth, Emacs and Gforth, Model, Top
5885: @chapter Integrating Gforth into C programs
5886:
5887: This is not yet implemented.
5888:
5889: Several people like to use Forth as scripting language for applications
5890: that are otherwise written in C, C++, or some other language.
5891:
5892: The Forth system ATLAST provides facilities for embedding it into
5893: applications; unfortunately it has several disadvantages: most
5894: importantly, it is not based on ANS Forth, and it is apparently dead
5895: (i.e., not developed further and not supported). The facilities
5896: provided by Gforth in this area are inspired by ATLASTs facilities, so
5897: making the switch should not be hard.
5898:
5899: We also tried to design the interface such that it can easily be
5900: implemented by other Forth systems, so that we may one day arrive at a
5901: standardized interface. Such a standard interface would allow you to
5902: replace the Forth system without having to rewrite C code.
5903:
5904: You embed the Gforth interpreter by linking with the library
5905: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
5906: global symbols in this library that belong to the interface, have the
5907: prefix @code{forth_}. (Global symbols that are used internally have the
5908: prefix @code{gforth_}).
5909:
5910: You can include the declarations of Forth types and the functions and
5911: variables of the interface with @code{#include <forth.h>}.
5912:
5913: Types.
5914:
5915: Variables.
5916:
5917: Data and FP Stack pointer. Area sizes.
5918:
5919: functions.
5920:
5921: forth_init(imagefile)
5922: forth_evaluate(string) exceptions?
5923: forth_goto(address) (or forth_execute(xt)?)
5924: forth_continue() (a corountining mechanism)
5925:
5926: Adding primitives.
5927:
5928: No checking.
5929:
5930: Signals?
5931:
5932: Accessing the Stacks
5933:
5934: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
5935: @chapter Emacs and Gforth
5936: @cindex Emacs and Gforth
5937:
5938: @cindex @file{gforth.el}
5939: @cindex @file{forth.el}
5940: @cindex Rydqvist, Goran
5941: @cindex comment editing commands
5942: @cindex @code{\}, editing with Emacs
5943: @cindex debug tracer editing commands
5944: @cindex @code{~~}, removal with Emacs
5945: @cindex Forth mode in Emacs
5946: Gforth comes with @file{gforth.el}, an improved version of
5947: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
5948: improvements are a better (but still not perfect) handling of
5949: indentation. I have also added comment paragraph filling (@kbd{M-q}),
5950: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
5951: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
5952: stuff I do not use alone, even though some of it only makes sense for
5953: TILE. To get a description of these features, enter Forth mode and type
5954: @kbd{C-h m}.
5955:
5956: @cindex source location of error or debugging output in Emacs
5957: @cindex error output, finding the source location in Emacs
5958: @cindex debugging output, finding the source location in Emacs
5959: In addition, Gforth supports Emacs quite well: The source code locations
5960: given in error messages, debugging output (from @code{~~}) and failed
5961: assertion messages are in the right format for Emacs' compilation mode
5962: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
5963: Manual}) so the source location corresponding to an error or other
5964: message is only a few keystrokes away (@kbd{C-x `} for the next error,
5965: @kbd{C-c C-c} for the error under the cursor).
5966:
5967: @cindex @file{TAGS} file
5968: @cindex @file{etags.fs}
5969: @cindex viewing the source of a word in Emacs
5970: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
5971: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
5972: contains the definitions of all words defined afterwards. You can then
5973: find the source for a word using @kbd{M-.}. Note that emacs can use
5974: several tags files at the same time (e.g., one for the Gforth sources
5975: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
5976: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
5977: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
5978: @file{/usr/local/share/gforth/0.2.0/TAGS}).
5979:
5980: @cindex @file{.emacs}
5981: To get all these benefits, add the following lines to your @file{.emacs}
5982: file:
5983:
5984: @example
5985: (autoload 'forth-mode "gforth.el")
5986: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
5987: @end example
5988:
5989: @node Image Files, Engine, Emacs and Gforth, Top
5990: @chapter Image Files
5991: @cindex image files
5992: @cindex @code{.fi} files
5993: @cindex precompiled Forth code
5994: @cindex dictionary in persistent form
5995: @cindex persistent form of dictionary
5996:
5997: An image file is a file containing an image of the Forth dictionary,
5998: i.e., compiled Forth code and data residing in the dictionary. By
5999: convention, we use the extension @code{.fi} for image files.
6000:
6001: @menu
6002: * Image File Background:: Why have image files?
6003: * Non-Relocatable Image Files:: don't always work.
6004: * Data-Relocatable Image Files:: are better.
6005: * Fully Relocatable Image Files:: better yet.
6006: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
6007: * Running Image Files:: @code{gforth -i @var{file}} or @var{file}.
6008: * Modifying the Startup Sequence:: and turnkey applications.
6009: @end menu
6010:
6011: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
6012: @section Image File Background
6013: @cindex image file background
6014:
6015: Our Forth system consists not only of primitives, but also of
6016: definitions written in Forth. Since the Forth compiler itself belongs to
6017: those definitions, it is not possible to start the system with the
6018: primitives and the Forth source alone. Therefore we provide the Forth
6019: code as an image file in nearly executable form. At the start of the
6020: system a C routine loads the image file into memory, optionally
6021: relocates the addresses, then sets up the memory (stacks etc.) according
6022: to information in the image file, and starts executing Forth code.
6023:
6024: The image file variants represent different compromises between the
6025: goals of making it easy to generate image files and making them
6026: portable.
6027:
6028: @cindex relocation at run-time
6029: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
6030: run-time. This avoids many of the complications discussed below (image
6031: files are data relocatable without further ado), but costs performance
6032: (one addition per memory access).
6033:
6034: @cindex relocation at load-time
6035: By contrast, our loader performs relocation at image load time. The
6036: loader also has to replace tokens standing for primitive calls with the
6037: appropriate code-field addresses (or code addresses in the case of
6038: direct threading).
6039:
6040: There are three kinds of image files, with different degrees of
6041: relocatability: non-relocatable, data-relocatable, and fully relocatable
6042: image files.
6043:
6044: @cindex image file loader
6045: @cindex relocating loader
6046: @cindex loader for image files
6047: These image file variants have several restrictions in common; they are
6048: caused by the design of the image file loader:
6049:
6050: @itemize @bullet
6051: @item
6052: There is only one segment; in particular, this means, that an image file
6053: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
6054: them). And the contents of the stacks are not represented, either.
6055:
6056: @item
6057: The only kinds of relocation supported are: adding the same offset to
6058: all cells that represent data addresses; and replacing special tokens
6059: with code addresses or with pieces of machine code.
6060:
6061: If any complex computations involving addresses are performed, the
6062: results cannot be represented in the image file. Several applications that
6063: use such computations come to mind:
6064: @itemize @minus
6065: @item
6066: Hashing addresses (or data structures which contain addresses) for table
6067: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
6068: purpose, you will have no problem, because the hash tables are
6069: recomputed automatically when the system is started. If you use your own
6070: hash tables, you will have to do something similar.
6071:
6072: @item
6073: There's a cute implementation of doubly-linked lists that uses
6074: @code{XOR}ed addresses. You could represent such lists as singly-linked
6075: in the image file, and restore the doubly-linked representation on
6076: startup.@footnote{In my opinion, though, you should think thrice before
6077: using a doubly-linked list (whatever implementation).}
6078:
6079: @item
6080: The code addresses of run-time routines like @code{docol:} cannot be
6081: represented in the image file (because their tokens would be replaced by
6082: machine code in direct threaded implementations). As a workaround,
6083: compute these addresses at run-time with @code{>code-address} from the
6084: executions tokens of appropriate words (see the definitions of
6085: @code{docol:} and friends in @file{kernel.fs}).
6086:
6087: @item
6088: On many architectures addresses are represented in machine code in some
6089: shifted or mangled form. You cannot put @code{CODE} words that contain
6090: absolute addresses in this form in a relocatable image file. Workarounds
6091: are representing the address in some relative form (e.g., relative to
6092: the CFA, which is present in some register), or loading the address from
6093: a place where it is stored in a non-mangled form.
6094: @end itemize
6095: @end itemize
6096:
6097: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
6098: @section Non-Relocatable Image Files
6099: @cindex non-relocatable image files
6100: @cindex image files, non-relocatable
6101:
6102: These files are simple memory dumps of the dictionary. They are specific
6103: to the executable (i.e., @file{gforth} file) they were created
6104: with. What's worse, they are specific to the place on which the
6105: dictionary resided when the image was created. Now, there is no
6106: guarantee that the dictionary will reside at the same place the next
6107: time you start Gforth, so there's no guarantee that a non-relocatable
6108: image will work the next time (Gforth will complain instead of crashing,
6109: though).
6110:
6111: You can create a non-relocatable image file with
6112:
6113: doc-savesystem
6114:
6115: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
6116: @section Data-Relocatable Image Files
6117: @cindex data-relocatable image files
6118: @cindex image files, data-relocatable
6119:
6120: These files contain relocatable data addresses, but fixed code addresses
6121: (instead of tokens). They are specific to the executable (i.e.,
6122: @file{gforth} file) they were created with. For direct threading on some
6123: architectures (e.g., the i386), data-relocatable images do not work. You
6124: get a data-relocatable image, if you use @file{gforthmi} with a
6125: Gforth binary that is not doubly indirect threaded (@pxref{Fully
6126: Relocatable Image Files}).
6127:
6128: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
6129: @section Fully Relocatable Image Files
6130: @cindex fully relocatable image files
6131: @cindex image files, fully relocatable
6132:
6133: @cindex @file{kern*.fi}, relocatability
6134: @cindex @file{gforth.fi}, relocatability
6135: These image files have relocatable data addresses, and tokens for code
6136: addresses. They can be used with different binaries (e.g., with and
6137: without debugging) on the same machine, and even across machines with
6138: the same data formats (byte order, cell size, floating point
6139: format). However, they are usually specific to the version of Gforth
6140: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
6141: are fully relocatable.
6142:
6143: There are two ways to create a fully relocatable image file:
6144:
6145: @menu
6146: * gforthmi:: The normal way
6147: * cross.fs:: The hard way
6148: @end menu
6149:
6150: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
6151: @subsection @file{gforthmi}
6152: @cindex @file{comp-i.fs}
6153: @cindex @file{gforthmi}
6154:
6155: You will usually use @file{gforthmi}. If you want to create an
6156: image @var{file} that contains everything you would load by invoking
6157: Gforth with @code{gforth @var{options}}, you simply say
6158: @example
6159: gforthmi @var{file} @var{options}
6160: @end example
6161:
6162: E.g., if you want to create an image @file{asm.fi} that has the file
6163: @file{asm.fs} loaded in addition to the usual stuff, you could do it
6164: like this:
6165:
6166: @example
6167: gforthmi asm.fi asm.fs
6168: @end example
6169:
6170: @file{gforthmi} works like this: It produces two non-relocatable
6171: images for different addresses and then compares them. Its output
6172: reflects this: first you see the output (if any) of the two Gforth
6173: invocations that produce the nonrelocatable image files, then you see
6174: the output of the comparing program: It displays the offset used for
6175: data addresses and the offset used for code addresses;
6176: moreover, for each cell that cannot be represented correctly in the
6177: image files, it displays a line like the following one:
6178:
6179: @example
6180: 78DC BFFFFA50 BFFFFA40
6181: @end example
6182:
6183: This means that at offset $78dc from @code{forthstart}, one input image
6184: contains $bffffa50, and the other contains $bffffa40. Since these cells
6185: cannot be represented correctly in the output image, you should examine
6186: these places in the dictionary and verify that these cells are dead
6187: (i.e., not read before they are written).
6188:
6189: @cindex @code{savesystem} during @file{gforthmi}
6190: @cindex @code{bye} during @file{gforthmi}
6191: @cindex doubly indirect threaded code
6192: @cindex environment variable @code{GFORTHD}
6193: @cindex @code{GFORTHD} environment variable
6194: @cindex @code{gforth-ditc}
6195: There are a few wrinkles: After processing the passed @var{options}, the
6196: words @code{savesystem} and @code{bye} must be visible. A special doubly
6197: indirect threaded version of the @file{gforth} executable is used for
6198: creating the nonrelocatable images; you can pass the exact filename of
6199: this executable through the environment variable @code{GFORTHD}
6200: (default: @file{gforth-ditc}); if you pass a version that is not doubly
6201: indirect threaded, you will not get a fully relocatable image, but a
6202: data-relocatable image (because there is no code address offset).
6203:
6204: @node cross.fs, , gforthmi, Fully Relocatable Image Files
6205: @subsection @file{cross.fs}
6206: @cindex @file{cross.fs}
6207: @cindex cross-compiler
6208: @cindex metacompiler
6209:
6210: You can also use @code{cross}, a batch compiler that accepts a Forth-like
6211: programming language. This @code{cross} language has to be documented
6212: yet.
6213:
6214: @cindex target compiler
6215: @code{cross} also allows you to create image files for machines with
6216: different data sizes and data formats than the one used for generating
6217: the image file. You can also use it to create an application image that
6218: does not contain a Forth compiler. These features are bought with
6219: restrictions and inconveniences in programming. E.g., addresses have to
6220: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
6221: order to make the code relocatable.
6222:
6223:
6224: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
6225: @section Stack and Dictionary Sizes
6226: @cindex image file, stack and dictionary sizes
6227: @cindex dictionary size default
6228: @cindex stack size default
6229:
6230: If you invoke Gforth with a command line flag for the size
6231: (@pxref{Invoking Gforth}), the size you specify is stored in the
6232: dictionary. If you save the dictionary with @code{savesystem} or create
6233: an image with @file{gforthmi}, this size will become the default
6234: for the resulting image file. E.g., the following will create a
6235: fully relocatable version of gforth.fi with a 1MB dictionary:
6236:
6237: @example
6238: gforthmi gforth.fi -m 1M
6239: @end example
6240:
6241: In other words, if you want to set the default size for the dictionary
6242: and the stacks of an image, just invoke @file{gforthmi} with the
6243: appropriate options when creating the image.
6244:
6245: @cindex stack size, cache-friendly
6246: Note: For cache-friendly behaviour (i.e., good performance), you should
6247: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
6248: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
6249: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
6250:
6251: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
6252: @section Running Image Files
6253: @cindex running image files
6254: @cindex invoking image files
6255: @cindex image file invocation
6256:
6257: @cindex -i, invoke image file
6258: @cindex --image file, invoke image file
6259: You can invoke Gforth with an image file @var{image} instead of the
6260: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
6261: @example
6262: gforth -i @var{image}
6263: @end example
6264:
6265: @cindex executable image file
6266: @cindex image files, executable
6267: If your operating system supports starting scripts with a line of the
6268: form @code{#! ...}, you just have to type the image file name to start
6269: Gforth with this image file (note that the file extension @code{.fi} is
6270: just a convention). I.e., to run Gforth with the image file @var{image},
6271: you can just type @var{image} instead of @code{gforth -i @var{image}}.
6272:
6273: doc-#!
6274:
6275: @node Modifying the Startup Sequence, , Running Image Files, Image Files
6276: @section Modifying the Startup Sequence
6277: @cindex startup sequence for image file
6278: @cindex image file initialization sequence
6279: @cindex initialization sequence of image file
6280:
6281: You can add your own initialization to the startup sequence through the
6282: deferred word
6283:
6284: doc-'cold
6285:
6286: @code{'cold} is invoked just before the image-specific command line
6287: processing (by default, loading files and evaluating (@code{-e}) strings)
6288: starts.
6289:
6290: A sequence for adding your initialization usually looks like this:
6291:
6292: @example
6293: :noname
6294: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
6295: ... \ your stuff
6296: ; IS 'cold
6297: @end example
6298:
6299: @cindex turnkey image files
6300: @cindex image files, turnkey applications
6301: You can make a turnkey image by letting @code{'cold} execute a word
6302: (your turnkey application) that never returns; instead, it exits Gforth
6303: via @code{bye} or @code{throw}.
6304:
6305: @cindex command-line arguments, access
6306: @cindex arguments on the command line, access
6307: You can access the (image-specific) command-line arguments through the
6308: variables @code{argc} and @code{argv}. @code{arg} provides conventient
6309: access to @code{argv}.
6310:
6311: doc-argc
6312: doc-argv
6313: doc-arg
6314:
6315: If @code{'cold} exits normally, Gforth processes the command-line
6316: arguments as files to be loaded and strings to be evaluated. Therefore,
6317: @code{'cold} should remove the arguments it has used in this case.
6318:
6319: @c ******************************************************************
6320: @node Engine, Bugs, Image Files, Top
6321: @chapter Engine
6322: @cindex engine
6323: @cindex virtual machine
6324:
6325: Reading this section is not necessary for programming with Gforth. It
6326: may be helpful for finding your way in the Gforth sources.
6327:
6328: The ideas in this section have also been published in the papers
6329: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
6330: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
6331: Ertl, presented at EuroForth '93; the latter is available at
6332: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
6333:
6334: @menu
6335: * Portability::
6336: * Threading::
6337: * Primitives::
6338: * Performance::
6339: @end menu
6340:
6341: @node Portability, Threading, Engine, Engine
6342: @section Portability
6343: @cindex engine portability
6344:
6345: One of the main goals of the effort is availability across a wide range
6346: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
6347: this goal by manually coding the engine in assembly language for several
6348: then-popular processors. This approach is very labor-intensive and the
6349: results are short-lived due to progress in computer architecture.
6350:
6351: @cindex C, using C for the engine
6352: Others have avoided this problem by coding in C, e.g., Mitch Bradley
6353: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
6354: particularly popular for UNIX-based Forths due to the large variety of
6355: architectures of UNIX machines. Unfortunately an implementation in C
6356: does not mix well with the goals of efficiency and with using
6357: traditional techniques: Indirect or direct threading cannot be expressed
6358: in C, and switch threading, the fastest technique available in C, is
6359: significantly slower. Another problem with C is that it is very
6360: cumbersome to express double integer arithmetic.
6361:
6362: @cindex GNU C for the engine
6363: @cindex long long
6364: Fortunately, there is a portable language that does not have these
6365: limitations: GNU C, the version of C processed by the GNU C compiler
6366: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
6367: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
6368: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
6369: threading possible, its @code{long long} type (@pxref{Long Long, ,
6370: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
6371: double numbers@footnote{Unfortunately, long longs are not implemented
6372: properly on all machines (e.g., on alpha-osf1, long longs are only 64
6373: bits, the same size as longs (and pointers), but they should be twice as
1.4 anton 6374: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
1.1 anton 6375: C Manual}). So, we had to implement doubles in C after all. Still, on
6376: most machines we can use long longs and achieve better performance than
6377: with the emulation package.}. GNU C is available for free on all
6378: important (and many unimportant) UNIX machines, VMS, 80386s running
6379: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
6380: on all these machines.
6381:
6382: Writing in a portable language has the reputation of producing code that
6383: is slower than assembly. For our Forth engine we repeatedly looked at
6384: the code produced by the compiler and eliminated most compiler-induced
6385: inefficiencies by appropriate changes in the source code.
6386:
6387: @cindex explicit register declarations
6388: @cindex --enable-force-reg, configuration flag
6389: @cindex -DFORCE_REG
6390: However, register allocation cannot be portably influenced by the
6391: programmer, leading to some inefficiencies on register-starved
6392: machines. We use explicit register declarations (@pxref{Explicit Reg
6393: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
6394: improve the speed on some machines. They are turned on by using the
6395: configuration flag @code{--enable-force-reg} (@code{gcc} switch
6396: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
6397: machine, but also on the compiler version: On some machines some
6398: compiler versions produce incorrect code when certain explicit register
6399: declarations are used. So by default @code{-DFORCE_REG} is not used.
6400:
6401: @node Threading, Primitives, Portability, Engine
6402: @section Threading
6403: @cindex inner interpreter implementation
6404: @cindex threaded code implementation
6405:
6406: @cindex labels as values
6407: GNU C's labels as values extension (available since @code{gcc-2.0},
6408: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
6409: makes it possible to take the address of @var{label} by writing
6410: @code{&&@var{label}}. This address can then be used in a statement like
6411: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
6412: @code{goto x}.
6413:
6414: @cindex NEXT, indirect threaded
6415: @cindex indirect threaded inner interpreter
6416: @cindex inner interpreter, indirect threaded
6417: With this feature an indirect threaded NEXT looks like:
6418: @example
6419: cfa = *ip++;
6420: ca = *cfa;
6421: goto *ca;
6422: @end example
6423: @cindex instruction pointer
6424: For those unfamiliar with the names: @code{ip} is the Forth instruction
6425: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
6426: execution token and points to the code field of the next word to be
6427: executed; The @code{ca} (code address) fetched from there points to some
6428: executable code, e.g., a primitive or the colon definition handler
6429: @code{docol}.
6430:
6431: @cindex NEXT, direct threaded
6432: @cindex direct threaded inner interpreter
6433: @cindex inner interpreter, direct threaded
6434: Direct threading is even simpler:
6435: @example
6436: ca = *ip++;
6437: goto *ca;
6438: @end example
6439:
6440: Of course we have packaged the whole thing neatly in macros called
6441: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
6442:
6443: @menu
6444: * Scheduling::
6445: * Direct or Indirect Threaded?::
6446: * DOES>::
6447: @end menu
6448:
6449: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
6450: @subsection Scheduling
6451: @cindex inner interpreter optimization
6452:
6453: There is a little complication: Pipelined and superscalar processors,
6454: i.e., RISC and some modern CISC machines can process independent
6455: instructions while waiting for the results of an instruction. The
6456: compiler usually reorders (schedules) the instructions in a way that
6457: achieves good usage of these delay slots. However, on our first tries
6458: the compiler did not do well on scheduling primitives. E.g., for
6459: @code{+} implemented as
6460: @example
6461: n=sp[0]+sp[1];
6462: sp++;
6463: sp[0]=n;
6464: NEXT;
6465: @end example
6466: the NEXT comes strictly after the other code, i.e., there is nearly no
6467: scheduling. After a little thought the problem becomes clear: The
6468: compiler cannot know that sp and ip point to different addresses (and
6469: the version of @code{gcc} we used would not know it even if it was
6470: possible), so it could not move the load of the cfa above the store to
6471: the TOS. Indeed the pointers could be the same, if code on or very near
6472: the top of stack were executed. In the interest of speed we chose to
6473: forbid this probably unused ``feature'' and helped the compiler in
6474: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
6475: the goto part (@code{NEXT_P2}). @code{+} now looks like:
6476: @example
6477: n=sp[0]+sp[1];
6478: sp++;
6479: NEXT_P1;
6480: sp[0]=n;
6481: NEXT_P2;
6482: @end example
6483: This can be scheduled optimally by the compiler.
6484:
6485: This division can be turned off with the switch @code{-DCISC_NEXT}. This
6486: switch is on by default on machines that do not profit from scheduling
6487: (e.g., the 80386), in order to preserve registers.
6488:
6489: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
6490: @subsection Direct or Indirect Threaded?
6491: @cindex threading, direct or indirect?
6492:
6493: @cindex -DDIRECT_THREADED
6494: Both! After packaging the nasty details in macro definitions we
6495: realized that we could switch between direct and indirect threading by
6496: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
6497: defining a few machine-specific macros for the direct-threading case.
6498: On the Forth level we also offer access words that hide the
6499: differences between the threading methods (@pxref{Threading Words}).
6500:
6501: Indirect threading is implemented completely machine-independently.
6502: Direct threading needs routines for creating jumps to the executable
6503: code (e.g. to docol or dodoes). These routines are inherently
6504: machine-dependent, but they do not amount to many source lines. I.e.,
6505: even porting direct threading to a new machine is a small effort.
6506:
6507: @cindex --enable-indirect-threaded, configuration flag
6508: @cindex --enable-direct-threaded, configuration flag
6509: The default threading method is machine-dependent. You can enforce a
6510: specific threading method when building Gforth with the configuration
6511: flag @code{--enable-direct-threaded} or
6512: @code{--enable-indirect-threaded}. Note that direct threading is not
6513: supported on all machines.
6514:
6515: @node DOES>, , Direct or Indirect Threaded?, Threading
6516: @subsection DOES>
6517: @cindex @code{DOES>} implementation
6518:
6519: @cindex dodoes routine
6520: @cindex DOES-code
6521: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
6522: the chunk of code executed by every word defined by a
6523: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
6524: the Forth code to be executed, i.e. the code after the
6525: @code{DOES>} (the DOES-code)? There are two solutions:
6526:
6527: In fig-Forth the code field points directly to the dodoes and the
6528: DOES-code address is stored in the cell after the code address (i.e. at
6529: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
6530: the Forth-79 and all later standards, because in fig-Forth this address
6531: lies in the body (which is illegal in these standards). However, by
6532: making the code field larger for all words this solution becomes legal
6533: again. We use this approach for the indirect threaded version and for
6534: direct threading on some machines. Leaving a cell unused in most words
6535: is a bit wasteful, but on the machines we are targeting this is hardly a
6536: problem. The other reason for having a code field size of two cells is
6537: to avoid having different image files for direct and indirect threaded
6538: systems (direct threaded systems require two-cell code fields on many
6539: machines).
6540:
6541: @cindex DOES-handler
6542: The other approach is that the code field points or jumps to the cell
6543: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
6544: this address (the DOES-handler). @code{dodoes} can then get the
6545: DOES-code address by computing the code address, i.e., the address of
6546: the jump to dodoes, and add the length of that jump field. A variant of
6547: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
6548: return address (which can be found in the return register on RISCs) is
6549: the DOES-code address. Since the two cells available in the code field
6550: are used up by the jump to the code address in direct threading on many
6551: architectures, we use this approach for direct threading on these
6552: architectures. We did not want to add another cell to the code field.
6553:
6554: @node Primitives, Performance, Threading, Engine
6555: @section Primitives
6556: @cindex primitives, implementation
6557: @cindex virtual machine instructions, implementation
6558:
6559: @menu
6560: * Automatic Generation::
6561: * TOS Optimization::
6562: * Produced code::
6563: @end menu
6564:
6565: @node Automatic Generation, TOS Optimization, Primitives, Primitives
6566: @subsection Automatic Generation
6567: @cindex primitives, automatic generation
6568:
6569: @cindex @file{prims2x.fs}
6570: Since the primitives are implemented in a portable language, there is no
6571: longer any need to minimize the number of primitives. On the contrary,
6572: having many primitives has an advantage: speed. In order to reduce the
6573: number of errors in primitives and to make programming them easier, we
6574: provide a tool, the primitive generator (@file{prims2x.fs}), that
6575: automatically generates most (and sometimes all) of the C code for a
6576: primitive from the stack effect notation. The source for a primitive
6577: has the following form:
6578:
6579: @cindex primitive source format
6580: @format
6581: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
6582: [@code{""}@var{glossary entry}@code{""}]
6583: @var{C code}
6584: [@code{:}
6585: @var{Forth code}]
6586: @end format
6587:
6588: The items in brackets are optional. The category and glossary fields
6589: are there for generating the documentation, the Forth code is there
6590: for manual implementations on machines without GNU C. E.g., the source
6591: for the primitive @code{+} is:
6592: @example
6593: + n1 n2 -- n core plus
6594: n = n1+n2;
6595: @end example
6596:
6597: This looks like a specification, but in fact @code{n = n1+n2} is C
6598: code. Our primitive generation tool extracts a lot of information from
6599: the stack effect notations@footnote{We use a one-stack notation, even
6600: though we have separate data and floating-point stacks; The separate
6601: notation can be generated easily from the unified notation.}: The number
6602: of items popped from and pushed on the stack, their type, and by what
6603: name they are referred to in the C code. It then generates a C code
6604: prelude and postlude for each primitive. The final C code for @code{+}
6605: looks like this:
6606:
6607: @example
6608: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
6609: /* */ /* documentation */
6610: @{
6611: DEF_CA /* definition of variable ca (indirect threading) */
6612: Cell n1; /* definitions of variables */
6613: Cell n2;
6614: Cell n;
6615: n1 = (Cell) sp[1]; /* input */
6616: n2 = (Cell) TOS;
6617: sp += 1; /* stack adjustment */
6618: NAME("+") /* debugging output (with -DDEBUG) */
6619: @{
6620: n = n1+n2; /* C code taken from the source */
6621: @}
6622: NEXT_P1; /* NEXT part 1 */
6623: TOS = (Cell)n; /* output */
6624: NEXT_P2; /* NEXT part 2 */
6625: @}
6626: @end example
6627:
6628: This looks long and inefficient, but the GNU C compiler optimizes quite
6629: well and produces optimal code for @code{+} on, e.g., the R3000 and the
6630: HP RISC machines: Defining the @code{n}s does not produce any code, and
6631: using them as intermediate storage also adds no cost.
6632:
6633: There are also other optimizations, that are not illustrated by this
6634: example: Assignments between simple variables are usually for free (copy
6635: propagation). If one of the stack items is not used by the primitive
6636: (e.g. in @code{drop}), the compiler eliminates the load from the stack
6637: (dead code elimination). On the other hand, there are some things that
6638: the compiler does not do, therefore they are performed by
6639: @file{prims2x.fs}: The compiler does not optimize code away that stores
6640: a stack item to the place where it just came from (e.g., @code{over}).
6641:
6642: While programming a primitive is usually easy, there are a few cases
6643: where the programmer has to take the actions of the generator into
6644: account, most notably @code{?dup}, but also words that do not (always)
6645: fall through to NEXT.
6646:
6647: @node TOS Optimization, Produced code, Automatic Generation, Primitives
6648: @subsection TOS Optimization
6649: @cindex TOS optimization for primitives
6650: @cindex primitives, keeping the TOS in a register
6651:
6652: An important optimization for stack machine emulators, e.g., Forth
6653: engines, is keeping one or more of the top stack items in
6654: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
6655: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
6656: @itemize @bullet
6657: @item
6658: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
6659: due to fewer loads from and stores to the stack.
6660: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
6661: @var{y<n}, due to additional moves between registers.
6662: @end itemize
6663:
6664: @cindex -DUSE_TOS
6665: @cindex -DUSE_NO_TOS
6666: In particular, keeping one item in a register is never a disadvantage,
6667: if there are enough registers. Keeping two items in registers is a
6668: disadvantage for frequent words like @code{?branch}, constants,
6669: variables, literals and @code{i}. Therefore our generator only produces
6670: code that keeps zero or one items in registers. The generated C code
6671: covers both cases; the selection between these alternatives is made at
6672: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
6673: code for @code{+} is just a simple variable name in the one-item case,
6674: otherwise it is a macro that expands into @code{sp[0]}. Note that the
6675: GNU C compiler tries to keep simple variables like @code{TOS} in
6676: registers, and it usually succeeds, if there are enough registers.
6677:
6678: @cindex -DUSE_FTOS
6679: @cindex -DUSE_NO_FTOS
6680: The primitive generator performs the TOS optimization for the
6681: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
6682: operations the benefit of this optimization is even larger:
6683: floating-point operations take quite long on most processors, but can be
6684: performed in parallel with other operations as long as their results are
6685: not used. If the FP-TOS is kept in a register, this works. If
6686: it is kept on the stack, i.e., in memory, the store into memory has to
6687: wait for the result of the floating-point operation, lengthening the
6688: execution time of the primitive considerably.
6689:
6690: The TOS optimization makes the automatic generation of primitives a
6691: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
6692: @code{TOS} is not sufficient. There are some special cases to
6693: consider:
6694: @itemize @bullet
6695: @item In the case of @code{dup ( w -- w w )} the generator must not
6696: eliminate the store to the original location of the item on the stack,
6697: if the TOS optimization is turned on.
6698: @item Primitives with stack effects of the form @code{--}
6699: @var{out1}...@var{outy} must store the TOS to the stack at the start.
6700: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
6701: must load the TOS from the stack at the end. But for the null stack
6702: effect @code{--} no stores or loads should be generated.
6703: @end itemize
6704:
6705: @node Produced code, , TOS Optimization, Primitives
6706: @subsection Produced code
6707: @cindex primitives, assembly code listing
6708:
6709: @cindex @file{engine.s}
6710: To see what assembly code is produced for the primitives on your machine
6711: with your compiler and your flag settings, type @code{make engine.s} and
6712: look at the resulting file @file{engine.s}.
6713:
6714: @node Performance, , Primitives, Engine
6715: @section Performance
6716: @cindex performance of some Forth interpreters
6717: @cindex engine performance
6718: @cindex benchmarking Forth systems
6719: @cindex Gforth performance
6720:
6721: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
6722: impossible to write a significantly faster engine.
6723:
6724: On register-starved machines like the 386 architecture processors
6725: improvements are possible, because @code{gcc} does not utilize the
6726: registers as well as a human, even with explicit register declarations;
6727: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
6728: and hand-tuned it for the 486; this system is 1.19 times faster on the
6729: Sieve benchmark on a 486DX2/66 than Gforth compiled with
6730: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
6731:
6732: @cindex Win32Forth performance
6733: @cindex NT Forth performance
6734: @cindex eforth performance
6735: @cindex ThisForth performance
6736: @cindex PFE performance
6737: @cindex TILE performance
6738: However, this potential advantage of assembly language implementations
6739: is not necessarily realized in complete Forth systems: We compared
6740: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
6741: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
6742: 1994) and Eforth (with and without peephole (aka pinhole) optimization
6743: of the threaded code); all these systems were written in assembly
6744: language. We also compared Gforth with three systems written in C:
6745: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
6746: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
6747: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
6748: -fomit-frame-pointer; ThisForth employs peephole optimization of the
6749: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
6750: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
6751: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
6752: 486DX2/66 with similar memory performance under Windows NT. Marcel
6753: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
6754: added the peephole optimizer, ran the benchmarks and reported the
6755: results.
6756:
6757: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
6758: matrix multiplication come from the Stanford integer benchmarks and have
6759: been translated into Forth by Martin Fraeman; we used the versions
6760: included in the TILE Forth package, but with bigger data set sizes; and
6761: a recursive Fibonacci number computation for benchmarking calling
6762: performance. The following table shows the time taken for the benchmarks
6763: scaled by the time taken by Gforth (in other words, it shows the speedup
6764: factor that Gforth achieved over the other systems).
6765:
6766: @example
6767: relative Win32- NT eforth This-
6768: time Gforth Forth Forth eforth +opt PFE Forth TILE
6769: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
6770: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
6771: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
6772: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
6773: @end example
6774:
6775: You may find the good performance of Gforth compared with the systems
6776: written in assembly language quite surprising. One important reason for
6777: the disappointing performance of these systems is probably that they are
6778: not written optimally for the 486 (e.g., they use the @code{lods}
6779: instruction). In addition, Win32Forth uses a comfortable, but costly
6780: method for relocating the Forth image: like @code{cforth}, it computes
6781: the actual addresses at run time, resulting in two address computations
6782: per NEXT (@pxref{Image File Background}).
6783:
6784: Only Eforth with the peephole optimizer performs comparable to
6785: Gforth. The speedups achieved with peephole optimization of threaded
6786: code are quite remarkable. Adding a peephole optimizer to Gforth should
6787: cause similar speedups.
6788:
6789: The speedup of Gforth over PFE, ThisForth and TILE can be easily
6790: explained with the self-imposed restriction of the latter systems to
6791: standard C, which makes efficient threading impossible (however, the
1.4 anton 6792: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
1.1 anton 6793: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
6794: Moreover, current C compilers have a hard time optimizing other aspects
6795: of the ThisForth and the TILE source.
6796:
6797: Note that the performance of Gforth on 386 architecture processors
6798: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
6799: failed to allocate any of the virtual machine registers into real
6800: machine registers by itself and would not work correctly with explicit
6801: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
6802: running the Sieve) than the one measured above.
6803:
6804: Note also that there have been several releases of Win32Forth since the
6805: release presented here, so the results presented here may have little
6806: predictive value for the performance of Win32Forth today.
6807:
6808: @cindex @file{Benchres}
6809: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
6810: Maierhofer (presented at EuroForth '95), an indirect threaded version of
6811: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
6812: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
6813: threaded version used here. The paper available at
6814: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
6815: it also contains numbers for some native code systems. You can find a
6816: newer version of these measurements at
6817: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
6818: find numbers for Gforth on various machines in @file{Benchres}.
6819:
6820: @node Bugs, Origin, Engine, Top
6821: @chapter Bugs
6822: @cindex bug reporting
6823:
6824: Known bugs are described in the file BUGS in the Gforth distribution.
6825:
6826: If you find a bug, please send a bug report to
6827: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should
6828: describe the Gforth version used (it is announced at the start of an
6829: interactive Gforth session), the machine and operating system (on Unix
6830: systems you can use @code{uname -a} to produce this information), the
6831: installation options (send the @file{config.status} file), and a
6832: complete list of changes you (or your installer) have made to the Gforth
6833: sources (if any); it should contain a program (or a sequence of keyboard
6834: commands) that reproduces the bug and a description of what you think
6835: constitutes the buggy behaviour.
6836:
6837: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
6838: to Report Bugs, gcc.info, GNU C Manual}.
6839:
6840:
6841: @node Origin, Word Index, Bugs, Top
6842: @chapter Authors and Ancestors of Gforth
6843:
6844: @section Authors and Contributors
6845: @cindex authors of Gforth
6846: @cindex contributors to Gforth
6847:
6848: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
6849: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
6850: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
6851: with their continuous feedback. Lennart Benshop contributed
6852: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
6853: support for calling C libraries. Helpful comments also came from Paul
6854: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
6855: Wavrik, Barrie Stott and Marc de Groot.
6856:
6857: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
6858: and autoconf, among others), and to the creators of the Internet: Gforth
6859: was developed across the Internet, and its authors have not met
6860: physically yet.
6861:
6862: @section Pedigree
6863: @cindex Pedigree of Gforth
6864:
6865: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
6866: Dirk Zoller) will cross-fertilize each other. Of course, a significant
6867: part of the design of Gforth was prescribed by ANS Forth.
6868:
6869: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
6870: 32 bit native code version of VolksForth for the Atari ST, written
6871: mostly by Dietrich Weineck.
6872:
6873: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
6874: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
6875: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
6876:
6877: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
6878: Forth-83 standard. !! Pedigree? When?
6879:
6880: A team led by Bill Ragsdale implemented fig-Forth on many processors in
6881: 1979. Robert Selzer and Bill Ragsdale developed the original
6882: implementation of fig-Forth for the 6502 based on microForth.
6883:
6884: The principal architect of microForth was Dean Sanderson. microForth was
6885: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
6886: the 1802, and subsequently implemented on the 8080, the 6800 and the
6887: Z80.
6888:
6889: All earlier Forth systems were custom-made, usually by Charles Moore,
6890: who discovered (as he puts it) Forth during the late 60s. The first full
6891: Forth existed in 1971.
6892:
6893: A part of the information in this section comes from @cite{The Evolution
6894: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
6895: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
6896: Notices 28(3), 1993. You can find more historical and genealogical
6897: information about Forth there.
6898:
6899: @node Word Index, Concept Index, Origin, Top
6900: @unnumbered Word Index
6901:
6902: This index is as incomplete as the manual. Each word is listed with
6903: stack effect and wordset.
6904:
6905: @printindex fn
6906:
6907: @node Concept Index, , Word Index, Top
6908: @unnumbered Concept and Word Index
6909:
6910: This index is as incomplete as the manual. Not all entries listed are
6911: present verbatim in the text. Only the names are listed for the words
6912: here.
6913:
6914: @printindex cp
6915:
6916: @contents
6917: @bye
6918:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>