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: @comment @setchapternewpage odd
7: @comment %**end of header (This is for running Texinfo on a region.)
8:
9: @ifinfo
10: This file documents Gforth 0.2
11:
12: Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
13:
14: Permission is granted to make and distribute verbatim copies of
15: this manual provided the copyright notice and this permission notice
16: are preserved on all copies.
17:
18: @ignore
19: Permission is granted to process this file through TeX and print the
20: results, provided the printed document carries a copying permission
21: notice identical to this one except for the removal of this paragraph
22: (this paragraph not being relevant to the printed manual).
23:
24: @end ignore
25: Permission is granted to copy and distribute modified versions of this
26: manual under the conditions for verbatim copying, provided also that the
27: sections entitled "Distribution" and "General Public License" are
28: included exactly as in the original, and provided that the entire
29: resulting derived work is distributed under the terms of a permission
30: notice identical to this one.
31:
32: Permission is granted to copy and distribute translations of this manual
33: into another language, under the above conditions for modified versions,
34: except that the sections entitled "Distribution" and "General Public
35: License" may be included in a translation approved by the author instead
36: of in the original English.
37: @end ifinfo
38:
39: @finalout
40: @titlepage
41: @sp 10
42: @center @titlefont{Gforth Manual}
43: @sp 2
44: @center for version 0.2
45: @sp 2
46: @center Anton Ertl
47: @center Bernd Paysan
48: @sp 3
49: @center This manual is under construction
50:
51: @comment The following two commands start the copyright page.
52: @page
53: @vskip 0pt plus 1filll
54: Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
55:
56: @comment !! Published by ... or You can get a copy of this manual ...
57:
58: Permission is granted to make and distribute verbatim copies of
59: this manual provided the copyright notice and this permission notice
60: are preserved on all copies.
61:
62: Permission is granted to copy and distribute modified versions of this
63: manual under the conditions for verbatim copying, provided also that the
64: sections entitled "Distribution" and "General Public License" are
65: included exactly as in the original, and provided that the entire
66: resulting derived work is distributed under the terms of a permission
67: notice identical to this one.
68:
69: Permission is granted to copy and distribute translations of this manual
70: into another language, under the above conditions for modified versions,
71: except that the sections entitled "Distribution" and "General Public
72: License" may be included in a translation approved by the author instead
73: of in the original English.
74: @end titlepage
75:
76:
77: @node Top, License, (dir), (dir)
78: @ifinfo
79: Gforth is a free implementation of ANS Forth available on many
80: personal machines. This manual corresponds to version 0.2.
81: @end ifinfo
82:
83: @menu
84: * License::
85: * Goals:: About the Gforth Project
86: * Other Books:: Things you might want to read
87: * Invocation:: Starting Gforth
88: * Words:: Forth words available in Gforth
89: * ANS conformance:: Implementation-defined options etc.
90: * Model:: The abstract machine of Gforth
91: * Integrating Gforth:: Forth as scripting language for applications.
92: * Emacs and Gforth:: The Gforth Mode
93: * Internals:: Implementation details
94: * Bugs:: How to report them
95: * Origin:: Authors and ancestors of Gforth
96: * Word Index:: An item for each Forth word
97: * Node Index:: An item for each node
98: @end menu
99:
100: @node License, Goals, Top, Top
101: @unnumbered GNU GENERAL PUBLIC LICENSE
102: @center Version 2, June 1991
103:
104: @display
105: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
106: 675 Mass Ave, Cambridge, MA 02139, USA
107:
108: Everyone is permitted to copy and distribute verbatim copies
109: of this license document, but changing it is not allowed.
110: @end display
111:
112: @unnumberedsec Preamble
113:
114: The licenses for most software are designed to take away your
115: freedom to share and change it. By contrast, the GNU General Public
116: License is intended to guarantee your freedom to share and change free
117: software---to make sure the software is free for all its users. This
118: General Public License applies to most of the Free Software
119: Foundation's software and to any other program whose authors commit to
120: using it. (Some other Free Software Foundation software is covered by
121: the GNU Library General Public License instead.) You can apply it to
122: your programs, too.
123:
124: When we speak of free software, we are referring to freedom, not
125: price. Our General Public Licenses are designed to make sure that you
126: have the freedom to distribute copies of free software (and charge for
127: this service if you wish), that you receive source code or can get it
128: if you want it, that you can change the software or use pieces of it
129: in new free programs; and that you know you can do these things.
130:
131: To protect your rights, we need to make restrictions that forbid
132: anyone to deny you these rights or to ask you to surrender the rights.
133: These restrictions translate to certain responsibilities for you if you
134: distribute copies of the software, or if you modify it.
135:
136: For example, if you distribute copies of such a program, whether
137: gratis or for a fee, you must give the recipients all the rights that
138: you have. You must make sure that they, too, receive or can get the
139: source code. And you must show them these terms so they know their
140: rights.
141:
142: We protect your rights with two steps: (1) copyright the software, and
143: (2) offer you this license which gives you legal permission to copy,
144: distribute and/or modify the software.
145:
146: Also, for each author's protection and ours, we want to make certain
147: that everyone understands that there is no warranty for this free
148: software. If the software is modified by someone else and passed on, we
149: want its recipients to know that what they have is not the original, so
150: that any problems introduced by others will not reflect on the original
151: authors' reputations.
152:
153: Finally, any free program is threatened constantly by software
154: patents. We wish to avoid the danger that redistributors of a free
155: program will individually obtain patent licenses, in effect making the
156: program proprietary. To prevent this, we have made it clear that any
157: patent must be licensed for everyone's free use or not licensed at all.
158:
159: The precise terms and conditions for copying, distribution and
160: modification follow.
161:
162: @iftex
163: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
164: @end iftex
165: @ifinfo
166: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
167: @end ifinfo
168:
169: @enumerate 0
170: @item
171: This License applies to any program or other work which contains
172: a notice placed by the copyright holder saying it may be distributed
173: under the terms of this General Public License. The ``Program'', below,
174: refers to any such program or work, and a ``work based on the Program''
175: means either the Program or any derivative work under copyright law:
176: that is to say, a work containing the Program or a portion of it,
177: either verbatim or with modifications and/or translated into another
178: language. (Hereinafter, translation is included without limitation in
179: the term ``modification''.) Each licensee is addressed as ``you''.
180:
181: Activities other than copying, distribution and modification are not
182: covered by this License; they are outside its scope. The act of
183: running the Program is not restricted, and the output from the Program
184: is covered only if its contents constitute a work based on the
185: Program (independent of having been made by running the Program).
186: Whether that is true depends on what the Program does.
187:
188: @item
189: You may copy and distribute verbatim copies of the Program's
190: source code as you receive it, in any medium, provided that you
191: conspicuously and appropriately publish on each copy an appropriate
192: copyright notice and disclaimer of warranty; keep intact all the
193: notices that refer to this License and to the absence of any warranty;
194: and give any other recipients of the Program a copy of this License
195: along with the Program.
196:
197: You may charge a fee for the physical act of transferring a copy, and
198: you may at your option offer warranty protection in exchange for a fee.
199:
200: @item
201: You may modify your copy or copies of the Program or any portion
202: of it, thus forming a work based on the Program, and copy and
203: distribute such modifications or work under the terms of Section 1
204: above, provided that you also meet all of these conditions:
205:
206: @enumerate a
207: @item
208: You must cause the modified files to carry prominent notices
209: stating that you changed the files and the date of any change.
210:
211: @item
212: You must cause any work that you distribute or publish, that in
213: whole or in part contains or is derived from the Program or any
214: part thereof, to be licensed as a whole at no charge to all third
215: parties under the terms of this License.
216:
217: @item
218: If the modified program normally reads commands interactively
219: when run, you must cause it, when started running for such
220: interactive use in the most ordinary way, to print or display an
221: announcement including an appropriate copyright notice and a
222: notice that there is no warranty (or else, saying that you provide
223: a warranty) and that users may redistribute the program under
224: these conditions, and telling the user how to view a copy of this
225: License. (Exception: if the Program itself is interactive but
226: does not normally print such an announcement, your work based on
227: the Program is not required to print an announcement.)
228: @end enumerate
229:
230: These requirements apply to the modified work as a whole. If
231: identifiable sections of that work are not derived from the Program,
232: and can be reasonably considered independent and separate works in
233: themselves, then this License, and its terms, do not apply to those
234: sections when you distribute them as separate works. But when you
235: distribute the same sections as part of a whole which is a work based
236: on the Program, the distribution of the whole must be on the terms of
237: this License, whose permissions for other licensees extend to the
238: entire whole, and thus to each and every part regardless of who wrote it.
239:
240: Thus, it is not the intent of this section to claim rights or contest
241: your rights to work written entirely by you; rather, the intent is to
242: exercise the right to control the distribution of derivative or
243: collective works based on the Program.
244:
245: In addition, mere aggregation of another work not based on the Program
246: with the Program (or with a work based on the Program) on a volume of
247: a storage or distribution medium does not bring the other work under
248: the scope of this License.
249:
250: @item
251: You may copy and distribute the Program (or a work based on it,
252: under Section 2) in object code or executable form under the terms of
253: Sections 1 and 2 above provided that you also do one of the following:
254:
255: @enumerate a
256: @item
257: Accompany it with the complete corresponding machine-readable
258: source code, which must be distributed under the terms of Sections
259: 1 and 2 above on a medium customarily used for software interchange; or,
260:
261: @item
262: Accompany it with a written offer, valid for at least three
263: years, to give any third party, for a charge no more than your
264: cost of physically performing source distribution, a complete
265: machine-readable copy of the corresponding source code, to be
266: distributed under the terms of Sections 1 and 2 above on a medium
267: customarily used for software interchange; or,
268:
269: @item
270: Accompany it with the information you received as to the offer
271: to distribute corresponding source code. (This alternative is
272: allowed only for noncommercial distribution and only if you
273: received the program in object code or executable form with such
274: an offer, in accord with Subsection b above.)
275: @end enumerate
276:
277: The source code for a work means the preferred form of the work for
278: making modifications to it. For an executable work, complete source
279: code means all the source code for all modules it contains, plus any
280: associated interface definition files, plus the scripts used to
281: control compilation and installation of the executable. However, as a
282: special exception, the source code distributed need not include
283: anything that is normally distributed (in either source or binary
284: form) with the major components (compiler, kernel, and so on) of the
285: operating system on which the executable runs, unless that component
286: itself accompanies the executable.
287:
288: If distribution of executable or object code is made by offering
289: access to copy from a designated place, then offering equivalent
290: access to copy the source code from the same place counts as
291: distribution of the source code, even though third parties are not
292: compelled to copy the source along with the object code.
293:
294: @item
295: You may not copy, modify, sublicense, or distribute the Program
296: except as expressly provided under this License. Any attempt
297: otherwise to copy, modify, sublicense or distribute the Program is
298: void, and will automatically terminate your rights under this License.
299: However, parties who have received copies, or rights, from you under
300: this License will not have their licenses terminated so long as such
301: parties remain in full compliance.
302:
303: @item
304: You are not required to accept this License, since you have not
305: signed it. However, nothing else grants you permission to modify or
306: distribute the Program or its derivative works. These actions are
307: prohibited by law if you do not accept this License. Therefore, by
308: modifying or distributing the Program (or any work based on the
309: Program), you indicate your acceptance of this License to do so, and
310: all its terms and conditions for copying, distributing or modifying
311: the Program or works based on it.
312:
313: @item
314: Each time you redistribute the Program (or any work based on the
315: Program), the recipient automatically receives a license from the
316: original licensor to copy, distribute or modify the Program subject to
317: these terms and conditions. You may not impose any further
318: restrictions on the recipients' exercise of the rights granted herein.
319: You are not responsible for enforcing compliance by third parties to
320: this License.
321:
322: @item
323: If, as a consequence of a court judgment or allegation of patent
324: infringement or for any other reason (not limited to patent issues),
325: conditions are imposed on you (whether by court order, agreement or
326: otherwise) that contradict the conditions of this License, they do not
327: excuse you from the conditions of this License. If you cannot
328: distribute so as to satisfy simultaneously your obligations under this
329: License and any other pertinent obligations, then as a consequence you
330: may not distribute the Program at all. For example, if a patent
331: license would not permit royalty-free redistribution of the Program by
332: all those who receive copies directly or indirectly through you, then
333: the only way you could satisfy both it and this License would be to
334: refrain entirely from distribution of the Program.
335:
336: If any portion of this section is held invalid or unenforceable under
337: any particular circumstance, the balance of the section is intended to
338: apply and the section as a whole is intended to apply in other
339: circumstances.
340:
341: It is not the purpose of this section to induce you to infringe any
342: patents or other property right claims or to contest validity of any
343: such claims; this section has the sole purpose of protecting the
344: integrity of the free software distribution system, which is
345: implemented by public license practices. Many people have made
346: generous contributions to the wide range of software distributed
347: through that system in reliance on consistent application of that
348: system; it is up to the author/donor to decide if he or she is willing
349: to distribute software through any other system and a licensee cannot
350: impose that choice.
351:
352: This section is intended to make thoroughly clear what is believed to
353: be a consequence of the rest of this License.
354:
355: @item
356: If the distribution and/or use of the Program is restricted in
357: certain countries either by patents or by copyrighted interfaces, the
358: original copyright holder who places the Program under this License
359: may add an explicit geographical distribution limitation excluding
360: those countries, so that distribution is permitted only in or among
361: countries not thus excluded. In such case, this License incorporates
362: the limitation as if written in the body of this License.
363:
364: @item
365: The Free Software Foundation may publish revised and/or new versions
366: of the General Public License from time to time. Such new versions will
367: be similar in spirit to the present version, but may differ in detail to
368: address new problems or concerns.
369:
370: Each version is given a distinguishing version number. If the Program
371: specifies a version number of this License which applies to it and ``any
372: later version'', you have the option of following the terms and conditions
373: either of that version or of any later version published by the Free
374: Software Foundation. If the Program does not specify a version number of
375: this License, you may choose any version ever published by the Free Software
376: Foundation.
377:
378: @item
379: If you wish to incorporate parts of the Program into other free
380: programs whose distribution conditions are different, write to the author
381: to ask for permission. For software which is copyrighted by the Free
382: Software Foundation, write to the Free Software Foundation; we sometimes
383: make exceptions for this. Our decision will be guided by the two goals
384: of preserving the free status of all derivatives of our free software and
385: of promoting the sharing and reuse of software generally.
386:
387: @iftex
388: @heading NO WARRANTY
389: @end iftex
390: @ifinfo
391: @center NO WARRANTY
392: @end ifinfo
393:
394: @item
395: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
396: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
397: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
398: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
399: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
400: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
401: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
402: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
403: REPAIR OR CORRECTION.
404:
405: @item
406: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
407: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
408: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
409: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
410: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
411: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
412: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
413: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
414: POSSIBILITY OF SUCH DAMAGES.
415: @end enumerate
416:
417: @iftex
418: @heading END OF TERMS AND CONDITIONS
419: @end iftex
420: @ifinfo
421: @center END OF TERMS AND CONDITIONS
422: @end ifinfo
423:
424: @page
425: @unnumberedsec How to Apply These Terms to Your New Programs
426:
427: If you develop a new program, and you want it to be of the greatest
428: possible use to the public, the best way to achieve this is to make it
429: free software which everyone can redistribute and change under these terms.
430:
431: To do so, attach the following notices to the program. It is safest
432: to attach them to the start of each source file to most effectively
433: convey the exclusion of warranty; and each file should have at least
434: the ``copyright'' line and a pointer to where the full notice is found.
435:
436: @smallexample
437: @var{one line to give the program's name and a brief idea of what it does.}
438: Copyright (C) 19@var{yy} @var{name of author}
439:
440: This program is free software; you can redistribute it and/or modify
441: it under the terms of the GNU General Public License as published by
442: the Free Software Foundation; either version 2 of the License, or
443: (at your option) any later version.
444:
445: This program is distributed in the hope that it will be useful,
446: but WITHOUT ANY WARRANTY; without even the implied warranty of
447: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
448: GNU General Public License for more details.
449:
450: You should have received a copy of the GNU General Public License
451: along with this program; if not, write to the Free Software
452: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
453: @end smallexample
454:
455: Also add information on how to contact you by electronic and paper mail.
456:
457: If the program is interactive, make it output a short notice like this
458: when it starts in an interactive mode:
459:
460: @smallexample
461: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
462: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
463: type `show w'.
464: This is free software, and you are welcome to redistribute it
465: under certain conditions; type `show c' for details.
466: @end smallexample
467:
468: The hypothetical commands @samp{show w} and @samp{show c} should show
469: the appropriate parts of the General Public License. Of course, the
470: commands you use may be called something other than @samp{show w} and
471: @samp{show c}; they could even be mouse-clicks or menu items---whatever
472: suits your program.
473:
474: You should also get your employer (if you work as a programmer) or your
475: school, if any, to sign a ``copyright disclaimer'' for the program, if
476: necessary. Here is a sample; alter the names:
477:
478: @smallexample
479: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
480: `Gnomovision' (which makes passes at compilers) written by James Hacker.
481:
482: @var{signature of Ty Coon}, 1 April 1989
483: Ty Coon, President of Vice
484: @end smallexample
485:
486: This General Public License does not permit incorporating your program into
487: proprietary programs. If your program is a subroutine library, you may
488: consider it more useful to permit linking proprietary applications with the
489: library. If this is what you want to do, use the GNU Library General
490: Public License instead of this License.
491:
492: @iftex
493: @node Preface
494: @comment node-name, next, previous, up
495: @unnumbered Preface
496: @cindex Preface
497: This manual documents Gforth. The reader is expected to know
498: Forth. This manual is primarily a reference manual. @xref{Other Books}
499: for introductory material.
500: @end iftex
501:
502: @node Goals, Other Books, License, Top
503: @comment node-name, next, previous, up
504: @chapter Goals of Gforth
505: @cindex Goals
506: The goal of the Gforth Project is to develop a standard model for
507: ANSI Forth. This can be split into several subgoals:
508:
509: @itemize @bullet
510: @item
511: Gforth should conform to the ANSI Forth standard.
512: @item
513: It should be a model, i.e. it should define all the
514: implementation-dependent things.
515: @item
516: It should become standard, i.e. widely accepted and used. This goal
517: is the most difficult one.
518: @end itemize
519:
520: To achieve these goals Gforth should be
521: @itemize @bullet
522: @item
523: Similar to previous models (fig-Forth, F83)
524: @item
525: Powerful. It should provide for all the things that are considered
526: necessary today and even some that are not yet considered necessary.
527: @item
528: Efficient. It should not get the reputation of being exceptionally
529: slow.
530: @item
531: Free.
532: @item
533: Available on many machines/easy to port.
534: @end itemize
535:
536: Have we achieved these goals? Gforth conforms to the ANS Forth
537: standard. It may be considered a model, but we have not yet documented
538: which parts of the model are stable and which parts we are likely to
539: change. It certainly has not yet become a de facto standard. It has some
540: similarities and some differences to previous models. It has some
541: powerful features, but not yet everything that we envisioned. We
542: certainly have achieved our execution speed goals (@pxref{Performance}).
543: It is free and available on many machines.
544:
545: @node Other Books, Invocation, Goals, Top
546: @chapter Other books on ANS Forth
547:
548: As the standard is relatively new, there are not many books out yet. It
549: is not recommended to learn Forth by using Gforth and a book that is
550: not written for ANS Forth, as you will not know your mistakes from the
551: deviations of the book.
552:
553: There is, of course, the standard, the definite reference if you want to
554: write ANS Forth programs. It is available in printed form from the
555: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
556: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
557: can also get it from Global Engineering Documents (Tel.: USA (800)
558: 854-7179; Fax.: (303) 843-9880) for about $300.
559:
560: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
561: for publication is available electronically and for free in some MS Word
562: format, and it has been converted to HTML. Some pointers to these
563: versions can be found through
564: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
565:
566: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
567: introductory book based on a draft version of the standard. It does not
568: cover the whole standard. It also contains interesting background
569: information (Jack Woehr was in the ANS Forth Technical Committe). It is
570: not appropriate for complete newbies, but programmers experienced in
571: other languages should find it ok.
572:
573: @node Invocation, Words, Other Books, Top
574: @chapter Invocation
575:
576: You will usually just say @code{gforth}. In many other cases the default
577: Gforth image will be invoked like this:
578:
579: @example
580: gforth [files] [-e forth-code]
581: @end example
582:
583: executing the contents of the files and the Forth code in the order they
584: are given.
585:
586: In general, the command line looks like this:
587:
588: @example
589: gforth [initialization options] [image-specific options]
590: @end example
591:
592: The initialization options must come before the rest of the command
593: line. They are:
594:
595: @table @code
596: @item --image-file @var{file}
597: @item -i @var{file}
598: Loads the Forth image @var{file} instead of the default
599: @file{gforth.fi}.
600:
601: @item --path @var{path}
602: @item -p @var{path}
603: Uses @var{path} for searching the image file and Forth source code
604: files instead of the default in the environment variable
605: @code{GFORTHPATH} or the path specified at installation time (typically
606: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
607: list.
608:
609: @item --dictionary-size @var{size}
610: @item -m @var{size}
611: Allocate @var{size} space for the Forth dictionary space instead of
612: using the default specified in the image (typically 256K). The
613: @var{size} specification consists of an integer and a unit (e.g.,
614: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
615: size, in this case Cells), @code{k} (kilobytes), and @code{M}
616: (Megabytes). If no unit is specified, @code{e} is used.
617:
618: @item --data-stack-size @var{size}
619: @item -d @var{size}
620: Allocate @var{size} space for the data stack instead of using the
621: default specified in the image (typically 16K).
622:
623: @item --return-stack-size @var{size}
624: @item -r @var{size}
625: Allocate @var{size} space for the return stack instead of using the
626: default specified in the image (typically 16K).
627:
628: @item --fp-stack-size @var{size}
629: @item -f @var{size}
630: Allocate @var{size} space for the floating point stack instead of
631: using the default specified in the image (typically 16K). In this case
632: the unit specifier @code{e} refers to floating point numbers.
633:
634: @item --locals-stack-size @var{size}
635: @item -l @var{size}
636: Allocate @var{size} space for the locals stack instead of using the
637: default specified in the image (typically 16K).
638:
639: @end table
640:
641: As explained above, the image-specific command-line arguments for the
642: default image @file{gforth.fi} consist of a sequence of filenames and
643: @code{-e @var{forth-code}} options that are interpreted in the seqence
644: in which they are given. The @code{-e @var{forth-code}} or
645: @code{--evaluate @var{forth-code}} option evaluates the forth
646: code. This option takes only one argument; if you want to evaluate more
647: Forth words, you have to quote them or use several @code{-e}s. To exit
648: after processing the command line (instead of entering interactive mode)
649: append @code{-e bye} to the command line.
650:
651: If you have several versions of Gforth installed, @code{gforth} will
652: invoke the version that was installed last. @code{gforth-@var{version}}
653: invokes a specific version. You may want to use the option
654: @code{--path}, if your environment contains the variable
655: @code{GFORTHPATH}.
656:
657: Not yet implemented:
658: On startup the system first executes the system initialization file
659: (unless the option @code{--no-init-file} is given; note that the system
660: resulting from using this option may not be ANS Forth conformant). Then
661: the user initialization file @file{.gforth.fs} is executed, unless the
662: option @code{--no-rc} is given; this file is first searched in @file{.},
663: then in @file{~}, then in the normal path (see above).
664:
665: @node Words, ANS conformance, Invocation, Top
666: @chapter Forth Words
667:
668: @menu
669: * Notation::
670: * Arithmetic::
671: * Stack Manipulation::
672: * Memory access::
673: * Control Structures::
674: * Locals::
675: * Defining Words::
676: * Tokens for Words::
677: * Wordlists::
678: * Files::
679: * Blocks::
680: * Other I/O::
681: * Programming Tools::
682: * Assembler and Code words::
683: * Threading Words::
684: @end menu
685:
686: @node Notation, Arithmetic, Words, Words
687: @section Notation
688:
689: The Forth words are described in this section in the glossary notation
690: that has become a de-facto standard for Forth texts, i.e.
691:
692: @format
693: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
694: @end format
695: @var{Description}
696:
697: @table @var
698: @item word
699: The name of the word. BTW, Gforth is case insensitive, so you can
700: type the words in in lower case (However, @pxref{core-idef}).
701:
702: @item Stack effect
703: The stack effect is written in the notation @code{@var{before} --
704: @var{after}}, where @var{before} and @var{after} describe the top of
705: stack entries before and after the execution of the word. The rest of
706: the stack is not touched by the word. The top of stack is rightmost,
707: i.e., a stack sequence is written as it is typed in. Note that Gforth
708: uses a separate floating point stack, but a unified stack
709: notation. Also, return stack effects are not shown in @var{stack
710: effect}, but in @var{Description}. The name of a stack item describes
711: the type and/or the function of the item. See below for a discussion of
712: the types.
713:
714: All words have two stack effects: A compile-time stack effect and a
715: run-time stack effect. The compile-time stack-effect of most words is
716: @var{ -- }. If the compile-time stack-effect of a word deviates from
717: this standard behaviour, or the word does other unusual things at
718: compile time, both stack effects are shown; otherwise only the run-time
719: stack effect is shown.
720:
721: @item pronunciation
722: How the word is pronounced
723:
724: @item wordset
725: The ANS Forth standard is divided into several wordsets. A standard
726: system need not support all of them. So, the fewer wordsets your program
727: uses the more portable it will be in theory. However, we suspect that
728: most ANS Forth systems on personal machines will feature all
729: wordsets. Words that are not defined in the ANS standard have
730: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
731: describes words that will work in future releases of Gforth;
732: @code{gforth-internal} words are more volatile. Environmental query
733: strings are also displayed like words; you can recognize them by the
734: @code{environment} in the wordset field.
735:
736: @item Description
737: A description of the behaviour of the word.
738: @end table
739:
740: The type of a stack item is specified by the character(s) the name
741: starts with:
742:
743: @table @code
744: @item f
745: Bool, i.e. @code{false} or @code{true}.
746: @item c
747: Char
748: @item w
749: Cell, can contain an integer or an address
750: @item n
751: signed integer
752: @item u
753: unsigned integer
754: @item d
755: double sized signed integer
756: @item ud
757: double sized unsigned integer
758: @item r
759: Float (on the FP stack)
760: @item a_
761: Cell-aligned address
762: @item c_
763: Char-aligned address (note that a Char may have two bytes in Windows NT)
764: @item f_
765: Float-aligned address
766: @item df_
767: Address aligned for IEEE double precision float
768: @item sf_
769: Address aligned for IEEE single precision float
770: @item xt
771: Execution token, same size as Cell
772: @item wid
773: Wordlist ID, same size as Cell
774: @item f83name
775: Pointer to a name structure
776: @item "
777: string in the input stream (not the stack). The terminating character is
778: a blank by default. If it is not a blank, it is shown in @code{<>}
779: quotes.
780:
781: @end table
782:
783: @node Arithmetic, Stack Manipulation, Notation, Words
784: @section Arithmetic
785: Forth arithmetic is not checked, i.e., you will not hear about integer
786: overflow on addition or multiplication, you may hear about division by
787: zero if you are lucky. The operator is written after the operands, but
788: the operands are still in the original order. I.e., the infix @code{2-1}
789: corresponds to @code{2 1 -}. Forth offers a variety of division
790: operators. If you perform division with potentially negative operands,
791: you do not want to use @code{/} or @code{/mod} with its undefined
792: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
793: former, @pxref{Mixed precision}).
794:
795: @menu
796: * Single precision::
797: * Bitwise operations::
798: * Mixed precision:: operations with single and double-cell integers
799: * Double precision:: Double-cell integer arithmetic
800: * Floating Point::
801: @end menu
802:
803: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
804: @subsection Single precision
805: doc-+
806: doc--
807: doc-*
808: doc-/
809: doc-mod
810: doc-/mod
811: doc-negate
812: doc-abs
813: doc-min
814: doc-max
815:
816: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
817: @subsection Bitwise operations
818: doc-and
819: doc-or
820: doc-xor
821: doc-invert
822: doc-2*
823: doc-2/
824:
825: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
826: @subsection Mixed precision
827: doc-m+
828: doc-*/
829: doc-*/mod
830: doc-m*
831: doc-um*
832: doc-m*/
833: doc-um/mod
834: doc-fm/mod
835: doc-sm/rem
836:
837: @node Double precision, Floating Point, Mixed precision, Arithmetic
838: @subsection Double precision
839:
840: The outer (aka text) interpreter converts numbers containing a dot into
841: a double precision number. Note that only numbers with the dot as last
842: character are standard-conforming.
843:
844: doc-d+
845: doc-d-
846: doc-dnegate
847: doc-dabs
848: doc-dmin
849: doc-dmax
850:
851: @node Floating Point, , Double precision, Arithmetic
852: @subsection Floating Point
853:
854: The format of floating point numbers recognized by the outer (aka text)
855: interpreter is: a signed decimal number, possibly containing a decimal
856: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
857: by a signed integer (the exponent). E.g., @code{1e} ist the same as
858: @code{+1.0e+0}. Note that a number without @code{e}
859: is not interpreted as floating-point number, but as double (if the
860: number contains a @code{.}) or single precision integer. Also,
861: conversions between string and floating point numbers always use base
862: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
863: value greater then 14, the @code{E} may be interpreted as digit and the
864: number will be interpreted as integer, unless it has a signed exponent
865: (both @code{+} and @code{-} are allowed as signs).
866:
867: Angles in floating point operations are given in radians (a full circle
868: has 2 pi radians). Note, that Gforth has a separate floating point
869: stack, but we use the unified notation.
870:
871: Floating point numbers have a number of unpleasant surprises for the
872: unwary (e.g., floating point addition is not associative) and even a few
873: for the wary. You should not use them unless you know what you are doing
874: or you don't care that the results you get are totally bogus. If you
875: want to learn about the problems of floating point numbers (and how to
876: avoid them), you might start with @cite{David Goldberg, What Every
877: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
878: Computing Surveys 23(1):5@minus{}48, March 1991}.
879:
880: doc-f+
881: doc-f-
882: doc-f*
883: doc-f/
884: doc-fnegate
885: doc-fabs
886: doc-fmax
887: doc-fmin
888: doc-floor
889: doc-fround
890: doc-f**
891: doc-fsqrt
892: doc-fexp
893: doc-fexpm1
894: doc-fln
895: doc-flnp1
896: doc-flog
897: doc-falog
898: doc-fsin
899: doc-fcos
900: doc-fsincos
901: doc-ftan
902: doc-fasin
903: doc-facos
904: doc-fatan
905: doc-fatan2
906: doc-fsinh
907: doc-fcosh
908: doc-ftanh
909: doc-fasinh
910: doc-facosh
911: doc-fatanh
912:
913: @node Stack Manipulation, Memory access, Arithmetic, Words
914: @section Stack Manipulation
915:
916: Gforth has a data stack (aka parameter stack) for characters, cells,
917: addresses, and double cells, a floating point stack for floating point
918: numbers, a return stack for storing the return addresses of colon
919: definitions and other data, and a locals stack for storing local
920: variables. Note that while every sane Forth has a separate floating
921: point stack, this is not strictly required; an ANS Forth system could
922: theoretically keep floating point numbers on the data stack. As an
923: additional difficulty, you don't know how many cells a floating point
924: number takes. It is reportedly possible to write words in a way that
925: they work also for a unified stack model, but we do not recommend trying
926: it. Instead, just say that your program has an environmental dependency
927: on a separate FP stack.
928:
929: Also, a Forth system is allowed to keep the local variables on the
930: return stack. This is reasonable, as local variables usually eliminate
931: the need to use the return stack explicitly. So, if you want to produce
932: a standard complying program and if you are using local variables in a
933: word, forget about return stack manipulations in that word (see the
934: standard document for the exact rules).
935:
936: @menu
937: * Data stack::
938: * Floating point stack::
939: * Return stack::
940: * Locals stack::
941: * Stack pointer manipulation::
942: @end menu
943:
944: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
945: @subsection Data stack
946: doc-drop
947: doc-nip
948: doc-dup
949: doc-over
950: doc-tuck
951: doc-swap
952: doc-rot
953: doc--rot
954: doc-?dup
955: doc-pick
956: doc-roll
957: doc-2drop
958: doc-2nip
959: doc-2dup
960: doc-2over
961: doc-2tuck
962: doc-2swap
963: doc-2rot
964:
965: @node Floating point stack, Return stack, Data stack, Stack Manipulation
966: @subsection Floating point stack
967: doc-fdrop
968: doc-fnip
969: doc-fdup
970: doc-fover
971: doc-ftuck
972: doc-fswap
973: doc-frot
974:
975: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
976: @subsection Return stack
977: doc->r
978: doc-r>
979: doc-r@
980: doc-rdrop
981: doc-2>r
982: doc-2r>
983: doc-2r@
984: doc-2rdrop
985:
986: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
987: @subsection Locals stack
988:
989: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
990: @subsection Stack pointer manipulation
991: doc-sp@
992: doc-sp!
993: doc-fp@
994: doc-fp!
995: doc-rp@
996: doc-rp!
997: doc-lp@
998: doc-lp!
999:
1000: @node Memory access, Control Structures, Stack Manipulation, Words
1001: @section Memory access
1002:
1003: @menu
1004: * Stack-Memory transfers::
1005: * Address arithmetic::
1006: * Memory block access::
1007: @end menu
1008:
1009: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1010: @subsection Stack-Memory transfers
1011:
1012: doc-@
1013: doc-!
1014: doc-+!
1015: doc-c@
1016: doc-c!
1017: doc-2@
1018: doc-2!
1019: doc-f@
1020: doc-f!
1021: doc-sf@
1022: doc-sf!
1023: doc-df@
1024: doc-df!
1025:
1026: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1027: @subsection Address arithmetic
1028:
1029: ANS Forth does not specify the sizes of the data types. Instead, it
1030: offers a number of words for computing sizes and doing address
1031: arithmetic. Basically, address arithmetic is performed in terms of
1032: address units (aus); on most systems the address unit is one byte. Note
1033: that a character may have more than one au, so @code{chars} is no noop
1034: (on systems where it is a noop, it compiles to nothing).
1035:
1036: ANS Forth also defines words for aligning addresses for specific
1037: addresses. Many computers require that accesses to specific data types
1038: must only occur at specific addresses; e.g., that cells may only be
1039: accessed at addresses divisible by 4. Even if a machine allows unaligned
1040: accesses, it can usually perform aligned accesses faster.
1041:
1042: For the performance-conscious: alignment operations are usually only
1043: necessary during the definition of a data structure, not during the
1044: (more frequent) accesses to it.
1045:
1046: ANS Forth defines no words for character-aligning addresses. This is not
1047: an oversight, but reflects the fact that addresses that are not
1048: char-aligned have no use in the standard and therefore will not be
1049: created.
1050:
1051: The standard guarantees that addresses returned by @code{CREATE}d words
1052: are cell-aligned; in addition, Gforth guarantees that these addresses
1053: are aligned for all purposes.
1054:
1055: Note that the standard defines a word @code{char}, which has nothing to
1056: do with address arithmetic.
1057:
1058: doc-chars
1059: doc-char+
1060: doc-cells
1061: doc-cell+
1062: doc-align
1063: doc-aligned
1064: doc-floats
1065: doc-float+
1066: doc-falign
1067: doc-faligned
1068: doc-sfloats
1069: doc-sfloat+
1070: doc-sfalign
1071: doc-sfaligned
1072: doc-dfloats
1073: doc-dfloat+
1074: doc-dfalign
1075: doc-dfaligned
1076: doc-maxalign
1077: doc-maxaligned
1078: doc-cfalign
1079: doc-cfaligned
1080: doc-address-unit-bits
1081:
1082: @node Memory block access, , Address arithmetic, Memory access
1083: @subsection Memory block access
1084:
1085: doc-move
1086: doc-erase
1087:
1088: While the previous words work on address units, the rest works on
1089: characters.
1090:
1091: doc-cmove
1092: doc-cmove>
1093: doc-fill
1094: doc-blank
1095:
1096: @node Control Structures, Locals, Memory access, Words
1097: @section Control Structures
1098:
1099: Control structures in Forth cannot be used in interpret state, only in
1100: compile state, i.e., in a colon definition. We do not like this
1101: limitation, but have not seen a satisfying way around it yet, although
1102: many schemes have been proposed.
1103:
1104: @menu
1105: * Selection::
1106: * Simple Loops::
1107: * Counted Loops::
1108: * Arbitrary control structures::
1109: * Calls and returns::
1110: * Exception Handling::
1111: @end menu
1112:
1113: @node Selection, Simple Loops, Control Structures, Control Structures
1114: @subsection Selection
1115:
1116: @example
1117: @var{flag}
1118: IF
1119: @var{code}
1120: ENDIF
1121: @end example
1122: or
1123: @example
1124: @var{flag}
1125: IF
1126: @var{code1}
1127: ELSE
1128: @var{code2}
1129: ENDIF
1130: @end example
1131:
1132: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1133: standard, and @code{ENDIF} is not, although it is quite popular. We
1134: recommend using @code{ENDIF}, because it is less confusing for people
1135: who also know other languages (and is not prone to reinforcing negative
1136: prejudices against Forth in these people). Adding @code{ENDIF} to a
1137: system that only supplies @code{THEN} is simple:
1138: @example
1139: : endif POSTPONE then ; immediate
1140: @end example
1141:
1142: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1143: (adv.)} has the following meanings:
1144: @quotation
1145: ... 2b: following next after in order ... 3d: as a necessary consequence
1146: (if you were there, then you saw them).
1147: @end quotation
1148: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1149: and many other programming languages has the meaning 3d.]
1150:
1151: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1152: you can avoid using @code{?dup}. Using these alternatives is also more
1153: efficient than using @code{?dup}. Definitions in plain standard Forth
1154: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1155: @file{compat/control.fs}.
1156:
1157: @example
1158: @var{n}
1159: CASE
1160: @var{n1} OF @var{code1} ENDOF
1161: @var{n2} OF @var{code2} ENDOF
1162: @dots{}
1163: ENDCASE
1164: @end example
1165:
1166: Executes the first @var{codei}, where the @var{ni} is equal to
1167: @var{n}. A default case can be added by simply writing the code after
1168: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1169: but must not consume it.
1170:
1171: @node Simple Loops, Counted Loops, Selection, Control Structures
1172: @subsection Simple Loops
1173:
1174: @example
1175: BEGIN
1176: @var{code1}
1177: @var{flag}
1178: WHILE
1179: @var{code2}
1180: REPEAT
1181: @end example
1182:
1183: @var{code1} is executed and @var{flag} is computed. If it is true,
1184: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1185:
1186: @example
1187: BEGIN
1188: @var{code}
1189: @var{flag}
1190: UNTIL
1191: @end example
1192:
1193: @var{code} is executed. The loop is restarted if @code{flag} is false.
1194:
1195: @example
1196: BEGIN
1197: @var{code}
1198: AGAIN
1199: @end example
1200:
1201: This is an endless loop.
1202:
1203: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1204: @subsection Counted Loops
1205:
1206: The basic counted loop is:
1207: @example
1208: @var{limit} @var{start}
1209: ?DO
1210: @var{body}
1211: LOOP
1212: @end example
1213:
1214: This performs one iteration for every integer, starting from @var{start}
1215: and up to, but excluding @var{limit}. The counter, aka index, can be
1216: accessed with @code{i}. E.g., the loop
1217: @example
1218: 10 0 ?DO
1219: i .
1220: LOOP
1221: @end example
1222: prints
1223: @example
1224: 0 1 2 3 4 5 6 7 8 9
1225: @end example
1226: The index of the innermost loop can be accessed with @code{i}, the index
1227: of the next loop with @code{j}, and the index of the third loop with
1228: @code{k}.
1229:
1230: The loop control data are kept on the return stack, so there are some
1231: restrictions on mixing return stack accesses and counted loop
1232: words. E.g., if you put values on the return stack outside the loop, you
1233: cannot read them inside the loop. If you put values on the return stack
1234: within a loop, you have to remove them before the end of the loop and
1235: before accessing the index of the loop.
1236:
1237: There are several variations on the counted loop:
1238:
1239: @code{LEAVE} leaves the innermost counted loop immediately.
1240:
1241: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1242: (and @code{LOOP} iterates until they become equal by wrap-around
1243: arithmetic). This behaviour is usually not what you want. Therefore,
1244: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1245: @code{?DO}), which do not enter the loop if @var{start} is greater than
1246: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1247: unsigned loop parameters.
1248:
1249: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1250: index by @var{n} instead of by 1. The loop is terminated when the border
1251: between @var{limit-1} and @var{limit} is crossed. E.g.:
1252:
1253: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1254:
1255: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1256:
1257: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1258:
1259: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1260:
1261: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1262:
1263: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1264: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1265: index by @var{u} each iteration. The loop is terminated when the border
1266: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1267: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1268:
1269: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1270:
1271: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1272:
1273: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1274:
1275: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1276: @code{-LOOP} are not in the ANS Forth standard. However, an
1277: implementation for these words that uses only standard words is provided
1278: in @file{compat/loops.fs}.
1279:
1280: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1281: the loop, independent of the loop parameters. Do not use @code{DO}, even
1282: if you know that the loop is entered in any case. Such knowledge tends
1283: to become invalid during maintenance of a program, and then the
1284: @code{DO} will make trouble.
1285:
1286: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1287: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1288: return stack so @code{EXIT} can get to its return address.
1289:
1290: Another counted loop is
1291: @example
1292: @var{n}
1293: FOR
1294: @var{body}
1295: NEXT
1296: @end example
1297: This is the preferred loop of native code compiler writers who are too
1298: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1299: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1300: and ending with 0. Other Forth systems may behave differently, even if
1301: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1302: loops.
1303:
1304: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1305: @subsection Arbitrary control structures
1306:
1307: ANS Forth permits and supports using control structures in a non-nested
1308: way. Information about incomplete control structures is stored on the
1309: control-flow stack. This stack may be implemented on the Forth data
1310: stack, and this is what we have done in Gforth.
1311:
1312: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1313: entry represents a backward branch target. A few words are the basis for
1314: building any control structure possible (except control structures that
1315: need storage, like calls, coroutines, and backtracking).
1316:
1317: doc-if
1318: doc-ahead
1319: doc-then
1320: doc-begin
1321: doc-until
1322: doc-again
1323: doc-cs-pick
1324: doc-cs-roll
1325:
1326: On many systems control-flow stack items take one word, in Gforth they
1327: currently take three (this may change in the future). Therefore it is a
1328: really good idea to manipulate the control flow stack with
1329: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1330: words.
1331:
1332: Some standard control structure words are built from these words:
1333:
1334: doc-else
1335: doc-while
1336: doc-repeat
1337:
1338: Gforth adds some more control-structure words:
1339:
1340: doc-endif
1341: doc-?dup-if
1342: doc-?dup-0=-if
1343:
1344: Counted loop words constitute a separate group of words:
1345:
1346: doc-?do
1347: doc-+do
1348: doc-u+do
1349: doc--do
1350: doc-u-do
1351: doc-do
1352: doc-for
1353: doc-loop
1354: doc-+loop
1355: doc--loop
1356: doc-next
1357: doc-leave
1358: doc-?leave
1359: doc-unloop
1360: doc-done
1361:
1362: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1363: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1364: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1365: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1366: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1367: resolved (by using one of the loop-ending words or @code{DONE}).
1368:
1369: Another group of control structure words are
1370:
1371: doc-case
1372: doc-endcase
1373: doc-of
1374: doc-endof
1375:
1376: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1377: @code{cs-roll}.
1378:
1379: @subsubsection Programming Style
1380:
1381: In order to ensure readability we recommend that you do not create
1382: arbitrary control structures directly, but define new control structure
1383: words for the control structure you want and use these words in your
1384: program.
1385:
1386: E.g., instead of writing
1387:
1388: @example
1389: begin
1390: ...
1391: if [ 1 cs-roll ]
1392: ...
1393: again then
1394: @end example
1395:
1396: we recommend defining control structure words, e.g.,
1397:
1398: @example
1399: : while ( dest -- orig dest )
1400: POSTPONE if
1401: 1 cs-roll ; immediate
1402:
1403: : repeat ( orig dest -- )
1404: POSTPONE again
1405: POSTPONE then ; immediate
1406: @end example
1407:
1408: and then using these to create the control structure:
1409:
1410: @example
1411: begin
1412: ...
1413: while
1414: ...
1415: repeat
1416: @end example
1417:
1418: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1419: @code{WHILE} are predefined, so in this example it would not be
1420: necessary to define them.
1421:
1422: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1423: @subsection Calls and returns
1424:
1425: A definition can be called simply be writing the name of the
1426: definition. When the end of the definition is reached, it returns. An
1427: earlier return can be forced using
1428:
1429: doc-exit
1430:
1431: Don't forget to clean up the return stack and @code{UNLOOP} any
1432: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1433: primitive compiled by @code{EXIT} is
1434:
1435: doc-;s
1436:
1437: @node Exception Handling, , Calls and returns, Control Structures
1438: @subsection Exception Handling
1439:
1440: doc-catch
1441: doc-throw
1442:
1443: @node Locals, Defining Words, Control Structures, Words
1444: @section Locals
1445:
1446: Local variables can make Forth programming more enjoyable and Forth
1447: programs easier to read. Unfortunately, the locals of ANS Forth are
1448: laden with restrictions. Therefore, we provide not only the ANS Forth
1449: locals wordset, but also our own, more powerful locals wordset (we
1450: implemented the ANS Forth locals wordset through our locals wordset).
1451:
1452: The ideas in this section have also been published in the paper
1453: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1454: at EuroForth '94; it is available at
1455: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1456:
1457: @menu
1458: * Gforth locals::
1459: * ANS Forth locals::
1460: @end menu
1461:
1462: @node Gforth locals, ANS Forth locals, Locals, Locals
1463: @subsection Gforth locals
1464:
1465: Locals can be defined with
1466:
1467: @example
1468: @{ local1 local2 ... -- comment @}
1469: @end example
1470: or
1471: @example
1472: @{ local1 local2 ... @}
1473: @end example
1474:
1475: E.g.,
1476: @example
1477: : max @{ n1 n2 -- n3 @}
1478: n1 n2 > if
1479: n1
1480: else
1481: n2
1482: endif ;
1483: @end example
1484:
1485: The similarity of locals definitions with stack comments is intended. A
1486: locals definition often replaces the stack comment of a word. The order
1487: of the locals corresponds to the order in a stack comment and everything
1488: after the @code{--} is really a comment.
1489:
1490: This similarity has one disadvantage: It is too easy to confuse locals
1491: declarations with stack comments, causing bugs and making them hard to
1492: find. However, this problem can be avoided by appropriate coding
1493: conventions: Do not use both notations in the same program. If you do,
1494: they should be distinguished using additional means, e.g. by position.
1495:
1496: The name of the local may be preceded by a type specifier, e.g.,
1497: @code{F:} for a floating point value:
1498:
1499: @example
1500: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1501: \ complex multiplication
1502: Ar Br f* Ai Bi f* f-
1503: Ar Bi f* Ai Br f* f+ ;
1504: @end example
1505:
1506: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1507: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1508: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1509: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1510: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1511: produces its address (which becomes invalid when the variable's scope is
1512: left). E.g., the standard word @code{emit} can be defined in therms of
1513: @code{type} like this:
1514:
1515: @example
1516: : emit @{ C^ char* -- @}
1517: char* 1 type ;
1518: @end example
1519:
1520: A local without type specifier is a @code{W:} local. Both flavours of
1521: locals are initialized with values from the data or FP stack.
1522:
1523: Currently there is no way to define locals with user-defined data
1524: structures, but we are working on it.
1525:
1526: Gforth allows defining locals everywhere in a colon definition. This
1527: poses the following questions:
1528:
1529: @menu
1530: * Where are locals visible by name?::
1531: * How long do locals live?::
1532: * Programming Style::
1533: * Implementation::
1534: @end menu
1535:
1536: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1537: @subsubsection Where are locals visible by name?
1538:
1539: Basically, the answer is that locals are visible where you would expect
1540: it in block-structured languages, and sometimes a little longer. If you
1541: want to restrict the scope of a local, enclose its definition in
1542: @code{SCOPE}...@code{ENDSCOPE}.
1543:
1544: doc-scope
1545: doc-endscope
1546:
1547: These words behave like control structure words, so you can use them
1548: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1549: arbitrary ways.
1550:
1551: If you want a more exact answer to the visibility question, here's the
1552: basic principle: A local is visible in all places that can only be
1553: reached through the definition of the local@footnote{In compiler
1554: construction terminology, all places dominated by the definition of the
1555: local.}. In other words, it is not visible in places that can be reached
1556: without going through the definition of the local. E.g., locals defined
1557: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1558: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1559: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1560:
1561: The reasoning behind this solution is: We want to have the locals
1562: visible as long as it is meaningful. The user can always make the
1563: visibility shorter by using explicit scoping. In a place that can
1564: only be reached through the definition of a local, the meaning of a
1565: local name is clear. In other places it is not: How is the local
1566: initialized at the control flow path that does not contain the
1567: definition? Which local is meant, if the same name is defined twice in
1568: two independent control flow paths?
1569:
1570: This should be enough detail for nearly all users, so you can skip the
1571: rest of this section. If you relly must know all the gory details and
1572: options, read on.
1573:
1574: In order to implement this rule, the compiler has to know which places
1575: are unreachable. It knows this automatically after @code{AHEAD},
1576: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1577: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1578: compiler that the control flow never reaches that place. If
1579: @code{UNREACHABLE} is not used where it could, the only consequence is
1580: that the visibility of some locals is more limited than the rule above
1581: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1582: lie to the compiler), buggy code will be produced.
1583:
1584: Another problem with this rule is that at @code{BEGIN}, the compiler
1585: does not know which locals will be visible on the incoming
1586: back-edge. All problems discussed in the following are due to this
1587: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1588: loops as examples; the discussion also applies to @code{?DO} and other
1589: loops). Perhaps the most insidious example is:
1590: @example
1591: AHEAD
1592: BEGIN
1593: x
1594: [ 1 CS-ROLL ] THEN
1595: @{ x @}
1596: ...
1597: UNTIL
1598: @end example
1599:
1600: This should be legal according to the visibility rule. The use of
1601: @code{x} can only be reached through the definition; but that appears
1602: textually below the use.
1603:
1604: From this example it is clear that the visibility rules cannot be fully
1605: implemented without major headaches. Our implementation treats common
1606: cases as advertised and the exceptions are treated in a safe way: The
1607: compiler makes a reasonable guess about the locals visible after a
1608: @code{BEGIN}; if it is too pessimistic, the
1609: user will get a spurious error about the local not being defined; if the
1610: compiler is too optimistic, it will notice this later and issue a
1611: warning. In the case above the compiler would complain about @code{x}
1612: being undefined at its use. You can see from the obscure examples in
1613: this section that it takes quite unusual control structures to get the
1614: compiler into trouble, and even then it will often do fine.
1615:
1616: If the @code{BEGIN} is reachable from above, the most optimistic guess
1617: is that all locals visible before the @code{BEGIN} will also be
1618: visible after the @code{BEGIN}. This guess is valid for all loops that
1619: are entered only through the @code{BEGIN}, in particular, for normal
1620: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1621: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1622: compiler. When the branch to the @code{BEGIN} is finally generated by
1623: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1624: warns the user if it was too optimisitic:
1625: @example
1626: IF
1627: @{ x @}
1628: BEGIN
1629: \ x ?
1630: [ 1 cs-roll ] THEN
1631: ...
1632: UNTIL
1633: @end example
1634:
1635: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1636: optimistically assumes that it lives until the @code{THEN}. It notices
1637: this difference when it compiles the @code{UNTIL} and issues a
1638: warning. The user can avoid the warning, and make sure that @code{x}
1639: is not used in the wrong area by using explicit scoping:
1640: @example
1641: IF
1642: SCOPE
1643: @{ x @}
1644: ENDSCOPE
1645: BEGIN
1646: [ 1 cs-roll ] THEN
1647: ...
1648: UNTIL
1649: @end example
1650:
1651: Since the guess is optimistic, there will be no spurious error messages
1652: about undefined locals.
1653:
1654: If the @code{BEGIN} is not reachable from above (e.g., after
1655: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1656: optimistic guess, as the locals visible after the @code{BEGIN} may be
1657: defined later. Therefore, the compiler assumes that no locals are
1658: visible after the @code{BEGIN}. However, the user can use
1659: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1660: visible at the BEGIN as at the point where the top control-flow stack
1661: item was created.
1662:
1663: doc-assume-live
1664:
1665: E.g.,
1666: @example
1667: @{ x @}
1668: AHEAD
1669: ASSUME-LIVE
1670: BEGIN
1671: x
1672: [ 1 CS-ROLL ] THEN
1673: ...
1674: UNTIL
1675: @end example
1676:
1677: Other cases where the locals are defined before the @code{BEGIN} can be
1678: handled by inserting an appropriate @code{CS-ROLL} before the
1679: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1680: behind the @code{ASSUME-LIVE}).
1681:
1682: Cases where locals are defined after the @code{BEGIN} (but should be
1683: visible immediately after the @code{BEGIN}) can only be handled by
1684: rearranging the loop. E.g., the ``most insidious'' example above can be
1685: arranged into:
1686: @example
1687: BEGIN
1688: @{ x @}
1689: ... 0=
1690: WHILE
1691: x
1692: REPEAT
1693: @end example
1694:
1695: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1696: @subsubsection How long do locals live?
1697:
1698: The right answer for the lifetime question would be: A local lives at
1699: least as long as it can be accessed. For a value-flavoured local this
1700: means: until the end of its visibility. However, a variable-flavoured
1701: local could be accessed through its address far beyond its visibility
1702: scope. Ultimately, this would mean that such locals would have to be
1703: garbage collected. Since this entails un-Forth-like implementation
1704: complexities, I adopted the same cowardly solution as some other
1705: languages (e.g., C): The local lives only as long as it is visible;
1706: afterwards its address is invalid (and programs that access it
1707: afterwards are erroneous).
1708:
1709: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1710: @subsubsection Programming Style
1711:
1712: The freedom to define locals anywhere has the potential to change
1713: programming styles dramatically. In particular, the need to use the
1714: return stack for intermediate storage vanishes. Moreover, all stack
1715: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1716: determined arguments) can be eliminated: If the stack items are in the
1717: wrong order, just write a locals definition for all of them; then
1718: write the items in the order you want.
1719:
1720: This seems a little far-fetched and eliminating stack manipulations is
1721: unlikely to become a conscious programming objective. Still, the number
1722: of stack manipulations will be reduced dramatically if local variables
1723: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1724: a traditional implementation of @code{max}).
1725:
1726: This shows one potential benefit of locals: making Forth programs more
1727: readable. Of course, this benefit will only be realized if the
1728: programmers continue to honour the principle of factoring instead of
1729: using the added latitude to make the words longer.
1730:
1731: Using @code{TO} can and should be avoided. Without @code{TO},
1732: every value-flavoured local has only a single assignment and many
1733: advantages of functional languages apply to Forth. I.e., programs are
1734: easier to analyse, to optimize and to read: It is clear from the
1735: definition what the local stands for, it does not turn into something
1736: different later.
1737:
1738: E.g., a definition using @code{TO} might look like this:
1739: @example
1740: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1741: u1 u2 min 0
1742: ?do
1743: addr1 c@@ addr2 c@@ -
1744: ?dup-if
1745: unloop exit
1746: then
1747: addr1 char+ TO addr1
1748: addr2 char+ TO addr2
1749: loop
1750: u1 u2 - ;
1751: @end example
1752: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1753: every loop iteration. @code{strcmp} is a typical example of the
1754: readability problems of using @code{TO}. When you start reading
1755: @code{strcmp}, you think that @code{addr1} refers to the start of the
1756: string. Only near the end of the loop you realize that it is something
1757: else.
1758:
1759: This can be avoided by defining two locals at the start of the loop that
1760: are initialized with the right value for the current iteration.
1761: @example
1762: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1763: addr1 addr2
1764: u1 u2 min 0
1765: ?do @{ s1 s2 @}
1766: s1 c@@ s2 c@@ -
1767: ?dup-if
1768: unloop exit
1769: then
1770: s1 char+ s2 char+
1771: loop
1772: 2drop
1773: u1 u2 - ;
1774: @end example
1775: Here it is clear from the start that @code{s1} has a different value
1776: in every loop iteration.
1777:
1778: @node Implementation, , Programming Style, Gforth locals
1779: @subsubsection Implementation
1780:
1781: Gforth uses an extra locals stack. The most compelling reason for
1782: this is that the return stack is not float-aligned; using an extra stack
1783: also eliminates the problems and restrictions of using the return stack
1784: as locals stack. Like the other stacks, the locals stack grows toward
1785: lower addresses. A few primitives allow an efficient implementation:
1786:
1787: doc-@local#
1788: doc-f@local#
1789: doc-laddr#
1790: doc-lp+!#
1791: doc-lp!
1792: doc->l
1793: doc-f>l
1794:
1795: In addition to these primitives, some specializations of these
1796: primitives for commonly occurring inline arguments are provided for
1797: efficiency reasons, e.g., @code{@@local0} as specialization of
1798: @code{@@local#} for the inline argument 0. The following compiling words
1799: compile the right specialized version, or the general version, as
1800: appropriate:
1801:
1802: doc-compile-@local
1803: doc-compile-f@local
1804: doc-compile-lp+!
1805:
1806: Combinations of conditional branches and @code{lp+!#} like
1807: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1808: is taken) are provided for efficiency and correctness in loops.
1809:
1810: A special area in the dictionary space is reserved for keeping the
1811: local variable names. @code{@{} switches the dictionary pointer to this
1812: area and @code{@}} switches it back and generates the locals
1813: initializing code. @code{W:} etc.@ are normal defining words. This
1814: special area is cleared at the start of every colon definition.
1815:
1816: A special feature of Gforth's dictionary is used to implement the
1817: definition of locals without type specifiers: every wordlist (aka
1818: vocabulary) has its own methods for searching
1819: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1820: with a special search method: When it is searched for a word, it
1821: actually creates that word using @code{W:}. @code{@{} changes the search
1822: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1823: and then the wordlist for defining locals without type specifiers.
1824:
1825: The lifetime rules support a stack discipline within a colon
1826: definition: The lifetime of a local is either nested with other locals
1827: lifetimes or it does not overlap them.
1828:
1829: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1830: pointer manipulation is generated. Between control structure words
1831: locals definitions can push locals onto the locals stack. @code{AGAIN}
1832: is the simplest of the other three control flow words. It has to
1833: restore the locals stack depth of the corresponding @code{BEGIN}
1834: before branching. The code looks like this:
1835: @format
1836: @code{lp+!#} current-locals-size @minus{} dest-locals-size
1837: @code{branch} <begin>
1838: @end format
1839:
1840: @code{UNTIL} is a little more complicated: If it branches back, it
1841: must adjust the stack just like @code{AGAIN}. But if it falls through,
1842: the locals stack must not be changed. The compiler generates the
1843: following code:
1844: @format
1845: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1846: @end format
1847: The locals stack pointer is only adjusted if the branch is taken.
1848:
1849: @code{THEN} can produce somewhat inefficient code:
1850: @format
1851: @code{lp+!#} current-locals-size @minus{} orig-locals-size
1852: <orig target>:
1853: @code{lp+!#} orig-locals-size @minus{} new-locals-size
1854: @end format
1855: The second @code{lp+!#} adjusts the locals stack pointer from the
1856: level at the @var{orig} point to the level after the @code{THEN}. The
1857: first @code{lp+!#} adjusts the locals stack pointer from the current
1858: level to the level at the orig point, so the complete effect is an
1859: adjustment from the current level to the right level after the
1860: @code{THEN}.
1861:
1862: In a conventional Forth implementation a dest control-flow stack entry
1863: is just the target address and an orig entry is just the address to be
1864: patched. Our locals implementation adds a wordlist to every orig or dest
1865: item. It is the list of locals visible (or assumed visible) at the point
1866: described by the entry. Our implementation also adds a tag to identify
1867: the kind of entry, in particular to differentiate between live and dead
1868: (reachable and unreachable) orig entries.
1869:
1870: A few unusual operations have to be performed on locals wordlists:
1871:
1872: doc-common-list
1873: doc-sub-list?
1874: doc-list-size
1875:
1876: Several features of our locals wordlist implementation make these
1877: operations easy to implement: The locals wordlists are organised as
1878: linked lists; the tails of these lists are shared, if the lists
1879: contain some of the same locals; and the address of a name is greater
1880: than the address of the names behind it in the list.
1881:
1882: Another important implementation detail is the variable
1883: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1884: determine if they can be reached directly or only through the branch
1885: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1886: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1887: definition, by @code{BEGIN} and usually by @code{THEN}.
1888:
1889: Counted loops are similar to other loops in most respects, but
1890: @code{LEAVE} requires special attention: It performs basically the same
1891: service as @code{AHEAD}, but it does not create a control-flow stack
1892: entry. Therefore the information has to be stored elsewhere;
1893: traditionally, the information was stored in the target fields of the
1894: branches created by the @code{LEAVE}s, by organizing these fields into a
1895: linked list. Unfortunately, this clever trick does not provide enough
1896: space for storing our extended control flow information. Therefore, we
1897: introduce another stack, the leave stack. It contains the control-flow
1898: stack entries for all unresolved @code{LEAVE}s.
1899:
1900: Local names are kept until the end of the colon definition, even if
1901: they are no longer visible in any control-flow path. In a few cases
1902: this may lead to increased space needs for the locals name area, but
1903: usually less than reclaiming this space would cost in code size.
1904:
1905:
1906: @node ANS Forth locals, , Gforth locals, Locals
1907: @subsection ANS Forth locals
1908:
1909: The ANS Forth locals wordset does not define a syntax for locals, but
1910: words that make it possible to define various syntaxes. One of the
1911: possible syntaxes is a subset of the syntax we used in the Gforth locals
1912: wordset, i.e.:
1913:
1914: @example
1915: @{ local1 local2 ... -- comment @}
1916: @end example
1917: or
1918: @example
1919: @{ local1 local2 ... @}
1920: @end example
1921:
1922: The order of the locals corresponds to the order in a stack comment. The
1923: restrictions are:
1924:
1925: @itemize @bullet
1926: @item
1927: Locals can only be cell-sized values (no type specifiers are allowed).
1928: @item
1929: Locals can be defined only outside control structures.
1930: @item
1931: Locals can interfere with explicit usage of the return stack. For the
1932: exact (and long) rules, see the standard. If you don't use return stack
1933: accessing words in a definition using locals, you will be all right. The
1934: purpose of this rule is to make locals implementation on the return
1935: stack easier.
1936: @item
1937: The whole definition must be in one line.
1938: @end itemize
1939:
1940: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
1941: Defining Words}). I.e., they are initialized from the stack. Using their
1942: name produces their value. Their value can be changed using @code{TO}.
1943:
1944: Since this syntax is supported by Gforth directly, you need not do
1945: anything to use it. If you want to port a program using this syntax to
1946: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
1947: syntax on the other system.
1948:
1949: Note that a syntax shown in the standard, section A.13 looks
1950: similar, but is quite different in having the order of locals
1951: reversed. Beware!
1952:
1953: The ANS Forth locals wordset itself consists of the following word
1954:
1955: doc-(local)
1956:
1957: The ANS Forth locals extension wordset defines a syntax, but it is so
1958: awful that we strongly recommend not to use it. We have implemented this
1959: syntax to make porting to Gforth easy, but do not document it here. The
1960: problem with this syntax is that the locals are defined in an order
1961: reversed with respect to the standard stack comment notation, making
1962: programs harder to read, and easier to misread and miswrite. The only
1963: merit of this syntax is that it is easy to implement using the ANS Forth
1964: locals wordset.
1965:
1966: @node Defining Words, Tokens for Words, Locals, Words
1967: @section Defining Words
1968:
1969: @menu
1970: * Simple Defining Words::
1971: * Colon Definitions::
1972: * User-defined Defining Words::
1973: * Supplying names::
1974: * Interpretation and Compilation Semantics::
1975: @end menu
1976:
1977: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
1978: @subsection Simple Defining Words
1979:
1980: doc-constant
1981: doc-2constant
1982: doc-fconstant
1983: doc-variable
1984: doc-2variable
1985: doc-fvariable
1986: doc-create
1987: doc-user
1988: doc-value
1989: doc-to
1990: doc-defer
1991: doc-is
1992:
1993: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
1994: @subsection Colon Definitions
1995:
1996: @example
1997: : name ( ... -- ... )
1998: word1 word2 word3 ;
1999: @end example
2000:
2001: creates a word called @code{name}, that, upon execution, executes
2002: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2003:
2004: The explanation above is somewhat superficial. @xref{Interpretation and
2005: Compilation Semantics} for an in-depth discussion of some of the issues
2006: involved.
2007:
2008: doc-:
2009: doc-;
2010:
2011: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2012: @subsection User-defined Defining Words
2013:
2014: You can create new defining words simply by wrapping defining-time code
2015: around existing defining words and putting the sequence in a colon
2016: definition.
2017:
2018: If you want the words defined with your defining words to behave
2019: differently from words defined with standard defining words, you can
2020: write your defining word like this:
2021:
2022: @example
2023: : def-word ( "name" -- )
2024: Create @var{code1}
2025: DOES> ( ... -- ... )
2026: @var{code2} ;
2027:
2028: def-word name
2029: @end example
2030:
2031: Technically, this fragment defines a defining word @code{def-word}, and
2032: a word @code{name}; when you execute @code{name}, the address of the
2033: body of @code{name} is put on the data stack and @var{code2} is executed
2034: (the address of the body of @code{name} is the address @code{HERE}
2035: returns immediately after the @code{CREATE}).
2036:
2037: In other words, if you make the following definitions:
2038:
2039: @example
2040: : def-word1 ( "name" -- )
2041: Create @var{code1} ;
2042:
2043: : action1 ( ... -- ... )
2044: @var{code2} ;
2045:
2046: def-word name1
2047: @end example
2048:
2049: Using @code{name1 action1} is equivalent to using @code{name}.
2050:
2051: E.g., you can implement @code{Constant} in this way:
2052:
2053: @example
2054: : constant ( w "name" -- )
2055: create ,
2056: DOES> ( -- w )
2057: @@ ;
2058: @end example
2059:
2060: When you create a constant with @code{5 constant five}, first a new word
2061: @code{five} is created, then the value 5 is laid down in the body of
2062: @code{five} with @code{,}. When @code{five} is invoked, the address of
2063: the body is put on the stack, and @code{@@} retrieves the value 5.
2064:
2065: In the example above the stack comment after the @code{DOES>} specifies
2066: the stack effect of the defined words, not the stack effect of the
2067: following code (the following code expects the address of the body on
2068: the top of stack, which is not reflected in the stack comment). This is
2069: the convention that I use and recommend (it clashes a bit with using
2070: locals declarations for stack effect specification, though).
2071:
2072: @subsubsection Applications of @code{CREATE..DOES>}
2073:
2074: You may wonder how to use this feature. Here are some usage patterns:
2075:
2076: When you see a sequence of code occurring several times, and you can
2077: identify a meaning, you will factor it out as a colon definition. When
2078: you see similar colon definitions, you can factor them using
2079: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
2080: that look very similar:
2081: @example
2082: : ori, ( reg-taget reg-source n -- )
2083: 0 asm-reg-reg-imm ;
2084: : andi, ( reg-taget reg-source n -- )
2085: 1 asm-reg-reg-imm ;
2086: @end example
2087:
2088: This could be factored with:
2089: @example
2090: : reg-reg-imm ( op-code -- )
2091: create ,
2092: DOES> ( reg-taget reg-source n -- )
2093: @@ asm-reg-reg-imm ;
2094:
2095: 0 reg-reg-imm ori,
2096: 1 reg-reg-imm andi,
2097: @end example
2098:
2099: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2100: supply a part of the parameters for a word (known as @dfn{currying} in
2101: the functional language community). E.g., @code{+} needs two
2102: parameters. Creating versions of @code{+} with one parameter fixed can
2103: be done like this:
2104: @example
2105: : curry+ ( n1 -- )
2106: create ,
2107: DOES> ( n2 -- n1+n2 )
2108: @@ + ;
2109:
2110: 3 curry+ 3+
2111: -2 curry+ 2-
2112: @end example
2113:
2114: @subsubsection The gory details of @code{CREATE..DOES>}
2115:
2116: doc-does>
2117:
2118: This means that you need not use @code{CREATE} and @code{DOES>} in the
2119: same definition; E.g., you can put the @code{DOES>}-part in a separate
2120: definition. This allows us to, e.g., select among different DOES>-parts:
2121: @example
2122: : does1
2123: DOES> ( ... -- ... )
2124: ... ;
2125:
2126: : does2
2127: DOES> ( ... -- ... )
2128: ... ;
2129:
2130: : def-word ( ... -- ... )
2131: create ...
2132: IF
2133: does1
2134: ELSE
2135: does2
2136: ENDIF ;
2137: @end example
2138:
2139: In a standard program you can apply a @code{DOES>}-part only if the last
2140: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2141: will override the behaviour of the last word defined in any case. In a
2142: standard program, you can use @code{DOES>} only in a colon
2143: definition. In Gforth, you can also use it in interpretation state, in a
2144: kind of one-shot mode:
2145: @example
2146: CREATE name ( ... -- ... )
2147: @var{initialization}
2148: DOES>
2149: @var{code} ;
2150: @end example
2151: This is equivalwent to the standard
2152: @example
2153: :noname
2154: DOES>
2155: @var{code} ;
2156: CREATE name EXECUTE ( ... -- ... )
2157: @var{initialization}
2158: @end example
2159:
2160: You can get the address of the body of a word with
2161:
2162: doc->body
2163:
2164: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2165: @subsection Supplying names for the defined words
2166:
2167: By default, defining words take the names for the defined words from the
2168: input stream. Sometimes you want to supply the name from a string. You
2169: can do this with
2170:
2171: doc-nextname
2172:
2173: E.g.,
2174:
2175: @example
2176: s" foo" nextname create
2177: @end example
2178: is equivalent to
2179: @example
2180: create foo
2181: @end example
2182:
2183: Sometimes you want to define a word without a name. You can do this with
2184:
2185: doc-noname
2186:
2187: To make any use of the newly defined word, you need its execution
2188: token. You can get it with
2189:
2190: doc-lastxt
2191:
2192: E.g., you can initialize a deferred word with an anonymous colon
2193: definition:
2194: @example
2195: Defer deferred
2196: noname : ( ... -- ... )
2197: ... ;
2198: lastxt IS deferred
2199: @end example
2200:
2201: @code{lastxt} also works when the last word was not defined as
2202: @code{noname}.
2203:
2204: The standard has also recognized the need for anonymous words and
2205: provides
2206:
2207: doc-:noname
2208:
2209: This leaves the execution token for the word on the stack after the
2210: closing @code{;}. You can rewrite the last example with @code{:noname}:
2211: @example
2212: Defer deferred
2213: :noname ( ... -- ... )
2214: ... ;
2215: IS deferred
2216: @end example
2217:
2218: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2219: @subsection Interpretation and Compilation Semantics
2220:
2221: The @dfn{interpretation semantics} of a word are what the text
2222: interpreter does when it encounters the word in interpret state. It also
2223: appears in some other contexts, e.g., the execution token returned by
2224: @code{' @var{word}} identifies the interpretation semantics of
2225: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
2226: interpret-state text interpretation of @code{@var{word}}).
2227:
2228: The @dfn{compilation semantics} of a word are what the text interpreter
2229: does when it encounters the word in compile state. It also appears in
2230: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2231: standard terminology, ``appends to the current definition''.} the
2232: compilation semantics of @var{word}.
2233:
2234: The standard also talks about @dfn{execution semantics}. They are used
2235: only for defining the interpretation and compilation semantics of many
2236: words. By default, the interpretation semantics of a word are to
2237: @code{execute} its execution semantics, and the compilation semantics of
2238: a word are to @code{compile,} its execution semantics.@footnote{In
2239: standard terminology: The default interpretation semantics are its
2240: execution semantics; the default compilation semantics are to append its
2241: execution semantics to the execution semantics of the current
2242: definition.}
2243:
2244: You can change the compilation semantics into @code{execute}ing the
2245: execution semantics with
2246:
2247: doc-immediate
2248:
2249: You can remove the interpretation semantics of a word with
2250:
2251: doc-compile-only
2252: doc-restrict
2253:
2254: Note that ticking (@code{'}) compile-only words gives an error
2255: (``Interpreting a compile-only word'').
2256:
2257: Gforth also allows you to define words with arbitrary combinations of
2258: interpretation and compilation semantics.
2259:
2260: doc-interpret/compile:
2261:
2262: This feature was introduced for implementing @code{TO} and @code{S"}. I
2263: recommend that you do not define such words, as cute as they may be:
2264: they make it hard to get at both parts of the word in some contexts.
2265: E.g., assume you want to get an execution token for the compilation
2266: part. Instead, define two words, one that embodies the interpretation
2267: part, and one that embodies the compilation part.
2268:
2269: There is, however, a potentially useful application of this feature:
2270: Providing differing implementations for the default semantics. While
2271: this introduces redundancy and is therefore usually a bad idea, a
2272: performance improvement may be worth the trouble. E.g., consider the
2273: word @code{foobar}:
2274:
2275: @example
2276: : foobar
2277: foo bar ;
2278: @end example
2279:
2280: Let us assume that @code{foobar} is called so frequently that the
2281: calling overhead would take a significant amount of the run-time. We can
2282: optimize it with @code{interpret/compile:}:
2283:
2284: @example
2285: :noname
2286: foo bar ;
2287: :noname
2288: POSTPONE foo POSTPONE bar ;
2289: interpret/compile: foobar
2290: @end example
2291:
2292: This definition has the same interpretation semantics and essentially
2293: the same compilation semantics as the simple definition of
2294: @code{foobar}, but the implementation of the compilation semantics is
2295: more efficient with respect to run-time.
2296:
2297: Some people try to use state-smart words to emulate the feature provided
2298: by @code{interpret/compile:} (words are state-smart if they check
2299: @code{STATE} during execution). E.g., they would try to code
2300: @code{foobar} like this:
2301:
2302: @example
2303: : foobar
2304: STATE @@
2305: IF ( compilation state )
2306: POSTPONE foo POSTPONE bar
2307: ELSE
2308: foo bar
2309: ENDIF ; immediate
2310: @end example
2311:
2312: While this works if @code{foobar} is processed only by the text
2313: interpreter, it does not work in other contexts (like @code{'} or
2314: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2315: for a state-smart word, not for the interpretation semantics of the
2316: original @code{foobar}; when you execute this execution token (directly
2317: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2318: state, the result will not be what you expected (i.e., it will not
2319: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2320: write them!
2321:
2322: It is also possible to write defining words that define words with
2323: arbitrary combinations of interpretation and compilation semantics (or,
2324: preferably, arbitrary combinations of implementations of the default
2325: semantics). In general, this looks like:
2326:
2327: @example
2328: : def-word
2329: create-interpret/compile
2330: @var{code1}
2331: interpretation>
2332: @var{code2}
2333: <interpretation
2334: compilation>
2335: @var{code3}
2336: <compilation ;
2337: @end example
2338:
2339: For a @var{word} defined with @code{def-word}, the interpretation
2340: semantics are to push the address of the body of @var{word} and perform
2341: @var{code2}, and the compilation semantics are to push the address of
2342: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2343: can also be defined like this:
2344:
2345: @example
2346: : constant ( n "name" -- )
2347: create-interpret/compile
2348: ,
2349: interpretation> ( -- n )
2350: @@
2351: <interpretation
2352: compilation> ( compilation. -- ; run-time. -- n )
2353: @@ postpone literal
2354: <compilation ;
2355: @end example
2356:
2357: doc-create-interpret/compile
2358: doc-interpretation>
2359: doc-<interpretation
2360: doc-compilation>
2361: doc-<compilation
2362:
2363: Note that words defined with @code{interpret/compile:} and
2364: @code{create-interpret/compile} have an extended header structure that
2365: differs from other words; however, unless you try to access them with
2366: plain address arithmetic, you should not notice this. Words for
2367: accessing the header structure usually know how to deal with this; e.g.,
2368: @code{' word >body} also gives you the body of a word created with
2369: @code{create-interpret/compile}.
2370:
2371: @node Tokens for Words, Wordlists, Defining Words, Words
2372: @section Tokens for Words
2373:
2374: This chapter describes the creation and use of tokens that represent
2375: words on the stack (and in data space).
2376:
2377: Named words have interpretation and compilation semantics. Unnamed words
2378: just have execution semantics.
2379:
2380: An @dfn{execution token} represents the execution semantics of an
2381: unnamed word. An execution token occupies one cell. As explained in
2382: section @ref{Supplying names}, the execution token of the last words
2383: defined can be produced with
2384:
2385: short-lastxt
2386:
2387: You can perform the semantics represented by an execution token with
2388: doc-execute
2389: You can compile the word with
2390: doc-compile,
2391:
2392: In Gforth, the abstract data type @emph{execution token} is implemented
2393: as CFA (code field address).
2394:
2395: The interpretation semantics of a named word are also represented by an
2396: execution token. You can get it with
2397:
2398: doc-[']
2399: doc-'
2400:
2401: For literals, you use @code{'} in interpreted code and @code{[']} in
2402: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
2403: by complaining about compile-only words. To get an execution token for a
2404: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
2405: @var{X} drop}.
2406:
2407: The compilation semantics are represented by a @dfn{compilation token}
2408: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
2409: execution token. The compilation semantics represented by the
2410: compilation token can be performed with @code{execute}, which consumes
2411: the whole compilation token, with an additional stack effect determined
2412: by the represented compilation semantics.
2413:
2414: doc-[comp']
2415: doc-comp'
2416:
2417: At present, the @var{w} part of a compilation token is an execution
2418: token, and the @var{xt} part represents either @code{execute} or
2419: @code{compile,}. However, don't rely on that kowledge, unless necessary;
2420: we may introduce unusual compilation tokens in the future (e.g.,
2421: compilation tokens representing the compilation semantics of literals).
2422:
2423: Named words are also represented by the @dfn{name token}. The abstract
2424: data type @emph{name token} is implemented as NFA (name field address).
2425:
2426: doc-find-name
2427: doc-name>int
2428: doc-name?int
2429: doc-name>comp
2430: doc-name>string
2431:
2432: @node Wordlists, Files, Tokens for Words, Words
2433: @section Wordlists
2434:
2435: @node Files, Blocks, Wordlists, Words
2436: @section Files
2437:
2438: @node Blocks, Other I/O, Files, Words
2439: @section Blocks
2440:
2441: @node Other I/O, Programming Tools, Blocks, Words
2442: @section Other I/O
2443:
2444: @node Programming Tools, Assembler and Code words, Other I/O, Words
2445: @section Programming Tools
2446:
2447: @menu
2448: * Debugging:: Simple and quick.
2449: * Assertions:: Making your programs self-checking.
2450: @end menu
2451:
2452: @node Debugging, Assertions, Programming Tools, Programming Tools
2453: @subsection Debugging
2454:
2455: The simple debugging aids provided in @file{debugging.fs}
2456: are meant to support a different style of debugging than the
2457: tracing/stepping debuggers used in languages with long turn-around
2458: times.
2459:
2460: A much better (faster) way in fast-compilig languages is to add
2461: printing code at well-selected places, let the program run, look at
2462: the output, see where things went wrong, add more printing code, etc.,
2463: until the bug is found.
2464:
2465: The word @code{~~} is easy to insert. It just prints debugging
2466: information (by default the source location and the stack contents). It
2467: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2468: query-replace them with nothing). The deferred words
2469: @code{printdebugdata} and @code{printdebugline} control the output of
2470: @code{~~}. The default source location output format works well with
2471: Emacs' compilation mode, so you can step through the program at the
2472: source level using @kbd{C-x `} (the advantage over a stepping debugger
2473: is that you can step in any direction and you know where the crash has
2474: happened or where the strange data has occurred).
2475:
2476: Note that the default actions clobber the contents of the pictured
2477: numeric output string, so you should not use @code{~~}, e.g., between
2478: @code{<#} and @code{#>}.
2479:
2480: doc-~~
2481: doc-printdebugdata
2482: doc-printdebugline
2483:
2484: @node Assertions, , Debugging, Programming Tools
2485: @subsection Assertions
2486:
2487: It is a good idea to make your programs self-checking, in particular, if
2488: you use an assumption (e.g., that a certain field of a data structure is
2489: never zero) that may become wrong during maintenance. Gforth supports
2490: assertions for this purpose. They are used like this:
2491:
2492: @example
2493: assert( @var{flag} )
2494: @end example
2495:
2496: The code between @code{assert(} and @code{)} should compute a flag, that
2497: should be true if everything is alright and false otherwise. It should
2498: not change anything else on the stack. The overall stack effect of the
2499: assertion is @code{( -- )}. E.g.
2500:
2501: @example
2502: assert( 1 1 + 2 = ) \ what we learn in school
2503: assert( dup 0<> ) \ assert that the top of stack is not zero
2504: assert( false ) \ this code should not be reached
2505: @end example
2506:
2507: The need for assertions is different at different times. During
2508: debugging, we want more checking, in production we sometimes care more
2509: for speed. Therefore, assertions can be turned off, i.e., the assertion
2510: becomes a comment. Depending on the importance of an assertion and the
2511: time it takes to check it, you may want to turn off some assertions and
2512: keep others turned on. Gforth provides several levels of assertions for
2513: this purpose:
2514:
2515: doc-assert0(
2516: doc-assert1(
2517: doc-assert2(
2518: doc-assert3(
2519: doc-assert(
2520: doc-)
2521:
2522: @code{Assert(} is the same as @code{assert1(}. The variable
2523: @code{assert-level} specifies the highest assertions that are turned
2524: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2525: @code{assert1(} assertions perform checking, while @code{assert2(} and
2526: @code{assert3(} assertions are treated as comments.
2527:
2528: Note that the @code{assert-level} is evaluated at compile-time, not at
2529: run-time. I.e., you cannot turn assertions on or off at run-time, you
2530: have to set the @code{assert-level} appropriately before compiling a
2531: piece of code. You can compile several pieces of code at several
2532: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2533: written code at level 3).
2534:
2535: doc-assert-level
2536:
2537: If an assertion fails, a message compatible with Emacs' compilation mode
2538: is produced and the execution is aborted (currently with @code{ABORT"}.
2539: If there is interest, we will introduce a special throw code. But if you
2540: intend to @code{catch} a specific condition, using @code{throw} is
2541: probably more appropriate than an assertion).
2542:
2543: @node Assembler and Code words, Threading Words, Programming Tools, Words
2544: @section Assembler and Code words
2545:
2546: Gforth provides some words for defining primitives (words written in
2547: machine code), and for defining the the machine-code equivalent of
2548: @code{DOES>}-based defining words. However, the machine-independent
2549: nature of Gforth poses a few problems: First of all. Gforth runs on
2550: several architectures, so it can provide no standard assembler. What's
2551: worse is that the register allocation not only depends on the processor,
2552: but also on the @code{gcc} version and options used.
2553:
2554: The words that Gforth offers encapsulate some system dependences (e.g., the
2555: header structure), so a system-independent assembler may be used in
2556: Gforth. If you do not have an assembler, you can compile machine code
2557: directly with @code{,} and @code{c,}.
2558:
2559: doc-assembler
2560: doc-code
2561: doc-end-code
2562: doc-;code
2563: doc-flush-icache
2564:
2565: If @code{flush-icache} does not work correctly, @code{code} words
2566: etc. will not work (reliably), either.
2567:
2568: These words are rarely used. Therefore they reside in @code{code.fs},
2569: which is usually not loaded (except @code{flush-icache}, which is always
2570: present). You can load them with @code{require code.fs}.
2571:
2572: In the assembly code you will want to refer to the inner interpreter's
2573: registers (e.g., the data stack pointer) and you may want to use other
2574: registers for temporary storage. Unfortunately, the register allocation
2575: is installation-dependent.
2576:
2577: The easiest solution is to use explicit register declarations
2578: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2579: GNU C Manual}) for all of the inner interpreter's registers: You have to
2580: compile Gforth with @code{-DFORCE_REG} (configure option
2581: @code{--enable-force-reg}) and the appropriate declarations must be
2582: present in the @code{machine.h} file (see @code{mips.h} for an example;
2583: you can find a full list of all declarable register symbols with
2584: @code{grep register engine.c}). If you give explicit registers to all
2585: variables that are declared at the beginning of @code{engine()}, you
2586: should be able to use the other caller-saved registers for temporary
2587: storage. Alternatively, you can use the @code{gcc} option
2588: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2589: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2590: (however, this restriction on register allocation may slow Gforth
2591: significantly).
2592:
2593: If this solution is not viable (e.g., because @code{gcc} does not allow
2594: you to explicitly declare all the registers you need), you have to find
2595: out by looking at the code where the inner interpreter's registers
2596: reside and which registers can be used for temporary storage. You can
2597: get an assembly listing of the engine's code with @code{make engine.s}.
2598:
2599: In any case, it is good practice to abstract your assembly code from the
2600: actual register allocation. E.g., if the data stack pointer resides in
2601: register @code{$17}, create an alias for this register called @code{sp},
2602: and use that in your assembly code.
2603:
2604: Another option for implementing normal and defining words efficiently
2605: is: adding the wanted functionality to the source of Gforth. For normal
2606: words you just have to edit @file{primitives} (@pxref{Automatic
2607: Generation}), defining words (equivalent to @code{;CODE} words, for fast
2608: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
2609: @file{prims2x.fs}, and possibly @file{cross.fs}.
2610:
2611:
2612: @node Threading Words, , Assembler and Code words, Words
2613: @section Threading Words
2614:
2615: These words provide access to code addresses and other threading stuff
2616: in Gforth (and, possibly, other interpretive Forths). It more or less
2617: abstracts away the differences between direct and indirect threading
2618: (and, for direct threading, the machine dependences). However, at
2619: present this wordset is still inclomplete. It is also pretty low-level;
2620: some day it will hopefully be made unnecessary by an internals words set
2621: that abstracts implementation details away completely.
2622:
2623: doc->code-address
2624: doc->does-code
2625: doc-code-address!
2626: doc-does-code!
2627: doc-does-handler!
2628: doc-/does-handler
2629:
2630: The code addresses produced by various defining words are produced by
2631: the following words:
2632:
2633: doc-docol:
2634: doc-docon:
2635: doc-dovar:
2636: doc-douser:
2637: doc-dodefer:
2638: doc-dofield:
2639:
2640: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
2641: with @code{>DOES-CODE}. If the word was defined in that way, the value
2642: returned is different from 0 and identifies the @code{DOES>} used by the
2643: defining word.
2644:
2645: @node ANS conformance, Model, Words, Top
2646: @chapter ANS conformance
2647:
2648: To the best of our knowledge, Gforth is an
2649:
2650: ANS Forth System
2651: @itemize @bullet
2652: @item providing the Core Extensions word set
2653: @item providing the Block word set
2654: @item providing the Block Extensions word set
2655: @item providing the Double-Number word set
2656: @item providing the Double-Number Extensions word set
2657: @item providing the Exception word set
2658: @item providing the Exception Extensions word set
2659: @item providing the Facility word set
2660: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2661: @item providing the File Access word set
2662: @item providing the File Access Extensions word set
2663: @item providing the Floating-Point word set
2664: @item providing the Floating-Point Extensions word set
2665: @item providing the Locals word set
2666: @item providing the Locals Extensions word set
2667: @item providing the Memory-Allocation word set
2668: @item providing the Memory-Allocation Extensions word set (that one's easy)
2669: @item providing the Programming-Tools word set
2670: @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
2671: @item providing the Search-Order word set
2672: @item providing the Search-Order Extensions word set
2673: @item providing the String word set
2674: @item providing the String Extensions word set (another easy one)
2675: @end itemize
2676:
2677: In addition, ANS Forth systems are required to document certain
2678: implementation choices. This chapter tries to meet these
2679: requirements. In many cases it gives a way to ask the system for the
2680: information instead of providing the information directly, in
2681: particular, if the information depends on the processor, the operating
2682: system or the installation options chosen, or if they are likely to
2683: change during the maintenance of Gforth.
2684:
2685: @comment The framework for the rest has been taken from pfe.
2686:
2687: @menu
2688: * The Core Words::
2689: * The optional Block word set::
2690: * The optional Double Number word set::
2691: * The optional Exception word set::
2692: * The optional Facility word set::
2693: * The optional File-Access word set::
2694: * The optional Floating-Point word set::
2695: * The optional Locals word set::
2696: * The optional Memory-Allocation word set::
2697: * The optional Programming-Tools word set::
2698: * The optional Search-Order word set::
2699: @end menu
2700:
2701:
2702: @c =====================================================================
2703: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2704: @comment node-name, next, previous, up
2705: @section The Core Words
2706: @c =====================================================================
2707:
2708: @menu
2709: * core-idef:: Implementation Defined Options
2710: * core-ambcond:: Ambiguous Conditions
2711: * core-other:: Other System Documentation
2712: @end menu
2713:
2714: @c ---------------------------------------------------------------------
2715: @node core-idef, core-ambcond, The Core Words, The Core Words
2716: @subsection Implementation Defined Options
2717: @c ---------------------------------------------------------------------
2718:
2719: @table @i
2720:
2721: @item (Cell) aligned addresses:
2722: processor-dependent. Gforth's alignment words perform natural alignment
2723: (e.g., an address aligned for a datum of size 8 is divisible by
2724: 8). Unaligned accesses usually result in a @code{-23 THROW}.
2725:
2726: @item @code{EMIT} and non-graphic characters:
2727: The character is output using the C library function (actually, macro)
2728: @code{putc}.
2729:
2730: @item character editing of @code{ACCEPT} and @code{EXPECT}:
2731: This is modeled on the GNU readline library (@pxref{Readline
2732: Interaction, , Command Line Editing, readline, The GNU Readline
2733: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2734: producing a full word completion every time you type it (instead of
2735: producing the common prefix of all completions).
2736:
2737: @item character set:
2738: The character set of your computer and display device. Gforth is
2739: 8-bit-clean (but some other component in your system may make trouble).
2740:
2741: @item Character-aligned address requirements:
2742: installation-dependent. Currently a character is represented by a C
2743: @code{unsigned char}; in the future we might switch to @code{wchar_t}
2744: (Comments on that requested).
2745:
2746: @item character-set extensions and matching of names:
2747: Any character except the ASCII NUL charcter can be used in a
2748: name. Matching is case-insensitive (except in @code{TABLE}s. The
2749: matching is performed using the C function @code{strncasecmp}, whose
2750: function is probably influenced by the locale. E.g., the @code{C} locale
2751: does not know about accents and umlauts, so they are matched
2752: case-sensitively in that locale. For portability reasons it is best to
2753: write programs such that they work in the @code{C} locale. Then one can
2754: use libraries written by a Polish programmer (who might use words
2755: containing ISO Latin-2 encoded characters) and by a French programmer
2756: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
2757: funny results for some of the words (which ones, depends on the font you
2758: are using)). Also, the locale you prefer may not be available in other
2759: operating systems. Hopefully, Unicode will solve these problems one day.
2760:
2761: @item conditions under which control characters match a space delimiter:
2762: If @code{WORD} is called with the space character as a delimiter, all
2763: white-space characters (as identified by the C macro @code{isspace()})
2764: are delimiters. @code{PARSE}, on the other hand, treats space like other
2765: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2766: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2767: interpreter (aka text interpreter) by default, treats all white-space
2768: characters as delimiters.
2769:
2770: @item format of the control flow stack:
2771: The data stack is used as control flow stack. The size of a control flow
2772: stack item in cells is given by the constant @code{cs-item-size}. At the
2773: time of this writing, an item consists of a (pointer to a) locals list
2774: (third), an address in the code (second), and a tag for identifying the
2775: item (TOS). The following tags are used: @code{defstart},
2776: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2777: @code{scopestart}.
2778:
2779: @item conversion of digits > 35
2780: The characters @code{[\]^_'} are the digits with the decimal value
2781: 36@minus{}41. There is no way to input many of the larger digits.
2782:
2783: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2784: The cursor is moved to the end of the entered string. If the input is
2785: terminated using the @kbd{Return} key, a space is typed.
2786:
2787: @item exception abort sequence of @code{ABORT"}:
2788: The error string is stored into the variable @code{"error} and a
2789: @code{-2 throw} is performed.
2790:
2791: @item input line terminator:
2792: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
2793: lines. One of these characters is typically produced when you type the
2794: @kbd{Enter} or @kbd{Return} key.
2795:
2796: @item maximum size of a counted string:
2797: @code{s" /counted-string" environment? drop .}. Currently 255 characters
2798: on all ports, but this may change.
2799:
2800: @item maximum size of a parsed string:
2801: Given by the constant @code{/line}. Currently 255 characters.
2802:
2803: @item maximum size of a definition name, in characters:
2804: 31
2805:
2806: @item maximum string length for @code{ENVIRONMENT?}, in characters:
2807: 31
2808:
2809: @item method of selecting the user input device:
2810: The user input device is the standard input. There is currently no way to
2811: change it from within Gforth. However, the input can typically be
2812: redirected in the command line that starts Gforth.
2813:
2814: @item method of selecting the user output device:
2815: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
2816: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
2817: output, so output on a terminal does not become visible before the next
2818: newline or buffer overflow. Output on non-terminals is invisible until
2819: the buffer overflows.
2820:
2821: @item methods of dictionary compilation:
2822: What are we expected to document here?
2823:
2824: @item number of bits in one address unit:
2825: @code{s" address-units-bits" environment? drop .}. 8 in all current
2826: ports.
2827:
2828: @item number representation and arithmetic:
2829: Processor-dependent. Binary two's complement on all current ports.
2830:
2831: @item ranges for integer types:
2832: Installation-dependent. Make environmental queries for @code{MAX-N},
2833: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2834: unsigned (and positive) types is 0. The lower bound for signed types on
2835: two's complement and one's complement machines machines can be computed
2836: by adding 1 to the upper bound.
2837:
2838: @item read-only data space regions:
2839: The whole Forth data space is writable.
2840:
2841: @item size of buffer at @code{WORD}:
2842: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2843: shared with the pictured numeric output string. If overwriting
2844: @code{PAD} is acceptable, it is as large as the remaining dictionary
2845: space, although only as much can be sensibly used as fits in a counted
2846: string.
2847:
2848: @item size of one cell in address units:
2849: @code{1 cells .}.
2850:
2851: @item size of one character in address units:
2852: @code{1 chars .}. 1 on all current ports.
2853:
2854: @item size of the keyboard terminal buffer:
2855: Varies. You can determine the size at a specific time using @code{lp@@
2856: tib - .}. It is shared with the locals stack and TIBs of files that
2857: include the current file. You can change the amount of space for TIBs
2858: and locals stack at Gforth startup with the command line option
2859: @code{-l}.
2860:
2861: @item size of the pictured numeric output buffer:
2862: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2863: shared with @code{WORD}.
2864:
2865: @item size of the scratch area returned by @code{PAD}:
2866: The remainder of dictionary space. You can even use the unused part of
2867: the data stack space. The current size can be computed with @code{sp@@
2868: pad - .}.
2869:
2870: @item system case-sensitivity characteristics:
2871: Dictionary searches are case insensitive (except in
2872: @code{TABLE}s). However, as explained above under @i{character-set
2873: extensions}, the matching for non-ASCII characters is determined by the
2874: locale you are using. In the default @code{C} locale all non-ASCII
2875: characters are matched case-sensitively.
2876:
2877: @item system prompt:
2878: @code{ ok} in interpret state, @code{ compiled} in compile state.
2879:
2880: @item division rounding:
2881: installation dependent. @code{s" floored" environment? drop .}. We leave
2882: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
2883: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
2884:
2885: @item values of @code{STATE} when true:
2886: -1.
2887:
2888: @item values returned after arithmetic overflow:
2889: On two's complement machines, arithmetic is performed modulo
2890: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2891: arithmetic (with appropriate mapping for signed types). Division by zero
2892: typically results in a @code{-55 throw} (Floating-point unidentified
2893: fault), although a @code{-10 throw} (divide by zero) would be more
2894: appropriate.
2895:
2896: @item whether the current definition can be found after @t{DOES>}:
2897: No.
2898:
2899: @end table
2900:
2901: @c ---------------------------------------------------------------------
2902: @node core-ambcond, core-other, core-idef, The Core Words
2903: @subsection Ambiguous conditions
2904: @c ---------------------------------------------------------------------
2905:
2906: @table @i
2907:
2908: @item a name is neither a word nor a number:
2909: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
2910: preserves the data and FP stack, so you don't lose more work than
2911: necessary.
2912:
2913: @item a definition name exceeds the maximum length allowed:
2914: @code{-19 throw} (Word name too long)
2915:
2916: @item addressing a region not inside the various data spaces of the forth system:
2917: The stacks, code space and name space are accessible. Machine code space is
2918: typically readable. Accessing other addresses gives results dependent on
2919: the operating system. On decent systems: @code{-9 throw} (Invalid memory
2920: address).
2921:
2922: @item argument type incompatible with parameter:
2923: This is usually not caught. Some words perform checks, e.g., the control
2924: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2925: mismatch).
2926:
2927: @item attempting to obtain the execution token of a word with undefined execution semantics:
2928: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
2929: get an execution token for @code{compile-only-error} (which performs a
2930: @code{-14 throw} when executed).
2931:
2932: @item dividing by zero:
2933: typically results in a @code{-55 throw} (floating point unidentified
2934: fault), although a @code{-10 throw} (divide by zero) would be more
2935: appropriate.
2936:
2937: @item insufficient data stack or return stack space:
2938: Not checked. This typically results in mysterious illegal memory
2939: accesses, producing @code{-9 throw} (Invalid memory address) or
2940: @code{-23 throw} (Address alignment exception).
2941:
2942: @item insufficient space for loop control parameters:
2943: like other return stack overflows.
2944:
2945: @item insufficient space in the dictionary:
2946: Not checked. Similar results as stack overflows. However, typically the
2947: error appears at a different place when one inserts or removes code.
2948:
2949: @item interpreting a word with undefined interpretation semantics:
2950: For some words, we defined interpretation semantics. For the others:
2951: @code{-14 throw} (Interpreting a compile-only word).
2952:
2953: @item modifying the contents of the input buffer or a string literal:
2954: These are located in writable memory and can be modified.
2955:
2956: @item overflow of the pictured numeric output string:
2957: Not checked.
2958:
2959: @item parsed string overflow:
2960: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
2961:
2962: @item producing a result out of range:
2963: On two's complement machines, arithmetic is performed modulo
2964: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2965: arithmetic (with appropriate mapping for signed types). Division by zero
2966: typically results in a @code{-55 throw} (floatingpoint unidentified
2967: fault), although a @code{-10 throw} (divide by zero) would be more
2968: appropriate. @code{convert} and @code{>number} currently overflow
2969: silently.
2970:
2971: @item reading from an empty data or return stack:
2972: The data stack is checked by the outer (aka text) interpreter after
2973: every word executed. If it has underflowed, a @code{-4 throw} (Stack
2974: underflow) is performed. Apart from that, the stacks are not checked and
2975: underflows can result in similar behaviour as overflows (of adjacent
2976: stacks).
2977:
2978: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
2979: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
2980: use zero-length string as a name). Words like @code{'} probably will not
2981: find what they search. Note that it is possible to create zero-length
2982: names with @code{nextname} (should it not?).
2983:
2984: @item @code{>IN} greater than input buffer:
2985: The next invocation of a parsing word returns a string wih length 0.
2986:
2987: @item @code{RECURSE} appears after @code{DOES>}:
2988: Compiles a recursive call to the defining word, not to the defined word.
2989:
2990: @item argument input source different than current input source for @code{RESTORE-INPUT}:
2991: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
2992: the end of the file was reached), its source-id may be
2993: reused. Therefore, restoring an input source specification referencing a
2994: closed file may lead to unpredictable results instead of a @code{-12
2995: THROW}.
2996:
2997: In the future, Gforth may be able to restore input source specifications
2998: from other than the current input soruce.
2999:
3000: @item data space containing definitions gets de-allocated:
3001: Deallocation with @code{allot} is not checked. This typically resuls in
3002: memory access faults or execution of illegal instructions.
3003:
3004: @item data space read/write with incorrect alignment:
3005: Processor-dependent. Typically results in a @code{-23 throw} (Address
3006: alignment exception). Under Linux on a 486 or later processor with
3007: alignment turned on, incorrect alignment results in a @code{-9 throw}
3008: (Invalid memory address). There are reportedly some processors with
3009: alignment restrictions that do not report them.
3010:
3011: @item data space pointer not properly aligned, @code{,}, @code{C,}:
3012: Like other alignment errors.
3013:
3014: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
3015: Not checked. May cause an illegal memory access.
3016:
3017: @item loop control parameters not available:
3018: Not checked. The counted loop words simply assume that the top of return
3019: stack items are loop control parameters and behave accordingly.
3020:
3021: @item most recent definition does not have a name (@code{IMMEDIATE}):
3022: @code{abort" last word was headerless"}.
3023:
3024: @item name not defined by @code{VALUE} used by @code{TO}:
3025: @code{-32 throw} (Invalid name argument) (unless name was defined by
3026: @code{CONSTANT}; then it just changes the constant).
3027:
3028: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
3029: @code{-13 throw} (Undefined word)
3030:
3031: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
3032: Gforth behaves as if they were of the same type. I.e., you can predict
3033: the behaviour by interpreting all parameters as, e.g., signed.
3034:
3035: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
3036: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
3037: compilation semantics of @code{TO}.
3038:
3039: @item String longer than a counted string returned by @code{WORD}:
3040: Not checked. The string will be ok, but the count will, of course,
3041: contain only the least significant bits of the length.
3042:
3043: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
3044: Processor-dependent. Typical behaviours are returning 0 and using only
3045: the low bits of the shift count.
3046:
3047: @item word not defined via @code{CREATE}:
3048: @code{>BODY} produces the PFA of the word no matter how it was defined.
3049:
3050: @code{DOES>} changes the execution semantics of the last defined word no
3051: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
3052: @code{CREATE , DOES>}.
3053:
3054: @item words improperly used outside @code{<#} and @code{#>}:
3055: Not checked. As usual, you can expect memory faults.
3056:
3057: @end table
3058:
3059:
3060: @c ---------------------------------------------------------------------
3061: @node core-other, , core-ambcond, The Core Words
3062: @subsection Other system documentation
3063: @c ---------------------------------------------------------------------
3064:
3065: @table @i
3066:
3067: @item nonstandard words using @code{PAD}:
3068: None.
3069:
3070: @item operator's terminal facilities available:
3071: After processing the command line, Gforth goes into interactive mode,
3072: and you can give commands to Gforth interactively. The actual facilities
3073: available depend on how you invoke Gforth.
3074:
3075: @item program data space available:
3076: @code{sp@@ here - .} gives the space remaining for dictionary and data
3077: stack together.
3078:
3079: @item return stack space available:
3080: By default 16 KBytes. The default can be overridden with the @code{-r}
3081: switch (@pxref{Invocation}) when Gforth starts up.
3082:
3083: @item stack space available:
3084: @code{sp@@ here - .} gives the space remaining for dictionary and data
3085: stack together.
3086:
3087: @item system dictionary space required, in address units:
3088: Type @code{here forthstart - .} after startup. At the time of this
3089: writing, this gives 70108 (bytes) on a 32-bit system.
3090: @end table
3091:
3092:
3093: @c =====================================================================
3094: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
3095: @section The optional Block word set
3096: @c =====================================================================
3097:
3098: @menu
3099: * block-idef:: Implementation Defined Options
3100: * block-ambcond:: Ambiguous Conditions
3101: * block-other:: Other System Documentation
3102: @end menu
3103:
3104:
3105: @c ---------------------------------------------------------------------
3106: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
3107: @subsection Implementation Defined Options
3108: @c ---------------------------------------------------------------------
3109:
3110: @table @i
3111:
3112: @item the format for display by @code{LIST}:
3113: First the screen number is displayed, then 16 lines of 64 characters,
3114: each line preceded by the line number.
3115:
3116: @item the length of a line affected by @code{\}:
3117: 64 characters.
3118: @end table
3119:
3120:
3121: @c ---------------------------------------------------------------------
3122: @node block-ambcond, block-other, block-idef, The optional Block word set
3123: @subsection Ambiguous conditions
3124: @c ---------------------------------------------------------------------
3125:
3126: @table @i
3127:
3128: @item correct block read was not possible:
3129: Typically results in a @code{throw} of some OS-derived value (between
3130: -512 and -2048). If the blocks file was just not long enough, blanks are
3131: supplied for the missing portion.
3132:
3133: @item I/O exception in block transfer:
3134: Typically results in a @code{throw} of some OS-derived value (between
3135: -512 and -2048).
3136:
3137: @item invalid block number:
3138: @code{-35 throw} (Invalid block number)
3139:
3140: @item a program directly alters the contents of @code{BLK}:
3141: The input stream is switched to that other block, at the same
3142: position. If the storing to @code{BLK} happens when interpreting
3143: non-block input, the system will get quite confused when the block ends.
3144:
3145: @item no current block buffer for @code{UPDATE}:
3146: @code{UPDATE} has no effect.
3147:
3148: @end table
3149:
3150:
3151: @c ---------------------------------------------------------------------
3152: @node block-other, , block-ambcond, The optional Block word set
3153: @subsection Other system documentation
3154: @c ---------------------------------------------------------------------
3155:
3156: @table @i
3157:
3158: @item any restrictions a multiprogramming system places on the use of buffer addresses:
3159: No restrictions (yet).
3160:
3161: @item the number of blocks available for source and data:
3162: depends on your disk space.
3163:
3164: @end table
3165:
3166:
3167: @c =====================================================================
3168: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
3169: @section The optional Double Number word set
3170: @c =====================================================================
3171:
3172: @menu
3173: * double-ambcond:: Ambiguous Conditions
3174: @end menu
3175:
3176:
3177: @c ---------------------------------------------------------------------
3178: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
3179: @subsection Ambiguous conditions
3180: @c ---------------------------------------------------------------------
3181:
3182: @table @i
3183:
3184: @item @var{d} outside of range of @var{n} in @code{D>S}:
3185: The least significant cell of @var{d} is produced.
3186:
3187: @end table
3188:
3189:
3190: @c =====================================================================
3191: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
3192: @section The optional Exception word set
3193: @c =====================================================================
3194:
3195: @menu
3196: * exception-idef:: Implementation Defined Options
3197: @end menu
3198:
3199:
3200: @c ---------------------------------------------------------------------
3201: @node exception-idef, , The optional Exception word set, The optional Exception word set
3202: @subsection Implementation Defined Options
3203: @c ---------------------------------------------------------------------
3204:
3205: @table @i
3206: @item @code{THROW}-codes used in the system:
3207: The codes -256@minus{}-511 are used for reporting signals (see
3208: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
3209: (for file and memory allocation operations). The mapping from OS error
3210: numbers to throw code is -512@minus{}@code{errno}. One side effect of
3211: this mapping is that undefined OS errors produce a message with a
3212: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
3213: 488} on my system.
3214: @end table
3215:
3216: @c =====================================================================
3217: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
3218: @section The optional Facility word set
3219: @c =====================================================================
3220:
3221: @menu
3222: * facility-idef:: Implementation Defined Options
3223: * facility-ambcond:: Ambiguous Conditions
3224: @end menu
3225:
3226:
3227: @c ---------------------------------------------------------------------
3228: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
3229: @subsection Implementation Defined Options
3230: @c ---------------------------------------------------------------------
3231:
3232: @table @i
3233:
3234: @item encoding of keyboard events (@code{EKEY}):
3235: Not yet implemeted.
3236:
3237: @item duration of a system clock tick
3238: System dependent. With respect to @code{MS}, the time is specified in
3239: microseconds. How well the OS and the hardware implement this, is
3240: another question.
3241:
3242: @item repeatability to be expected from the execution of @code{MS}:
3243: System dependent. On Unix, a lot depends on load. If the system is
3244: lightly loaded, and the delay is short enough that Gforth does not get
3245: swapped out, the performance should be acceptable. Under MS-DOS and
3246: other single-tasking systems, it should be good.
3247:
3248: @end table
3249:
3250:
3251: @c ---------------------------------------------------------------------
3252: @node facility-ambcond, , facility-idef, The optional Facility word set
3253: @subsection Ambiguous conditions
3254: @c ---------------------------------------------------------------------
3255:
3256: @table @i
3257:
3258: @item @code{AT-XY} can't be performed on user output device:
3259: Largely terminal dependant. No range checks are done on the arguments.
3260: No errors are reported. You may see some garbage appearing, you may see
3261: simply nothing happen.
3262:
3263: @end table
3264:
3265:
3266: @c =====================================================================
3267: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
3268: @section The optional File-Access word set
3269: @c =====================================================================
3270:
3271: @menu
3272: * file-idef:: Implementation Defined Options
3273: * file-ambcond:: Ambiguous Conditions
3274: @end menu
3275:
3276:
3277: @c ---------------------------------------------------------------------
3278: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
3279: @subsection Implementation Defined Options
3280: @c ---------------------------------------------------------------------
3281:
3282: @table @i
3283:
3284: @item File access methods used:
3285: @code{R/O}, @code{R/W} and @code{BIN} work as you would
3286: expect. @code{W/O} translates into the C file opening mode @code{w} (or
3287: @code{wb}): The file is cleared, if it exists, and created, if it does
3288: not (both with @code{open-file} and @code{create-file}). Under Unix
3289: @code{create-file} creates a file with 666 permissions modified by your
3290: umask.
3291:
3292: @item file exceptions:
3293: The file words do not raise exceptions (except, perhaps, memory access
3294: faults when you pass illegal addresses or file-ids).
3295:
3296: @item file line terminator:
3297: System-dependent. Gforth uses C's newline character as line
3298: terminator. What the actual character code(s) of this are is
3299: system-dependent.
3300:
3301: @item file name format
3302: System dependent. Gforth just uses the file name format of your OS.
3303:
3304: @item information returned by @code{FILE-STATUS}:
3305: @code{FILE-STATUS} returns the most powerful file access mode allowed
3306: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
3307: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
3308: along with the retured mode.
3309:
3310: @item input file state after an exception when including source:
3311: All files that are left via the exception are closed.
3312:
3313: @item @var{ior} values and meaning:
3314: The @var{ior}s returned by the file and memory allocation words are
3315: intended as throw codes. They typically are in the range
3316: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3317: @var{ior}s is -512@minus{}@var{errno}.
3318:
3319: @item maximum depth of file input nesting:
3320: limited by the amount of return stack, locals/TIB stack, and the number
3321: of open files available. This should not give you troubles.
3322:
3323: @item maximum size of input line:
3324: @code{/line}. Currently 255.
3325:
3326: @item methods of mapping block ranges to files:
3327: By default, blocks are accessed in the file @file{blocks.fb} in the
3328: current working directory. The file can be switched with @code{USE}.
3329:
3330: @item number of string buffers provided by @code{S"}:
3331: 1
3332:
3333: @item size of string buffer used by @code{S"}:
3334: @code{/line}. currently 255.
3335:
3336: @end table
3337:
3338: @c ---------------------------------------------------------------------
3339: @node file-ambcond, , file-idef, The optional File-Access word set
3340: @subsection Ambiguous conditions
3341: @c ---------------------------------------------------------------------
3342:
3343: @table @i
3344:
3345: @item attempting to position a file outside it's boundaries:
3346: @code{REPOSITION-FILE} is performed as usual: Afterwards,
3347: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
3348:
3349: @item attempting to read from file positions not yet written:
3350: End-of-file, i.e., zero characters are read and no error is reported.
3351:
3352: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
3353: An appropriate exception may be thrown, but a memory fault or other
3354: problem is more probable.
3355:
3356: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
3357: The @var{ior} produced by the operation, that discovered the problem, is
3358: thrown.
3359:
3360: @item named file cannot be opened (@code{included}):
3361: The @var{ior} produced by @code{open-file} is thrown.
3362:
3363: @item requesting an unmapped block number:
3364: There are no unmapped legal block numbers. On some operating systems,
3365: writing a block with a large number may overflow the file system and
3366: have an error message as consequence.
3367:
3368: @item using @code{source-id} when @code{blk} is non-zero:
3369: @code{source-id} performs its function. Typically it will give the id of
3370: the source which loaded the block. (Better ideas?)
3371:
3372: @end table
3373:
3374:
3375: @c =====================================================================
3376: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
3377: @section The optional Floating-Point word set
3378: @c =====================================================================
3379:
3380: @menu
3381: * floating-idef:: Implementation Defined Options
3382: * floating-ambcond:: Ambiguous Conditions
3383: @end menu
3384:
3385:
3386: @c ---------------------------------------------------------------------
3387: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
3388: @subsection Implementation Defined Options
3389: @c ---------------------------------------------------------------------
3390:
3391: @table @i
3392:
3393: @item format and range of floating point numbers:
3394: System-dependent; the @code{double} type of C.
3395:
3396: @item results of @code{REPRESENT} when @var{float} is out of range:
3397: System dependent; @code{REPRESENT} is implemented using the C library
3398: function @code{ecvt()} and inherits its behaviour in this respect.
3399:
3400: @item rounding or truncation of floating-point numbers:
3401: System dependent; the rounding behaviour is inherited from the hosting C
3402: compiler. IEEE-FP-based (i.e., most) systems by default round to
3403: nearest, and break ties by rounding to even (i.e., such that the last
3404: bit of the mantissa is 0).
3405:
3406: @item size of floating-point stack:
3407: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
3408: with the command-line option @code{-f}.
3409:
3410: @item width of floating-point stack:
3411: @code{1 floats}.
3412:
3413: @end table
3414:
3415:
3416: @c ---------------------------------------------------------------------
3417: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
3418: @subsection Ambiguous conditions
3419: @c ---------------------------------------------------------------------
3420:
3421: @table @i
3422:
3423: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
3424: System-dependent. Typically results in a @code{-23 THROW} like other
3425: alignment violations.
3426:
3427: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
3428: System-dependent. Typically results in a @code{-23 THROW} like other
3429: alignment violations.
3430:
3431: @item Floating-point result out of range:
3432: System-dependent. Can result in a @code{-55 THROW} (Floating-point
3433: unidentified fault), or can produce a special value representing, e.g.,
3434: Infinity.
3435:
3436: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
3437: System-dependent. Typically results in an alignment fault like other
3438: alignment violations.
3439:
3440: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
3441: The floating-point number is converted into decimal nonetheless.
3442:
3443: @item Both arguments are equal to zero (@code{FATAN2}):
3444: System-dependent. @code{FATAN2} is implemented using the C library
3445: function @code{atan2()}.
3446:
3447: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
3448: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
3449: because of small errors and the tan will be a very large (or very small)
3450: but finite number.
3451:
3452: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
3453: The result is rounded to the nearest float.
3454:
3455: @item dividing by zero:
3456: @code{-55 throw} (Floating-point unidentified fault)
3457:
3458: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
3459: System dependent. On IEEE-FP based systems the number is converted into
3460: an infinity.
3461:
3462: @item @var{float}<1 (@code{facosh}):
3463: @code{-55 throw} (Floating-point unidentified fault)
3464:
3465: @item @var{float}=<-1 (@code{flnp1}):
3466: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3467: negative infinity is typically produced for @var{float}=-1.
3468:
3469: @item @var{float}=<0 (@code{fln}, @code{flog}):
3470: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3471: negative infinity is typically produced for @var{float}=0.
3472:
3473: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
3474: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
3475: produces values for these inputs on my Linux box (Bug in the C library?)
3476:
3477: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
3478: @code{-55 throw} (Floating-point unidentified fault).
3479:
3480: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
3481: @code{-55 throw} (Floating-point unidentified fault).
3482:
3483: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
3484: This does not happen.
3485: @end table
3486:
3487:
3488:
3489: @c =====================================================================
3490: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
3491: @section The optional Locals word set
3492: @c =====================================================================
3493:
3494: @menu
3495: * locals-idef:: Implementation Defined Options
3496: * locals-ambcond:: Ambiguous Conditions
3497: @end menu
3498:
3499:
3500: @c ---------------------------------------------------------------------
3501: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
3502: @subsection Implementation Defined Options
3503: @c ---------------------------------------------------------------------
3504:
3505: @table @i
3506:
3507: @item maximum number of locals in a definition:
3508: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
3509: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
3510: characters. The number of locals in a definition is bounded by the size
3511: of locals-buffer, which contains the names of the locals.
3512:
3513: @end table
3514:
3515:
3516: @c ---------------------------------------------------------------------
3517: @node locals-ambcond, , locals-idef, The optional Locals word set
3518: @subsection Ambiguous conditions
3519: @c ---------------------------------------------------------------------
3520:
3521: @table @i
3522:
3523: @item executing a named local in interpretation state:
3524: @code{-14 throw} (Interpreting a compile-only word).
3525:
3526: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3527: @code{-32 throw} (Invalid name argument)
3528:
3529: @end table
3530:
3531:
3532: @c =====================================================================
3533: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3534: @section The optional Memory-Allocation word set
3535: @c =====================================================================
3536:
3537: @menu
3538: * memory-idef:: Implementation Defined Options
3539: @end menu
3540:
3541:
3542: @c ---------------------------------------------------------------------
3543: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
3544: @subsection Implementation Defined Options
3545: @c ---------------------------------------------------------------------
3546:
3547: @table @i
3548:
3549: @item values and meaning of @var{ior}:
3550: The @var{ior}s returned by the file and memory allocation words are
3551: intended as throw codes. They typically are in the range
3552: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3553: @var{ior}s is -512@minus{}@var{errno}.
3554:
3555: @end table
3556:
3557: @c =====================================================================
3558: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3559: @section The optional Programming-Tools word set
3560: @c =====================================================================
3561:
3562: @menu
3563: * programming-idef:: Implementation Defined Options
3564: * programming-ambcond:: Ambiguous Conditions
3565: @end menu
3566:
3567:
3568: @c ---------------------------------------------------------------------
3569: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
3570: @subsection Implementation Defined Options
3571: @c ---------------------------------------------------------------------
3572:
3573: @table @i
3574:
3575: @item ending sequence for input following @code{;code} and @code{code}:
3576: @code{end-code}
3577:
3578: @item manner of processing input following @code{;code} and @code{code}:
3579: The @code{assembler} vocabulary is pushed on the search order stack, and
3580: the input is processed by the text interpreter, (starting) in interpret
3581: state.
3582:
3583: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3584: The ANS Forth search order word set.
3585:
3586: @item source and format of display by @code{SEE}:
3587: The source for @code{see} is the intermediate code used by the inner
3588: interpreter. The current @code{see} tries to output Forth source code
3589: as well as possible.
3590:
3591: @end table
3592:
3593: @c ---------------------------------------------------------------------
3594: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
3595: @subsection Ambiguous conditions
3596: @c ---------------------------------------------------------------------
3597:
3598: @table @i
3599:
3600: @item deleting the compilation wordlist (@code{FORGET}):
3601: Not implemented (yet).
3602:
3603: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3604: This typically results in an @code{abort"} with a descriptive error
3605: message (may change into a @code{-22 throw} (Control structure mismatch)
3606: in the future). You may also get a memory access error. If you are
3607: unlucky, this ambiguous condition is not caught.
3608:
3609: @item @var{name} can't be found (@code{forget}):
3610: Not implemented (yet).
3611:
3612: @item @var{name} not defined via @code{CREATE}:
3613: @code{;code} behaves like @code{DOES>} in this respect, i.e., it changes
3614: the execution semantics of the last defined word no matter how it was
3615: defined.
3616:
3617: @item @code{POSTPONE} applied to @code{[IF]}:
3618: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3619: equivalent to @code{[IF]}.
3620:
3621: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3622: Continue in the same state of conditional compilation in the next outer
3623: input source. Currently there is no warning to the user about this.
3624:
3625: @item removing a needed definition (@code{FORGET}):
3626: Not implemented (yet).
3627:
3628: @end table
3629:
3630:
3631: @c =====================================================================
3632: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3633: @section The optional Search-Order word set
3634: @c =====================================================================
3635:
3636: @menu
3637: * search-idef:: Implementation Defined Options
3638: * search-ambcond:: Ambiguous Conditions
3639: @end menu
3640:
3641:
3642: @c ---------------------------------------------------------------------
3643: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
3644: @subsection Implementation Defined Options
3645: @c ---------------------------------------------------------------------
3646:
3647: @table @i
3648:
3649: @item maximum number of word lists in search order:
3650: @code{s" wordlists" environment? drop .}. Currently 16.
3651:
3652: @item minimum search order:
3653: @code{root root}.
3654:
3655: @end table
3656:
3657: @c ---------------------------------------------------------------------
3658: @node search-ambcond, , search-idef, The optional Search-Order word set
3659: @subsection Ambiguous conditions
3660: @c ---------------------------------------------------------------------
3661:
3662: @table @i
3663:
3664: @item changing the compilation wordlist (during compilation):
3665: The word is entered into the wordlist that was the compilation wordlist
3666: at the start of the definition. Any changes to the name field (e.g.,
3667: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
3668: are applied to the latest defined word (as reported by @code{last} or
3669: @code{lastxt}), if possible, irrespective of the compilation wordlist.
3670:
3671: @item search order empty (@code{previous}):
3672: @code{abort" Vocstack empty"}.
3673:
3674: @item too many word lists in search order (@code{also}):
3675: @code{abort" Vocstack full"}.
3676:
3677: @end table
3678:
3679: @node Model, Integrating Gforth, ANS conformance, Top
3680: @chapter Model
3681:
3682: This chapter has yet to be written. It will contain information, on
3683: which internal structures you can rely.
3684:
3685: @node Integrating Gforth, Emacs and Gforth, Model, Top
3686: @chapter Integrating Gforth into C programs
3687:
3688: This is not yet implemented.
3689:
3690: Several people like to use Forth as scripting language for applications
3691: that are otherwise written in C, C++, or some other language.
3692:
3693: The Forth system ATLAST provides facilities for embedding it into
3694: applications; unfortunately it has several disadvantages: most
3695: importantly, it is not based on ANS Forth, and it is apparently dead
3696: (i.e., not developed further and not supported). The facilities
3697: provided by Gforth in this area are inspired by ATLASTs facilities, so
3698: making the switch should not be hard.
3699:
3700: We also tried to design the interface such that it can easily be
3701: implemented by other Forth systems, so that we may one day arrive at a
3702: standardized interface. Such a standard interface would allow you to
3703: replace the Forth system without having to rewrite C code.
3704:
3705: You embed the Gforth interpreter by linking with the library
3706: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
3707: global symbols in this library that belong to the interface, have the
3708: prefix @code{forth_}. (Global symbols that are used internally have the
3709: prefix @code{gforth_}).
3710:
3711: You can include the declarations of Forth types and the functions and
3712: variables of the interface with @code{#include <forth.h>}.
3713:
3714: Types.
3715:
3716: Variables.
3717:
3718: Data and FP Stack pointer. Area sizes.
3719:
3720: functions.
3721:
3722: forth_init(imagefile)
3723: forth_evaluate(string) exceptions?
3724: forth_goto(address) (or forth_execute(xt)?)
3725: forth_continue() (a corountining mechanism)
3726:
3727: Adding primitives.
3728:
3729: No checking.
3730:
3731: Signals?
3732:
3733: Accessing the Stacks
3734:
3735: @node Emacs and Gforth, Internals, Integrating Gforth, Top
3736: @chapter Emacs and Gforth
3737:
3738: Gforth comes with @file{gforth.el}, an improved version of
3739: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
3740: improvements are a better (but still not perfect) handling of
3741: indentation. I have also added comment paragraph filling (@kbd{M-q}),
3742: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3743: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3744: stuff I do not use alone, even though some of it only makes sense for
3745: TILE. To get a description of these features, enter Forth mode and type
3746: @kbd{C-h m}.
3747:
3748: In addition, Gforth supports Emacs quite well: The source code locations
3749: given in error messages, debugging output (from @code{~~}) and failed
3750: assertion messages are in the right format for Emacs' compilation mode
3751: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3752: Manual}) so the source location corresponding to an error or other
3753: message is only a few keystrokes away (@kbd{C-x `} for the next error,
3754: @kbd{C-c C-c} for the error under the cursor).
3755:
3756: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3757: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3758: contains the definitions of all words defined afterwards. You can then
3759: find the source for a word using @kbd{M-.}. Note that emacs can use
3760: several tags files at the same time (e.g., one for the Gforth sources
3761: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
3762: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
3763: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
3764: @file{/usr/local/share/gforth/0.2.0/TAGS}).
3765:
3766: To get all these benefits, add the following lines to your @file{.emacs}
3767: file:
3768:
3769: @example
3770: (autoload 'forth-mode "gforth.el")
3771: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3772: @end example
3773:
3774: @node Internals, Bugs, Emacs and Gforth, Top
3775: @chapter Internals
3776:
3777: Reading this section is not necessary for programming with Gforth. It
3778: should be helpful for finding your way in the Gforth sources.
3779:
3780: The ideas in this section have also been published in the papers
3781: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
3782: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
3783: Ertl, presented at EuroForth '93; the latter is available at
3784: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
3785:
3786: @menu
3787: * Portability::
3788: * Threading::
3789: * Primitives::
3790: * System Architecture::
3791: * Performance::
3792: @end menu
3793:
3794: @node Portability, Threading, Internals, Internals
3795: @section Portability
3796:
3797: One of the main goals of the effort is availability across a wide range
3798: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3799: this goal by manually coding the engine in assembly language for several
3800: then-popular processors. This approach is very labor-intensive and the
3801: results are short-lived due to progress in computer architecture.
3802:
3803: Others have avoided this problem by coding in C, e.g., Mitch Bradley
3804: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3805: particularly popular for UNIX-based Forths due to the large variety of
3806: architectures of UNIX machines. Unfortunately an implementation in C
3807: does not mix well with the goals of efficiency and with using
3808: traditional techniques: Indirect or direct threading cannot be expressed
3809: in C, and switch threading, the fastest technique available in C, is
3810: significantly slower. Another problem with C is that it's very
3811: cumbersome to express double integer arithmetic.
3812:
3813: Fortunately, there is a portable language that does not have these
3814: limitations: GNU C, the version of C processed by the GNU C compiler
3815: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3816: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3817: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3818: threading possible, its @code{long long} type (@pxref{Long Long, ,
3819: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
3820: double numbers@footnote{Unfortunately, long longs are not implemented
3821: properly on all machines (e.g., on alpha-osf1, long longs are only 64
3822: bits, the same size as longs (and pointers), but they should be twice as
3823: long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
3824: C Manual}). So, we had to implement doubles in C after all. Still, on
3825: most machines we can use long longs and achieve better performance than
3826: with the emulation package.}. GNU C is available for free on all
3827: important (and many unimportant) UNIX machines, VMS, 80386s running
3828: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
3829: on all these machines.
3830:
3831: Writing in a portable language has the reputation of producing code that
3832: is slower than assembly. For our Forth engine we repeatedly looked at
3833: the code produced by the compiler and eliminated most compiler-induced
3834: inefficiencies by appropriate changes in the source-code.
3835:
3836: However, register allocation cannot be portably influenced by the
3837: programmer, leading to some inefficiencies on register-starved
3838: machines. We use explicit register declarations (@pxref{Explicit Reg
3839: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3840: improve the speed on some machines. They are turned on by using the
3841: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3842: only depends on the machine, but also on the compiler version: On some
3843: machines some compiler versions produce incorrect code when certain
3844: explicit register declarations are used. So by default
3845: @code{-DFORCE_REG} is not used.
3846:
3847: @node Threading, Primitives, Portability, Internals
3848: @section Threading
3849:
3850: GNU C's labels as values extension (available since @code{gcc-2.0},
3851: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3852: makes it possible to take the address of @var{label} by writing
3853: @code{&&@var{label}}. This address can then be used in a statement like
3854: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
3855: @code{goto x}.
3856:
3857: With this feature an indirect threaded NEXT looks like:
3858: @example
3859: cfa = *ip++;
3860: ca = *cfa;
3861: goto *ca;
3862: @end example
3863: For those unfamiliar with the names: @code{ip} is the Forth instruction
3864: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3865: execution token and points to the code field of the next word to be
3866: executed; The @code{ca} (code address) fetched from there points to some
3867: executable code, e.g., a primitive or the colon definition handler
3868: @code{docol}.
3869:
3870: Direct threading is even simpler:
3871: @example
3872: ca = *ip++;
3873: goto *ca;
3874: @end example
3875:
3876: Of course we have packaged the whole thing neatly in macros called
3877: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3878:
3879: @menu
3880: * Scheduling::
3881: * Direct or Indirect Threaded?::
3882: * DOES>::
3883: @end menu
3884:
3885: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
3886: @subsection Scheduling
3887:
3888: There is a little complication: Pipelined and superscalar processors,
3889: i.e., RISC and some modern CISC machines can process independent
3890: instructions while waiting for the results of an instruction. The
3891: compiler usually reorders (schedules) the instructions in a way that
3892: achieves good usage of these delay slots. However, on our first tries
3893: the compiler did not do well on scheduling primitives. E.g., for
3894: @code{+} implemented as
3895: @example
3896: n=sp[0]+sp[1];
3897: sp++;
3898: sp[0]=n;
3899: NEXT;
3900: @end example
3901: the NEXT comes strictly after the other code, i.e., there is nearly no
3902: scheduling. After a little thought the problem becomes clear: The
3903: compiler cannot know that sp and ip point to different addresses (and
3904: the version of @code{gcc} we used would not know it even if it was
3905: possible), so it could not move the load of the cfa above the store to
3906: the TOS. Indeed the pointers could be the same, if code on or very near
3907: the top of stack were executed. In the interest of speed we chose to
3908: forbid this probably unused ``feature'' and helped the compiler in
3909: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3910: the goto part (@code{NEXT_P2}). @code{+} now looks like:
3911: @example
3912: n=sp[0]+sp[1];
3913: sp++;
3914: NEXT_P1;
3915: sp[0]=n;
3916: NEXT_P2;
3917: @end example
3918: This can be scheduled optimally by the compiler.
3919:
3920: This division can be turned off with the switch @code{-DCISC_NEXT}. This
3921: switch is on by default on machines that do not profit from scheduling
3922: (e.g., the 80386), in order to preserve registers.
3923:
3924: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
3925: @subsection Direct or Indirect Threaded?
3926:
3927: Both! After packaging the nasty details in macro definitions we
3928: realized that we could switch between direct and indirect threading by
3929: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3930: defining a few machine-specific macros for the direct-threading case.
3931: On the Forth level we also offer access words that hide the
3932: differences between the threading methods (@pxref{Threading Words}).
3933:
3934: Indirect threading is implemented completely
3935: machine-independently. Direct threading needs routines for creating
3936: jumps to the executable code (e.g. to docol or dodoes). These routines
3937: are inherently machine-dependent, but they do not amount to many source
3938: lines. I.e., even porting direct threading to a new machine is a small
3939: effort.
3940:
3941: @node DOES>, , Direct or Indirect Threaded?, Threading
3942: @subsection DOES>
3943: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
3944: the chunk of code executed by every word defined by a
3945: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
3946: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
3947: DOES-code)? There are two solutions:
3948:
3949: In fig-Forth the code field points directly to the dodoes and the
3950: DOES-code address is stored in the cell after the code address
3951: (i.e. at cfa cell+). It may seem that this solution is illegal in the
3952: Forth-79 and all later standards, because in fig-Forth this address
3953: lies in the body (which is illegal in these standards). However, by
3954: making the code field larger for all words this solution becomes legal
3955: again. We use this approach for the indirect threaded version. Leaving
3956: a cell unused in most words is a bit wasteful, but on the machines we
3957: are targetting this is hardly a problem. The other reason for having a
3958: code field size of two cells is to avoid having different image files
3959: for direct and indirect threaded systems (@pxref{System Architecture}).
3960:
3961: The other approach is that the code field points or jumps to the cell
3962: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
3963: this address. @code{dodoes} can then get the DOES-code address by
3964: computing the code address, i.e., the address of the jump to dodoes,
3965: and add the length of that jump field. A variant of this is to have a
3966: call to @code{dodoes} after the @code{DOES>}; then the return address
3967: (which can be found in the return register on RISCs) is the DOES-code
3968: address. Since the two cells available in the code field are usually
3969: used up by the jump to the code address in direct threading, we use
3970: this approach for direct threading. We did not want to add another
3971: cell to the code field.
3972:
3973: @node Primitives, System Architecture, Threading, Internals
3974: @section Primitives
3975:
3976: @menu
3977: * Automatic Generation::
3978: * TOS Optimization::
3979: * Produced code::
3980: @end menu
3981:
3982: @node Automatic Generation, TOS Optimization, Primitives, Primitives
3983: @subsection Automatic Generation
3984:
3985: Since the primitives are implemented in a portable language, there is no
3986: longer any need to minimize the number of primitives. On the contrary,
3987: having many primitives is an advantage: speed. In order to reduce the
3988: number of errors in primitives and to make programming them easier, we
3989: provide a tool, the primitive generator (@file{prims2x.fs}), that
3990: automatically generates most (and sometimes all) of the C code for a
3991: primitive from the stack effect notation. The source for a primitive
3992: has the following form:
3993:
3994: @format
3995: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
3996: [@code{""}@var{glossary entry}@code{""}]
3997: @var{C code}
3998: [@code{:}
3999: @var{Forth code}]
4000: @end format
4001:
4002: The items in brackets are optional. The category and glossary fields
4003: are there for generating the documentation, the Forth code is there
4004: for manual implementations on machines without GNU C. E.g., the source
4005: for the primitive @code{+} is:
4006: @example
4007: + n1 n2 -- n core plus
4008: n = n1+n2;
4009: @end example
4010:
4011: This looks like a specification, but in fact @code{n = n1+n2} is C
4012: code. Our primitive generation tool extracts a lot of information from
4013: the stack effect notations@footnote{We use a one-stack notation, even
4014: though we have separate data and floating-point stacks; The separate
4015: notation can be generated easily from the unified notation.}: The number
4016: of items popped from and pushed on the stack, their type, and by what
4017: name they are referred to in the C code. It then generates a C code
4018: prelude and postlude for each primitive. The final C code for @code{+}
4019: looks like this:
4020:
4021: @example
4022: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
4023: /* */ /* documentation */
4024: @{
4025: DEF_CA /* definition of variable ca (indirect threading) */
4026: Cell n1; /* definitions of variables */
4027: Cell n2;
4028: Cell n;
4029: n1 = (Cell) sp[1]; /* input */
4030: n2 = (Cell) TOS;
4031: sp += 1; /* stack adjustment */
4032: NAME("+") /* debugging output (with -DDEBUG) */
4033: @{
4034: n = n1+n2; /* C code taken from the source */
4035: @}
4036: NEXT_P1; /* NEXT part 1 */
4037: TOS = (Cell)n; /* output */
4038: NEXT_P2; /* NEXT part 2 */
4039: @}
4040: @end example
4041:
4042: This looks long and inefficient, but the GNU C compiler optimizes quite
4043: well and produces optimal code for @code{+} on, e.g., the R3000 and the
4044: HP RISC machines: Defining the @code{n}s does not produce any code, and
4045: using them as intermediate storage also adds no cost.
4046:
4047: There are also other optimizations, that are not illustrated by this
4048: example: Assignments between simple variables are usually for free (copy
4049: propagation). If one of the stack items is not used by the primitive
4050: (e.g. in @code{drop}), the compiler eliminates the load from the stack
4051: (dead code elimination). On the other hand, there are some things that
4052: the compiler does not do, therefore they are performed by
4053: @file{prims2x.fs}: The compiler does not optimize code away that stores
4054: a stack item to the place where it just came from (e.g., @code{over}).
4055:
4056: While programming a primitive is usually easy, there are a few cases
4057: where the programmer has to take the actions of the generator into
4058: account, most notably @code{?dup}, but also words that do not (always)
4059: fall through to NEXT.
4060:
4061: @node TOS Optimization, Produced code, Automatic Generation, Primitives
4062: @subsection TOS Optimization
4063:
4064: An important optimization for stack machine emulators, e.g., Forth
4065: engines, is keeping one or more of the top stack items in
4066: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
4067: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
4068: @itemize @bullet
4069: @item
4070: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
4071: due to fewer loads from and stores to the stack.
4072: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
4073: @var{y<n}, due to additional moves between registers.
4074: @end itemize
4075:
4076: In particular, keeping one item in a register is never a disadvantage,
4077: if there are enough registers. Keeping two items in registers is a
4078: disadvantage for frequent words like @code{?branch}, constants,
4079: variables, literals and @code{i}. Therefore our generator only produces
4080: code that keeps zero or one items in registers. The generated C code
4081: covers both cases; the selection between these alternatives is made at
4082: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
4083: code for @code{+} is just a simple variable name in the one-item case,
4084: otherwise it is a macro that expands into @code{sp[0]}. Note that the
4085: GNU C compiler tries to keep simple variables like @code{TOS} in
4086: registers, and it usually succeeds, if there are enough registers.
4087:
4088: The primitive generator performs the TOS optimization for the
4089: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
4090: operations the benefit of this optimization is even larger:
4091: floating-point operations take quite long on most processors, but can be
4092: performed in parallel with other operations as long as their results are
4093: not used. If the FP-TOS is kept in a register, this works. If
4094: it is kept on the stack, i.e., in memory, the store into memory has to
4095: wait for the result of the floating-point operation, lengthening the
4096: execution time of the primitive considerably.
4097:
4098: The TOS optimization makes the automatic generation of primitives a
4099: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
4100: @code{TOS} is not sufficient. There are some special cases to
4101: consider:
4102: @itemize @bullet
4103: @item In the case of @code{dup ( w -- w w )} the generator must not
4104: eliminate the store to the original location of the item on the stack,
4105: if the TOS optimization is turned on.
4106: @item Primitives with stack effects of the form @code{--}
4107: @var{out1}...@var{outy} must store the TOS to the stack at the start.
4108: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
4109: must load the TOS from the stack at the end. But for the null stack
4110: effect @code{--} no stores or loads should be generated.
4111: @end itemize
4112:
4113: @node Produced code, , TOS Optimization, Primitives
4114: @subsection Produced code
4115:
4116: To see what assembly code is produced for the primitives on your machine
4117: with your compiler and your flag settings, type @code{make engine.s} and
4118: look at the resulting file @file{engine.s}.
4119:
4120: @node System Architecture, Performance, Primitives, Internals
4121: @section System Architecture
4122:
4123: Our Forth system consists not only of primitives, but also of
4124: definitions written in Forth. Since the Forth compiler itself belongs
4125: to those definitions, it is not possible to start the system with the
4126: primitives and the Forth source alone. Therefore we provide the Forth
4127: code as an image file in nearly executable form. At the start of the
4128: system a C routine loads the image file into memory, sets up the
4129: memory (stacks etc.) according to information in the image file, and
4130: starts executing Forth code.
4131:
4132: The image file format is a compromise between the goals of making it
4133: easy to generate image files and making them portable. The easiest way
4134: to generate an image file is to just generate a memory dump. However,
4135: this kind of image file cannot be used on a different machine, or on
4136: the next version of the engine on the same machine, it even might not
4137: work with the same engine compiled by a different version of the C
4138: compiler. We would like to have as few versions of the image file as
4139: possible, because we do not want to distribute many versions of the
4140: same image file, and to make it easy for the users to use their image
4141: files on many machines. We currently need to create a different image
4142: file for machines with different cell sizes and different byte order
4143: (little- or big-endian)@footnote{We are considering adding information to the
4144: image file that enables the loader to change the byte order.}.
4145:
4146: Forth code that is going to end up in a portable image file has to
4147: comply to some restrictions: addresses have to be stored in memory with
4148: special words (@code{A!}, @code{A,}, etc.) in order to make the code
4149: relocatable. Cells, floats, etc., have to be stored at the natural
4150: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
4151: dividable by~8. This happens automatically in our system when you use
4152: the ANS Forth alignment words.}, in order to avoid alignment faults on
4153: machines with stricter alignment. The image file is produced by a
4154: metacompiler (@file{cross.fs}).
4155:
4156: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
4157: file is not directly executable, but has to undergo some manipulations
4158: during loading. Address relocation is performed at image load-time, not
4159: at run-time. The loader also has to replace tokens standing for
4160: primitive calls with the appropriate code-field addresses (or code
4161: addresses in the case of direct threading).
4162:
4163: @node Performance, , System Architecture, Internals
4164: @section Performance
4165:
4166: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
4167: impossible to write a significantly faster engine.
4168:
4169: On register-starved machines like the 386 architecture processors
4170: improvements are possible, because @code{gcc} does not utilize the
4171: registers as well as a human, even with explicit register declarations;
4172: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
4173: and hand-tuned it for the 486; this system is 1.19 times faster on the
4174: Sieve benchmark on a 486DX2/66 than Gforth compiled with
4175: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
4176:
4177: However, this potential advantage of assembly language implementations
4178: is not necessarily realized in complete Forth systems: We compared
4179: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
4180: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
4181: 1994) and Eforth (with and without peephole (aka pinhole) optimization
4182: of the threaded code); all these systems were written in assembly
4183: language. We also compared Gforth with three systems written in C:
4184: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
4185: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
4186: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
4187: -fomit-frame-pointer; ThisForth employs peephole optimization of the
4188: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
4189: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
4190: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
4191: 486DX2/66 with similar memory performance under Windows NT. Marcel
4192: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
4193: added the peephole optimizer, ran the benchmarks and reported the
4194: results.
4195:
4196: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
4197: matrix multiplication come from the Stanford integer benchmarks and have
4198: been translated into Forth by Martin Fraeman; we used the versions
4199: included in the TILE Forth package, but with bigger data set sizes; and
4200: a recursive Fibonacci number computation for benchmarking calling
4201: performance. The following table shows the time taken for the benchmarks
4202: scaled by the time taken by Gforth (in other words, it shows the speedup
4203: factor that Gforth achieved over the other systems).
4204:
4205: @example
4206: relative Win32- NT eforth This-
4207: time Gforth Forth Forth eforth +opt PFE Forth TILE
4208: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
4209: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
4210: matmul 1.00 1.47 1.35 1.46 1.16 1.58 4.09
4211: fib 1.00 1.52 1.34 1.22 1.13 1.74 2.99 4.30
4212: @end example
4213:
4214: You may find the good performance of Gforth compared with the systems
4215: written in assembly language quite surprising. One important reason for
4216: the disappointing performance of these systems is probably that they are
4217: not written optimally for the 486 (e.g., they use the @code{lods}
4218: instruction). In addition, Win32Forth uses a comfortable, but costly
4219: method for relocating the Forth image: like @code{cforth}, it computes
4220: the actual addresses at run time, resulting in two address computations
4221: per NEXT (@pxref{System Architecture}).
4222:
4223: Only Eforth with the peephole optimizer performs comparable to
4224: Gforth. The speedups achieved with peephole optimization of threaded
4225: code are quite remarkable. Adding a peephole optimizer to Gforth should
4226: cause similar speedups.
4227:
4228: The speedup of Gforth over PFE, ThisForth and TILE can be easily
4229: explained with the self-imposed restriction to standard C, which makes
4230: efficient threading impossible (however, the measured implementation of
4231: PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
4232: Register Variables, gcc.info, GNU C Manual}). Moreover, current C
4233: compilers have a hard time optimizing other aspects of the ThisForth
4234: and the TILE source.
4235:
4236: Note that the performance of Gforth on 386 architecture processors
4237: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
4238: failed to allocate any of the virtual machine registers into real
4239: machine registers by itself and would not work correctly with explicit
4240: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
4241: running the Sieve) than the one measured above.
4242:
4243: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
4244: Maierhofer (presented at EuroForth '95), an indirect threaded version of
4245: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
4246: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
4247: used here. The paper available at
4248: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
4249: it also contains numbers for some native code systems. You can find
4250: numbers for Gforth on various machines in @file{Benchres}.
4251:
4252: @node Bugs, Origin, Internals, Top
4253: @chapter Bugs
4254:
4255: Known bugs are described in the file BUGS in the Gforth distribution.
4256:
4257: If you find a bug, please send a bug report to
4258: @code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
4259: describe the Gforth version used (it is announced at the start of an
4260: interactive Gforth session), the machine and operating system (on Unix
4261: systems you can use @code{uname -a} to produce this information), the
4262: installation options (send the @code{config.status} file), and a
4263: complete list of changes you (or your installer) have made to the Gforth
4264: sources (if any); it should contain a program (or a sequence of keyboard
4265: commands) that reproduces the bug and a description of what you think
4266: constitutes the buggy behaviour.
4267:
4268: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
4269: to Report Bugs, gcc.info, GNU C Manual}.
4270:
4271:
4272: @node Origin, Word Index, Bugs, Top
4273: @chapter Authors and Ancestors of Gforth
4274:
4275: @section Authors and Contributors
4276:
4277: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
4278: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
4279: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
4280: with their continuous feedback. Lennart Benshop contributed
4281: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
4282: support for calling C libraries. Helpful comments also came from Paul
4283: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
4284: Wavrik, Stott Bolton and Marc de Groot.
4285:
4286: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
4287: and autoconf, among others), and to the creators of the Internet: Gforth
4288: was developed across the Internet, and its authors have not met
4289: physically yet.
4290:
4291: @section Pedigree
4292:
4293: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
4294: Dirk Zoller) will cross-fertilize each other. Of course, a significant
4295: part of the design of Gforth was prescribed by ANS Forth.
4296:
4297: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
4298: 32 bit native code version of VolksForth for the Atari ST, written
4299: mostly by Dietrich Weineck.
4300:
4301: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
4302: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
4303: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
4304:
4305: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
4306: Forth-83 standard. !! Pedigree? When?
4307:
4308: A team led by Bill Ragsdale implemented fig-Forth on many processors in
4309: 1979. Robert Selzer and Bill Ragsdale developed the original
4310: implementation of fig-Forth for the 6502 based on microForth.
4311:
4312: The principal architect of microForth was Dean Sanderson. microForth was
4313: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
4314: the 1802, and subsequently implemented on the 8080, the 6800 and the
4315: Z80.
4316:
4317: All earlier Forth systems were custom-made, usually by Charles Moore,
4318: who discovered (as he puts it) Forth during the late 60s. The first full
4319: Forth existed in 1971.
4320:
4321: A part of the information in this section comes from @cite{The Evolution
4322: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
4323: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
4324: Notices 28(3), 1993. You can find more historical and genealogical
4325: information about Forth there.
4326:
4327: @node Word Index, Node Index, Origin, Top
4328: @chapter Word Index
4329:
4330: This index is as incomplete as the manual. Each word is listed with
4331: stack effect and wordset.
4332:
4333: @printindex fn
4334:
4335: @node Node Index, , Word Index, Top
4336: @chapter Node Index
4337:
4338: This index is even less complete than the manual.
4339:
4340: @contents
4341: @bye
4342:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>