Annotation of gforth/gforth.ds, revision 1.48
1.1 anton 1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3: @comment %**start of header (This is for running Texinfo on a region.)
1.4 anton 4: @setfilename gforth.info
1.17 anton 5: @settitle Gforth Manual
1.48 ! anton 6: @dircategory GNU programming tools
! 7: @direntry
! 8: * Gforth: (gforth). A fast interpreter for the Forth language.
! 9: @end direntry
1.4 anton 10: @comment @setchapternewpage odd
1.1 anton 11: @comment %**end of header (This is for running Texinfo on a region.)
12:
13: @ifinfo
1.43 anton 14: This file documents Gforth 0.3
1.1 anton 15:
1.43 anton 16: Copyright @copyright{} 1995-1997 Free Software Foundation, Inc.
1.1 anton 17:
18: Permission is granted to make and distribute verbatim copies of
19: this manual provided the copyright notice and this permission notice
20: are preserved on all copies.
21:
1.4 anton 22: @ignore
1.1 anton 23: Permission is granted to process this file through TeX and print the
24: results, provided the printed document carries a copying permission
25: notice identical to this one except for the removal of this paragraph
26: (this paragraph not being relevant to the printed manual).
27:
1.4 anton 28: @end ignore
1.1 anton 29: Permission is granted to copy and distribute modified versions of this
30: manual under the conditions for verbatim copying, provided also that the
31: sections entitled "Distribution" and "General Public License" are
32: included exactly as in the original, and provided that the entire
33: resulting derived work is distributed under the terms of a permission
34: notice identical to this one.
35:
36: Permission is granted to copy and distribute translations of this manual
37: into another language, under the above conditions for modified versions,
38: except that the sections entitled "Distribution" and "General Public
39: License" may be included in a translation approved by the author instead
40: of in the original English.
41: @end ifinfo
42:
1.24 anton 43: @finalout
1.1 anton 44: @titlepage
45: @sp 10
1.17 anton 46: @center @titlefont{Gforth Manual}
1.1 anton 47: @sp 2
1.43 anton 48: @center for version 0.3
1.1 anton 49: @sp 2
50: @center Anton Ertl
1.25 anton 51: @center Bernd Paysan
1.17 anton 52: @sp 3
53: @center This manual is under construction
1.1 anton 54:
55: @comment The following two commands start the copyright page.
56: @page
57: @vskip 0pt plus 1filll
1.43 anton 58: Copyright @copyright{} 1995--1997 Free Software Foundation, Inc.
1.1 anton 59:
60: @comment !! Published by ... or You can get a copy of this manual ...
61:
62: Permission is granted to make and distribute verbatim copies of
63: this manual provided the copyright notice and this permission notice
64: are preserved on all copies.
65:
66: Permission is granted to copy and distribute modified versions of this
67: manual under the conditions for verbatim copying, provided also that the
68: sections entitled "Distribution" and "General Public License" are
69: included exactly as in the original, and provided that the entire
70: resulting derived work is distributed under the terms of a permission
71: notice identical to this one.
72:
73: Permission is granted to copy and distribute translations of this manual
74: into another language, under the above conditions for modified versions,
75: except that the sections entitled "Distribution" and "General Public
76: License" may be included in a translation approved by the author instead
77: of in the original English.
78: @end titlepage
79:
80:
81: @node Top, License, (dir), (dir)
82: @ifinfo
1.17 anton 83: Gforth is a free implementation of ANS Forth available on many
1.43 anton 84: personal machines. This manual corresponds to version 0.3.
1.1 anton 85: @end ifinfo
86:
87: @menu
1.4 anton 88: * License::
1.17 anton 89: * Goals:: About the Gforth Project
1.4 anton 90: * Other Books:: Things you might want to read
1.43 anton 91: * Invoking Gforth:: Starting Gforth
1.17 anton 92: * Words:: Forth words available in Gforth
1.40 anton 93: * Tools:: Programming tools
1.4 anton 94: * ANS conformance:: Implementation-defined options etc.
1.17 anton 95: * Model:: The abstract machine of Gforth
1.43 anton 96: * Integrating Gforth:: Forth as scripting language for applications
1.17 anton 97: * Emacs and Gforth:: The Gforth Mode
1.43 anton 98: * Image Files:: @code{.fi} files contain compiled code
99: * Engine:: The inner interpreter and the primitives
1.4 anton 100: * Bugs:: How to report them
1.29 anton 101: * Origin:: Authors and ancestors of Gforth
1.4 anton 102: * Word Index:: An item for each Forth word
1.43 anton 103: * Concept Index:: A menu covering many topics
1.1 anton 104: @end menu
105:
1.47 anton 106: @node License, Goals, Top, Top
1.20 pazsan 107: @unnumbered GNU GENERAL PUBLIC LICENSE
108: @center Version 2, June 1991
109:
110: @display
111: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
112: 675 Mass Ave, Cambridge, MA 02139, USA
113:
114: Everyone is permitted to copy and distribute verbatim copies
115: of this license document, but changing it is not allowed.
116: @end display
117:
118: @unnumberedsec Preamble
119:
120: The licenses for most software are designed to take away your
121: freedom to share and change it. By contrast, the GNU General Public
122: License is intended to guarantee your freedom to share and change free
123: software---to make sure the software is free for all its users. This
124: General Public License applies to most of the Free Software
125: Foundation's software and to any other program whose authors commit to
126: using it. (Some other Free Software Foundation software is covered by
127: the GNU Library General Public License instead.) You can apply it to
128: your programs, too.
129:
130: When we speak of free software, we are referring to freedom, not
131: price. Our General Public Licenses are designed to make sure that you
132: have the freedom to distribute copies of free software (and charge for
133: this service if you wish), that you receive source code or can get it
134: if you want it, that you can change the software or use pieces of it
135: in new free programs; and that you know you can do these things.
136:
137: To protect your rights, we need to make restrictions that forbid
138: anyone to deny you these rights or to ask you to surrender the rights.
139: These restrictions translate to certain responsibilities for you if you
140: distribute copies of the software, or if you modify it.
141:
142: For example, if you distribute copies of such a program, whether
143: gratis or for a fee, you must give the recipients all the rights that
144: you have. You must make sure that they, too, receive or can get the
145: source code. And you must show them these terms so they know their
146: rights.
147:
148: We protect your rights with two steps: (1) copyright the software, and
149: (2) offer you this license which gives you legal permission to copy,
150: distribute and/or modify the software.
151:
152: Also, for each author's protection and ours, we want to make certain
153: that everyone understands that there is no warranty for this free
154: software. If the software is modified by someone else and passed on, we
155: want its recipients to know that what they have is not the original, so
156: that any problems introduced by others will not reflect on the original
157: authors' reputations.
158:
159: Finally, any free program is threatened constantly by software
160: patents. We wish to avoid the danger that redistributors of a free
161: program will individually obtain patent licenses, in effect making the
162: program proprietary. To prevent this, we have made it clear that any
163: patent must be licensed for everyone's free use or not licensed at all.
164:
165: The precise terms and conditions for copying, distribution and
166: modification follow.
167:
168: @iftex
169: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
170: @end iftex
171: @ifinfo
172: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
173: @end ifinfo
174:
175: @enumerate 0
176: @item
177: This License applies to any program or other work which contains
178: a notice placed by the copyright holder saying it may be distributed
179: under the terms of this General Public License. The ``Program'', below,
180: refers to any such program or work, and a ``work based on the Program''
181: means either the Program or any derivative work under copyright law:
182: that is to say, a work containing the Program or a portion of it,
183: either verbatim or with modifications and/or translated into another
184: language. (Hereinafter, translation is included without limitation in
185: the term ``modification''.) Each licensee is addressed as ``you''.
186:
187: Activities other than copying, distribution and modification are not
188: covered by this License; they are outside its scope. The act of
189: running the Program is not restricted, and the output from the Program
190: is covered only if its contents constitute a work based on the
191: Program (independent of having been made by running the Program).
192: Whether that is true depends on what the Program does.
193:
194: @item
195: You may copy and distribute verbatim copies of the Program's
196: source code as you receive it, in any medium, provided that you
197: conspicuously and appropriately publish on each copy an appropriate
198: copyright notice and disclaimer of warranty; keep intact all the
199: notices that refer to this License and to the absence of any warranty;
200: and give any other recipients of the Program a copy of this License
201: along with the Program.
202:
203: You may charge a fee for the physical act of transferring a copy, and
204: you may at your option offer warranty protection in exchange for a fee.
205:
206: @item
207: You may modify your copy or copies of the Program or any portion
208: of it, thus forming a work based on the Program, and copy and
209: distribute such modifications or work under the terms of Section 1
210: above, provided that you also meet all of these conditions:
211:
212: @enumerate a
213: @item
214: You must cause the modified files to carry prominent notices
215: stating that you changed the files and the date of any change.
216:
217: @item
218: You must cause any work that you distribute or publish, that in
219: whole or in part contains or is derived from the Program or any
220: part thereof, to be licensed as a whole at no charge to all third
221: parties under the terms of this License.
222:
223: @item
224: If the modified program normally reads commands interactively
225: when run, you must cause it, when started running for such
226: interactive use in the most ordinary way, to print or display an
227: announcement including an appropriate copyright notice and a
228: notice that there is no warranty (or else, saying that you provide
229: a warranty) and that users may redistribute the program under
230: these conditions, and telling the user how to view a copy of this
231: License. (Exception: if the Program itself is interactive but
232: does not normally print such an announcement, your work based on
233: the Program is not required to print an announcement.)
234: @end enumerate
235:
236: These requirements apply to the modified work as a whole. If
237: identifiable sections of that work are not derived from the Program,
238: and can be reasonably considered independent and separate works in
239: themselves, then this License, and its terms, do not apply to those
240: sections when you distribute them as separate works. But when you
241: distribute the same sections as part of a whole which is a work based
242: on the Program, the distribution of the whole must be on the terms of
243: this License, whose permissions for other licensees extend to the
244: entire whole, and thus to each and every part regardless of who wrote it.
245:
246: Thus, it is not the intent of this section to claim rights or contest
247: your rights to work written entirely by you; rather, the intent is to
248: exercise the right to control the distribution of derivative or
249: collective works based on the Program.
250:
251: In addition, mere aggregation of another work not based on the Program
252: with the Program (or with a work based on the Program) on a volume of
253: a storage or distribution medium does not bring the other work under
254: the scope of this License.
255:
256: @item
257: You may copy and distribute the Program (or a work based on it,
258: under Section 2) in object code or executable form under the terms of
259: Sections 1 and 2 above provided that you also do one of the following:
260:
261: @enumerate a
262: @item
263: Accompany it with the complete corresponding machine-readable
264: source code, which must be distributed under the terms of Sections
265: 1 and 2 above on a medium customarily used for software interchange; or,
266:
267: @item
268: Accompany it with a written offer, valid for at least three
269: years, to give any third party, for a charge no more than your
270: cost of physically performing source distribution, a complete
271: machine-readable copy of the corresponding source code, to be
272: distributed under the terms of Sections 1 and 2 above on a medium
273: customarily used for software interchange; or,
274:
275: @item
276: Accompany it with the information you received as to the offer
277: to distribute corresponding source code. (This alternative is
278: allowed only for noncommercial distribution and only if you
279: received the program in object code or executable form with such
280: an offer, in accord with Subsection b above.)
281: @end enumerate
282:
283: The source code for a work means the preferred form of the work for
284: making modifications to it. For an executable work, complete source
285: code means all the source code for all modules it contains, plus any
286: associated interface definition files, plus the scripts used to
287: control compilation and installation of the executable. However, as a
288: special exception, the source code distributed need not include
289: anything that is normally distributed (in either source or binary
290: form) with the major components (compiler, kernel, and so on) of the
291: operating system on which the executable runs, unless that component
292: itself accompanies the executable.
293:
294: If distribution of executable or object code is made by offering
295: access to copy from a designated place, then offering equivalent
296: access to copy the source code from the same place counts as
297: distribution of the source code, even though third parties are not
298: compelled to copy the source along with the object code.
299:
300: @item
301: You may not copy, modify, sublicense, or distribute the Program
302: except as expressly provided under this License. Any attempt
303: otherwise to copy, modify, sublicense or distribute the Program is
304: void, and will automatically terminate your rights under this License.
305: However, parties who have received copies, or rights, from you under
306: this License will not have their licenses terminated so long as such
307: parties remain in full compliance.
308:
309: @item
310: You are not required to accept this License, since you have not
311: signed it. However, nothing else grants you permission to modify or
312: distribute the Program or its derivative works. These actions are
313: prohibited by law if you do not accept this License. Therefore, by
314: modifying or distributing the Program (or any work based on the
315: Program), you indicate your acceptance of this License to do so, and
316: all its terms and conditions for copying, distributing or modifying
317: the Program or works based on it.
318:
319: @item
320: Each time you redistribute the Program (or any work based on the
321: Program), the recipient automatically receives a license from the
322: original licensor to copy, distribute or modify the Program subject to
323: these terms and conditions. You may not impose any further
324: restrictions on the recipients' exercise of the rights granted herein.
325: You are not responsible for enforcing compliance by third parties to
326: this License.
327:
328: @item
329: If, as a consequence of a court judgment or allegation of patent
330: infringement or for any other reason (not limited to patent issues),
331: conditions are imposed on you (whether by court order, agreement or
332: otherwise) that contradict the conditions of this License, they do not
333: excuse you from the conditions of this License. If you cannot
334: distribute so as to satisfy simultaneously your obligations under this
335: License and any other pertinent obligations, then as a consequence you
336: may not distribute the Program at all. For example, if a patent
337: license would not permit royalty-free redistribution of the Program by
338: all those who receive copies directly or indirectly through you, then
339: the only way you could satisfy both it and this License would be to
340: refrain entirely from distribution of the Program.
341:
342: If any portion of this section is held invalid or unenforceable under
343: any particular circumstance, the balance of the section is intended to
344: apply and the section as a whole is intended to apply in other
345: circumstances.
346:
347: It is not the purpose of this section to induce you to infringe any
348: patents or other property right claims or to contest validity of any
349: such claims; this section has the sole purpose of protecting the
350: integrity of the free software distribution system, which is
351: implemented by public license practices. Many people have made
352: generous contributions to the wide range of software distributed
353: through that system in reliance on consistent application of that
354: system; it is up to the author/donor to decide if he or she is willing
355: to distribute software through any other system and a licensee cannot
356: impose that choice.
357:
358: This section is intended to make thoroughly clear what is believed to
359: be a consequence of the rest of this License.
360:
361: @item
362: If the distribution and/or use of the Program is restricted in
363: certain countries either by patents or by copyrighted interfaces, the
364: original copyright holder who places the Program under this License
365: may add an explicit geographical distribution limitation excluding
366: those countries, so that distribution is permitted only in or among
367: countries not thus excluded. In such case, this License incorporates
368: the limitation as if written in the body of this License.
369:
370: @item
371: The Free Software Foundation may publish revised and/or new versions
372: of the General Public License from time to time. Such new versions will
373: be similar in spirit to the present version, but may differ in detail to
374: address new problems or concerns.
375:
376: Each version is given a distinguishing version number. If the Program
377: specifies a version number of this License which applies to it and ``any
378: later version'', you have the option of following the terms and conditions
379: either of that version or of any later version published by the Free
380: Software Foundation. If the Program does not specify a version number of
381: this License, you may choose any version ever published by the Free Software
382: Foundation.
383:
384: @item
385: If you wish to incorporate parts of the Program into other free
386: programs whose distribution conditions are different, write to the author
387: to ask for permission. For software which is copyrighted by the Free
388: Software Foundation, write to the Free Software Foundation; we sometimes
389: make exceptions for this. Our decision will be guided by the two goals
390: of preserving the free status of all derivatives of our free software and
391: of promoting the sharing and reuse of software generally.
392:
393: @iftex
394: @heading NO WARRANTY
395: @end iftex
396: @ifinfo
397: @center NO WARRANTY
398: @end ifinfo
399:
400: @item
401: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
402: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
403: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
404: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
405: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
406: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
407: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
408: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
409: REPAIR OR CORRECTION.
410:
411: @item
412: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
413: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
414: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
415: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
416: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
417: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
418: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
419: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
420: POSSIBILITY OF SUCH DAMAGES.
421: @end enumerate
422:
423: @iftex
424: @heading END OF TERMS AND CONDITIONS
425: @end iftex
426: @ifinfo
427: @center END OF TERMS AND CONDITIONS
428: @end ifinfo
429:
430: @page
431: @unnumberedsec How to Apply These Terms to Your New Programs
432:
433: If you develop a new program, and you want it to be of the greatest
434: possible use to the public, the best way to achieve this is to make it
435: free software which everyone can redistribute and change under these terms.
436:
437: To do so, attach the following notices to the program. It is safest
438: to attach them to the start of each source file to most effectively
439: convey the exclusion of warranty; and each file should have at least
440: the ``copyright'' line and a pointer to where the full notice is found.
441:
442: @smallexample
443: @var{one line to give the program's name and a brief idea of what it does.}
444: Copyright (C) 19@var{yy} @var{name of author}
445:
446: This program is free software; you can redistribute it and/or modify
447: it under the terms of the GNU General Public License as published by
448: the Free Software Foundation; either version 2 of the License, or
449: (at your option) any later version.
450:
451: This program is distributed in the hope that it will be useful,
452: but WITHOUT ANY WARRANTY; without even the implied warranty of
453: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
454: GNU General Public License for more details.
455:
456: You should have received a copy of the GNU General Public License
457: along with this program; if not, write to the Free Software
458: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
459: @end smallexample
460:
461: Also add information on how to contact you by electronic and paper mail.
462:
463: If the program is interactive, make it output a short notice like this
464: when it starts in an interactive mode:
465:
466: @smallexample
467: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
468: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
469: type `show w'.
470: This is free software, and you are welcome to redistribute it
471: under certain conditions; type `show c' for details.
472: @end smallexample
473:
474: The hypothetical commands @samp{show w} and @samp{show c} should show
475: the appropriate parts of the General Public License. Of course, the
476: commands you use may be called something other than @samp{show w} and
477: @samp{show c}; they could even be mouse-clicks or menu items---whatever
478: suits your program.
479:
480: You should also get your employer (if you work as a programmer) or your
481: school, if any, to sign a ``copyright disclaimer'' for the program, if
482: necessary. Here is a sample; alter the names:
483:
484: @smallexample
485: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
486: `Gnomovision' (which makes passes at compilers) written by James Hacker.
487:
488: @var{signature of Ty Coon}, 1 April 1989
489: Ty Coon, President of Vice
490: @end smallexample
491:
492: This General Public License does not permit incorporating your program into
493: proprietary programs. If your program is a subroutine library, you may
494: consider it more useful to permit linking proprietary applications with the
495: library. If this is what you want to do, use the GNU Library General
496: Public License instead of this License.
1.1 anton 497:
498: @iftex
499: @unnumbered Preface
1.23 pazsan 500: @cindex Preface
1.17 anton 501: This manual documents Gforth. The reader is expected to know
1.1 anton 502: Forth. This manual is primarily a reference manual. @xref{Other Books}
503: for introductory material.
504: @end iftex
505:
1.47 anton 506: @node Goals, Other Books, License, Top
1.1 anton 507: @comment node-name, next, previous, up
1.17 anton 508: @chapter Goals of Gforth
1.1 anton 509: @cindex Goals
1.17 anton 510: The goal of the Gforth Project is to develop a standard model for
1.43 anton 511: ANS Forth. This can be split into several subgoals:
1.1 anton 512:
513: @itemize @bullet
514: @item
1.43 anton 515: Gforth should conform to the Forth standard (ANS Forth).
1.1 anton 516: @item
517: It should be a model, i.e. it should define all the
518: implementation-dependent things.
519: @item
520: It should become standard, i.e. widely accepted and used. This goal
521: is the most difficult one.
522: @end itemize
523:
1.17 anton 524: To achieve these goals Gforth should be
1.1 anton 525: @itemize @bullet
526: @item
527: Similar to previous models (fig-Forth, F83)
528: @item
529: Powerful. It should provide for all the things that are considered
530: necessary today and even some that are not yet considered necessary.
531: @item
532: Efficient. It should not get the reputation of being exceptionally
533: slow.
534: @item
535: Free.
536: @item
537: Available on many machines/easy to port.
538: @end itemize
539:
1.17 anton 540: Have we achieved these goals? Gforth conforms to the ANS Forth
541: standard. It may be considered a model, but we have not yet documented
1.1 anton 542: which parts of the model are stable and which parts we are likely to
1.17 anton 543: change. It certainly has not yet become a de facto standard. It has some
544: similarities and some differences to previous models. It has some
545: powerful features, but not yet everything that we envisioned. We
546: certainly have achieved our execution speed goals (@pxref{Performance}).
547: It is free and available on many machines.
1.1 anton 548:
1.43 anton 549: @node Other Books, Invoking Gforth, Goals, Top
1.1 anton 550: @chapter Other books on ANS Forth
1.43 anton 551: @cindex books on Forth
1.1 anton 552:
553: As the standard is relatively new, there are not many books out yet. It
1.17 anton 554: is not recommended to learn Forth by using Gforth and a book that is
1.1 anton 555: not written for ANS Forth, as you will not know your mistakes from the
556: deviations of the book.
557:
1.43 anton 558: @cindex standard document for ANS Forth
559: @cindex ANS Forth document
1.1 anton 560: There is, of course, the standard, the definite reference if you want to
1.19 anton 561: write ANS Forth programs. It is available in printed form from the
562: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
563: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
564: can also get it from Global Engineering Documents (Tel.: USA (800)
565: 854-7179; Fax.: (303) 843-9880) for about $300.
566:
567: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
568: for publication is available electronically and for free in some MS Word
569: format, and it has been converted to HTML. Some pointers to these
570: versions can be found through
1.48 ! anton 571: @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
1.1 anton 572:
1.43 anton 573: @cindex introductory book
574: @cindex book, introductory
575: @cindex Woehr, Jack: @cite{Forth: The New Model}
576: @cindex @cite{Forth: The new model} (book)
577: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
1.1 anton 578: introductory book based on a draft version of the standard. It does not
579: cover the whole standard. It also contains interesting background
1.41 anton 580: information (Jack Woehr was in the ANS Forth Technical Committee). It is
1.1 anton 581: not appropriate for complete newbies, but programmers experienced in
582: other languages should find it ok.
583:
1.43 anton 584: @node Invoking Gforth, Words, Other Books, Top
585: @chapter Invoking Gforth
586: @cindex invoking Gforth
587: @cindex running Gforth
588: @cindex command-line options
589: @cindex options on the command line
590: @cindex flags on the command line
1.1 anton 591:
592: You will usually just say @code{gforth}. In many other cases the default
1.17 anton 593: Gforth image will be invoked like this:
1.1 anton 594:
595: @example
596: gforth [files] [-e forth-code]
597: @end example
598:
599: executing the contents of the files and the Forth code in the order they
600: are given.
601:
602: In general, the command line looks like this:
603:
604: @example
605: gforth [initialization options] [image-specific options]
606: @end example
607:
608: The initialization options must come before the rest of the command
609: line. They are:
610:
611: @table @code
1.43 anton 612: @cindex -i, command-line option
613: @cindex --image-file, command-line option
1.1 anton 614: @item --image-file @var{file}
1.43 anton 615: @itemx -i @var{file}
1.1 anton 616: Loads the Forth image @var{file} instead of the default
1.43 anton 617: @file{gforth.fi} (@pxref{Image Files}).
1.1 anton 618:
1.43 anton 619: @cindex --path, command-line option
620: @cindex -p, command-line option
1.1 anton 621: @item --path @var{path}
1.43 anton 622: @itemx -p @var{path}
1.39 anton 623: Uses @var{path} for searching the image file and Forth source code files
624: instead of the default in the environment variable @code{GFORTHPATH} or
625: the path specified at installation time (e.g.,
626: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
627: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1.1 anton 628:
1.43 anton 629: @cindex --dictionary-size, command-line option
630: @cindex -m, command-line option
631: @cindex @var{size} parameters for command-line options
632: @cindex size of the dictionary and the stacks
1.1 anton 633: @item --dictionary-size @var{size}
1.43 anton 634: @itemx -m @var{size}
1.1 anton 635: Allocate @var{size} space for the Forth dictionary space instead of
636: using the default specified in the image (typically 256K). The
637: @var{size} specification consists of an integer and a unit (e.g.,
638: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
639: size, in this case Cells), @code{k} (kilobytes), and @code{M}
640: (Megabytes). If no unit is specified, @code{e} is used.
641:
1.43 anton 642: @cindex --data-stack-size, command-line option
643: @cindex -d, command-line option
1.1 anton 644: @item --data-stack-size @var{size}
1.43 anton 645: @itemx -d @var{size}
1.1 anton 646: Allocate @var{size} space for the data stack instead of using the
647: default specified in the image (typically 16K).
648:
1.43 anton 649: @cindex --return-stack-size, command-line option
650: @cindex -r, command-line option
1.1 anton 651: @item --return-stack-size @var{size}
1.43 anton 652: @itemx -r @var{size}
1.1 anton 653: Allocate @var{size} space for the return stack instead of using the
1.43 anton 654: default specified in the image (typically 15K).
1.1 anton 655:
1.43 anton 656: @cindex --fp-stack-size, command-line option
657: @cindex -f, command-line option
1.1 anton 658: @item --fp-stack-size @var{size}
1.43 anton 659: @itemx -f @var{size}
1.1 anton 660: Allocate @var{size} space for the floating point stack instead of
1.43 anton 661: using the default specified in the image (typically 15.5K). In this case
1.1 anton 662: the unit specifier @code{e} refers to floating point numbers.
663:
1.43 anton 664: @cindex --locals-stack-size, command-line option
665: @cindex -l, command-line option
1.1 anton 666: @item --locals-stack-size @var{size}
1.43 anton 667: @itemx -l @var{size}
1.1 anton 668: Allocate @var{size} space for the locals stack instead of using the
1.43 anton 669: default specified in the image (typically 14.5K).
1.1 anton 670:
1.43 anton 671: @cindex -h, command-line option
672: @cindex --help, command-line option
673: @item --help
674: @itemx -h
675: Print a message about the command-line options
676:
677: @cindex -v, command-line option
678: @cindex --version, command-line option
679: @item --version
680: @itemx -v
681: Print version and exit
682:
683: @cindex --debug, command-line option
684: @item --debug
685: Print some information useful for debugging on startup.
686:
687: @cindex --offset-image, command-line option
688: @item --offset-image
689: Start the dictionary at a slightly different position than would be used
690: otherwise (useful for creating data-relocatable images,
691: @pxref{Data-Relocatable Image Files}).
692:
693: @cindex --clear-dictionary, command-line option
694: @item --clear-dictionary
695: Initialize all bytes in the dictionary to 0 before loading the image
696: (@pxref{Data-Relocatable Image Files}).
1.1 anton 697: @end table
698:
1.43 anton 699: @cindex loading files at startup
700: @cindex executing code on startup
701: @cindex batch processing with Gforth
1.1 anton 702: As explained above, the image-specific command-line arguments for the
703: default image @file{gforth.fi} consist of a sequence of filenames and
1.41 anton 704: @code{-e @var{forth-code}} options that are interpreted in the sequence
1.1 anton 705: in which they are given. The @code{-e @var{forth-code}} or
706: @code{--evaluate @var{forth-code}} option evaluates the forth
707: code. This option takes only one argument; if you want to evaluate more
708: Forth words, you have to quote them or use several @code{-e}s. To exit
709: after processing the command line (instead of entering interactive mode)
710: append @code{-e bye} to the command line.
711:
1.43 anton 712: @cindex versions, invoking other versions of Gforth
1.22 anton 713: If you have several versions of Gforth installed, @code{gforth} will
714: invoke the version that was installed last. @code{gforth-@var{version}}
715: invokes a specific version. You may want to use the option
716: @code{--path}, if your environment contains the variable
717: @code{GFORTHPATH}.
718:
1.1 anton 719: Not yet implemented:
720: On startup the system first executes the system initialization file
721: (unless the option @code{--no-init-file} is given; note that the system
722: resulting from using this option may not be ANS Forth conformant). Then
723: the user initialization file @file{.gforth.fs} is executed, unless the
724: option @code{--no-rc} is given; this file is first searched in @file{.},
725: then in @file{~}, then in the normal path (see above).
726:
1.43 anton 727: @node Words, Tools, Invoking Gforth, Top
1.1 anton 728: @chapter Forth Words
1.43 anton 729: @cindex Words
1.1 anton 730:
731: @menu
1.4 anton 732: * Notation::
733: * Arithmetic::
734: * Stack Manipulation::
1.43 anton 735: * Memory::
1.4 anton 736: * Control Structures::
737: * Locals::
738: * Defining Words::
1.37 anton 739: * Tokens for Words::
1.4 anton 740: * Wordlists::
741: * Files::
742: * Blocks::
743: * Other I/O::
744: * Programming Tools::
1.18 anton 745: * Assembler and Code words::
1.4 anton 746: * Threading Words::
1.1 anton 747: @end menu
748:
749: @node Notation, Arithmetic, Words, Words
750: @section Notation
1.43 anton 751: @cindex notation of glossary entries
752: @cindex format of glossary entries
753: @cindex glossary notation format
754: @cindex word glossary entry format
1.1 anton 755:
756: The Forth words are described in this section in the glossary notation
1.43 anton 757: that has become a de-facto standard for Forth texts, i.e.,
1.1 anton 758:
1.4 anton 759: @format
1.1 anton 760: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
1.4 anton 761: @end format
1.1 anton 762: @var{Description}
763:
764: @table @var
765: @item word
1.43 anton 766: @cindex case insensitivity
1.17 anton 767: The name of the word. BTW, Gforth is case insensitive, so you can
1.14 anton 768: type the words in in lower case (However, @pxref{core-idef}).
1.1 anton 769:
770: @item Stack effect
1.43 anton 771: @cindex stack effect
1.1 anton 772: The stack effect is written in the notation @code{@var{before} --
773: @var{after}}, where @var{before} and @var{after} describe the top of
774: stack entries before and after the execution of the word. The rest of
775: the stack is not touched by the word. The top of stack is rightmost,
1.17 anton 776: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1 anton 777: uses a separate floating point stack, but a unified stack
778: notation. Also, return stack effects are not shown in @var{stack
779: effect}, but in @var{Description}. The name of a stack item describes
780: the type and/or the function of the item. See below for a discussion of
781: the types.
782:
1.19 anton 783: All words have two stack effects: A compile-time stack effect and a
784: run-time stack effect. The compile-time stack-effect of most words is
785: @var{ -- }. If the compile-time stack-effect of a word deviates from
786: this standard behaviour, or the word does other unusual things at
787: compile time, both stack effects are shown; otherwise only the run-time
788: stack effect is shown.
789:
1.43 anton 790: @cindex pronounciation of words
1.1 anton 791: @item pronunciation
1.43 anton 792: How the word is pronounced.
1.1 anton 793:
1.43 anton 794: @cindex wordset
1.1 anton 795: @item wordset
796: The ANS Forth standard is divided into several wordsets. A standard
797: system need not support all of them. So, the fewer wordsets your program
798: uses the more portable it will be in theory. However, we suspect that
799: most ANS Forth systems on personal machines will feature all
800: wordsets. Words that are not defined in the ANS standard have
1.19 anton 801: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
802: describes words that will work in future releases of Gforth;
803: @code{gforth-internal} words are more volatile. Environmental query
804: strings are also displayed like words; you can recognize them by the
805: @code{environment} in the wordset field.
1.1 anton 806:
807: @item Description
808: A description of the behaviour of the word.
809: @end table
810:
1.43 anton 811: @cindex types of stack items
812: @cindex stack item types
1.4 anton 813: The type of a stack item is specified by the character(s) the name
814: starts with:
1.1 anton 815:
816: @table @code
817: @item f
1.43 anton 818: @cindex @code{f}, stack item type
819: Boolean flags, i.e. @code{false} or @code{true}.
1.1 anton 820: @item c
1.43 anton 821: @cindex @code{c}, stack item type
1.1 anton 822: Char
823: @item w
1.43 anton 824: @cindex @code{w}, stack item type
1.1 anton 825: Cell, can contain an integer or an address
826: @item n
1.43 anton 827: @cindex @code{n}, stack item type
1.1 anton 828: signed integer
829: @item u
1.43 anton 830: @cindex @code{u}, stack item type
1.1 anton 831: unsigned integer
832: @item d
1.43 anton 833: @cindex @code{d}, stack item type
1.1 anton 834: double sized signed integer
835: @item ud
1.43 anton 836: @cindex @code{ud}, stack item type
1.1 anton 837: double sized unsigned integer
838: @item r
1.43 anton 839: @cindex @code{r}, stack item type
1.36 anton 840: Float (on the FP stack)
1.1 anton 841: @item a_
1.43 anton 842: @cindex @code{a_}, stack item type
1.1 anton 843: Cell-aligned address
844: @item c_
1.43 anton 845: @cindex @code{c_}, stack item type
1.36 anton 846: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.1 anton 847: @item f_
1.43 anton 848: @cindex @code{f_}, stack item type
1.1 anton 849: Float-aligned address
850: @item df_
1.43 anton 851: @cindex @code{df_}, stack item type
1.1 anton 852: Address aligned for IEEE double precision float
853: @item sf_
1.43 anton 854: @cindex @code{sf_}, stack item type
1.1 anton 855: Address aligned for IEEE single precision float
856: @item xt
1.43 anton 857: @cindex @code{xt}, stack item type
1.1 anton 858: Execution token, same size as Cell
859: @item wid
1.43 anton 860: @cindex @code{wid}, stack item type
1.1 anton 861: Wordlist ID, same size as Cell
862: @item f83name
1.43 anton 863: @cindex @code{f83name}, stack item type
1.1 anton 864: Pointer to a name structure
1.36 anton 865: @item "
1.43 anton 866: @cindex @code{"}, stack item type
1.36 anton 867: string in the input stream (not the stack). The terminating character is
868: a blank by default. If it is not a blank, it is shown in @code{<>}
869: quotes.
1.1 anton 870: @end table
871:
1.4 anton 872: @node Arithmetic, Stack Manipulation, Notation, Words
1.1 anton 873: @section Arithmetic
1.43 anton 874: @cindex arithmetic words
875:
876: @cindex division with potentially negative operands
1.1 anton 877: Forth arithmetic is not checked, i.e., you will not hear about integer
878: overflow on addition or multiplication, you may hear about division by
879: zero if you are lucky. The operator is written after the operands, but
880: the operands are still in the original order. I.e., the infix @code{2-1}
881: corresponds to @code{2 1 -}. Forth offers a variety of division
882: operators. If you perform division with potentially negative operands,
883: you do not want to use @code{/} or @code{/mod} with its undefined
884: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4 anton 885: former, @pxref{Mixed precision}).
886:
887: @menu
888: * Single precision::
889: * Bitwise operations::
890: * Mixed precision:: operations with single and double-cell integers
891: * Double precision:: Double-cell integer arithmetic
892: * Floating Point::
893: @end menu
1.1 anton 894:
1.4 anton 895: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1 anton 896: @subsection Single precision
1.43 anton 897: @cindex single precision arithmetic words
898:
1.1 anton 899: doc-+
900: doc--
901: doc-*
902: doc-/
903: doc-mod
904: doc-/mod
905: doc-negate
906: doc-abs
907: doc-min
908: doc-max
909:
1.4 anton 910: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1 anton 911: @subsection Bitwise operations
1.43 anton 912: @cindex bitwise operation words
913:
1.1 anton 914: doc-and
915: doc-or
916: doc-xor
917: doc-invert
918: doc-2*
919: doc-2/
920:
1.4 anton 921: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1 anton 922: @subsection Mixed precision
1.43 anton 923: @cindex mixed precision arithmetic words
924:
1.1 anton 925: doc-m+
926: doc-*/
927: doc-*/mod
928: doc-m*
929: doc-um*
930: doc-m*/
931: doc-um/mod
932: doc-fm/mod
933: doc-sm/rem
934:
1.4 anton 935: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1 anton 936: @subsection Double precision
1.43 anton 937: @cindex double precision arithmetic words
1.16 anton 938:
1.43 anton 939: @cindex double-cell numbers, input format
940: @cindex input format for double-cell numbers
1.16 anton 941: The outer (aka text) interpreter converts numbers containing a dot into
942: a double precision number. Note that only numbers with the dot as last
943: character are standard-conforming.
944:
1.1 anton 945: doc-d+
946: doc-d-
947: doc-dnegate
948: doc-dabs
949: doc-dmin
950: doc-dmax
951:
1.4 anton 952: @node Floating Point, , Double precision, Arithmetic
953: @subsection Floating Point
1.43 anton 954: @cindex floating point arithmetic words
1.16 anton 955:
1.43 anton 956: @cindex floating-point numbers, input format
957: @cindex input format for floating-point numbers
1.16 anton 958: The format of floating point numbers recognized by the outer (aka text)
959: interpreter is: a signed decimal number, possibly containing a decimal
960: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
1.41 anton 961: by a signed integer (the exponent). E.g., @code{1e} is the same as
1.35 anton 962: @code{+1.0e+0}. Note that a number without @code{e}
1.16 anton 963: is not interpreted as floating-point number, but as double (if the
964: number contains a @code{.}) or single precision integer. Also,
965: conversions between string and floating point numbers always use base
966: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
967: value greater then 14, the @code{E} may be interpreted as digit and the
968: number will be interpreted as integer, unless it has a signed exponent
969: (both @code{+} and @code{-} are allowed as signs).
1.4 anton 970:
1.43 anton 971: @cindex angles in trigonometric operations
972: @cindex trigonometric operations
1.4 anton 973: Angles in floating point operations are given in radians (a full circle
1.17 anton 974: has 2 pi radians). Note, that Gforth has a separate floating point
1.4 anton 975: stack, but we use the unified notation.
976:
1.43 anton 977: @cindex floating-point arithmetic, pitfalls
1.4 anton 978: Floating point numbers have a number of unpleasant surprises for the
979: unwary (e.g., floating point addition is not associative) and even a few
980: for the wary. You should not use them unless you know what you are doing
981: or you don't care that the results you get are totally bogus. If you
982: want to learn about the problems of floating point numbers (and how to
1.11 anton 983: avoid them), you might start with @cite{David Goldberg, What Every
1.6 anton 984: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
985: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4 anton 986:
987: doc-f+
988: doc-f-
989: doc-f*
990: doc-f/
991: doc-fnegate
992: doc-fabs
993: doc-fmax
994: doc-fmin
995: doc-floor
996: doc-fround
997: doc-f**
998: doc-fsqrt
999: doc-fexp
1000: doc-fexpm1
1001: doc-fln
1002: doc-flnp1
1003: doc-flog
1.6 anton 1004: doc-falog
1.4 anton 1005: doc-fsin
1006: doc-fcos
1007: doc-fsincos
1008: doc-ftan
1009: doc-fasin
1010: doc-facos
1011: doc-fatan
1012: doc-fatan2
1013: doc-fsinh
1014: doc-fcosh
1015: doc-ftanh
1016: doc-fasinh
1017: doc-facosh
1018: doc-fatanh
1019:
1.43 anton 1020: @node Stack Manipulation, Memory, Arithmetic, Words
1.1 anton 1021: @section Stack Manipulation
1.43 anton 1022: @cindex stack manipulation words
1.1 anton 1023:
1.43 anton 1024: @cindex floating-point stack in the standard
1.17 anton 1025: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1 anton 1026: addresses, and double cells, a floating point stack for floating point
1027: numbers, a return stack for storing the return addresses of colon
1028: definitions and other data, and a locals stack for storing local
1029: variables. Note that while every sane Forth has a separate floating
1030: point stack, this is not strictly required; an ANS Forth system could
1031: theoretically keep floating point numbers on the data stack. As an
1032: additional difficulty, you don't know how many cells a floating point
1033: number takes. It is reportedly possible to write words in a way that
1034: they work also for a unified stack model, but we do not recommend trying
1.4 anton 1035: it. Instead, just say that your program has an environmental dependency
1036: on a separate FP stack.
1037:
1.43 anton 1038: @cindex return stack and locals
1039: @cindex locals and return stack
1.4 anton 1040: Also, a Forth system is allowed to keep the local variables on the
1.1 anton 1041: return stack. This is reasonable, as local variables usually eliminate
1042: the need to use the return stack explicitly. So, if you want to produce
1043: a standard complying program and if you are using local variables in a
1044: word, forget about return stack manipulations in that word (see the
1045: standard document for the exact rules).
1046:
1.4 anton 1047: @menu
1048: * Data stack::
1049: * Floating point stack::
1050: * Return stack::
1051: * Locals stack::
1052: * Stack pointer manipulation::
1053: @end menu
1054:
1055: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1 anton 1056: @subsection Data stack
1.43 anton 1057: @cindex data stack manipulation words
1058: @cindex stack manipulations words, data stack
1059:
1.1 anton 1060: doc-drop
1061: doc-nip
1062: doc-dup
1063: doc-over
1064: doc-tuck
1065: doc-swap
1066: doc-rot
1067: doc--rot
1068: doc-?dup
1069: doc-pick
1070: doc-roll
1071: doc-2drop
1072: doc-2nip
1073: doc-2dup
1074: doc-2over
1075: doc-2tuck
1076: doc-2swap
1077: doc-2rot
1078:
1.4 anton 1079: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1 anton 1080: @subsection Floating point stack
1.43 anton 1081: @cindex floating-point stack manipulation words
1082: @cindex stack manipulation words, floating-point stack
1083:
1.1 anton 1084: doc-fdrop
1085: doc-fnip
1086: doc-fdup
1087: doc-fover
1088: doc-ftuck
1089: doc-fswap
1090: doc-frot
1091:
1.4 anton 1092: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1 anton 1093: @subsection Return stack
1.43 anton 1094: @cindex return stack manipulation words
1095: @cindex stack manipulation words, return stack
1096:
1.1 anton 1097: doc->r
1098: doc-r>
1099: doc-r@
1100: doc-rdrop
1101: doc-2>r
1102: doc-2r>
1103: doc-2r@
1104: doc-2rdrop
1105:
1.4 anton 1106: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1 anton 1107: @subsection Locals stack
1108:
1.4 anton 1109: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1.1 anton 1110: @subsection Stack pointer manipulation
1.43 anton 1111: @cindex stack pointer manipulation words
1112:
1.1 anton 1113: doc-sp@
1114: doc-sp!
1115: doc-fp@
1116: doc-fp!
1117: doc-rp@
1118: doc-rp!
1119: doc-lp@
1120: doc-lp!
1121:
1.43 anton 1122: @node Memory, Control Structures, Stack Manipulation, Words
1123: @section Memory
1124: @cindex Memory words
1.1 anton 1125:
1.4 anton 1126: @menu
1.43 anton 1127: * Memory Access::
1.4 anton 1128: * Address arithmetic::
1.43 anton 1129: * Memory Blocks::
1.4 anton 1130: @end menu
1131:
1.43 anton 1132: @node Memory Access, Address arithmetic, Memory, Memory
1133: @subsection Memory Access
1134: @cindex memory access words
1.1 anton 1135:
1136: doc-@
1137: doc-!
1138: doc-+!
1139: doc-c@
1140: doc-c!
1141: doc-2@
1142: doc-2!
1143: doc-f@
1144: doc-f!
1145: doc-sf@
1146: doc-sf!
1147: doc-df@
1148: doc-df!
1149:
1.43 anton 1150: @node Address arithmetic, Memory Blocks, Memory Access, Memory
1.1 anton 1151: @subsection Address arithmetic
1.43 anton 1152: @cindex address arithmetic words
1.1 anton 1153:
1154: ANS Forth does not specify the sizes of the data types. Instead, it
1155: offers a number of words for computing sizes and doing address
1156: arithmetic. Basically, address arithmetic is performed in terms of
1157: address units (aus); on most systems the address unit is one byte. Note
1158: that a character may have more than one au, so @code{chars} is no noop
1159: (on systems where it is a noop, it compiles to nothing).
1160:
1.43 anton 1161: @cindex alignment of addresses for types
1.1 anton 1162: ANS Forth also defines words for aligning addresses for specific
1.43 anton 1163: types. Many computers require that accesses to specific data types
1.1 anton 1164: must only occur at specific addresses; e.g., that cells may only be
1165: accessed at addresses divisible by 4. Even if a machine allows unaligned
1166: accesses, it can usually perform aligned accesses faster.
1167:
1.17 anton 1168: For the performance-conscious: alignment operations are usually only
1.1 anton 1169: necessary during the definition of a data structure, not during the
1170: (more frequent) accesses to it.
1171:
1172: ANS Forth defines no words for character-aligning addresses. This is not
1173: an oversight, but reflects the fact that addresses that are not
1174: char-aligned have no use in the standard and therefore will not be
1175: created.
1176:
1.43 anton 1177: @cindex @code{CREATE} and alignment
1.1 anton 1178: The standard guarantees that addresses returned by @code{CREATE}d words
1.17 anton 1179: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1 anton 1180: are aligned for all purposes.
1181:
1.9 anton 1182: Note that the standard defines a word @code{char}, which has nothing to
1183: do with address arithmetic.
1184:
1.1 anton 1185: doc-chars
1186: doc-char+
1187: doc-cells
1188: doc-cell+
1.43 anton 1189: doc-cell
1.1 anton 1190: doc-align
1191: doc-aligned
1192: doc-floats
1193: doc-float+
1.43 anton 1194: doc-float
1.1 anton 1195: doc-falign
1196: doc-faligned
1197: doc-sfloats
1198: doc-sfloat+
1199: doc-sfalign
1200: doc-sfaligned
1201: doc-dfloats
1202: doc-dfloat+
1203: doc-dfalign
1204: doc-dfaligned
1.10 anton 1205: doc-maxalign
1206: doc-maxaligned
1207: doc-cfalign
1208: doc-cfaligned
1.1 anton 1209: doc-address-unit-bits
1210:
1.43 anton 1211: @node Memory Blocks, , Address arithmetic, Memory
1212: @subsection Memory Blocks
1213: @cindex memory block words
1.1 anton 1214:
1215: doc-move
1216: doc-erase
1217:
1218: While the previous words work on address units, the rest works on
1219: characters.
1220:
1221: doc-cmove
1222: doc-cmove>
1223: doc-fill
1224: doc-blank
1225:
1.43 anton 1226: @node Control Structures, Locals, Memory, Words
1.1 anton 1227: @section Control Structures
1.43 anton 1228: @cindex control structures
1.1 anton 1229:
1230: Control structures in Forth cannot be used in interpret state, only in
1.43 anton 1231: compile state@footnote{More precisely, they have no interpretation
1232: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
1233: a colon definition. We do not like this limitation, but have not seen a
1234: satisfying way around it yet, although many schemes have been proposed.
1.1 anton 1235:
1.4 anton 1236: @menu
1237: * Selection::
1238: * Simple Loops::
1239: * Counted Loops::
1240: * Arbitrary control structures::
1241: * Calls and returns::
1242: * Exception Handling::
1243: @end menu
1244:
1245: @node Selection, Simple Loops, Control Structures, Control Structures
1.1 anton 1246: @subsection Selection
1.43 anton 1247: @cindex selection control structures
1248: @cindex control structures for selection
1.1 anton 1249:
1.43 anton 1250: @cindex @code{IF} control structure
1.1 anton 1251: @example
1252: @var{flag}
1253: IF
1254: @var{code}
1255: ENDIF
1256: @end example
1257: or
1258: @example
1259: @var{flag}
1260: IF
1261: @var{code1}
1262: ELSE
1263: @var{code2}
1264: ENDIF
1265: @end example
1266:
1.4 anton 1267: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1 anton 1268: standard, and @code{ENDIF} is not, although it is quite popular. We
1269: recommend using @code{ENDIF}, because it is less confusing for people
1270: who also know other languages (and is not prone to reinforcing negative
1271: prejudices against Forth in these people). Adding @code{ENDIF} to a
1272: system that only supplies @code{THEN} is simple:
1273: @example
1274: : endif POSTPONE then ; immediate
1275: @end example
1276:
1277: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1278: (adv.)} has the following meanings:
1279: @quotation
1280: ... 2b: following next after in order ... 3d: as a necessary consequence
1281: (if you were there, then you saw them).
1282: @end quotation
1283: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1284: and many other programming languages has the meaning 3d.]
1285:
1.31 anton 1286: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1287: you can avoid using @code{?dup}. Using these alternatives is also more
1288: efficient than using @code{?dup}. Definitions in plain standard Forth
1289: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1290: @file{compat/control.fs}.
1.1 anton 1291:
1.43 anton 1292: @cindex @code{CASE} control structure
1.1 anton 1293: @example
1294: @var{n}
1295: CASE
1296: @var{n1} OF @var{code1} ENDOF
1297: @var{n2} OF @var{code2} ENDOF
1.4 anton 1298: @dots{}
1.1 anton 1299: ENDCASE
1300: @end example
1301:
1302: Executes the first @var{codei}, where the @var{ni} is equal to
1303: @var{n}. A default case can be added by simply writing the code after
1304: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1305: but must not consume it.
1306:
1.4 anton 1307: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1 anton 1308: @subsection Simple Loops
1.43 anton 1309: @cindex simple loops
1310: @cindex loops without count
1.1 anton 1311:
1.43 anton 1312: @cindex @code{WHILE} loop
1.1 anton 1313: @example
1314: BEGIN
1315: @var{code1}
1316: @var{flag}
1317: WHILE
1318: @var{code2}
1319: REPEAT
1320: @end example
1321:
1322: @var{code1} is executed and @var{flag} is computed. If it is true,
1.43 anton 1323: @var{code2} is executed and the loop is restarted; If @var{flag} is
1324: false, execution continues after the @code{REPEAT}.
1.1 anton 1325:
1.43 anton 1326: @cindex @code{UNTIL} loop
1.1 anton 1327: @example
1328: BEGIN
1329: @var{code}
1330: @var{flag}
1331: UNTIL
1332: @end example
1333:
1334: @var{code} is executed. The loop is restarted if @code{flag} is false.
1335:
1.43 anton 1336: @cindex endless loop
1337: @cindex loops, endless
1.1 anton 1338: @example
1339: BEGIN
1340: @var{code}
1341: AGAIN
1342: @end example
1343:
1344: This is an endless loop.
1345:
1.4 anton 1346: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1 anton 1347: @subsection Counted Loops
1.43 anton 1348: @cindex counted loops
1349: @cindex loops, counted
1350: @cindex @code{DO} loops
1.1 anton 1351:
1352: The basic counted loop is:
1353: @example
1354: @var{limit} @var{start}
1355: ?DO
1356: @var{body}
1357: LOOP
1358: @end example
1359:
1360: This performs one iteration for every integer, starting from @var{start}
1361: and up to, but excluding @var{limit}. The counter, aka index, can be
1362: accessed with @code{i}. E.g., the loop
1363: @example
1364: 10 0 ?DO
1365: i .
1366: LOOP
1367: @end example
1368: prints
1369: @example
1370: 0 1 2 3 4 5 6 7 8 9
1371: @end example
1372: The index of the innermost loop can be accessed with @code{i}, the index
1373: of the next loop with @code{j}, and the index of the third loop with
1374: @code{k}.
1375:
1.46 anton 1376: doc-i
1377: doc-j
1378: doc-k
1379:
1.1 anton 1380: The loop control data are kept on the return stack, so there are some
1381: restrictions on mixing return stack accesses and counted loop
1382: words. E.g., if you put values on the return stack outside the loop, you
1383: cannot read them inside the loop. If you put values on the return stack
1384: within a loop, you have to remove them before the end of the loop and
1385: before accessing the index of the loop.
1386:
1387: There are several variations on the counted loop:
1388:
1389: @code{LEAVE} leaves the innermost counted loop immediately.
1390:
1.18 anton 1391: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1392: (and @code{LOOP} iterates until they become equal by wrap-around
1393: arithmetic). This behaviour is usually not what you want. Therefore,
1394: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1395: @code{?DO}), which do not enter the loop if @var{start} is greater than
1396: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.30 anton 1397: unsigned loop parameters.
1.18 anton 1398:
1.1 anton 1399: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1400: index by @var{n} instead of by 1. The loop is terminated when the border
1401: between @var{limit-1} and @var{limit} is crossed. E.g.:
1402:
1.18 anton 1403: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1.1 anton 1404:
1.18 anton 1405: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1.1 anton 1406:
1.43 anton 1407: @cindex negative increment for counted loops
1408: @cindex counted loops with negative increment
1.1 anton 1409: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1410:
1.2 anton 1411: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1.1 anton 1412:
1.2 anton 1413: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1.1 anton 1414:
1.18 anton 1415: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1416: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1417: index by @var{u} each iteration. The loop is terminated when the border
1418: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1419: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1 anton 1420:
1.18 anton 1421: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1.1 anton 1422:
1.18 anton 1423: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1.1 anton 1424:
1.18 anton 1425: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1.1 anton 1426:
1.30 anton 1427: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1428: @code{-LOOP} are not in the ANS Forth standard. However, an
1429: implementation for these words that uses only standard words is provided
1430: in @file{compat/loops.fs}.
1.18 anton 1431:
1432: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1433: the loop, independent of the loop parameters. Do not use @code{DO}, even
1434: if you know that the loop is entered in any case. Such knowledge tends
1435: to become invalid during maintenance of a program, and then the
1436: @code{DO} will make trouble.
1.1 anton 1437:
1438: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1439: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1440: return stack so @code{EXIT} can get to its return address.
1441:
1.43 anton 1442: @cindex @code{FOR} loops
1.1 anton 1443: Another counted loop is
1444: @example
1445: @var{n}
1446: FOR
1447: @var{body}
1448: NEXT
1449: @end example
1450: This is the preferred loop of native code compiler writers who are too
1.17 anton 1451: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1 anton 1452: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1453: and ending with 0. Other Forth systems may behave differently, even if
1.30 anton 1454: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1455: loops.
1.1 anton 1456:
1.4 anton 1457: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2 anton 1458: @subsection Arbitrary control structures
1.43 anton 1459: @cindex control structures, user-defined
1.2 anton 1460:
1.43 anton 1461: @cindex control-flow stack
1.2 anton 1462: ANS Forth permits and supports using control structures in a non-nested
1463: way. Information about incomplete control structures is stored on the
1464: control-flow stack. This stack may be implemented on the Forth data
1.17 anton 1465: stack, and this is what we have done in Gforth.
1.2 anton 1466:
1.43 anton 1467: @cindex @code{orig}, control-flow stack item
1468: @cindex @code{dest}, control-flow stack item
1.2 anton 1469: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1470: entry represents a backward branch target. A few words are the basis for
1471: building any control structure possible (except control structures that
1472: need storage, like calls, coroutines, and backtracking).
1473:
1.3 anton 1474: doc-if
1475: doc-ahead
1476: doc-then
1477: doc-begin
1478: doc-until
1479: doc-again
1480: doc-cs-pick
1481: doc-cs-roll
1.2 anton 1482:
1.17 anton 1483: On many systems control-flow stack items take one word, in Gforth they
1.2 anton 1484: currently take three (this may change in the future). Therefore it is a
1485: really good idea to manipulate the control flow stack with
1486: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1487: words.
1488:
1489: Some standard control structure words are built from these words:
1490:
1.3 anton 1491: doc-else
1492: doc-while
1493: doc-repeat
1.2 anton 1494:
1.31 anton 1495: Gforth adds some more control-structure words:
1496:
1497: doc-endif
1498: doc-?dup-if
1499: doc-?dup-0=-if
1500:
1.2 anton 1501: Counted loop words constitute a separate group of words:
1502:
1.3 anton 1503: doc-?do
1.18 anton 1504: doc-+do
1505: doc-u+do
1506: doc--do
1507: doc-u-do
1.3 anton 1508: doc-do
1509: doc-for
1510: doc-loop
1511: doc-+loop
1.18 anton 1512: doc--loop
1.3 anton 1513: doc-next
1514: doc-leave
1515: doc-?leave
1516: doc-unloop
1.10 anton 1517: doc-done
1.2 anton 1518:
1519: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1520: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1521: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3 anton 1522: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1523: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7 pazsan 1524: resolved (by using one of the loop-ending words or @code{DONE}).
1.2 anton 1525:
1526: Another group of control structure words are
1527:
1.3 anton 1528: doc-case
1529: doc-endcase
1530: doc-of
1531: doc-endof
1.2 anton 1532:
1533: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1534: @code{cs-roll}.
1535:
1.3 anton 1536: @subsubsection Programming Style
1537:
1538: In order to ensure readability we recommend that you do not create
1539: arbitrary control structures directly, but define new control structure
1540: words for the control structure you want and use these words in your
1541: program.
1542:
1543: E.g., instead of writing
1544:
1545: @example
1546: begin
1547: ...
1548: if [ 1 cs-roll ]
1549: ...
1550: again then
1551: @end example
1552:
1553: we recommend defining control structure words, e.g.,
1554:
1555: @example
1556: : while ( dest -- orig dest )
1557: POSTPONE if
1558: 1 cs-roll ; immediate
1559:
1560: : repeat ( orig dest -- )
1561: POSTPONE again
1562: POSTPONE then ; immediate
1563: @end example
1564:
1565: and then using these to create the control structure:
1566:
1567: @example
1568: begin
1569: ...
1570: while
1571: ...
1572: repeat
1573: @end example
1574:
1.30 anton 1575: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1.3 anton 1576: @code{WHILE} are predefined, so in this example it would not be
1577: necessary to define them.
1578:
1.4 anton 1579: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3 anton 1580: @subsection Calls and returns
1.43 anton 1581: @cindex calling a definition
1582: @cindex returning from a definition
1.3 anton 1583:
1584: A definition can be called simply be writing the name of the
1.17 anton 1585: definition. When the end of the definition is reached, it returns. An
1586: earlier return can be forced using
1.3 anton 1587:
1588: doc-exit
1589:
1590: Don't forget to clean up the return stack and @code{UNLOOP} any
1591: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1592: primitive compiled by @code{EXIT} is
1593:
1594: doc-;s
1595:
1.4 anton 1596: @node Exception Handling, , Calls and returns, Control Structures
1.3 anton 1597: @subsection Exception Handling
1.43 anton 1598: @cindex Exceptions
1.3 anton 1599:
1600: doc-catch
1601: doc-throw
1602:
1.4 anton 1603: @node Locals, Defining Words, Control Structures, Words
1.1 anton 1604: @section Locals
1.43 anton 1605: @cindex locals
1.1 anton 1606:
1.2 anton 1607: Local variables can make Forth programming more enjoyable and Forth
1608: programs easier to read. Unfortunately, the locals of ANS Forth are
1609: laden with restrictions. Therefore, we provide not only the ANS Forth
1610: locals wordset, but also our own, more powerful locals wordset (we
1611: implemented the ANS Forth locals wordset through our locals wordset).
1612:
1.24 anton 1613: The ideas in this section have also been published in the paper
1614: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1615: at EuroForth '94; it is available at
1.48 ! anton 1616: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1.24 anton 1617:
1.2 anton 1618: @menu
1.17 anton 1619: * Gforth locals::
1.4 anton 1620: * ANS Forth locals::
1.2 anton 1621: @end menu
1622:
1.17 anton 1623: @node Gforth locals, ANS Forth locals, Locals, Locals
1624: @subsection Gforth locals
1.43 anton 1625: @cindex Gforth locals
1626: @cindex locals, Gforth style
1.2 anton 1627:
1628: Locals can be defined with
1629:
1630: @example
1631: @{ local1 local2 ... -- comment @}
1632: @end example
1633: or
1634: @example
1635: @{ local1 local2 ... @}
1636: @end example
1637:
1638: E.g.,
1639: @example
1640: : max @{ n1 n2 -- n3 @}
1641: n1 n2 > if
1642: n1
1643: else
1644: n2
1645: endif ;
1646: @end example
1647:
1648: The similarity of locals definitions with stack comments is intended. A
1649: locals definition often replaces the stack comment of a word. The order
1650: of the locals corresponds to the order in a stack comment and everything
1651: after the @code{--} is really a comment.
1652:
1653: This similarity has one disadvantage: It is too easy to confuse locals
1654: declarations with stack comments, causing bugs and making them hard to
1655: find. However, this problem can be avoided by appropriate coding
1656: conventions: Do not use both notations in the same program. If you do,
1657: they should be distinguished using additional means, e.g. by position.
1658:
1.43 anton 1659: @cindex types of locals
1660: @cindex locals types
1.2 anton 1661: The name of the local may be preceded by a type specifier, e.g.,
1662: @code{F:} for a floating point value:
1663:
1664: @example
1665: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1666: \ complex multiplication
1667: Ar Br f* Ai Bi f* f-
1668: Ar Bi f* Ai Br f* f+ ;
1669: @end example
1670:
1.43 anton 1671: @cindex flavours of locals
1672: @cindex locals flavours
1673: @cindex value-flavoured locals
1674: @cindex variable-flavoured locals
1.17 anton 1675: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2 anton 1676: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1677: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1678: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1679: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1680: produces its address (which becomes invalid when the variable's scope is
1.41 anton 1681: left). E.g., the standard word @code{emit} can be defined in terms of
1.2 anton 1682: @code{type} like this:
1683:
1684: @example
1685: : emit @{ C^ char* -- @}
1686: char* 1 type ;
1687: @end example
1688:
1.43 anton 1689: @cindex default type of locals
1690: @cindex locals, default type
1.2 anton 1691: A local without type specifier is a @code{W:} local. Both flavours of
1692: locals are initialized with values from the data or FP stack.
1693:
1694: Currently there is no way to define locals with user-defined data
1695: structures, but we are working on it.
1696:
1.17 anton 1697: Gforth allows defining locals everywhere in a colon definition. This
1.7 pazsan 1698: poses the following questions:
1.2 anton 1699:
1.4 anton 1700: @menu
1701: * Where are locals visible by name?::
1.14 anton 1702: * How long do locals live?::
1.4 anton 1703: * Programming Style::
1704: * Implementation::
1705: @end menu
1706:
1.17 anton 1707: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2 anton 1708: @subsubsection Where are locals visible by name?
1.43 anton 1709: @cindex locals visibility
1710: @cindex visibility of locals
1711: @cindex scope of locals
1.2 anton 1712:
1713: Basically, the answer is that locals are visible where you would expect
1714: it in block-structured languages, and sometimes a little longer. If you
1715: want to restrict the scope of a local, enclose its definition in
1716: @code{SCOPE}...@code{ENDSCOPE}.
1717:
1718: doc-scope
1719: doc-endscope
1720:
1721: These words behave like control structure words, so you can use them
1722: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1723: arbitrary ways.
1724:
1725: If you want a more exact answer to the visibility question, here's the
1726: basic principle: A local is visible in all places that can only be
1727: reached through the definition of the local@footnote{In compiler
1728: construction terminology, all places dominated by the definition of the
1729: local.}. In other words, it is not visible in places that can be reached
1730: without going through the definition of the local. E.g., locals defined
1731: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1732: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1733: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1734:
1735: The reasoning behind this solution is: We want to have the locals
1736: visible as long as it is meaningful. The user can always make the
1737: visibility shorter by using explicit scoping. In a place that can
1738: only be reached through the definition of a local, the meaning of a
1739: local name is clear. In other places it is not: How is the local
1740: initialized at the control flow path that does not contain the
1741: definition? Which local is meant, if the same name is defined twice in
1742: two independent control flow paths?
1743:
1744: This should be enough detail for nearly all users, so you can skip the
1.41 anton 1745: rest of this section. If you really must know all the gory details and
1.2 anton 1746: options, read on.
1747:
1748: In order to implement this rule, the compiler has to know which places
1749: are unreachable. It knows this automatically after @code{AHEAD},
1750: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1751: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1752: compiler that the control flow never reaches that place. If
1753: @code{UNREACHABLE} is not used where it could, the only consequence is
1754: that the visibility of some locals is more limited than the rule above
1755: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1756: lie to the compiler), buggy code will be produced.
1757:
1.43 anton 1758: doc-unreachable
1759:
1.2 anton 1760: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3 anton 1761: does not know which locals will be visible on the incoming
1762: back-edge. All problems discussed in the following are due to this
1763: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1764: loops as examples; the discussion also applies to @code{?DO} and other
1.2 anton 1765: loops). Perhaps the most insidious example is:
1766: @example
1767: AHEAD
1768: BEGIN
1769: x
1770: [ 1 CS-ROLL ] THEN
1.4 anton 1771: @{ x @}
1.2 anton 1772: ...
1773: UNTIL
1774: @end example
1775:
1776: This should be legal according to the visibility rule. The use of
1777: @code{x} can only be reached through the definition; but that appears
1778: textually below the use.
1779:
1780: From this example it is clear that the visibility rules cannot be fully
1781: implemented without major headaches. Our implementation treats common
1782: cases as advertised and the exceptions are treated in a safe way: The
1783: compiler makes a reasonable guess about the locals visible after a
1784: @code{BEGIN}; if it is too pessimistic, the
1785: user will get a spurious error about the local not being defined; if the
1786: compiler is too optimistic, it will notice this later and issue a
1787: warning. In the case above the compiler would complain about @code{x}
1788: being undefined at its use. You can see from the obscure examples in
1789: this section that it takes quite unusual control structures to get the
1790: compiler into trouble, and even then it will often do fine.
1791:
1792: If the @code{BEGIN} is reachable from above, the most optimistic guess
1793: is that all locals visible before the @code{BEGIN} will also be
1794: visible after the @code{BEGIN}. This guess is valid for all loops that
1795: are entered only through the @code{BEGIN}, in particular, for normal
1796: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1797: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1798: compiler. When the branch to the @code{BEGIN} is finally generated by
1799: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1.41 anton 1800: warns the user if it was too optimistic:
1.2 anton 1801: @example
1802: IF
1.4 anton 1803: @{ x @}
1.2 anton 1804: BEGIN
1805: \ x ?
1806: [ 1 cs-roll ] THEN
1807: ...
1808: UNTIL
1809: @end example
1810:
1811: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1812: optimistically assumes that it lives until the @code{THEN}. It notices
1813: this difference when it compiles the @code{UNTIL} and issues a
1814: warning. The user can avoid the warning, and make sure that @code{x}
1815: is not used in the wrong area by using explicit scoping:
1816: @example
1817: IF
1818: SCOPE
1.4 anton 1819: @{ x @}
1.2 anton 1820: ENDSCOPE
1821: BEGIN
1822: [ 1 cs-roll ] THEN
1823: ...
1824: UNTIL
1825: @end example
1826:
1827: Since the guess is optimistic, there will be no spurious error messages
1828: about undefined locals.
1829:
1830: If the @code{BEGIN} is not reachable from above (e.g., after
1831: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1832: optimistic guess, as the locals visible after the @code{BEGIN} may be
1833: defined later. Therefore, the compiler assumes that no locals are
1.17 anton 1834: visible after the @code{BEGIN}. However, the user can use
1.2 anton 1835: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17 anton 1836: visible at the BEGIN as at the point where the top control-flow stack
1837: item was created.
1.2 anton 1838:
1839: doc-assume-live
1840:
1841: E.g.,
1842: @example
1.4 anton 1843: @{ x @}
1.2 anton 1844: AHEAD
1845: ASSUME-LIVE
1846: BEGIN
1847: x
1848: [ 1 CS-ROLL ] THEN
1849: ...
1850: UNTIL
1851: @end example
1852:
1853: Other cases where the locals are defined before the @code{BEGIN} can be
1854: handled by inserting an appropriate @code{CS-ROLL} before the
1855: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1856: behind the @code{ASSUME-LIVE}).
1857:
1858: Cases where locals are defined after the @code{BEGIN} (but should be
1859: visible immediately after the @code{BEGIN}) can only be handled by
1860: rearranging the loop. E.g., the ``most insidious'' example above can be
1861: arranged into:
1862: @example
1863: BEGIN
1.4 anton 1864: @{ x @}
1.2 anton 1865: ... 0=
1866: WHILE
1867: x
1868: REPEAT
1869: @end example
1870:
1.17 anton 1871: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2 anton 1872: @subsubsection How long do locals live?
1.43 anton 1873: @cindex locals lifetime
1874: @cindex lifetime of locals
1.2 anton 1875:
1876: The right answer for the lifetime question would be: A local lives at
1877: least as long as it can be accessed. For a value-flavoured local this
1878: means: until the end of its visibility. However, a variable-flavoured
1879: local could be accessed through its address far beyond its visibility
1880: scope. Ultimately, this would mean that such locals would have to be
1881: garbage collected. Since this entails un-Forth-like implementation
1882: complexities, I adopted the same cowardly solution as some other
1883: languages (e.g., C): The local lives only as long as it is visible;
1884: afterwards its address is invalid (and programs that access it
1885: afterwards are erroneous).
1886:
1.17 anton 1887: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2 anton 1888: @subsubsection Programming Style
1.43 anton 1889: @cindex locals programming style
1890: @cindex programming style, locals
1.2 anton 1891:
1892: The freedom to define locals anywhere has the potential to change
1893: programming styles dramatically. In particular, the need to use the
1894: return stack for intermediate storage vanishes. Moreover, all stack
1895: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1896: determined arguments) can be eliminated: If the stack items are in the
1897: wrong order, just write a locals definition for all of them; then
1898: write the items in the order you want.
1899:
1900: This seems a little far-fetched and eliminating stack manipulations is
1.4 anton 1901: unlikely to become a conscious programming objective. Still, the number
1902: of stack manipulations will be reduced dramatically if local variables
1.17 anton 1903: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4 anton 1904: a traditional implementation of @code{max}).
1.2 anton 1905:
1906: This shows one potential benefit of locals: making Forth programs more
1907: readable. Of course, this benefit will only be realized if the
1908: programmers continue to honour the principle of factoring instead of
1909: using the added latitude to make the words longer.
1910:
1.43 anton 1911: @cindex single-assignment style for locals
1.2 anton 1912: Using @code{TO} can and should be avoided. Without @code{TO},
1913: every value-flavoured local has only a single assignment and many
1914: advantages of functional languages apply to Forth. I.e., programs are
1915: easier to analyse, to optimize and to read: It is clear from the
1916: definition what the local stands for, it does not turn into something
1917: different later.
1918:
1919: E.g., a definition using @code{TO} might look like this:
1920: @example
1921: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1922: u1 u2 min 0
1923: ?do
1.36 anton 1924: addr1 c@@ addr2 c@@ -
1.31 anton 1925: ?dup-if
1.2 anton 1926: unloop exit
1927: then
1928: addr1 char+ TO addr1
1929: addr2 char+ TO addr2
1930: loop
1931: u1 u2 - ;
1932: @end example
1933: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1934: every loop iteration. @code{strcmp} is a typical example of the
1935: readability problems of using @code{TO}. When you start reading
1936: @code{strcmp}, you think that @code{addr1} refers to the start of the
1937: string. Only near the end of the loop you realize that it is something
1938: else.
1939:
1940: This can be avoided by defining two locals at the start of the loop that
1941: are initialized with the right value for the current iteration.
1942: @example
1943: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1944: addr1 addr2
1945: u1 u2 min 0
1946: ?do @{ s1 s2 @}
1.36 anton 1947: s1 c@@ s2 c@@ -
1.31 anton 1948: ?dup-if
1.2 anton 1949: unloop exit
1950: then
1951: s1 char+ s2 char+
1952: loop
1953: 2drop
1954: u1 u2 - ;
1955: @end example
1956: Here it is clear from the start that @code{s1} has a different value
1957: in every loop iteration.
1958:
1.17 anton 1959: @node Implementation, , Programming Style, Gforth locals
1.2 anton 1960: @subsubsection Implementation
1.43 anton 1961: @cindex locals implementation
1962: @cindex implementation of locals
1.2 anton 1963:
1.43 anton 1964: @cindex locals stack
1.17 anton 1965: Gforth uses an extra locals stack. The most compelling reason for
1.2 anton 1966: this is that the return stack is not float-aligned; using an extra stack
1967: also eliminates the problems and restrictions of using the return stack
1968: as locals stack. Like the other stacks, the locals stack grows toward
1969: lower addresses. A few primitives allow an efficient implementation:
1970:
1971: doc-@local#
1972: doc-f@local#
1973: doc-laddr#
1974: doc-lp+!#
1975: doc-lp!
1976: doc->l
1977: doc-f>l
1978:
1979: In addition to these primitives, some specializations of these
1980: primitives for commonly occurring inline arguments are provided for
1981: efficiency reasons, e.g., @code{@@local0} as specialization of
1982: @code{@@local#} for the inline argument 0. The following compiling words
1983: compile the right specialized version, or the general version, as
1984: appropriate:
1985:
1.12 anton 1986: doc-compile-@local
1987: doc-compile-f@local
1.2 anton 1988: doc-compile-lp+!
1989:
1990: Combinations of conditional branches and @code{lp+!#} like
1991: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1992: is taken) are provided for efficiency and correctness in loops.
1993:
1994: A special area in the dictionary space is reserved for keeping the
1995: local variable names. @code{@{} switches the dictionary pointer to this
1996: area and @code{@}} switches it back and generates the locals
1997: initializing code. @code{W:} etc.@ are normal defining words. This
1998: special area is cleared at the start of every colon definition.
1999:
1.43 anton 2000: @cindex wordlist for defining locals
1.17 anton 2001: A special feature of Gforth's dictionary is used to implement the
1.2 anton 2002: definition of locals without type specifiers: every wordlist (aka
2003: vocabulary) has its own methods for searching
1.4 anton 2004: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2 anton 2005: with a special search method: When it is searched for a word, it
2006: actually creates that word using @code{W:}. @code{@{} changes the search
2007: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
2008: and then the wordlist for defining locals without type specifiers.
2009:
2010: The lifetime rules support a stack discipline within a colon
2011: definition: The lifetime of a local is either nested with other locals
2012: lifetimes or it does not overlap them.
2013:
2014: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
2015: pointer manipulation is generated. Between control structure words
2016: locals definitions can push locals onto the locals stack. @code{AGAIN}
2017: is the simplest of the other three control flow words. It has to
2018: restore the locals stack depth of the corresponding @code{BEGIN}
2019: before branching. The code looks like this:
2020: @format
2021: @code{lp+!#} current-locals-size @minus{} dest-locals-size
2022: @code{branch} <begin>
2023: @end format
2024:
2025: @code{UNTIL} is a little more complicated: If it branches back, it
2026: must adjust the stack just like @code{AGAIN}. But if it falls through,
2027: the locals stack must not be changed. The compiler generates the
2028: following code:
2029: @format
2030: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
2031: @end format
2032: The locals stack pointer is only adjusted if the branch is taken.
2033:
2034: @code{THEN} can produce somewhat inefficient code:
2035: @format
2036: @code{lp+!#} current-locals-size @minus{} orig-locals-size
2037: <orig target>:
2038: @code{lp+!#} orig-locals-size @minus{} new-locals-size
2039: @end format
2040: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4 anton 2041: level at the @var{orig} point to the level after the @code{THEN}. The
1.2 anton 2042: first @code{lp+!#} adjusts the locals stack pointer from the current
2043: level to the level at the orig point, so the complete effect is an
2044: adjustment from the current level to the right level after the
2045: @code{THEN}.
2046:
1.43 anton 2047: @cindex locals information on the control-flow stack
2048: @cindex control-flow stack items, locals information
1.2 anton 2049: In a conventional Forth implementation a dest control-flow stack entry
2050: is just the target address and an orig entry is just the address to be
2051: patched. Our locals implementation adds a wordlist to every orig or dest
2052: item. It is the list of locals visible (or assumed visible) at the point
2053: described by the entry. Our implementation also adds a tag to identify
2054: the kind of entry, in particular to differentiate between live and dead
2055: (reachable and unreachable) orig entries.
2056:
2057: A few unusual operations have to be performed on locals wordlists:
2058:
2059: doc-common-list
2060: doc-sub-list?
2061: doc-list-size
2062:
2063: Several features of our locals wordlist implementation make these
2064: operations easy to implement: The locals wordlists are organised as
2065: linked lists; the tails of these lists are shared, if the lists
2066: contain some of the same locals; and the address of a name is greater
2067: than the address of the names behind it in the list.
2068:
2069: Another important implementation detail is the variable
2070: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
2071: determine if they can be reached directly or only through the branch
2072: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
2073: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
2074: definition, by @code{BEGIN} and usually by @code{THEN}.
2075:
2076: Counted loops are similar to other loops in most respects, but
2077: @code{LEAVE} requires special attention: It performs basically the same
2078: service as @code{AHEAD}, but it does not create a control-flow stack
2079: entry. Therefore the information has to be stored elsewhere;
2080: traditionally, the information was stored in the target fields of the
2081: branches created by the @code{LEAVE}s, by organizing these fields into a
2082: linked list. Unfortunately, this clever trick does not provide enough
2083: space for storing our extended control flow information. Therefore, we
2084: introduce another stack, the leave stack. It contains the control-flow
2085: stack entries for all unresolved @code{LEAVE}s.
2086:
2087: Local names are kept until the end of the colon definition, even if
2088: they are no longer visible in any control-flow path. In a few cases
2089: this may lead to increased space needs for the locals name area, but
2090: usually less than reclaiming this space would cost in code size.
2091:
2092:
1.17 anton 2093: @node ANS Forth locals, , Gforth locals, Locals
1.2 anton 2094: @subsection ANS Forth locals
1.43 anton 2095: @cindex locals, ANS Forth style
1.2 anton 2096:
2097: The ANS Forth locals wordset does not define a syntax for locals, but
2098: words that make it possible to define various syntaxes. One of the
1.17 anton 2099: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2 anton 2100: wordset, i.e.:
2101:
2102: @example
2103: @{ local1 local2 ... -- comment @}
2104: @end example
2105: or
2106: @example
2107: @{ local1 local2 ... @}
2108: @end example
2109:
2110: The order of the locals corresponds to the order in a stack comment. The
2111: restrictions are:
1.1 anton 2112:
1.2 anton 2113: @itemize @bullet
2114: @item
1.17 anton 2115: Locals can only be cell-sized values (no type specifiers are allowed).
1.2 anton 2116: @item
2117: Locals can be defined only outside control structures.
2118: @item
2119: Locals can interfere with explicit usage of the return stack. For the
2120: exact (and long) rules, see the standard. If you don't use return stack
1.17 anton 2121: accessing words in a definition using locals, you will be all right. The
1.2 anton 2122: purpose of this rule is to make locals implementation on the return
2123: stack easier.
2124: @item
2125: The whole definition must be in one line.
2126: @end itemize
2127:
1.35 anton 2128: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
2129: Defining Words}). I.e., they are initialized from the stack. Using their
1.2 anton 2130: name produces their value. Their value can be changed using @code{TO}.
2131:
1.17 anton 2132: Since this syntax is supported by Gforth directly, you need not do
1.2 anton 2133: anything to use it. If you want to port a program using this syntax to
1.30 anton 2134: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
2135: syntax on the other system.
1.2 anton 2136:
2137: Note that a syntax shown in the standard, section A.13 looks
2138: similar, but is quite different in having the order of locals
2139: reversed. Beware!
2140:
2141: The ANS Forth locals wordset itself consists of the following word
2142:
2143: doc-(local)
2144:
2145: The ANS Forth locals extension wordset defines a syntax, but it is so
2146: awful that we strongly recommend not to use it. We have implemented this
1.17 anton 2147: syntax to make porting to Gforth easy, but do not document it here. The
1.2 anton 2148: problem with this syntax is that the locals are defined in an order
2149: reversed with respect to the standard stack comment notation, making
2150: programs harder to read, and easier to misread and miswrite. The only
2151: merit of this syntax is that it is easy to implement using the ANS Forth
2152: locals wordset.
1.3 anton 2153:
1.37 anton 2154: @node Defining Words, Tokens for Words, Locals, Words
1.4 anton 2155: @section Defining Words
1.43 anton 2156: @cindex defining words
1.4 anton 2157:
1.14 anton 2158: @menu
1.35 anton 2159: * Simple Defining Words::
2160: * Colon Definitions::
2161: * User-defined Defining Words::
2162: * Supplying names::
2163: * Interpretation and Compilation Semantics::
1.14 anton 2164: @end menu
2165:
1.35 anton 2166: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
2167: @subsection Simple Defining Words
1.43 anton 2168: @cindex simple defining words
2169: @cindex defining words, simple
1.35 anton 2170:
2171: doc-constant
2172: doc-2constant
2173: doc-fconstant
2174: doc-variable
2175: doc-2variable
2176: doc-fvariable
2177: doc-create
2178: doc-user
2179: doc-value
2180: doc-to
2181: doc-defer
2182: doc-is
2183:
2184: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
2185: @subsection Colon Definitions
1.43 anton 2186: @cindex colon definitions
1.35 anton 2187:
2188: @example
2189: : name ( ... -- ... )
2190: word1 word2 word3 ;
2191: @end example
2192:
2193: creates a word called @code{name}, that, upon execution, executes
2194: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2195:
2196: The explanation above is somewhat superficial. @xref{Interpretation and
2197: Compilation Semantics} for an in-depth discussion of some of the issues
2198: involved.
2199:
2200: doc-:
2201: doc-;
2202:
2203: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2204: @subsection User-defined Defining Words
1.43 anton 2205: @cindex user-defined defining words
2206: @cindex defining words, user-defined
1.35 anton 2207:
2208: You can create new defining words simply by wrapping defining-time code
2209: around existing defining words and putting the sequence in a colon
2210: definition.
2211:
1.43 anton 2212: @cindex @code{CREATE} ... @code{DOES>}
1.36 anton 2213: If you want the words defined with your defining words to behave
2214: differently from words defined with standard defining words, you can
1.35 anton 2215: write your defining word like this:
2216:
2217: @example
2218: : def-word ( "name" -- )
2219: Create @var{code1}
2220: DOES> ( ... -- ... )
2221: @var{code2} ;
2222:
2223: def-word name
2224: @end example
2225:
2226: Technically, this fragment defines a defining word @code{def-word}, and
2227: a word @code{name}; when you execute @code{name}, the address of the
2228: body of @code{name} is put on the data stack and @var{code2} is executed
2229: (the address of the body of @code{name} is the address @code{HERE}
1.36 anton 2230: returns immediately after the @code{CREATE}).
2231:
2232: In other words, if you make the following definitions:
2233:
2234: @example
2235: : def-word1 ( "name" -- )
2236: Create @var{code1} ;
2237:
2238: : action1 ( ... -- ... )
2239: @var{code2} ;
2240:
2241: def-word name1
2242: @end example
2243:
2244: Using @code{name1 action1} is equivalent to using @code{name}.
2245:
2246: E.g., you can implement @code{Constant} in this way:
1.35 anton 2247:
2248: @example
2249: : constant ( w "name" -- )
2250: create ,
2251: DOES> ( -- w )
1.36 anton 2252: @@ ;
1.35 anton 2253: @end example
2254:
2255: When you create a constant with @code{5 constant five}, first a new word
2256: @code{five} is created, then the value 5 is laid down in the body of
2257: @code{five} with @code{,}. When @code{five} is invoked, the address of
2258: the body is put on the stack, and @code{@@} retrieves the value 5.
2259:
1.43 anton 2260: @cindex stack effect of @code{DOES>}-parts
2261: @cindex @code{DOES>}-parts, stack effect
1.35 anton 2262: In the example above the stack comment after the @code{DOES>} specifies
2263: the stack effect of the defined words, not the stack effect of the
2264: following code (the following code expects the address of the body on
2265: the top of stack, which is not reflected in the stack comment). This is
2266: the convention that I use and recommend (it clashes a bit with using
2267: locals declarations for stack effect specification, though).
2268:
2269: @subsubsection Applications of @code{CREATE..DOES>}
1.43 anton 2270: @cindex @code{CREATE} ... @code{DOES>}, applications
1.35 anton 2271:
1.36 anton 2272: You may wonder how to use this feature. Here are some usage patterns:
1.35 anton 2273:
1.43 anton 2274: @cindex factoring similar colon definitions
1.35 anton 2275: When you see a sequence of code occurring several times, and you can
2276: identify a meaning, you will factor it out as a colon definition. When
2277: you see similar colon definitions, you can factor them using
2278: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
2279: that look very similar:
2280: @example
1.41 anton 2281: : ori, ( reg-target reg-source n -- )
1.35 anton 2282: 0 asm-reg-reg-imm ;
1.41 anton 2283: : andi, ( reg-target reg-source n -- )
1.35 anton 2284: 1 asm-reg-reg-imm ;
2285: @end example
2286:
2287: This could be factored with:
2288: @example
2289: : reg-reg-imm ( op-code -- )
2290: create ,
1.41 anton 2291: DOES> ( reg-target reg-source n -- )
1.36 anton 2292: @@ asm-reg-reg-imm ;
1.35 anton 2293:
2294: 0 reg-reg-imm ori,
2295: 1 reg-reg-imm andi,
2296: @end example
2297:
1.43 anton 2298: @cindex currying
1.35 anton 2299: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2300: supply a part of the parameters for a word (known as @dfn{currying} in
2301: the functional language community). E.g., @code{+} needs two
2302: parameters. Creating versions of @code{+} with one parameter fixed can
2303: be done like this:
2304: @example
2305: : curry+ ( n1 -- )
2306: create ,
2307: DOES> ( n2 -- n1+n2 )
1.36 anton 2308: @@ + ;
1.35 anton 2309:
2310: 3 curry+ 3+
2311: -2 curry+ 2-
2312: @end example
2313:
2314: @subsubsection The gory details of @code{CREATE..DOES>}
1.43 anton 2315: @cindex @code{CREATE} ... @code{DOES>}, details
1.35 anton 2316:
2317: doc-does>
2318:
1.43 anton 2319: @cindex @code{DOES>} in a separate definition
1.35 anton 2320: This means that you need not use @code{CREATE} and @code{DOES>} in the
2321: same definition; E.g., you can put the @code{DOES>}-part in a separate
2322: definition. This allows us to, e.g., select among different DOES>-parts:
2323: @example
2324: : does1
2325: DOES> ( ... -- ... )
2326: ... ;
2327:
2328: : does2
2329: DOES> ( ... -- ... )
2330: ... ;
2331:
2332: : def-word ( ... -- ... )
2333: create ...
2334: IF
2335: does1
2336: ELSE
2337: does2
2338: ENDIF ;
2339: @end example
2340:
1.43 anton 2341: @cindex @code{DOES>} in interpretation state
1.35 anton 2342: In a standard program you can apply a @code{DOES>}-part only if the last
2343: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2344: will override the behaviour of the last word defined in any case. In a
2345: standard program, you can use @code{DOES>} only in a colon
2346: definition. In Gforth, you can also use it in interpretation state, in a
2347: kind of one-shot mode:
2348: @example
2349: CREATE name ( ... -- ... )
2350: @var{initialization}
2351: DOES>
2352: @var{code} ;
2353: @end example
1.41 anton 2354: This is equivalent to the standard
1.35 anton 2355: @example
2356: :noname
2357: DOES>
2358: @var{code} ;
2359: CREATE name EXECUTE ( ... -- ... )
2360: @var{initialization}
2361: @end example
2362:
2363: You can get the address of the body of a word with
2364:
2365: doc->body
2366:
2367: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2368: @subsection Supplying names for the defined words
1.43 anton 2369: @cindex names for defined words
2370: @cindex defining words, name parameter
1.35 anton 2371:
1.43 anton 2372: @cindex defining words, name given in a string
1.35 anton 2373: By default, defining words take the names for the defined words from the
2374: input stream. Sometimes you want to supply the name from a string. You
2375: can do this with
2376:
2377: doc-nextname
2378:
2379: E.g.,
2380:
2381: @example
2382: s" foo" nextname create
2383: @end example
2384: is equivalent to
2385: @example
2386: create foo
2387: @end example
2388:
1.43 anton 2389: @cindex defining words without name
1.35 anton 2390: Sometimes you want to define a word without a name. You can do this with
2391:
2392: doc-noname
2393:
1.43 anton 2394: @cindex execution token of last defined word
1.35 anton 2395: To make any use of the newly defined word, you need its execution
2396: token. You can get it with
2397:
2398: doc-lastxt
2399:
2400: E.g., you can initialize a deferred word with an anonymous colon
2401: definition:
2402: @example
2403: Defer deferred
2404: noname : ( ... -- ... )
2405: ... ;
2406: lastxt IS deferred
2407: @end example
2408:
2409: @code{lastxt} also works when the last word was not defined as
2410: @code{noname}.
2411:
2412: The standard has also recognized the need for anonymous words and
2413: provides
2414:
2415: doc-:noname
2416:
2417: This leaves the execution token for the word on the stack after the
2418: closing @code{;}. You can rewrite the last example with @code{:noname}:
2419: @example
2420: Defer deferred
2421: :noname ( ... -- ... )
2422: ... ;
2423: IS deferred
2424: @end example
2425:
2426: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2427: @subsection Interpretation and Compilation Semantics
1.43 anton 2428: @cindex semantics, interpretation and compilation
1.35 anton 2429:
1.43 anton 2430: @cindex interpretation semantics
1.36 anton 2431: The @dfn{interpretation semantics} of a word are what the text
2432: interpreter does when it encounters the word in interpret state. It also
2433: appears in some other contexts, e.g., the execution token returned by
2434: @code{' @var{word}} identifies the interpretation semantics of
2435: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
2436: interpret-state text interpretation of @code{@var{word}}).
2437:
1.43 anton 2438: @cindex compilation semantics
1.36 anton 2439: The @dfn{compilation semantics} of a word are what the text interpreter
2440: does when it encounters the word in compile state. It also appears in
2441: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2442: standard terminology, ``appends to the current definition''.} the
2443: compilation semantics of @var{word}.
2444:
1.43 anton 2445: @cindex execution semantics
1.36 anton 2446: The standard also talks about @dfn{execution semantics}. They are used
2447: only for defining the interpretation and compilation semantics of many
2448: words. By default, the interpretation semantics of a word are to
2449: @code{execute} its execution semantics, and the compilation semantics of
2450: a word are to @code{compile,} its execution semantics.@footnote{In
2451: standard terminology: The default interpretation semantics are its
2452: execution semantics; the default compilation semantics are to append its
2453: execution semantics to the execution semantics of the current
2454: definition.}
2455:
1.43 anton 2456: @cindex immediate words
1.36 anton 2457: You can change the compilation semantics into @code{execute}ing the
2458: execution semantics with
2459:
1.35 anton 2460: doc-immediate
1.36 anton 2461:
1.43 anton 2462: @cindex compile-only words
1.36 anton 2463: You can remove the interpretation semantics of a word with
2464:
2465: doc-compile-only
2466: doc-restrict
2467:
2468: Note that ticking (@code{'}) compile-only words gives an error
2469: (``Interpreting a compile-only word'').
2470:
2471: Gforth also allows you to define words with arbitrary combinations of
2472: interpretation and compilation semantics.
2473:
1.35 anton 2474: doc-interpret/compile:
2475:
1.36 anton 2476: This feature was introduced for implementing @code{TO} and @code{S"}. I
2477: recommend that you do not define such words, as cute as they may be:
2478: they make it hard to get at both parts of the word in some contexts.
2479: E.g., assume you want to get an execution token for the compilation
2480: part. Instead, define two words, one that embodies the interpretation
2481: part, and one that embodies the compilation part.
2482:
2483: There is, however, a potentially useful application of this feature:
2484: Providing differing implementations for the default semantics. While
2485: this introduces redundancy and is therefore usually a bad idea, a
2486: performance improvement may be worth the trouble. E.g., consider the
2487: word @code{foobar}:
2488:
2489: @example
2490: : foobar
2491: foo bar ;
2492: @end example
2493:
2494: Let us assume that @code{foobar} is called so frequently that the
2495: calling overhead would take a significant amount of the run-time. We can
2496: optimize it with @code{interpret/compile:}:
1.35 anton 2497:
1.36 anton 2498: @example
2499: :noname
2500: foo bar ;
2501: :noname
2502: POSTPONE foo POSTPONE bar ;
2503: interpret/compile: foobar
2504: @end example
2505:
2506: This definition has the same interpretation semantics and essentially
2507: the same compilation semantics as the simple definition of
2508: @code{foobar}, but the implementation of the compilation semantics is
2509: more efficient with respect to run-time.
2510:
1.43 anton 2511: @cindex state-smart words are a bad idea
1.36 anton 2512: Some people try to use state-smart words to emulate the feature provided
2513: by @code{interpret/compile:} (words are state-smart if they check
2514: @code{STATE} during execution). E.g., they would try to code
2515: @code{foobar} like this:
2516:
2517: @example
2518: : foobar
2519: STATE @@
2520: IF ( compilation state )
2521: POSTPONE foo POSTPONE bar
2522: ELSE
2523: foo bar
2524: ENDIF ; immediate
2525: @end example
2526:
2527: While this works if @code{foobar} is processed only by the text
2528: interpreter, it does not work in other contexts (like @code{'} or
2529: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2530: for a state-smart word, not for the interpretation semantics of the
2531: original @code{foobar}; when you execute this execution token (directly
2532: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2533: state, the result will not be what you expected (i.e., it will not
2534: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2535: write them!
2536:
1.43 anton 2537: @cindex defining words with arbitrary semantics combinations
1.36 anton 2538: It is also possible to write defining words that define words with
2539: arbitrary combinations of interpretation and compilation semantics (or,
2540: preferably, arbitrary combinations of implementations of the default
2541: semantics). In general, this looks like:
2542:
2543: @example
2544: : def-word
2545: create-interpret/compile
2546: @var{code1}
2547: interpretation>
2548: @var{code2}
2549: <interpretation
2550: compilation>
2551: @var{code3}
2552: <compilation ;
2553: @end example
2554:
2555: For a @var{word} defined with @code{def-word}, the interpretation
2556: semantics are to push the address of the body of @var{word} and perform
2557: @var{code2}, and the compilation semantics are to push the address of
2558: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2559: can also be defined like this:
2560:
2561: @example
2562: : constant ( n "name" -- )
2563: create-interpret/compile
2564: ,
2565: interpretation> ( -- n )
2566: @@
2567: <interpretation
2568: compilation> ( compilation. -- ; run-time. -- n )
2569: @@ postpone literal
2570: <compilation ;
2571: @end example
2572:
2573: doc-create-interpret/compile
2574: doc-interpretation>
2575: doc-<interpretation
2576: doc-compilation>
2577: doc-<compilation
2578:
2579: Note that words defined with @code{interpret/compile:} and
2580: @code{create-interpret/compile} have an extended header structure that
2581: differs from other words; however, unless you try to access them with
2582: plain address arithmetic, you should not notice this. Words for
2583: accessing the header structure usually know how to deal with this; e.g.,
2584: @code{' word >body} also gives you the body of a word created with
2585: @code{create-interpret/compile}.
1.4 anton 2586:
1.37 anton 2587: @node Tokens for Words, Wordlists, Defining Words, Words
2588: @section Tokens for Words
1.43 anton 2589: @cindex tokens for words
1.37 anton 2590:
2591: This chapter describes the creation and use of tokens that represent
2592: words on the stack (and in data space).
2593:
2594: Named words have interpretation and compilation semantics. Unnamed words
2595: just have execution semantics.
2596:
1.43 anton 2597: @cindex execution token
1.37 anton 2598: An @dfn{execution token} represents the execution semantics of an
2599: unnamed word. An execution token occupies one cell. As explained in
2600: section @ref{Supplying names}, the execution token of the last words
2601: defined can be produced with
2602:
2603: short-lastxt
2604:
2605: You can perform the semantics represented by an execution token with
2606: doc-execute
2607: You can compile the word with
2608: doc-compile,
2609:
1.43 anton 2610: @cindex code field address
2611: @cindex CFA
1.37 anton 2612: In Gforth, the abstract data type @emph{execution token} is implemented
2613: as CFA (code field address).
2614:
2615: The interpretation semantics of a named word are also represented by an
2616: execution token. You can get it with
2617:
2618: doc-[']
2619: doc-'
2620:
2621: For literals, you use @code{'} in interpreted code and @code{[']} in
2622: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
2623: by complaining about compile-only words. To get an execution token for a
2624: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
2625: @var{X} drop}.
2626:
1.43 anton 2627: @cindex compilation token
1.37 anton 2628: The compilation semantics are represented by a @dfn{compilation token}
2629: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
2630: execution token. The compilation semantics represented by the
2631: compilation token can be performed with @code{execute}, which consumes
2632: the whole compilation token, with an additional stack effect determined
2633: by the represented compilation semantics.
2634:
2635: doc-[comp']
2636: doc-comp'
2637:
1.38 anton 2638: You can compile the compilation semantics with @code{postpone,}. I.e.,
2639: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
2640: @var{word}}.
2641:
2642: doc-postpone,
2643:
1.37 anton 2644: At present, the @var{w} part of a compilation token is an execution
2645: token, and the @var{xt} part represents either @code{execute} or
1.41 anton 2646: @code{compile,}. However, don't rely on that knowledge, unless necessary;
1.37 anton 2647: we may introduce unusual compilation tokens in the future (e.g.,
2648: compilation tokens representing the compilation semantics of literals).
2649:
1.43 anton 2650: @cindex name token
2651: @cindex name field address
2652: @cindex NFA
1.37 anton 2653: Named words are also represented by the @dfn{name token}. The abstract
2654: data type @emph{name token} is implemented as NFA (name field address).
2655:
2656: doc-find-name
2657: doc-name>int
2658: doc-name?int
2659: doc-name>comp
2660: doc-name>string
2661:
2662: @node Wordlists, Files, Tokens for Words, Words
1.4 anton 2663: @section Wordlists
2664:
2665: @node Files, Blocks, Wordlists, Words
2666: @section Files
2667:
2668: @node Blocks, Other I/O, Files, Words
2669: @section Blocks
2670:
2671: @node Other I/O, Programming Tools, Blocks, Words
2672: @section Other I/O
2673:
1.18 anton 2674: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4 anton 2675: @section Programming Tools
1.43 anton 2676: @cindex programming tools
1.4 anton 2677:
1.5 anton 2678: @menu
2679: * Debugging:: Simple and quick.
2680: * Assertions:: Making your programs self-checking.
2681: @end menu
2682:
2683: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4 anton 2684: @subsection Debugging
1.43 anton 2685: @cindex debugging
1.4 anton 2686:
2687: The simple debugging aids provided in @file{debugging.fs}
2688: are meant to support a different style of debugging than the
2689: tracing/stepping debuggers used in languages with long turn-around
2690: times.
2691:
1.41 anton 2692: A much better (faster) way in fast-compiling languages is to add
1.4 anton 2693: printing code at well-selected places, let the program run, look at
2694: the output, see where things went wrong, add more printing code, etc.,
2695: until the bug is found.
2696:
2697: The word @code{~~} is easy to insert. It just prints debugging
2698: information (by default the source location and the stack contents). It
2699: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2700: query-replace them with nothing). The deferred words
2701: @code{printdebugdata} and @code{printdebugline} control the output of
2702: @code{~~}. The default source location output format works well with
2703: Emacs' compilation mode, so you can step through the program at the
1.5 anton 2704: source level using @kbd{C-x `} (the advantage over a stepping debugger
2705: is that you can step in any direction and you know where the crash has
2706: happened or where the strange data has occurred).
1.4 anton 2707:
2708: Note that the default actions clobber the contents of the pictured
2709: numeric output string, so you should not use @code{~~}, e.g., between
2710: @code{<#} and @code{#>}.
2711:
2712: doc-~~
2713: doc-printdebugdata
2714: doc-printdebugline
2715:
1.5 anton 2716: @node Assertions, , Debugging, Programming Tools
1.4 anton 2717: @subsection Assertions
1.43 anton 2718: @cindex assertions
1.4 anton 2719:
1.5 anton 2720: It is a good idea to make your programs self-checking, in particular, if
2721: you use an assumption (e.g., that a certain field of a data structure is
1.17 anton 2722: never zero) that may become wrong during maintenance. Gforth supports
1.5 anton 2723: assertions for this purpose. They are used like this:
2724:
2725: @example
2726: assert( @var{flag} )
2727: @end example
2728:
2729: The code between @code{assert(} and @code{)} should compute a flag, that
2730: should be true if everything is alright and false otherwise. It should
2731: not change anything else on the stack. The overall stack effect of the
2732: assertion is @code{( -- )}. E.g.
2733:
2734: @example
2735: assert( 1 1 + 2 = ) \ what we learn in school
2736: assert( dup 0<> ) \ assert that the top of stack is not zero
2737: assert( false ) \ this code should not be reached
2738: @end example
2739:
2740: The need for assertions is different at different times. During
2741: debugging, we want more checking, in production we sometimes care more
2742: for speed. Therefore, assertions can be turned off, i.e., the assertion
2743: becomes a comment. Depending on the importance of an assertion and the
2744: time it takes to check it, you may want to turn off some assertions and
1.17 anton 2745: keep others turned on. Gforth provides several levels of assertions for
1.5 anton 2746: this purpose:
2747:
2748: doc-assert0(
2749: doc-assert1(
2750: doc-assert2(
2751: doc-assert3(
2752: doc-assert(
2753: doc-)
2754:
2755: @code{Assert(} is the same as @code{assert1(}. The variable
2756: @code{assert-level} specifies the highest assertions that are turned
2757: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2758: @code{assert1(} assertions perform checking, while @code{assert2(} and
2759: @code{assert3(} assertions are treated as comments.
2760:
2761: Note that the @code{assert-level} is evaluated at compile-time, not at
2762: run-time. I.e., you cannot turn assertions on or off at run-time, you
2763: have to set the @code{assert-level} appropriately before compiling a
2764: piece of code. You can compile several pieces of code at several
2765: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2766: written code at level 3).
2767:
2768: doc-assert-level
2769:
2770: If an assertion fails, a message compatible with Emacs' compilation mode
2771: is produced and the execution is aborted (currently with @code{ABORT"}.
2772: If there is interest, we will introduce a special throw code. But if you
2773: intend to @code{catch} a specific condition, using @code{throw} is
2774: probably more appropriate than an assertion).
2775:
1.18 anton 2776: @node Assembler and Code words, Threading Words, Programming Tools, Words
2777: @section Assembler and Code words
1.43 anton 2778: @cindex assembler
2779: @cindex code words
1.18 anton 2780:
2781: Gforth provides some words for defining primitives (words written in
2782: machine code), and for defining the the machine-code equivalent of
2783: @code{DOES>}-based defining words. However, the machine-independent
1.40 anton 2784: nature of Gforth poses a few problems: First of all, Gforth runs on
1.18 anton 2785: several architectures, so it can provide no standard assembler. What's
2786: worse is that the register allocation not only depends on the processor,
1.25 anton 2787: but also on the @code{gcc} version and options used.
1.18 anton 2788:
1.25 anton 2789: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18 anton 2790: header structure), so a system-independent assembler may be used in
2791: Gforth. If you do not have an assembler, you can compile machine code
2792: directly with @code{,} and @code{c,}.
2793:
2794: doc-assembler
2795: doc-code
2796: doc-end-code
2797: doc-;code
2798: doc-flush-icache
2799:
2800: If @code{flush-icache} does not work correctly, @code{code} words
2801: etc. will not work (reliably), either.
2802:
2803: These words are rarely used. Therefore they reside in @code{code.fs},
2804: which is usually not loaded (except @code{flush-icache}, which is always
1.19 anton 2805: present). You can load them with @code{require code.fs}.
1.18 anton 2806:
1.43 anton 2807: @cindex registers of the inner interpreter
1.25 anton 2808: In the assembly code you will want to refer to the inner interpreter's
2809: registers (e.g., the data stack pointer) and you may want to use other
2810: registers for temporary storage. Unfortunately, the register allocation
2811: is installation-dependent.
2812:
2813: The easiest solution is to use explicit register declarations
2814: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2815: GNU C Manual}) for all of the inner interpreter's registers: You have to
2816: compile Gforth with @code{-DFORCE_REG} (configure option
2817: @code{--enable-force-reg}) and the appropriate declarations must be
2818: present in the @code{machine.h} file (see @code{mips.h} for an example;
2819: you can find a full list of all declarable register symbols with
2820: @code{grep register engine.c}). If you give explicit registers to all
2821: variables that are declared at the beginning of @code{engine()}, you
2822: should be able to use the other caller-saved registers for temporary
2823: storage. Alternatively, you can use the @code{gcc} option
2824: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2825: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2826: (however, this restriction on register allocation may slow Gforth
2827: significantly).
2828:
2829: If this solution is not viable (e.g., because @code{gcc} does not allow
2830: you to explicitly declare all the registers you need), you have to find
2831: out by looking at the code where the inner interpreter's registers
2832: reside and which registers can be used for temporary storage. You can
2833: get an assembly listing of the engine's code with @code{make engine.s}.
2834:
2835: In any case, it is good practice to abstract your assembly code from the
2836: actual register allocation. E.g., if the data stack pointer resides in
2837: register @code{$17}, create an alias for this register called @code{sp},
2838: and use that in your assembly code.
2839:
1.43 anton 2840: @cindex code words, portable
1.18 anton 2841: Another option for implementing normal and defining words efficiently
2842: is: adding the wanted functionality to the source of Gforth. For normal
1.35 anton 2843: words you just have to edit @file{primitives} (@pxref{Automatic
2844: Generation}), defining words (equivalent to @code{;CODE} words, for fast
2845: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
2846: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.18 anton 2847:
2848:
2849: @node Threading Words, , Assembler and Code words, Words
1.4 anton 2850: @section Threading Words
1.43 anton 2851: @cindex threading words
1.4 anton 2852:
1.43 anton 2853: @cindex code address
1.4 anton 2854: These words provide access to code addresses and other threading stuff
1.17 anton 2855: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4 anton 2856: abstracts away the differences between direct and indirect threading
2857: (and, for direct threading, the machine dependences). However, at
1.43 anton 2858: present this wordset is still incomplete. It is also pretty low-level;
2859: some day it will hopefully be made unnecessary by an internals wordset
1.4 anton 2860: that abstracts implementation details away completely.
2861:
2862: doc->code-address
2863: doc->does-code
2864: doc-code-address!
2865: doc-does-code!
2866: doc-does-handler!
2867: doc-/does-handler
2868:
1.18 anton 2869: The code addresses produced by various defining words are produced by
2870: the following words:
1.14 anton 2871:
1.18 anton 2872: doc-docol:
2873: doc-docon:
2874: doc-dovar:
2875: doc-douser:
2876: doc-dodefer:
2877: doc-dofield:
2878:
1.35 anton 2879: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
2880: with @code{>DOES-CODE}. If the word was defined in that way, the value
2881: returned is different from 0 and identifies the @code{DOES>} used by the
2882: defining word.
1.14 anton 2883:
1.40 anton 2884: @node Tools, ANS conformance, Words, Top
2885: @chapter Tools
2886:
2887: @menu
1.43 anton 2888: * ANS Report:: Report the words used, sorted by wordset.
1.40 anton 2889: @end menu
2890:
2891: See also @ref{Emacs and Gforth}.
2892:
2893: @node ANS Report, , Tools, Tools
2894: @section @file{ans-report.fs}: Report the words used, sorted by wordset
1.43 anton 2895: @cindex @file{ans-report.fs}
2896: @cindex report the words used in your program
2897: @cindex words used in your program
1.40 anton 2898:
2899: If you want to label a Forth program as ANS Forth Program, you must
2900: document which wordsets the program uses; for extension wordsets, it is
2901: helpful to list the words the program requires from these wordsets
2902: (because Forth systems are allowed to provide only some words of them).
2903:
2904: The @file{ans-report.fs} tool makes it easy for you to determine which
2905: words from which wordset and which non-ANS words your application
2906: uses. You simply have to include @file{ans-report.fs} before loading the
2907: program you want to check. After loading your program, you can get the
2908: report with @code{print-ans-report}. A typical use is to run this as
2909: batch job like this:
2910: @example
2911: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
2912: @end example
2913:
2914: The output looks like this (for @file{compat/control.fs}):
2915: @example
2916: The program uses the following words
2917: from CORE :
2918: : POSTPONE THEN ; immediate ?dup IF 0=
2919: from BLOCK-EXT :
2920: \
2921: from FILE :
2922: (
2923: @end example
2924:
2925: @subsection Caveats
2926:
2927: Note that @file{ans-report.fs} just checks which words are used, not whether
2928: they are used in an ANS Forth conforming way!
2929:
2930: Some words are defined in several wordsets in the
2931: standard. @file{ans-report.fs} reports them for only one of the
2932: wordsets, and not necessarily the one you expect. It depends on usage
2933: which wordset is the right one to specify. E.g., if you only use the
2934: compilation semantics of @code{S"}, it is a Core word; if you also use
2935: its interpretation semantics, it is a File word.
2936:
1.43 anton 2937: @c ******************************************************************
1.40 anton 2938: @node ANS conformance, Model, Tools, Top
1.4 anton 2939: @chapter ANS conformance
1.43 anton 2940: @cindex ANS conformance of Gforth
1.4 anton 2941:
1.17 anton 2942: To the best of our knowledge, Gforth is an
1.14 anton 2943:
1.15 anton 2944: ANS Forth System
1.34 anton 2945: @itemize @bullet
1.15 anton 2946: @item providing the Core Extensions word set
2947: @item providing the Block word set
2948: @item providing the Block Extensions word set
2949: @item providing the Double-Number word set
2950: @item providing the Double-Number Extensions word set
2951: @item providing the Exception word set
2952: @item providing the Exception Extensions word set
2953: @item providing the Facility word set
2954: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2955: @item providing the File Access word set
2956: @item providing the File Access Extensions word set
2957: @item providing the Floating-Point word set
2958: @item providing the Floating-Point Extensions word set
2959: @item providing the Locals word set
2960: @item providing the Locals Extensions word set
2961: @item providing the Memory-Allocation word set
2962: @item providing the Memory-Allocation Extensions word set (that one's easy)
2963: @item providing the Programming-Tools word set
1.34 anton 2964: @item providing @code{;CODE}, @code{AHEAD}, @code{ASSEMBLER}, @code{BYE}, @code{CODE}, @code{CS-PICK}, @code{CS-ROLL}, @code{STATE}, @code{[ELSE]}, @code{[IF]}, @code{[THEN]} from the Programming-Tools Extensions word set
1.15 anton 2965: @item providing the Search-Order word set
2966: @item providing the Search-Order Extensions word set
2967: @item providing the String word set
2968: @item providing the String Extensions word set (another easy one)
2969: @end itemize
2970:
1.43 anton 2971: @cindex system documentation
1.15 anton 2972: In addition, ANS Forth systems are required to document certain
2973: implementation choices. This chapter tries to meet these
2974: requirements. In many cases it gives a way to ask the system for the
2975: information instead of providing the information directly, in
2976: particular, if the information depends on the processor, the operating
2977: system or the installation options chosen, or if they are likely to
1.17 anton 2978: change during the maintenance of Gforth.
1.15 anton 2979:
1.14 anton 2980: @comment The framework for the rest has been taken from pfe.
2981:
2982: @menu
2983: * The Core Words::
2984: * The optional Block word set::
2985: * The optional Double Number word set::
2986: * The optional Exception word set::
2987: * The optional Facility word set::
2988: * The optional File-Access word set::
2989: * The optional Floating-Point word set::
2990: * The optional Locals word set::
2991: * The optional Memory-Allocation word set::
2992: * The optional Programming-Tools word set::
2993: * The optional Search-Order word set::
2994: @end menu
2995:
2996:
2997: @c =====================================================================
2998: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2999: @comment node-name, next, previous, up
3000: @section The Core Words
3001: @c =====================================================================
1.43 anton 3002: @cindex core words, system documentation
3003: @cindex system documentation, core words
1.14 anton 3004:
3005: @menu
1.15 anton 3006: * core-idef:: Implementation Defined Options
3007: * core-ambcond:: Ambiguous Conditions
3008: * core-other:: Other System Documentation
1.14 anton 3009: @end menu
3010:
3011: @c ---------------------------------------------------------------------
3012: @node core-idef, core-ambcond, The Core Words, The Core Words
3013: @subsection Implementation Defined Options
3014: @c ---------------------------------------------------------------------
1.43 anton 3015: @cindex core words, implementation-defined options
3016: @cindex implementation-defined options, core words
3017:
1.14 anton 3018:
3019: @table @i
3020: @item (Cell) aligned addresses:
1.43 anton 3021: @cindex cell-aligned addresses
3022: @cindex aligned addresses
1.17 anton 3023: processor-dependent. Gforth's alignment words perform natural alignment
1.14 anton 3024: (e.g., an address aligned for a datum of size 8 is divisible by
3025: 8). Unaligned accesses usually result in a @code{-23 THROW}.
3026:
3027: @item @code{EMIT} and non-graphic characters:
1.43 anton 3028: @cindex @code{EMIT} and non-graphic characters
3029: @cindex non-graphic characters and @code{EMIT}
1.14 anton 3030: The character is output using the C library function (actually, macro)
1.36 anton 3031: @code{putc}.
1.14 anton 3032:
3033: @item character editing of @code{ACCEPT} and @code{EXPECT}:
1.43 anton 3034: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
3035: @cindex editing in @code{ACCEPT} and @code{EXPECT}
3036: @cindex @code{ACCEPT}, editing
3037: @cindex @code{EXPECT}, editing
1.14 anton 3038: This is modeled on the GNU readline library (@pxref{Readline
3039: Interaction, , Command Line Editing, readline, The GNU Readline
3040: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
3041: producing a full word completion every time you type it (instead of
3042: producing the common prefix of all completions).
3043:
3044: @item character set:
1.43 anton 3045: @cindex character set
1.14 anton 3046: The character set of your computer and display device. Gforth is
3047: 8-bit-clean (but some other component in your system may make trouble).
3048:
3049: @item Character-aligned address requirements:
1.43 anton 3050: @cindex character-aligned address requirements
1.14 anton 3051: installation-dependent. Currently a character is represented by a C
3052: @code{unsigned char}; in the future we might switch to @code{wchar_t}
3053: (Comments on that requested).
3054:
3055: @item character-set extensions and matching of names:
1.43 anton 3056: @cindex character-set extensions and matching of names
3057: @cindex case sensitivity for name lookup
3058: @cindex name lookup, case sensitivity
3059: @cindex locale and case sensitivity
1.17 anton 3060: Any character except the ASCII NUL charcter can be used in a
1.43 anton 3061: name. Matching is case-insensitive (except in @code{TABLE}s). The
1.36 anton 3062: matching is performed using the C function @code{strncasecmp}, whose
3063: function is probably influenced by the locale. E.g., the @code{C} locale
3064: does not know about accents and umlauts, so they are matched
3065: case-sensitively in that locale. For portability reasons it is best to
3066: write programs such that they work in the @code{C} locale. Then one can
3067: use libraries written by a Polish programmer (who might use words
3068: containing ISO Latin-2 encoded characters) and by a French programmer
3069: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
3070: funny results for some of the words (which ones, depends on the font you
3071: are using)). Also, the locale you prefer may not be available in other
3072: operating systems. Hopefully, Unicode will solve these problems one day.
1.14 anton 3073:
3074: @item conditions under which control characters match a space delimiter:
1.43 anton 3075: @cindex space delimiters
3076: @cindex control characters as delimiters
1.14 anton 3077: If @code{WORD} is called with the space character as a delimiter, all
3078: white-space characters (as identified by the C macro @code{isspace()})
3079: are delimiters. @code{PARSE}, on the other hand, treats space like other
3080: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
3081: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
3082: interpreter (aka text interpreter) by default, treats all white-space
3083: characters as delimiters.
3084:
3085: @item format of the control flow stack:
1.43 anton 3086: @cindex control flow stack, format
1.14 anton 3087: The data stack is used as control flow stack. The size of a control flow
3088: stack item in cells is given by the constant @code{cs-item-size}. At the
3089: time of this writing, an item consists of a (pointer to a) locals list
3090: (third), an address in the code (second), and a tag for identifying the
3091: item (TOS). The following tags are used: @code{defstart},
3092: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
3093: @code{scopestart}.
3094:
3095: @item conversion of digits > 35
1.43 anton 3096: @cindex digits > 35
1.14 anton 3097: The characters @code{[\]^_'} are the digits with the decimal value
3098: 36@minus{}41. There is no way to input many of the larger digits.
3099:
3100: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
1.43 anton 3101: @cindex @code{EXPECT}, display after end of input
3102: @cindex @code{ACCEPT}, display after end of input
1.14 anton 3103: The cursor is moved to the end of the entered string. If the input is
3104: terminated using the @kbd{Return} key, a space is typed.
3105:
3106: @item exception abort sequence of @code{ABORT"}:
1.43 anton 3107: @cindex exception abort sequence of @code{ABORT"}
3108: @cindex @code{ABORT"}, exception abort sequence
1.14 anton 3109: The error string is stored into the variable @code{"error} and a
3110: @code{-2 throw} is performed.
3111:
3112: @item input line terminator:
1.43 anton 3113: @cindex input line terminator
3114: @cindex line terminator on input
3115: @cindex newline charcter on input
1.36 anton 3116: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
3117: lines. One of these characters is typically produced when you type the
3118: @kbd{Enter} or @kbd{Return} key.
1.14 anton 3119:
3120: @item maximum size of a counted string:
1.43 anton 3121: @cindex maximum size of a counted string
3122: @cindex counted string, maximum size
1.14 anton 3123: @code{s" /counted-string" environment? drop .}. Currently 255 characters
3124: on all ports, but this may change.
3125:
3126: @item maximum size of a parsed string:
1.43 anton 3127: @cindex maximum size of a parsed string
3128: @cindex parsed string, maximum size
1.14 anton 3129: Given by the constant @code{/line}. Currently 255 characters.
3130:
3131: @item maximum size of a definition name, in characters:
1.43 anton 3132: @cindex maximum size of a definition name, in characters
3133: @cindex name, maximum length
1.14 anton 3134: 31
3135:
3136: @item maximum string length for @code{ENVIRONMENT?}, in characters:
1.43 anton 3137: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
3138: @cindex @code{ENVIRONMENT?} string length, maximum
1.14 anton 3139: 31
3140:
3141: @item method of selecting the user input device:
1.43 anton 3142: @cindex user input device, method of selecting
1.17 anton 3143: The user input device is the standard input. There is currently no way to
3144: change it from within Gforth. However, the input can typically be
3145: redirected in the command line that starts Gforth.
1.14 anton 3146:
3147: @item method of selecting the user output device:
1.43 anton 3148: @cindex user output device, method of selecting
1.36 anton 3149: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
3150: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
3151: output, so output on a terminal does not become visible before the next
3152: newline or buffer overflow. Output on non-terminals is invisible until
3153: the buffer overflows.
1.14 anton 3154:
3155: @item methods of dictionary compilation:
1.17 anton 3156: What are we expected to document here?
1.14 anton 3157:
3158: @item number of bits in one address unit:
1.43 anton 3159: @cindex number of bits in one address unit
3160: @cindex address unit, size in bits
1.14 anton 3161: @code{s" address-units-bits" environment? drop .}. 8 in all current
3162: ports.
3163:
3164: @item number representation and arithmetic:
1.43 anton 3165: @cindex number representation and arithmetic
1.14 anton 3166: Processor-dependent. Binary two's complement on all current ports.
3167:
3168: @item ranges for integer types:
1.43 anton 3169: @cindex ranges for integer types
3170: @cindex integer types, ranges
1.14 anton 3171: Installation-dependent. Make environmental queries for @code{MAX-N},
3172: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
3173: unsigned (and positive) types is 0. The lower bound for signed types on
3174: two's complement and one's complement machines machines can be computed
3175: by adding 1 to the upper bound.
3176:
3177: @item read-only data space regions:
1.43 anton 3178: @cindex read-only data space regions
3179: @cindex data-space, read-only regions
1.14 anton 3180: The whole Forth data space is writable.
3181:
3182: @item size of buffer at @code{WORD}:
1.43 anton 3183: @cindex size of buffer at @code{WORD}
3184: @cindex @code{WORD} buffer size
1.14 anton 3185: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
3186: shared with the pictured numeric output string. If overwriting
3187: @code{PAD} is acceptable, it is as large as the remaining dictionary
3188: space, although only as much can be sensibly used as fits in a counted
3189: string.
3190:
3191: @item size of one cell in address units:
1.43 anton 3192: @cindex cell size
1.14 anton 3193: @code{1 cells .}.
3194:
3195: @item size of one character in address units:
1.43 anton 3196: @cindex char size
1.14 anton 3197: @code{1 chars .}. 1 on all current ports.
3198:
3199: @item size of the keyboard terminal buffer:
1.43 anton 3200: @cindex size of the keyboard terminal buffer
3201: @cindex terminal buffer, size
1.36 anton 3202: Varies. You can determine the size at a specific time using @code{lp@@
1.14 anton 3203: tib - .}. It is shared with the locals stack and TIBs of files that
3204: include the current file. You can change the amount of space for TIBs
1.17 anton 3205: and locals stack at Gforth startup with the command line option
1.14 anton 3206: @code{-l}.
3207:
3208: @item size of the pictured numeric output buffer:
1.43 anton 3209: @cindex size of the pictured numeric output buffer
3210: @cindex pictured numeric output buffer, size
1.14 anton 3211: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
3212: shared with @code{WORD}.
3213:
3214: @item size of the scratch area returned by @code{PAD}:
1.43 anton 3215: @cindex size of the scratch area returned by @code{PAD}
3216: @cindex @code{PAD} size
3217: The remainder of dictionary space. @code{unused pad here - - .}.
1.14 anton 3218:
3219: @item system case-sensitivity characteristics:
1.43 anton 3220: @cindex case-sensitivity characteristics
1.36 anton 3221: Dictionary searches are case insensitive (except in
3222: @code{TABLE}s). However, as explained above under @i{character-set
3223: extensions}, the matching for non-ASCII characters is determined by the
3224: locale you are using. In the default @code{C} locale all non-ASCII
3225: characters are matched case-sensitively.
1.14 anton 3226:
3227: @item system prompt:
1.43 anton 3228: @cindex system prompt
3229: @cindex prompt
1.14 anton 3230: @code{ ok} in interpret state, @code{ compiled} in compile state.
3231:
3232: @item division rounding:
1.43 anton 3233: @cindex division rounding
1.14 anton 3234: installation dependent. @code{s" floored" environment? drop .}. We leave
1.43 anton 3235: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
3236: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
1.14 anton 3237:
3238: @item values of @code{STATE} when true:
1.43 anton 3239: @cindex @code{STATE} values
1.14 anton 3240: -1.
3241:
3242: @item values returned after arithmetic overflow:
3243: On two's complement machines, arithmetic is performed modulo
3244: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
3245: arithmetic (with appropriate mapping for signed types). Division by zero
1.36 anton 3246: typically results in a @code{-55 throw} (Floating-point unidentified
1.14 anton 3247: fault), although a @code{-10 throw} (divide by zero) would be more
3248: appropriate.
3249:
3250: @item whether the current definition can be found after @t{DOES>}:
1.43 anton 3251: @cindex @t{DOES>}, visibility of current definition
1.14 anton 3252: No.
3253:
3254: @end table
3255:
3256: @c ---------------------------------------------------------------------
3257: @node core-ambcond, core-other, core-idef, The Core Words
3258: @subsection Ambiguous conditions
3259: @c ---------------------------------------------------------------------
1.43 anton 3260: @cindex core words, ambiguous conditions
3261: @cindex ambiguous conditions, core words
1.14 anton 3262:
3263: @table @i
3264:
3265: @item a name is neither a word nor a number:
1.43 anton 3266: @cindex name not found
3267: @cindex Undefined word
1.36 anton 3268: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
3269: preserves the data and FP stack, so you don't lose more work than
3270: necessary.
1.14 anton 3271:
3272: @item a definition name exceeds the maximum length allowed:
1.43 anton 3273: @cindex Word name too long
1.14 anton 3274: @code{-19 throw} (Word name too long)
3275:
3276: @item addressing a region not inside the various data spaces of the forth system:
1.43 anton 3277: @cindex Invalid memory address
1.14 anton 3278: The stacks, code space and name space are accessible. Machine code space is
3279: typically readable. Accessing other addresses gives results dependent on
3280: the operating system. On decent systems: @code{-9 throw} (Invalid memory
3281: address).
3282:
3283: @item argument type incompatible with parameter:
1.43 anton 3284: @cindex Argument type mismatch
1.14 anton 3285: This is usually not caught. Some words perform checks, e.g., the control
3286: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
3287: mismatch).
3288:
3289: @item attempting to obtain the execution token of a word with undefined execution semantics:
1.43 anton 3290: @cindex Interpreting a compile-only word, for @code{'} etc.
3291: @cindex execution token of words with undefined execution semantics
1.36 anton 3292: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
3293: get an execution token for @code{compile-only-error} (which performs a
3294: @code{-14 throw} when executed).
1.14 anton 3295:
3296: @item dividing by zero:
1.43 anton 3297: @cindex dividing by zero
3298: @cindex floating point unidentified fault, integer division
3299: @cindex divide by zero
1.14 anton 3300: typically results in a @code{-55 throw} (floating point unidentified
3301: fault), although a @code{-10 throw} (divide by zero) would be more
3302: appropriate.
3303:
3304: @item insufficient data stack or return stack space:
1.43 anton 3305: @cindex insufficient data stack or return stack space
3306: @cindex stack overflow
3307: @cindex Address alignment exception, stack overflow
3308: @cindex Invalid memory address, stack overflow
3309: Depending on the operating system, the installation, and the invocation
3310: of Gforth, this is either checked by the memory management hardware, or
3311: it is not checked. If it is checked, you typically get a @code{-9 throw}
3312: (Invalid memory address) as soon as the overflow happens. If it is not
3313: check, overflows typically result in mysterious illegal memory accesses,
3314: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
3315: (Address alignment exception); they might also destroy the internal data
3316: structure of @code{ALLOCATE} and friends, resulting in various errors in
3317: these words.
1.14 anton 3318:
3319: @item insufficient space for loop control parameters:
1.43 anton 3320: @cindex insufficient space for loop control parameters
1.14 anton 3321: like other return stack overflows.
3322:
3323: @item insufficient space in the dictionary:
1.43 anton 3324: @cindex insufficient space in the dictionary
3325: @cindex dictionary overflow
3326: Depending on the operating system, the installation, and the invocation
3327: of Gforth, this is either checked by the memory management hardware, or
3328: it is not checked. Similar results as stack overflows. However,
3329: typically the error appears at a different place when one inserts or
3330: removes code. Also, the @code{THROW} does not relieve the situation (it
3331: does for stack overflows).
1.14 anton 3332:
3333: @item interpreting a word with undefined interpretation semantics:
1.43 anton 3334: @cindex interpreting a word with undefined interpretation semantics
3335: @cindex Interpreting a compile-only word
3336: For some words, we have defined interpretation semantics. For the
3337: others: @code{-14 throw} (Interpreting a compile-only word).
1.14 anton 3338:
3339: @item modifying the contents of the input buffer or a string literal:
1.43 anton 3340: @cindex modifying the contents of the input buffer or a string literal
1.14 anton 3341: These are located in writable memory and can be modified.
3342:
3343: @item overflow of the pictured numeric output string:
1.43 anton 3344: @cindex overflow of the pictured numeric output string
3345: @cindex pictured numeric output string, overflow
3346: Not checked. Runs into the dictionary and destroys it (at least,
3347: partially).
1.14 anton 3348:
3349: @item parsed string overflow:
1.43 anton 3350: @cindex parsed string overflow
1.14 anton 3351: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
3352:
3353: @item producing a result out of range:
1.43 anton 3354: @cindex result out of range
1.14 anton 3355: On two's complement machines, arithmetic is performed modulo
3356: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
3357: arithmetic (with appropriate mapping for signed types). Division by zero
3358: typically results in a @code{-55 throw} (floatingpoint unidentified
3359: fault), although a @code{-10 throw} (divide by zero) would be more
3360: appropriate. @code{convert} and @code{>number} currently overflow
3361: silently.
3362:
3363: @item reading from an empty data or return stack:
1.43 anton 3364: @cindex stack empty
3365: @cindex stack underflow
1.14 anton 3366: The data stack is checked by the outer (aka text) interpreter after
3367: every word executed. If it has underflowed, a @code{-4 throw} (Stack
1.43 anton 3368: underflow) is performed. Apart from that, stacks may be checked or not,
3369: depending on operating system, installation, and invocation. The
3370: consequences of stack underflows are similar to the consequences of
3371: stack overflows. Note that even if the system uses checking (through the
3372: MMU), your program may have to underflow by a significant number of
3373: stack items to trigger the reaction (the reason for this is that the
3374: MMU, and therefore the checking, works with a page-size granularity).
1.14 anton 3375:
1.36 anton 3376: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
1.43 anton 3377: @cindex unexpected end of the input buffer
3378: @cindex zero-length string as a name
3379: @cindex Attempt to use zero-length string as a name
1.14 anton 3380: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
3381: use zero-length string as a name). Words like @code{'} probably will not
3382: find what they search. Note that it is possible to create zero-length
3383: names with @code{nextname} (should it not?).
3384:
3385: @item @code{>IN} greater than input buffer:
1.43 anton 3386: @cindex @code{>IN} greater than input buffer
1.41 anton 3387: The next invocation of a parsing word returns a string with length 0.
1.14 anton 3388:
3389: @item @code{RECURSE} appears after @code{DOES>}:
1.43 anton 3390: @cindex @code{RECURSE} appears after @code{DOES>}
1.36 anton 3391: Compiles a recursive call to the defining word, not to the defined word.
1.14 anton 3392:
3393: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.43 anton 3394: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
3395: @cindex Argument type mismatch, @code{RESTORE-INPUT}
3396: @cindex @code{RESTORE-INPUT}, Argument type mismatch
1.27 anton 3397: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
3398: the end of the file was reached), its source-id may be
3399: reused. Therefore, restoring an input source specification referencing a
3400: closed file may lead to unpredictable results instead of a @code{-12
3401: THROW}.
3402:
1.36 anton 3403: In the future, Gforth may be able to restore input source specifications
1.41 anton 3404: from other than the current input source.
1.14 anton 3405:
3406: @item data space containing definitions gets de-allocated:
1.43 anton 3407: @cindex data space containing definitions gets de-allocated
1.41 anton 3408: Deallocation with @code{allot} is not checked. This typically results in
1.14 anton 3409: memory access faults or execution of illegal instructions.
3410:
3411: @item data space read/write with incorrect alignment:
1.43 anton 3412: @cindex data space read/write with incorrect alignment
3413: @cindex alignment faults
3414: @cindex Address alignment exception
1.14 anton 3415: Processor-dependent. Typically results in a @code{-23 throw} (Address
3416: alignment exception). Under Linux on a 486 or later processor with
3417: alignment turned on, incorrect alignment results in a @code{-9 throw}
3418: (Invalid memory address). There are reportedly some processors with
3419: alignment restrictions that do not report them.
3420:
3421: @item data space pointer not properly aligned, @code{,}, @code{C,}:
1.43 anton 3422: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
1.14 anton 3423: Like other alignment errors.
3424:
3425: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
1.43 anton 3426: Like other stack underflows.
1.14 anton 3427:
3428: @item loop control parameters not available:
1.43 anton 3429: @cindex loop control parameters not available
1.14 anton 3430: Not checked. The counted loop words simply assume that the top of return
3431: stack items are loop control parameters and behave accordingly.
3432:
3433: @item most recent definition does not have a name (@code{IMMEDIATE}):
1.43 anton 3434: @cindex most recent definition does not have a name (@code{IMMEDIATE})
3435: @cindex last word was headerless
1.14 anton 3436: @code{abort" last word was headerless"}.
3437:
3438: @item name not defined by @code{VALUE} used by @code{TO}:
1.43 anton 3439: @cindex name not defined by @code{VALUE} used by @code{TO}
3440: @cindex @code{TO} on non-@code{VALUE}s
3441: @cindex Invalid name argument, @code{TO}
3442: @code{-32 throw} (Invalid name argument) (unless name is a local or was
3443: defined by @code{CONSTANT}; in the latter case it just changes the constant).
1.14 anton 3444:
1.15 anton 3445: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.43 anton 3446: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
3447: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
1.14 anton 3448: @code{-13 throw} (Undefined word)
3449:
3450: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
1.43 anton 3451: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
1.14 anton 3452: Gforth behaves as if they were of the same type. I.e., you can predict
3453: the behaviour by interpreting all parameters as, e.g., signed.
3454:
3455: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
1.43 anton 3456: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
1.36 anton 3457: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
3458: compilation semantics of @code{TO}.
1.14 anton 3459:
3460: @item String longer than a counted string returned by @code{WORD}:
1.43 anton 3461: @cindex String longer than a counted string returned by @code{WORD}
3462: @cindex @code{WORD}, string overflow
1.14 anton 3463: Not checked. The string will be ok, but the count will, of course,
3464: contain only the least significant bits of the length.
3465:
1.15 anton 3466: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.43 anton 3467: @cindex @code{LSHIFT}, large shift counts
3468: @cindex @code{RSHIFT}, large shift counts
1.14 anton 3469: Processor-dependent. Typical behaviours are returning 0 and using only
3470: the low bits of the shift count.
3471:
3472: @item word not defined via @code{CREATE}:
1.43 anton 3473: @cindex @code{>BODY} of non-@code{CREATE}d words
1.14 anton 3474: @code{>BODY} produces the PFA of the word no matter how it was defined.
3475:
1.43 anton 3476: @cindex @code{DOES>} of non-@code{CREATE}d words
1.14 anton 3477: @code{DOES>} changes the execution semantics of the last defined word no
3478: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
3479: @code{CREATE , DOES>}.
3480:
3481: @item words improperly used outside @code{<#} and @code{#>}:
3482: Not checked. As usual, you can expect memory faults.
3483:
3484: @end table
3485:
3486:
3487: @c ---------------------------------------------------------------------
3488: @node core-other, , core-ambcond, The Core Words
3489: @subsection Other system documentation
3490: @c ---------------------------------------------------------------------
1.43 anton 3491: @cindex other system documentation, core words
3492: @cindex core words, other system documentation
1.14 anton 3493:
3494: @table @i
3495: @item nonstandard words using @code{PAD}:
1.43 anton 3496: @cindex @code{PAD} use by nonstandard words
1.14 anton 3497: None.
3498:
3499: @item operator's terminal facilities available:
1.43 anton 3500: @cindex operator's terminal facilities available
1.26 anton 3501: After processing the command line, Gforth goes into interactive mode,
3502: and you can give commands to Gforth interactively. The actual facilities
3503: available depend on how you invoke Gforth.
1.14 anton 3504:
3505: @item program data space available:
1.43 anton 3506: @cindex program data space available
3507: @cindex data space available
1.42 anton 3508: @code{UNUSED .} gives the remaining dictionary space. The total
3509: dictionary space can be specified with the @code{-m} switch
1.43 anton 3510: (@pxref{Invoking Gforth}) when Gforth starts up.
1.14 anton 3511:
3512: @item return stack space available:
1.43 anton 3513: @cindex return stack space available
1.42 anton 3514: You can compute the total return stack space in cells with
3515: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
1.43 anton 3516: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
1.14 anton 3517:
3518: @item stack space available:
1.43 anton 3519: @cindex stack space available
1.42 anton 3520: You can compute the total data stack space in cells with
3521: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
1.43 anton 3522: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
1.14 anton 3523:
3524: @item system dictionary space required, in address units:
1.43 anton 3525: @cindex system dictionary space required, in address units
1.14 anton 3526: Type @code{here forthstart - .} after startup. At the time of this
1.42 anton 3527: writing, this gives 80080 (bytes) on a 32-bit system.
1.14 anton 3528: @end table
3529:
3530:
3531: @c =====================================================================
3532: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
3533: @section The optional Block word set
3534: @c =====================================================================
1.43 anton 3535: @cindex system documentation, block words
3536: @cindex block words, system documentation
1.14 anton 3537:
3538: @menu
1.43 anton 3539: * block-idef:: Implementation Defined Options
1.15 anton 3540: * block-ambcond:: Ambiguous Conditions
3541: * block-other:: Other System Documentation
1.14 anton 3542: @end menu
3543:
3544:
3545: @c ---------------------------------------------------------------------
3546: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
3547: @subsection Implementation Defined Options
3548: @c ---------------------------------------------------------------------
1.43 anton 3549: @cindex implementation-defined options, block words
3550: @cindex block words, implementation-defined options
1.14 anton 3551:
3552: @table @i
3553: @item the format for display by @code{LIST}:
1.43 anton 3554: @cindex @code{LIST} display format
1.14 anton 3555: First the screen number is displayed, then 16 lines of 64 characters,
3556: each line preceded by the line number.
3557:
3558: @item the length of a line affected by @code{\}:
1.43 anton 3559: @cindex length of a line affected by @code{\}
3560: @cindex @code{\}, line length in blocks
1.14 anton 3561: 64 characters.
3562: @end table
3563:
3564:
3565: @c ---------------------------------------------------------------------
3566: @node block-ambcond, block-other, block-idef, The optional Block word set
3567: @subsection Ambiguous conditions
3568: @c ---------------------------------------------------------------------
1.43 anton 3569: @cindex block words, ambiguous conditions
3570: @cindex ambiguous conditions, block words
1.14 anton 3571:
3572: @table @i
3573: @item correct block read was not possible:
1.43 anton 3574: @cindex block read not possible
1.14 anton 3575: Typically results in a @code{throw} of some OS-derived value (between
3576: -512 and -2048). If the blocks file was just not long enough, blanks are
3577: supplied for the missing portion.
3578:
3579: @item I/O exception in block transfer:
1.43 anton 3580: @cindex I/O exception in block transfer
3581: @cindex block transfer, I/O exception
1.14 anton 3582: Typically results in a @code{throw} of some OS-derived value (between
3583: -512 and -2048).
3584:
3585: @item invalid block number:
1.43 anton 3586: @cindex invalid block number
3587: @cindex block number invalid
1.14 anton 3588: @code{-35 throw} (Invalid block number)
3589:
3590: @item a program directly alters the contents of @code{BLK}:
1.43 anton 3591: @cindex @code{BLK}, altering @code{BLK}
1.14 anton 3592: The input stream is switched to that other block, at the same
3593: position. If the storing to @code{BLK} happens when interpreting
3594: non-block input, the system will get quite confused when the block ends.
3595:
3596: @item no current block buffer for @code{UPDATE}:
1.43 anton 3597: @cindex @code{UPDATE}, no current block buffer
1.14 anton 3598: @code{UPDATE} has no effect.
3599:
3600: @end table
3601:
3602: @c ---------------------------------------------------------------------
3603: @node block-other, , block-ambcond, The optional Block word set
3604: @subsection Other system documentation
3605: @c ---------------------------------------------------------------------
1.43 anton 3606: @cindex other system documentation, block words
3607: @cindex block words, other system documentation
1.14 anton 3608:
3609: @table @i
3610: @item any restrictions a multiprogramming system places on the use of buffer addresses:
3611: No restrictions (yet).
3612:
3613: @item the number of blocks available for source and data:
3614: depends on your disk space.
3615:
3616: @end table
3617:
3618:
3619: @c =====================================================================
3620: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
3621: @section The optional Double Number word set
3622: @c =====================================================================
1.43 anton 3623: @cindex system documentation, double words
3624: @cindex double words, system documentation
1.14 anton 3625:
3626: @menu
1.15 anton 3627: * double-ambcond:: Ambiguous Conditions
1.14 anton 3628: @end menu
3629:
3630:
3631: @c ---------------------------------------------------------------------
1.15 anton 3632: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
1.14 anton 3633: @subsection Ambiguous conditions
3634: @c ---------------------------------------------------------------------
1.43 anton 3635: @cindex double words, ambiguous conditions
3636: @cindex ambiguous conditions, double words
1.14 anton 3637:
3638: @table @i
1.15 anton 3639: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.43 anton 3640: @cindex @code{D>S}, @var{d} out of range of @var{n}
1.14 anton 3641: The least significant cell of @var{d} is produced.
3642:
3643: @end table
3644:
3645:
3646: @c =====================================================================
3647: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
3648: @section The optional Exception word set
3649: @c =====================================================================
1.43 anton 3650: @cindex system documentation, exception words
3651: @cindex exception words, system documentation
1.14 anton 3652:
3653: @menu
1.15 anton 3654: * exception-idef:: Implementation Defined Options
1.14 anton 3655: @end menu
3656:
3657:
3658: @c ---------------------------------------------------------------------
1.15 anton 3659: @node exception-idef, , The optional Exception word set, The optional Exception word set
1.14 anton 3660: @subsection Implementation Defined Options
3661: @c ---------------------------------------------------------------------
1.43 anton 3662: @cindex implementation-defined options, exception words
3663: @cindex exception words, implementation-defined options
1.14 anton 3664:
3665: @table @i
3666: @item @code{THROW}-codes used in the system:
1.43 anton 3667: @cindex @code{THROW}-codes used in the system
3668: The codes -256@minus{}-511 are used for reporting signals. The mapping
3669: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
3670: codes -512@minus{}-2047 are used for OS errors (for file and memory
3671: allocation operations). The mapping from OS error numbers to throw codes
3672: is -512@minus{}@code{errno}. One side effect of this mapping is that
3673: undefined OS errors produce a message with a strange number; e.g.,
3674: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
1.14 anton 3675: @end table
3676:
3677: @c =====================================================================
3678: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
3679: @section The optional Facility word set
3680: @c =====================================================================
1.43 anton 3681: @cindex system documentation, facility words
3682: @cindex facility words, system documentation
1.14 anton 3683:
3684: @menu
1.15 anton 3685: * facility-idef:: Implementation Defined Options
3686: * facility-ambcond:: Ambiguous Conditions
1.14 anton 3687: @end menu
3688:
3689:
3690: @c ---------------------------------------------------------------------
3691: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
3692: @subsection Implementation Defined Options
3693: @c ---------------------------------------------------------------------
1.43 anton 3694: @cindex implementation-defined options, facility words
3695: @cindex facility words, implementation-defined options
1.14 anton 3696:
3697: @table @i
3698: @item encoding of keyboard events (@code{EKEY}):
1.43 anton 3699: @cindex keyboard events, encoding in @code{EKEY}
3700: @cindex @code{EKEY}, encoding of keyboard events
1.41 anton 3701: Not yet implemented.
1.14 anton 3702:
1.43 anton 3703: @item duration of a system clock tick:
3704: @cindex duration of a system clock tick
3705: @cindex clock tick duration
1.14 anton 3706: System dependent. With respect to @code{MS}, the time is specified in
3707: microseconds. How well the OS and the hardware implement this, is
3708: another question.
3709:
3710: @item repeatability to be expected from the execution of @code{MS}:
1.43 anton 3711: @cindex repeatability to be expected from the execution of @code{MS}
3712: @cindex @code{MS}, repeatability to be expected
1.14 anton 3713: System dependent. On Unix, a lot depends on load. If the system is
1.17 anton 3714: lightly loaded, and the delay is short enough that Gforth does not get
1.14 anton 3715: swapped out, the performance should be acceptable. Under MS-DOS and
3716: other single-tasking systems, it should be good.
3717:
3718: @end table
3719:
3720:
3721: @c ---------------------------------------------------------------------
1.15 anton 3722: @node facility-ambcond, , facility-idef, The optional Facility word set
1.14 anton 3723: @subsection Ambiguous conditions
3724: @c ---------------------------------------------------------------------
1.43 anton 3725: @cindex facility words, ambiguous conditions
3726: @cindex ambiguous conditions, facility words
1.14 anton 3727:
3728: @table @i
3729: @item @code{AT-XY} can't be performed on user output device:
1.43 anton 3730: @cindex @code{AT-XY} can't be performed on user output device
1.41 anton 3731: Largely terminal dependent. No range checks are done on the arguments.
1.14 anton 3732: No errors are reported. You may see some garbage appearing, you may see
3733: simply nothing happen.
3734:
3735: @end table
3736:
3737:
3738: @c =====================================================================
3739: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
3740: @section The optional File-Access word set
3741: @c =====================================================================
1.43 anton 3742: @cindex system documentation, file words
3743: @cindex file words, system documentation
1.14 anton 3744:
3745: @menu
1.43 anton 3746: * file-idef:: Implementation Defined Options
1.15 anton 3747: * file-ambcond:: Ambiguous Conditions
1.14 anton 3748: @end menu
3749:
3750: @c ---------------------------------------------------------------------
3751: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
3752: @subsection Implementation Defined Options
3753: @c ---------------------------------------------------------------------
1.43 anton 3754: @cindex implementation-defined options, file words
3755: @cindex file words, implementation-defined options
1.14 anton 3756:
3757: @table @i
1.43 anton 3758: @item file access methods used:
3759: @cindex file access methods used
1.14 anton 3760: @code{R/O}, @code{R/W} and @code{BIN} work as you would
3761: expect. @code{W/O} translates into the C file opening mode @code{w} (or
3762: @code{wb}): The file is cleared, if it exists, and created, if it does
1.43 anton 3763: not (with both @code{open-file} and @code{create-file}). Under Unix
1.14 anton 3764: @code{create-file} creates a file with 666 permissions modified by your
3765: umask.
3766:
3767: @item file exceptions:
1.43 anton 3768: @cindex file exceptions
1.14 anton 3769: The file words do not raise exceptions (except, perhaps, memory access
3770: faults when you pass illegal addresses or file-ids).
3771:
3772: @item file line terminator:
1.43 anton 3773: @cindex file line terminator
1.14 anton 3774: System-dependent. Gforth uses C's newline character as line
3775: terminator. What the actual character code(s) of this are is
3776: system-dependent.
3777:
1.43 anton 3778: @item file name format:
3779: @cindex file name format
1.14 anton 3780: System dependent. Gforth just uses the file name format of your OS.
3781:
3782: @item information returned by @code{FILE-STATUS}:
1.43 anton 3783: @cindex @code{FILE-STATUS}, returned information
1.14 anton 3784: @code{FILE-STATUS} returns the most powerful file access mode allowed
3785: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
3786: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
1.41 anton 3787: along with the returned mode.
1.14 anton 3788:
3789: @item input file state after an exception when including source:
1.43 anton 3790: @cindex exception when including source
1.14 anton 3791: All files that are left via the exception are closed.
3792:
3793: @item @var{ior} values and meaning:
1.43 anton 3794: @cindex @var{ior} values and meaning
1.15 anton 3795: The @var{ior}s returned by the file and memory allocation words are
3796: intended as throw codes. They typically are in the range
3797: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3798: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 3799:
3800: @item maximum depth of file input nesting:
1.43 anton 3801: @cindex maximum depth of file input nesting
3802: @cindex file input nesting, maximum depth
1.14 anton 3803: limited by the amount of return stack, locals/TIB stack, and the number
3804: of open files available. This should not give you troubles.
3805:
3806: @item maximum size of input line:
1.43 anton 3807: @cindex maximum size of input line
3808: @cindex input line size, maximum
1.14 anton 3809: @code{/line}. Currently 255.
3810:
3811: @item methods of mapping block ranges to files:
1.43 anton 3812: @cindex mapping block ranges to files
3813: @cindex files containing blocks
3814: @cindex blocks in files
1.37 anton 3815: By default, blocks are accessed in the file @file{blocks.fb} in the
3816: current working directory. The file can be switched with @code{USE}.
1.14 anton 3817:
3818: @item number of string buffers provided by @code{S"}:
1.43 anton 3819: @cindex @code{S"}, number of string buffers
1.14 anton 3820: 1
3821:
3822: @item size of string buffer used by @code{S"}:
1.43 anton 3823: @cindex @code{S"}, size of string buffer
1.14 anton 3824: @code{/line}. currently 255.
3825:
3826: @end table
3827:
3828: @c ---------------------------------------------------------------------
1.15 anton 3829: @node file-ambcond, , file-idef, The optional File-Access word set
1.14 anton 3830: @subsection Ambiguous conditions
3831: @c ---------------------------------------------------------------------
1.43 anton 3832: @cindex file words, ambiguous conditions
3833: @cindex ambiguous conditions, file words
1.14 anton 3834:
3835: @table @i
1.43 anton 3836: @item attempting to position a file outside its boundaries:
3837: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
1.14 anton 3838: @code{REPOSITION-FILE} is performed as usual: Afterwards,
3839: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
3840:
3841: @item attempting to read from file positions not yet written:
1.43 anton 3842: @cindex reading from file positions not yet written
1.14 anton 3843: End-of-file, i.e., zero characters are read and no error is reported.
3844:
3845: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
1.43 anton 3846: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid
1.14 anton 3847: An appropriate exception may be thrown, but a memory fault or other
3848: problem is more probable.
3849:
1.43 anton 3850: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
3851: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
3852: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
1.14 anton 3853: The @var{ior} produced by the operation, that discovered the problem, is
3854: thrown.
3855:
1.43 anton 3856: @item named file cannot be opened (@code{INCLUDED}):
3857: @cindex @code{INCLUDED}, named file cannot be opened
1.14 anton 3858: The @var{ior} produced by @code{open-file} is thrown.
3859:
3860: @item requesting an unmapped block number:
1.43 anton 3861: @cindex unmapped block numbers
1.14 anton 3862: There are no unmapped legal block numbers. On some operating systems,
3863: writing a block with a large number may overflow the file system and
3864: have an error message as consequence.
3865:
3866: @item using @code{source-id} when @code{blk} is non-zero:
1.43 anton 3867: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
1.14 anton 3868: @code{source-id} performs its function. Typically it will give the id of
3869: the source which loaded the block. (Better ideas?)
3870:
3871: @end table
3872:
3873:
3874: @c =====================================================================
3875: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15 anton 3876: @section The optional Floating-Point word set
1.14 anton 3877: @c =====================================================================
1.43 anton 3878: @cindex system documentation, floating-point words
3879: @cindex floating-point words, system documentation
1.14 anton 3880:
3881: @menu
1.15 anton 3882: * floating-idef:: Implementation Defined Options
3883: * floating-ambcond:: Ambiguous Conditions
1.14 anton 3884: @end menu
3885:
3886:
3887: @c ---------------------------------------------------------------------
3888: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
3889: @subsection Implementation Defined Options
3890: @c ---------------------------------------------------------------------
1.43 anton 3891: @cindex implementation-defined options, floating-point words
3892: @cindex floating-point words, implementation-defined options
1.14 anton 3893:
3894: @table @i
1.15 anton 3895: @item format and range of floating point numbers:
1.43 anton 3896: @cindex format and range of floating point numbers
3897: @cindex floating point numbers, format and range
1.15 anton 3898: System-dependent; the @code{double} type of C.
1.14 anton 3899:
1.15 anton 3900: @item results of @code{REPRESENT} when @var{float} is out of range:
1.43 anton 3901: @cindex @code{REPRESENT}, results when @var{float} is out of range
1.15 anton 3902: System dependent; @code{REPRESENT} is implemented using the C library
3903: function @code{ecvt()} and inherits its behaviour in this respect.
1.14 anton 3904:
1.15 anton 3905: @item rounding or truncation of floating-point numbers:
1.43 anton 3906: @cindex rounding of floating-point numbers
3907: @cindex truncation of floating-point numbers
3908: @cindex floating-point numbers, rounding or truncation
1.26 anton 3909: System dependent; the rounding behaviour is inherited from the hosting C
3910: compiler. IEEE-FP-based (i.e., most) systems by default round to
3911: nearest, and break ties by rounding to even (i.e., such that the last
3912: bit of the mantissa is 0).
1.14 anton 3913:
1.15 anton 3914: @item size of floating-point stack:
1.43 anton 3915: @cindex floating-point stack size
1.42 anton 3916: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
3917: the floating-point stack (in floats). You can specify this on startup
1.43 anton 3918: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
1.14 anton 3919:
1.15 anton 3920: @item width of floating-point stack:
1.43 anton 3921: @cindex floating-point stack width
1.15 anton 3922: @code{1 floats}.
1.14 anton 3923:
3924: @end table
3925:
3926:
3927: @c ---------------------------------------------------------------------
1.15 anton 3928: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
3929: @subsection Ambiguous conditions
1.14 anton 3930: @c ---------------------------------------------------------------------
1.43 anton 3931: @cindex floating-point words, ambiguous conditions
3932: @cindex ambiguous conditions, floating-point words
1.14 anton 3933:
3934: @table @i
1.15 anton 3935: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
1.43 anton 3936: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
1.37 anton 3937: System-dependent. Typically results in a @code{-23 THROW} like other
1.15 anton 3938: alignment violations.
1.14 anton 3939:
1.15 anton 3940: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
1.43 anton 3941: @cindex @code{f@@} used with an address that is not float aligned
3942: @cindex @code{f!} used with an address that is not float aligned
1.37 anton 3943: System-dependent. Typically results in a @code{-23 THROW} like other
1.15 anton 3944: alignment violations.
1.14 anton 3945:
1.43 anton 3946: @item floating-point result out of range:
3947: @cindex floating-point result out of range
1.15 anton 3948: System-dependent. Can result in a @code{-55 THROW} (Floating-point
3949: unidentified fault), or can produce a special value representing, e.g.,
3950: Infinity.
1.14 anton 3951:
1.15 anton 3952: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
1.43 anton 3953: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
1.15 anton 3954: System-dependent. Typically results in an alignment fault like other
3955: alignment violations.
1.14 anton 3956:
1.43 anton 3957: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
3958: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
1.15 anton 3959: The floating-point number is converted into decimal nonetheless.
1.14 anton 3960:
1.15 anton 3961: @item Both arguments are equal to zero (@code{FATAN2}):
1.43 anton 3962: @cindex @code{FATAN2}, both arguments are equal to zero
1.15 anton 3963: System-dependent. @code{FATAN2} is implemented using the C library
3964: function @code{atan2()}.
1.14 anton 3965:
1.43 anton 3966: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
3967: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
1.15 anton 3968: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
3969: because of small errors and the tan will be a very large (or very small)
3970: but finite number.
1.14 anton 3971:
1.15 anton 3972: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
1.43 anton 3973: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
1.15 anton 3974: The result is rounded to the nearest float.
1.14 anton 3975:
1.15 anton 3976: @item dividing by zero:
1.43 anton 3977: @cindex dividing by zero, floating-point
3978: @cindex floating-point dividing by zero
3979: @cindex floating-point unidentified fault, FP divide-by-zero
1.15 anton 3980: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 3981:
1.15 anton 3982: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
1.43 anton 3983: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
1.15 anton 3984: System dependent. On IEEE-FP based systems the number is converted into
3985: an infinity.
1.14 anton 3986:
1.43 anton 3987: @item @var{float}<1 (@code{FACOSH}):
3988: @cindex @code{FACOSH}, @var{float}<1
3989: @cindex floating-point unidentified fault, @code{FACOSH}
1.15 anton 3990: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 3991:
1.43 anton 3992: @item @var{float}=<-1 (@code{FLNP1}):
3993: @cindex @code{FLNP1}, @var{float}=<-1
3994: @cindex floating-point unidentified fault, @code{FLNP1}
1.15 anton 3995: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3996: negative infinity is typically produced for @var{float}=-1.
1.14 anton 3997:
1.43 anton 3998: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
3999: @cindex @code{FLN}, @var{float}=<0
4000: @cindex @code{FLOG}, @var{float}=<0
4001: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
1.15 anton 4002: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
4003: negative infinity is typically produced for @var{float}=0.
1.14 anton 4004:
1.43 anton 4005: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
4006: @cindex @code{FASINH}, @var{float}<0
4007: @cindex @code{FSQRT}, @var{float}<0
4008: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
1.15 anton 4009: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
4010: produces values for these inputs on my Linux box (Bug in the C library?)
1.14 anton 4011:
1.43 anton 4012: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
4013: @cindex @code{FACOS}, |@var{float}|>1
4014: @cindex @code{FASIN}, |@var{float}|>1
4015: @cindex @code{FATANH}, |@var{float}|>1
4016: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
1.15 anton 4017: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 4018:
1.43 anton 4019: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
4020: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
4021: @cindex floating-point unidentified fault, @code{F>D}
1.15 anton 4022: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 4023:
1.15 anton 4024: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
1.43 anton 4025: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
1.15 anton 4026: This does not happen.
4027: @end table
1.14 anton 4028:
4029: @c =====================================================================
1.15 anton 4030: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
4031: @section The optional Locals word set
1.14 anton 4032: @c =====================================================================
1.43 anton 4033: @cindex system documentation, locals words
4034: @cindex locals words, system documentation
1.14 anton 4035:
4036: @menu
1.15 anton 4037: * locals-idef:: Implementation Defined Options
4038: * locals-ambcond:: Ambiguous Conditions
1.14 anton 4039: @end menu
4040:
4041:
4042: @c ---------------------------------------------------------------------
1.15 anton 4043: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14 anton 4044: @subsection Implementation Defined Options
4045: @c ---------------------------------------------------------------------
1.43 anton 4046: @cindex implementation-defined options, locals words
4047: @cindex locals words, implementation-defined options
1.14 anton 4048:
4049: @table @i
1.15 anton 4050: @item maximum number of locals in a definition:
1.43 anton 4051: @cindex maximum number of locals in a definition
4052: @cindex locals, maximum number in a definition
1.15 anton 4053: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
4054: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
4055: characters. The number of locals in a definition is bounded by the size
4056: of locals-buffer, which contains the names of the locals.
1.14 anton 4057:
4058: @end table
4059:
4060:
4061: @c ---------------------------------------------------------------------
1.15 anton 4062: @node locals-ambcond, , locals-idef, The optional Locals word set
1.14 anton 4063: @subsection Ambiguous conditions
4064: @c ---------------------------------------------------------------------
1.43 anton 4065: @cindex locals words, ambiguous conditions
4066: @cindex ambiguous conditions, locals words
1.14 anton 4067:
4068: @table @i
1.15 anton 4069: @item executing a named local in interpretation state:
1.43 anton 4070: @cindex local in interpretation state
4071: @cindex Interpreting a compile-only word, for a local
4072: Locals have no interpretation semantics. If you try to perform the
4073: interpretation semantics, you will get a @code{-14 throw} somewhere
4074: (Interpreting a compile-only word). If you perform the compilation
4075: semantics, the locals access will be compiled (irrespective of state).
1.14 anton 4076:
1.15 anton 4077: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
1.43 anton 4078: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
4079: @cindex @code{TO} on non-@code{VALUE}s and non-locals
4080: @cindex Invalid name argument, @code{TO}
1.15 anton 4081: @code{-32 throw} (Invalid name argument)
1.14 anton 4082:
4083: @end table
4084:
4085:
4086: @c =====================================================================
1.15 anton 4087: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
4088: @section The optional Memory-Allocation word set
1.14 anton 4089: @c =====================================================================
1.43 anton 4090: @cindex system documentation, memory-allocation words
4091: @cindex memory-allocation words, system documentation
1.14 anton 4092:
4093: @menu
1.15 anton 4094: * memory-idef:: Implementation Defined Options
1.14 anton 4095: @end menu
4096:
4097:
4098: @c ---------------------------------------------------------------------
1.15 anton 4099: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14 anton 4100: @subsection Implementation Defined Options
4101: @c ---------------------------------------------------------------------
1.43 anton 4102: @cindex implementation-defined options, memory-allocation words
4103: @cindex memory-allocation words, implementation-defined options
1.14 anton 4104:
4105: @table @i
1.15 anton 4106: @item values and meaning of @var{ior}:
1.43 anton 4107: @cindex @var{ior} values and meaning
1.15 anton 4108: The @var{ior}s returned by the file and memory allocation words are
4109: intended as throw codes. They typically are in the range
4110: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
4111: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 4112:
4113: @end table
4114:
4115: @c =====================================================================
1.15 anton 4116: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
4117: @section The optional Programming-Tools word set
1.14 anton 4118: @c =====================================================================
1.43 anton 4119: @cindex system documentation, programming-tools words
4120: @cindex programming-tools words, system documentation
1.14 anton 4121:
4122: @menu
1.15 anton 4123: * programming-idef:: Implementation Defined Options
4124: * programming-ambcond:: Ambiguous Conditions
1.14 anton 4125: @end menu
4126:
4127:
4128: @c ---------------------------------------------------------------------
1.15 anton 4129: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14 anton 4130: @subsection Implementation Defined Options
4131: @c ---------------------------------------------------------------------
1.43 anton 4132: @cindex implementation-defined options, programming-tools words
4133: @cindex programming-tools words, implementation-defined options
1.14 anton 4134:
4135: @table @i
1.43 anton 4136: @item ending sequence for input following @code{;CODE} and @code{CODE}:
4137: @cindex @code{;CODE} ending sequence
4138: @cindex @code{CODE} ending sequence
4139: @code{END-CODE}
4140:
4141: @item manner of processing input following @code{;CODE} and @code{CODE}:
4142: @cindex @code{;CODE}, processing input
4143: @cindex @code{CODE}, processing input
4144: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
1.37 anton 4145: the input is processed by the text interpreter, (starting) in interpret
4146: state.
1.15 anton 4147:
4148: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
1.43 anton 4149: @cindex @code{ASSEMBLER}, search order capability
1.37 anton 4150: The ANS Forth search order word set.
1.15 anton 4151:
4152: @item source and format of display by @code{SEE}:
1.43 anton 4153: @cindex @code{SEE}, source and format of output
1.15 anton 4154: The source for @code{see} is the intermediate code used by the inner
4155: interpreter. The current @code{see} tries to output Forth source code
4156: as well as possible.
4157:
1.14 anton 4158: @end table
4159:
4160: @c ---------------------------------------------------------------------
1.15 anton 4161: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
1.14 anton 4162: @subsection Ambiguous conditions
4163: @c ---------------------------------------------------------------------
1.43 anton 4164: @cindex programming-tools words, ambiguous conditions
4165: @cindex ambiguous conditions, programming-tools words
1.14 anton 4166:
4167: @table @i
4168:
1.15 anton 4169: @item deleting the compilation wordlist (@code{FORGET}):
1.43 anton 4170: @cindex @code{FORGET}, deleting the compilation wordlist
1.15 anton 4171: Not implemented (yet).
1.14 anton 4172:
1.15 anton 4173: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
1.43 anton 4174: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
4175: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
4176: @cindex control-flow stack underflow
1.15 anton 4177: This typically results in an @code{abort"} with a descriptive error
4178: message (may change into a @code{-22 throw} (Control structure mismatch)
4179: in the future). You may also get a memory access error. If you are
4180: unlucky, this ambiguous condition is not caught.
4181:
1.43 anton 4182: @item @var{name} can't be found (@code{FORGET}):
4183: @cindex @code{FORGET}, @var{name} can't be found
1.15 anton 4184: Not implemented (yet).
1.14 anton 4185:
1.15 anton 4186: @item @var{name} not defined via @code{CREATE}:
1.43 anton 4187: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
4188: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
1.37 anton 4189: the execution semantics of the last defined word no matter how it was
4190: defined.
1.14 anton 4191:
1.15 anton 4192: @item @code{POSTPONE} applied to @code{[IF]}:
1.43 anton 4193: @cindex @code{POSTPONE} applied to @code{[IF]}
4194: @cindex @code{[IF]} and @code{POSTPONE}
1.15 anton 4195: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
4196: equivalent to @code{[IF]}.
1.14 anton 4197:
1.15 anton 4198: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
1.43 anton 4199: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
1.15 anton 4200: Continue in the same state of conditional compilation in the next outer
4201: input source. Currently there is no warning to the user about this.
1.14 anton 4202:
1.15 anton 4203: @item removing a needed definition (@code{FORGET}):
1.43 anton 4204: @cindex @code{FORGET}, removing a needed definition
1.15 anton 4205: Not implemented (yet).
1.14 anton 4206:
4207: @end table
4208:
4209:
4210: @c =====================================================================
1.15 anton 4211: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
4212: @section The optional Search-Order word set
1.14 anton 4213: @c =====================================================================
1.43 anton 4214: @cindex system documentation, search-order words
4215: @cindex search-order words, system documentation
1.14 anton 4216:
4217: @menu
1.15 anton 4218: * search-idef:: Implementation Defined Options
4219: * search-ambcond:: Ambiguous Conditions
1.14 anton 4220: @end menu
4221:
4222:
4223: @c ---------------------------------------------------------------------
1.15 anton 4224: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14 anton 4225: @subsection Implementation Defined Options
4226: @c ---------------------------------------------------------------------
1.43 anton 4227: @cindex implementation-defined options, search-order words
4228: @cindex search-order words, implementation-defined options
1.14 anton 4229:
4230: @table @i
1.15 anton 4231: @item maximum number of word lists in search order:
1.43 anton 4232: @cindex maximum number of word lists in search order
4233: @cindex search order, maximum depth
1.15 anton 4234: @code{s" wordlists" environment? drop .}. Currently 16.
4235:
4236: @item minimum search order:
1.43 anton 4237: @cindex minimum search order
4238: @cindex search order, minimum
1.15 anton 4239: @code{root root}.
1.14 anton 4240:
4241: @end table
4242:
4243: @c ---------------------------------------------------------------------
1.15 anton 4244: @node search-ambcond, , search-idef, The optional Search-Order word set
1.14 anton 4245: @subsection Ambiguous conditions
4246: @c ---------------------------------------------------------------------
1.43 anton 4247: @cindex search-order words, ambiguous conditions
4248: @cindex ambiguous conditions, search-order words
1.14 anton 4249:
4250: @table @i
1.15 anton 4251: @item changing the compilation wordlist (during compilation):
1.43 anton 4252: @cindex changing the compilation wordlist (during compilation)
4253: @cindex compilation wordlist, change before definition ends
1.33 anton 4254: The word is entered into the wordlist that was the compilation wordlist
4255: at the start of the definition. Any changes to the name field (e.g.,
4256: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
4257: are applied to the latest defined word (as reported by @code{last} or
4258: @code{lastxt}), if possible, irrespective of the compilation wordlist.
1.14 anton 4259:
1.15 anton 4260: @item search order empty (@code{previous}):
1.43 anton 4261: @cindex @code{previous}, search order empty
4262: @cindex Vocstack empty, @code{previous}
1.15 anton 4263: @code{abort" Vocstack empty"}.
1.14 anton 4264:
1.15 anton 4265: @item too many word lists in search order (@code{also}):
1.43 anton 4266: @cindex @code{also}, too many word lists in search order
4267: @cindex Vocstack full, @code{also}
1.15 anton 4268: @code{abort" Vocstack full"}.
1.14 anton 4269:
4270: @end table
1.13 anton 4271:
1.43 anton 4272: @c ***************************************************************
1.34 anton 4273: @node Model, Integrating Gforth, ANS conformance, Top
4274: @chapter Model
4275:
4276: This chapter has yet to be written. It will contain information, on
4277: which internal structures you can rely.
4278:
1.43 anton 4279: @c ***************************************************************
1.34 anton 4280: @node Integrating Gforth, Emacs and Gforth, Model, Top
4281: @chapter Integrating Gforth into C programs
4282:
4283: This is not yet implemented.
4284:
4285: Several people like to use Forth as scripting language for applications
4286: that are otherwise written in C, C++, or some other language.
4287:
4288: The Forth system ATLAST provides facilities for embedding it into
4289: applications; unfortunately it has several disadvantages: most
1.36 anton 4290: importantly, it is not based on ANS Forth, and it is apparently dead
1.34 anton 4291: (i.e., not developed further and not supported). The facilities
4292: provided by Gforth in this area are inspired by ATLASTs facilities, so
4293: making the switch should not be hard.
4294:
4295: We also tried to design the interface such that it can easily be
4296: implemented by other Forth systems, so that we may one day arrive at a
4297: standardized interface. Such a standard interface would allow you to
4298: replace the Forth system without having to rewrite C code.
4299:
4300: You embed the Gforth interpreter by linking with the library
4301: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
4302: global symbols in this library that belong to the interface, have the
4303: prefix @code{forth_}. (Global symbols that are used internally have the
4304: prefix @code{gforth_}).
4305:
4306: You can include the declarations of Forth types and the functions and
1.36 anton 4307: variables of the interface with @code{#include <forth.h>}.
1.34 anton 4308:
4309: Types.
1.13 anton 4310:
1.34 anton 4311: Variables.
4312:
4313: Data and FP Stack pointer. Area sizes.
4314:
4315: functions.
4316:
4317: forth_init(imagefile)
4318: forth_evaluate(string) exceptions?
4319: forth_goto(address) (or forth_execute(xt)?)
4320: forth_continue() (a corountining mechanism)
4321:
4322: Adding primitives.
4323:
4324: No checking.
4325:
4326: Signals?
4327:
4328: Accessing the Stacks
1.4 anton 4329:
1.43 anton 4330: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
1.17 anton 4331: @chapter Emacs and Gforth
1.43 anton 4332: @cindex Emacs and Gforth
1.4 anton 4333:
1.43 anton 4334: @cindex @file{gforth.el}
4335: @cindex @file{forth.el}
4336: @cindex Rydqvist, Goran
4337: @cindex comment editing commands
4338: @cindex @code{\}, editing with Emacs
4339: @cindex debug tracer editing commands
4340: @cindex @code{~~}, removal with Emacs
4341: @cindex Forth mode in Emacs
1.17 anton 4342: Gforth comes with @file{gforth.el}, an improved version of
1.33 anton 4343: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.4 anton 4344: improvements are a better (but still not perfect) handling of
4345: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8 anton 4346: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
4347: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
4348: stuff I do not use alone, even though some of it only makes sense for
4349: TILE. To get a description of these features, enter Forth mode and type
4350: @kbd{C-h m}.
1.4 anton 4351:
1.43 anton 4352: @cindex source location of error or debugging output in Emacs
4353: @cindex error output, finding the source location in Emacs
4354: @cindex debugging output, finding the source location in Emacs
1.17 anton 4355: In addition, Gforth supports Emacs quite well: The source code locations
1.4 anton 4356: given in error messages, debugging output (from @code{~~}) and failed
4357: assertion messages are in the right format for Emacs' compilation mode
4358: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
4359: Manual}) so the source location corresponding to an error or other
4360: message is only a few keystrokes away (@kbd{C-x `} for the next error,
4361: @kbd{C-c C-c} for the error under the cursor).
4362:
1.43 anton 4363: @cindex @file{TAGS} file
4364: @cindex @file{etags.fs}
4365: @cindex viewing the source of a word in Emacs
1.4 anton 4366: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
4367: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
4368: contains the definitions of all words defined afterwards. You can then
4369: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17 anton 4370: several tags files at the same time (e.g., one for the Gforth sources
1.28 anton 4371: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
4372: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
4373: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
1.33 anton 4374: @file{/usr/local/share/gforth/0.2.0/TAGS}).
1.4 anton 4375:
1.43 anton 4376: @cindex @file{.emacs}
1.4 anton 4377: To get all these benefits, add the following lines to your @file{.emacs}
4378: file:
4379:
4380: @example
4381: (autoload 'forth-mode "gforth.el")
4382: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
4383: @end example
4384:
1.43 anton 4385: @node Image Files, Engine, Emacs and Gforth, Top
4386: @chapter Image Files
4387: @cindex image files
4388: @cindex @code{.fi} files
4389: @cindex precompiled Forth code
4390: @cindex dictionary in persistent form
4391: @cindex persistent form of dictionary
4392:
4393: An image file is a file containing an image of the Forth dictionary,
4394: i.e., compiled Forth code and data residing in the dictionary. By
4395: convention, we use the extension @code{.fi} for image files.
4396:
4397: @menu
4398: * Image File Background:: Why have image files?
4399: * Non-Relocatable Image Files:: don't always work.
4400: * Data-Relocatable Image Files:: are better.
1.45 anton 4401: * Fully Relocatable Image Files:: better yet.
1.43 anton 4402: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
4403: * Running Image Files:: @code{gforth -i @var{file}} or @var{file}.
4404: * Modifying the Startup Sequence:: and turnkey applications.
4405: @end menu
4406:
4407: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
4408: @section Image File Background
4409: @cindex image file background
4410:
4411: Our Forth system consists not only of primitives, but also of
1.45 anton 4412: definitions written in Forth. Since the Forth compiler itself belongs to
4413: those definitions, it is not possible to start the system with the
1.43 anton 4414: primitives and the Forth source alone. Therefore we provide the Forth
4415: code as an image file in nearly executable form. At the start of the
1.45 anton 4416: system a C routine loads the image file into memory, optionally
4417: relocates the addresses, then sets up the memory (stacks etc.) according
4418: to information in the image file, and starts executing Forth code.
1.43 anton 4419:
4420: The image file variants represent different compromises between the
4421: goals of making it easy to generate image files and making them
4422: portable.
4423:
4424: @cindex relocation at run-time
4425: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
4426: run-time. This avoids many of the complications discussed below (image
4427: files are data relocatable without further ado), but costs performance
4428: (one addition per memory access).
4429:
4430: @cindex relocation at load-time
1.45 anton 4431: By contrast, our loader performs relocation at image load time. The
4432: loader also has to replace tokens standing for primitive calls with the
4433: appropriate code-field addresses (or code addresses in the case of
4434: direct threading).
1.43 anton 4435:
4436: There are three kinds of image files, with different degrees of
4437: relocatability: non-relocatable, data-relocatable, and fully relocatable
4438: image files.
4439:
4440: @cindex image file loader
4441: @cindex relocating loader
4442: @cindex loader for image files
4443: These image file variants have several restrictions in common; they are
4444: caused by the design of the image file loader:
4445:
4446: @itemize @bullet
4447: @item
4448: There is only one segment; in particular, this means, that an image file
4449: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
4450: them). And the contents of the stacks are not represented, either.
4451:
4452: @item
1.44 anton 4453: The only kinds of relocation supported are: adding the same offset to
4454: all cells that represent data addresses; and replacing special tokens
4455: with code addresses or with pieces of machine code.
1.43 anton 4456:
4457: If any complex computations involving addresses are performed, the
4458: results cannot be represented in the image file. Several applications that
4459: use such computations come to mind:
4460: @itemize @minus
4461: @item
4462: Hashing addresses (or data structures which contain addresses) for table
4463: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
4464: purpose, you will have no problem, because the hash tables are
4465: recomputed automatically when the system is started. If you use your own
4466: hash tables, you will have to do something similar.
4467:
4468: @item
4469: There's a cute implementation of doubly-linked lists that uses
4470: @code{XOR}ed addresses. You could represent such lists as singly-linked
4471: in the image file, and restore the doubly-linked representation on
4472: startup.@footnote{In my opinion, though, you should think thrice before
4473: using a doubly-linked list (whatever implementation).}
1.44 anton 4474:
4475: @item
4476: The code addresses of run-time routines like @code{docol:} cannot be
4477: represented in the image file (because their tokens would be replaced by
4478: machine code in direct threaded implementations). As a workaround,
4479: compute these addresses at run-time with @code{>code-address} from the
1.45 anton 4480: executions tokens of appropriate words (see the definitions of
1.44 anton 4481: @code{docol:} and friends in @file{kernel.fs}).
1.43 anton 4482:
4483: @item
4484: On many architectures addresses are represented in machine code in some
4485: shifted or mangled form. You cannot put @code{CODE} words that contain
4486: absolute addresses in this form in a relocatable image file. Workarounds
4487: are representing the address in some relative form (e.g., relative to
4488: the CFA, which is present in some register), or loading the address from
4489: a place where it is stored in a non-mangled form.
4490: @end itemize
4491: @end itemize
4492:
4493: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
4494: @section Non-Relocatable Image Files
4495: @cindex non-relocatable image files
4496: @cindex image files, non-relocatable
4497:
4498: These files are simple memory dumps of the dictionary. They are specific
4499: to the executable (i.e., @file{gforth} file) they were created
4500: with. What's worse, they are specific to the place on which the
4501: dictionary resided when the image was created. Now, there is no
4502: guarantee that the dictionary will reside at the same place the next
4503: time you start Gforth, so there's no guarantee that a non-relocatable
4504: image will work the next time (Gforth will complain instead of crashing,
4505: though).
4506:
4507: You can create a non-relocatable image file with
4508:
4509: doc-savesystem
4510:
4511: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
4512: @section Data-Relocatable Image Files
4513: @cindex data-relocatable image files
4514: @cindex image files, data-relocatable
4515:
4516: These files contain relocatable data addresses, but fixed code addresses
4517: (instead of tokens). They are specific to the executable (i.e.,
1.45 anton 4518: @file{gforth} file) they were created with. For direct threading on some
4519: architectures (e.g., the i386), data-relocatable images do not work. You
4520: get a data-relocatable image, if you use @file{gforth-makeimage} with a
4521: Gforth binary that is not doubly indirect threaded (@pxref{Fully
4522: Relocatable Image Files}).
4523:
4524: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
4525: @section Fully Relocatable Image Files
4526: @cindex fully relocatable image files
4527: @cindex image files, fully relocatable
1.43 anton 4528:
1.45 anton 4529: @cindex @file{kern*.fi}, relocatability
4530: @cindex @file{gforth.fi}, relocatability
4531: These image files have relocatable data addresses, and tokens for code
4532: addresses. They can be used with different binaries (e.g., with and
4533: without debugging) on the same machine, and even across machines with
4534: the same data formats (byte order, cell size, floating point
4535: format). However, they are usually specific to the version of Gforth
4536: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
4537: are fully relocatable.
4538:
4539: There are two ways to create a fully relocatable image file:
4540:
4541: @menu
4542: * gforth-makeimage:: The normal way
4543: * cross.fs:: The hard way
4544: @end menu
4545:
4546: @node gforth-makeimage, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
4547: @subsection @file{gforth-makeimage}
1.43 anton 4548: @cindex @file{comp-image.fs}
1.45 anton 4549: @cindex @file{gforth-makeimage}
1.43 anton 4550:
1.45 anton 4551: You will usually use @file{gforth-makeimage}. If you want to create an
4552: image @var{file} that contains everything you would load by invoking
4553: Gforth with @code{gforth @var{options}}, you simply say
1.43 anton 4554: @example
1.45 anton 4555: gforth-makeimage @var{file} @var{options}
1.43 anton 4556: @end example
4557:
4558: E.g., if you want to create an image @file{asm.fi} that has the file
4559: @file{asm.fs} loaded in addition to the usual stuff, you could do it
4560: like this:
4561:
4562: @example
1.45 anton 4563: gforth-makeimage asm.fi asm.fs
1.43 anton 4564: @end example
4565:
1.45 anton 4566: @file{gforth-makeimage} works like this: It produces two non-relocatable
4567: images for different addresses and then compares them. Its output
4568: reflects this: first you see the output (if any) of the two Gforth
4569: invocations that produce the nonrelocatable image files, then you see
4570: the output of the comparing program: It displays the offset used for
4571: data addresses and the offset used for code addresses;
4572: moreover, for each cell that cannot be represented correctly in the
4573: image files, it displays a line like the following one:
1.43 anton 4574:
1.45 anton 4575: @example
4576: 78DC BFFFFA50 BFFFFA40
4577: @end example
1.43 anton 4578:
1.45 anton 4579: This means that at offset $78dc from @code{forthstart}, one input image
4580: contains $bffffa50, and the other contains $bffffa40. Since these cells
4581: cannot be represented correctly in the output image, you should examine
4582: these places in the dictionary and verify that these cells are dead
4583: (i.e., not read before they are written).
1.43 anton 4584:
1.48 ! anton 4585: @cindex @code{savesystem} during @file{gforth-makeimage}
! 4586: @cindex @code{bye} during @file{gforth-makeimage}
! 4587: @cindex doubly indirect threaded code
! 4588: @cindex environment variable @code{GFORTHD}
! 4589: @cindex @code{GFORTHD} environment variable
! 4590: @cindex @code{gforth-ditc}
1.45 anton 4591: There are a few wrinkles: After processing the passed @var{options}, the
4592: words @code{savesystem} and @code{bye} must be visible. A special doubly
4593: indirect threaded version of the @file{gforth} executable is used for
4594: creating the nonrelocatable images; you can pass the exact filename of
4595: this executable through the environment variable @code{GFORTHD}
4596: (default: @file{gforth-ditc}); if you pass a version that is not doubly
4597: indirect threaded, you will not get a fully relocatable image, but a
4598: data-relocatable image (because there is no code address offset).
4599:
4600: @node cross.fs, , gforth-makeimage, Fully Relocatable Image Files
4601: @subsection @file{cross.fs}
4602: @cindex @file{cross.fs}
1.43 anton 4603: @cindex cross-compiler
4604: @cindex metacompiler
1.45 anton 4605:
4606: You can also use @code{cross}, a batch compiler that accepts a Forth-like
4607: programming language. This @code{cross} language has to be documented
1.43 anton 4608: yet.
4609:
4610: @cindex target compiler
1.45 anton 4611: @code{cross} also allows you to create image files for machines with
4612: different data sizes and data formats than the one used for generating
4613: the image file. You can also use it to create an application image that
4614: does not contain a Forth compiler. These features are bought with
4615: restrictions and inconveniences in programming. E.g., addresses have to
4616: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
4617: order to make the code relocatable.
4618:
1.43 anton 4619:
4620: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
4621: @section Stack and Dictionary Sizes
4622: @cindex image file, stack and dictionary sizes
4623: @cindex dictionary size default
4624: @cindex stack size default
4625:
4626: If you invoke Gforth with a command line flag for the size
4627: (@pxref{Invoking Gforth}), the size you specify is stored in the
1.45 anton 4628: dictionary. If you save the dictionary with @code{savesystem} or create
4629: an image with @file{gforth-makeimage}, this size will become the default
4630: for the resulting image file. E.g., the following will create a
4631: fully relocatable version of gforth.fi with a 1MB dictionary:
1.43 anton 4632:
4633: @example
1.45 anton 4634: gforth-makeimage gforth.fi -m 1M
1.43 anton 4635: @end example
4636:
4637: In other words, if you want to set the default size for the dictionary
1.45 anton 4638: and the stacks of an image, just invoke @file{gforth-makeimage} with the
4639: appropriate options when creating the image.
1.43 anton 4640:
4641: @cindex stack size, cache-friendly
4642: Note: For cache-friendly behaviour (i.e., good performance), you should
4643: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
4644: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
4645: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
4646:
4647: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
4648: @section Running Image Files
4649: @cindex running image files
4650: @cindex invoking image files
4651: @cindex image file invocation
4652:
4653: @cindex -i, invoke image file
4654: @cindex --image file, invoke image file
4655: You can invoke Gforth with an image file @var{image} instead of the
4656: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
4657: @example
4658: gforth -i @var{image}
4659: @end example
4660:
4661: @cindex executable image file
4662: @cindex image files, executable
4663: If your operating system supports starting scripts with a line of the
1.48 ! anton 4664: form @code{#! ...}, you just have to type the image file name to start
! 4665: Gforth with this image file (note that the file extension @code{.fi} is
! 4666: just a convention). I.e., to run Gforth with the image file @var{image},
! 4667: you can just type @var{image} instead of @code{gforth -i @var{image}}.
1.43 anton 4668:
1.48 ! anton 4669: doc-#!
1.43 anton 4670:
4671: @node Modifying the Startup Sequence, , Running Image Files, Image Files
4672: @section Modifying the Startup Sequence
4673: @cindex startup sequence for image file
4674: @cindex image file initialization sequence
4675: @cindex initialization sequence of image file
4676:
4677: You can add your own initialization to the startup sequence through the
4678: deferred word
4679:
4680: doc-'cold
4681:
4682: @code{'cold} is invoked just before the image-specific command line
1.45 anton 4683: processing (by default, loading files and evaluating (@code{-e}) strings)
1.43 anton 4684: starts.
4685:
4686: A sequence for adding your initialization usually looks like this:
4687:
4688: @example
4689: :noname
4690: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
4691: ... \ your stuff
4692: ; IS 'cold
4693: @end example
4694:
4695: @cindex turnkey image files
4696: @cindex image files, turnkey applications
4697: You can make a turnkey image by letting @code{'cold} execute a word
4698: (your turnkey application) that never returns; instead, it exits Gforth
4699: via @code{bye} or @code{throw}.
4700:
1.48 ! anton 4701: @cindex command-line arguments, access
! 4702: @cindex arguments on the command line, access
! 4703: You can access the (image-specific) command-line arguments through the
! 4704: variables @code{argc} and @code{argv}. @code{arg} provides conventient
! 4705: access to @code{argv}.
! 4706:
! 4707: doc-argc
! 4708: doc-argv
! 4709: doc-arg
! 4710:
! 4711: If @code{'cold} exits normally, Gforth processes the command-line
! 4712: arguments as files to be loaded and strings to be evaluated. Therefore,
! 4713: @code{'cold} should remove the arguments it has used in this case.
! 4714:
1.43 anton 4715: @c ******************************************************************
4716: @node Engine, Bugs, Image Files, Top
4717: @chapter Engine
4718: @cindex engine
4719: @cindex virtual machine
1.3 anton 4720:
1.17 anton 4721: Reading this section is not necessary for programming with Gforth. It
1.43 anton 4722: may be helpful for finding your way in the Gforth sources.
1.3 anton 4723:
1.24 anton 4724: The ideas in this section have also been published in the papers
4725: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
4726: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
4727: Ertl, presented at EuroForth '93; the latter is available at
1.48 ! anton 4728: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
1.24 anton 4729:
1.4 anton 4730: @menu
4731: * Portability::
4732: * Threading::
4733: * Primitives::
1.17 anton 4734: * Performance::
1.4 anton 4735: @end menu
4736:
1.43 anton 4737: @node Portability, Threading, Engine, Engine
1.3 anton 4738: @section Portability
1.43 anton 4739: @cindex engine portability
1.3 anton 4740:
4741: One of the main goals of the effort is availability across a wide range
4742: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
4743: this goal by manually coding the engine in assembly language for several
4744: then-popular processors. This approach is very labor-intensive and the
4745: results are short-lived due to progress in computer architecture.
4746:
1.43 anton 4747: @cindex C, using C for the engine
1.3 anton 4748: Others have avoided this problem by coding in C, e.g., Mitch Bradley
4749: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
4750: particularly popular for UNIX-based Forths due to the large variety of
4751: architectures of UNIX machines. Unfortunately an implementation in C
4752: does not mix well with the goals of efficiency and with using
4753: traditional techniques: Indirect or direct threading cannot be expressed
4754: in C, and switch threading, the fastest technique available in C, is
1.43 anton 4755: significantly slower. Another problem with C is that it is very
1.3 anton 4756: cumbersome to express double integer arithmetic.
4757:
1.43 anton 4758: @cindex GNU C for the engine
4759: @cindex long long
1.3 anton 4760: Fortunately, there is a portable language that does not have these
4761: limitations: GNU C, the version of C processed by the GNU C compiler
4762: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
4763: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
4764: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
4765: threading possible, its @code{long long} type (@pxref{Long Long, ,
1.33 anton 4766: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
1.32 anton 4767: double numbers@footnote{Unfortunately, long longs are not implemented
4768: properly on all machines (e.g., on alpha-osf1, long longs are only 64
4769: bits, the same size as longs (and pointers), but they should be twice as
4770: long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
4771: C Manual}). So, we had to implement doubles in C after all. Still, on
4772: most machines we can use long longs and achieve better performance than
4773: with the emulation package.}. GNU C is available for free on all
4774: important (and many unimportant) UNIX machines, VMS, 80386s running
4775: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
4776: on all these machines.
1.3 anton 4777:
4778: Writing in a portable language has the reputation of producing code that
4779: is slower than assembly. For our Forth engine we repeatedly looked at
4780: the code produced by the compiler and eliminated most compiler-induced
1.43 anton 4781: inefficiencies by appropriate changes in the source code.
1.3 anton 4782:
1.43 anton 4783: @cindex explicit register declarations
4784: @cindex --enable-force-reg, configuration flag
4785: @cindex -DFORCE_REG
1.3 anton 4786: However, register allocation cannot be portably influenced by the
4787: programmer, leading to some inefficiencies on register-starved
4788: machines. We use explicit register declarations (@pxref{Explicit Reg
4789: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
4790: improve the speed on some machines. They are turned on by using the
1.43 anton 4791: configuration flag @code{--enable-force-reg} (@code{gcc} switch
4792: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
4793: machine, but also on the compiler version: On some machines some
4794: compiler versions produce incorrect code when certain explicit register
4795: declarations are used. So by default @code{-DFORCE_REG} is not used.
1.3 anton 4796:
1.43 anton 4797: @node Threading, Primitives, Portability, Engine
1.3 anton 4798: @section Threading
1.43 anton 4799: @cindex inner interpreter implementation
4800: @cindex threaded code implementation
1.3 anton 4801:
1.43 anton 4802: @cindex labels as values
1.3 anton 4803: GNU C's labels as values extension (available since @code{gcc-2.0},
4804: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
4805: makes it possible to take the address of @var{label} by writing
4806: @code{&&@var{label}}. This address can then be used in a statement like
4807: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
4808: @code{goto x}.
4809:
1.43 anton 4810: @cindex NEXT, indirect threaded
4811: @cindex indirect threaded inner interpreter
4812: @cindex inner interpreter, indirect threaded
1.3 anton 4813: With this feature an indirect threaded NEXT looks like:
4814: @example
4815: cfa = *ip++;
4816: ca = *cfa;
4817: goto *ca;
4818: @end example
1.43 anton 4819: @cindex instruction pointer
1.3 anton 4820: For those unfamiliar with the names: @code{ip} is the Forth instruction
4821: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
4822: execution token and points to the code field of the next word to be
4823: executed; The @code{ca} (code address) fetched from there points to some
4824: executable code, e.g., a primitive or the colon definition handler
4825: @code{docol}.
4826:
1.43 anton 4827: @cindex NEXT, direct threaded
4828: @cindex direct threaded inner interpreter
4829: @cindex inner interpreter, direct threaded
1.3 anton 4830: Direct threading is even simpler:
4831: @example
4832: ca = *ip++;
4833: goto *ca;
4834: @end example
4835:
4836: Of course we have packaged the whole thing neatly in macros called
4837: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
4838:
1.4 anton 4839: @menu
4840: * Scheduling::
4841: * Direct or Indirect Threaded?::
4842: * DOES>::
4843: @end menu
4844:
4845: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3 anton 4846: @subsection Scheduling
1.43 anton 4847: @cindex inner interpreter optimization
1.3 anton 4848:
4849: There is a little complication: Pipelined and superscalar processors,
4850: i.e., RISC and some modern CISC machines can process independent
4851: instructions while waiting for the results of an instruction. The
4852: compiler usually reorders (schedules) the instructions in a way that
4853: achieves good usage of these delay slots. However, on our first tries
4854: the compiler did not do well on scheduling primitives. E.g., for
4855: @code{+} implemented as
4856: @example
4857: n=sp[0]+sp[1];
4858: sp++;
4859: sp[0]=n;
4860: NEXT;
4861: @end example
4862: the NEXT comes strictly after the other code, i.e., there is nearly no
4863: scheduling. After a little thought the problem becomes clear: The
4864: compiler cannot know that sp and ip point to different addresses (and
1.4 anton 4865: the version of @code{gcc} we used would not know it even if it was
4866: possible), so it could not move the load of the cfa above the store to
4867: the TOS. Indeed the pointers could be the same, if code on or very near
4868: the top of stack were executed. In the interest of speed we chose to
4869: forbid this probably unused ``feature'' and helped the compiler in
4870: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
4871: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3 anton 4872: @example
4873: n=sp[0]+sp[1];
4874: sp++;
4875: NEXT_P1;
4876: sp[0]=n;
4877: NEXT_P2;
4878: @end example
1.4 anton 4879: This can be scheduled optimally by the compiler.
1.3 anton 4880:
4881: This division can be turned off with the switch @code{-DCISC_NEXT}. This
4882: switch is on by default on machines that do not profit from scheduling
4883: (e.g., the 80386), in order to preserve registers.
4884:
1.4 anton 4885: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3 anton 4886: @subsection Direct or Indirect Threaded?
1.43 anton 4887: @cindex threading, direct or indirect?
1.3 anton 4888:
1.43 anton 4889: @cindex -DDIRECT_THREADED
1.3 anton 4890: Both! After packaging the nasty details in macro definitions we
4891: realized that we could switch between direct and indirect threading by
4892: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
4893: defining a few machine-specific macros for the direct-threading case.
4894: On the Forth level we also offer access words that hide the
4895: differences between the threading methods (@pxref{Threading Words}).
4896:
1.43 anton 4897: Indirect threading is implemented completely machine-independently.
4898: Direct threading needs routines for creating jumps to the executable
4899: code (e.g. to docol or dodoes). These routines are inherently
4900: machine-dependent, but they do not amount to many source lines. I.e.,
4901: even porting direct threading to a new machine is a small effort.
4902:
4903: @cindex --enable-indirect-threaded, configuration flag
4904: @cindex --enable-direct-threaded, configuration flag
4905: The default threading method is machine-dependent. You can enforce a
4906: specific threading method when building Gforth with the configuration
4907: flag @code{--enable-direct-threaded} or
4908: @code{--enable-indirect-threaded}. Note that direct threading is not
4909: supported on all machines.
1.3 anton 4910:
1.4 anton 4911: @node DOES>, , Direct or Indirect Threaded?, Threading
1.3 anton 4912: @subsection DOES>
1.43 anton 4913: @cindex @code{DOES>} implementation
4914:
4915: @cindex dodoes routine
4916: @cindex DOES-code
1.3 anton 4917: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
4918: the chunk of code executed by every word defined by a
4919: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
1.43 anton 4920: the Forth code to be executed, i.e. the code after the
4921: @code{DOES>} (the DOES-code)? There are two solutions:
1.3 anton 4922:
4923: In fig-Forth the code field points directly to the dodoes and the
1.43 anton 4924: DOES-code address is stored in the cell after the code address (i.e. at
4925: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
4926: the Forth-79 and all later standards, because in fig-Forth this address
1.3 anton 4927: lies in the body (which is illegal in these standards). However, by
4928: making the code field larger for all words this solution becomes legal
1.43 anton 4929: again. We use this approach for the indirect threaded version and for
4930: direct threading on some machines. Leaving a cell unused in most words
4931: is a bit wasteful, but on the machines we are targeting this is hardly a
4932: problem. The other reason for having a code field size of two cells is
4933: to avoid having different image files for direct and indirect threaded
4934: systems (direct threaded systems require two-cell code fields on many
4935: machines).
1.3 anton 4936:
1.43 anton 4937: @cindex DOES-handler
1.3 anton 4938: The other approach is that the code field points or jumps to the cell
4939: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
1.43 anton 4940: this address (the DOES-handler). @code{dodoes} can then get the
4941: DOES-code address by computing the code address, i.e., the address of
4942: the jump to dodoes, and add the length of that jump field. A variant of
4943: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
4944: return address (which can be found in the return register on RISCs) is
4945: the DOES-code address. Since the two cells available in the code field
4946: are used up by the jump to the code address in direct threading on many
4947: architectures, we use this approach for direct threading on these
4948: architectures. We did not want to add another cell to the code field.
1.3 anton 4949:
1.43 anton 4950: @node Primitives, Performance, Threading, Engine
1.3 anton 4951: @section Primitives
1.43 anton 4952: @cindex primitives, implementation
4953: @cindex virtual machine instructions, implementation
1.3 anton 4954:
1.4 anton 4955: @menu
4956: * Automatic Generation::
4957: * TOS Optimization::
4958: * Produced code::
4959: @end menu
4960:
4961: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3 anton 4962: @subsection Automatic Generation
1.43 anton 4963: @cindex primitives, automatic generation
1.3 anton 4964:
1.43 anton 4965: @cindex @file{prims2x.fs}
1.3 anton 4966: Since the primitives are implemented in a portable language, there is no
4967: longer any need to minimize the number of primitives. On the contrary,
1.43 anton 4968: having many primitives has an advantage: speed. In order to reduce the
1.3 anton 4969: number of errors in primitives and to make programming them easier, we
4970: provide a tool, the primitive generator (@file{prims2x.fs}), that
4971: automatically generates most (and sometimes all) of the C code for a
4972: primitive from the stack effect notation. The source for a primitive
4973: has the following form:
4974:
1.43 anton 4975: @cindex primitive source format
1.3 anton 4976: @format
4977: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
4978: [@code{""}@var{glossary entry}@code{""}]
4979: @var{C code}
4980: [@code{:}
4981: @var{Forth code}]
4982: @end format
4983:
4984: The items in brackets are optional. The category and glossary fields
4985: are there for generating the documentation, the Forth code is there
4986: for manual implementations on machines without GNU C. E.g., the source
4987: for the primitive @code{+} is:
4988: @example
4989: + n1 n2 -- n core plus
4990: n = n1+n2;
4991: @end example
4992:
4993: This looks like a specification, but in fact @code{n = n1+n2} is C
4994: code. Our primitive generation tool extracts a lot of information from
4995: the stack effect notations@footnote{We use a one-stack notation, even
4996: though we have separate data and floating-point stacks; The separate
4997: notation can be generated easily from the unified notation.}: The number
4998: of items popped from and pushed on the stack, their type, and by what
4999: name they are referred to in the C code. It then generates a C code
5000: prelude and postlude for each primitive. The final C code for @code{+}
5001: looks like this:
5002:
5003: @example
5004: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
5005: /* */ /* documentation */
1.4 anton 5006: @{
1.3 anton 5007: DEF_CA /* definition of variable ca (indirect threading) */
5008: Cell n1; /* definitions of variables */
5009: Cell n2;
5010: Cell n;
5011: n1 = (Cell) sp[1]; /* input */
5012: n2 = (Cell) TOS;
5013: sp += 1; /* stack adjustment */
5014: NAME("+") /* debugging output (with -DDEBUG) */
1.4 anton 5015: @{
1.3 anton 5016: n = n1+n2; /* C code taken from the source */
1.4 anton 5017: @}
1.3 anton 5018: NEXT_P1; /* NEXT part 1 */
5019: TOS = (Cell)n; /* output */
5020: NEXT_P2; /* NEXT part 2 */
1.4 anton 5021: @}
1.3 anton 5022: @end example
5023:
5024: This looks long and inefficient, but the GNU C compiler optimizes quite
5025: well and produces optimal code for @code{+} on, e.g., the R3000 and the
5026: HP RISC machines: Defining the @code{n}s does not produce any code, and
5027: using them as intermediate storage also adds no cost.
5028:
5029: There are also other optimizations, that are not illustrated by this
5030: example: Assignments between simple variables are usually for free (copy
5031: propagation). If one of the stack items is not used by the primitive
5032: (e.g. in @code{drop}), the compiler eliminates the load from the stack
5033: (dead code elimination). On the other hand, there are some things that
5034: the compiler does not do, therefore they are performed by
5035: @file{prims2x.fs}: The compiler does not optimize code away that stores
5036: a stack item to the place where it just came from (e.g., @code{over}).
5037:
5038: While programming a primitive is usually easy, there are a few cases
5039: where the programmer has to take the actions of the generator into
5040: account, most notably @code{?dup}, but also words that do not (always)
5041: fall through to NEXT.
5042:
1.4 anton 5043: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3 anton 5044: @subsection TOS Optimization
1.43 anton 5045: @cindex TOS optimization for primitives
5046: @cindex primitives, keeping the TOS in a register
1.3 anton 5047:
5048: An important optimization for stack machine emulators, e.g., Forth
5049: engines, is keeping one or more of the top stack items in
1.4 anton 5050: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
5051: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.34 anton 5052: @itemize @bullet
1.3 anton 5053: @item
5054: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
5055: due to fewer loads from and stores to the stack.
5056: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
5057: @var{y<n}, due to additional moves between registers.
5058: @end itemize
5059:
1.43 anton 5060: @cindex -DUSE_TOS
5061: @cindex -DUSE_NO_TOS
1.3 anton 5062: In particular, keeping one item in a register is never a disadvantage,
5063: if there are enough registers. Keeping two items in registers is a
5064: disadvantage for frequent words like @code{?branch}, constants,
5065: variables, literals and @code{i}. Therefore our generator only produces
5066: code that keeps zero or one items in registers. The generated C code
5067: covers both cases; the selection between these alternatives is made at
5068: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
5069: code for @code{+} is just a simple variable name in the one-item case,
5070: otherwise it is a macro that expands into @code{sp[0]}. Note that the
5071: GNU C compiler tries to keep simple variables like @code{TOS} in
5072: registers, and it usually succeeds, if there are enough registers.
5073:
1.43 anton 5074: @cindex -DUSE_FTOS
5075: @cindex -DUSE_NO_FTOS
1.3 anton 5076: The primitive generator performs the TOS optimization for the
5077: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
5078: operations the benefit of this optimization is even larger:
5079: floating-point operations take quite long on most processors, but can be
5080: performed in parallel with other operations as long as their results are
5081: not used. If the FP-TOS is kept in a register, this works. If
5082: it is kept on the stack, i.e., in memory, the store into memory has to
5083: wait for the result of the floating-point operation, lengthening the
5084: execution time of the primitive considerably.
5085:
5086: The TOS optimization makes the automatic generation of primitives a
5087: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
5088: @code{TOS} is not sufficient. There are some special cases to
5089: consider:
1.34 anton 5090: @itemize @bullet
1.3 anton 5091: @item In the case of @code{dup ( w -- w w )} the generator must not
5092: eliminate the store to the original location of the item on the stack,
5093: if the TOS optimization is turned on.
1.4 anton 5094: @item Primitives with stack effects of the form @code{--}
5095: @var{out1}...@var{outy} must store the TOS to the stack at the start.
5096: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3 anton 5097: must load the TOS from the stack at the end. But for the null stack
5098: effect @code{--} no stores or loads should be generated.
5099: @end itemize
5100:
1.4 anton 5101: @node Produced code, , TOS Optimization, Primitives
1.3 anton 5102: @subsection Produced code
1.43 anton 5103: @cindex primitives, assembly code listing
1.3 anton 5104:
1.43 anton 5105: @cindex @file{engine.s}
1.3 anton 5106: To see what assembly code is produced for the primitives on your machine
5107: with your compiler and your flag settings, type @code{make engine.s} and
1.4 anton 5108: look at the resulting file @file{engine.s}.
1.3 anton 5109:
1.43 anton 5110: @node Performance, , Primitives, Engine
1.17 anton 5111: @section Performance
1.43 anton 5112: @cindex performance of some Forth interpreters
5113: @cindex engine performance
5114: @cindex benchmarking Forth systems
5115: @cindex Gforth performance
1.17 anton 5116:
5117: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
5118: impossible to write a significantly faster engine.
5119:
5120: On register-starved machines like the 386 architecture processors
5121: improvements are possible, because @code{gcc} does not utilize the
5122: registers as well as a human, even with explicit register declarations;
5123: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
5124: and hand-tuned it for the 486; this system is 1.19 times faster on the
5125: Sieve benchmark on a 486DX2/66 than Gforth compiled with
5126: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
5127:
1.43 anton 5128: @cindex Win32Forth performance
5129: @cindex NT Forth performance
5130: @cindex eforth performance
5131: @cindex ThisForth performance
5132: @cindex PFE performance
5133: @cindex TILE performance
1.17 anton 5134: However, this potential advantage of assembly language implementations
5135: is not necessarily realized in complete Forth systems: We compared
1.26 anton 5136: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
5137: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
5138: 1994) and Eforth (with and without peephole (aka pinhole) optimization
5139: of the threaded code); all these systems were written in assembly
1.30 anton 5140: language. We also compared Gforth with three systems written in C:
1.32 anton 5141: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
5142: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
5143: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
5144: -fomit-frame-pointer; ThisForth employs peephole optimization of the
5145: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
5146: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
5147: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
5148: 486DX2/66 with similar memory performance under Windows NT. Marcel
5149: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
5150: added the peephole optimizer, ran the benchmarks and reported the
5151: results.
1.17 anton 5152:
5153: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
5154: matrix multiplication come from the Stanford integer benchmarks and have
5155: been translated into Forth by Martin Fraeman; we used the versions
1.30 anton 5156: included in the TILE Forth package, but with bigger data set sizes; and
5157: a recursive Fibonacci number computation for benchmarking calling
5158: performance. The following table shows the time taken for the benchmarks
5159: scaled by the time taken by Gforth (in other words, it shows the speedup
5160: factor that Gforth achieved over the other systems).
1.17 anton 5161:
5162: @example
1.30 anton 5163: relative Win32- NT eforth This-
5164: time Gforth Forth Forth eforth +opt PFE Forth TILE
1.32 anton 5165: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
5166: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
1.38 anton 5167: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
5168: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
1.17 anton 5169: @end example
5170:
5171: You may find the good performance of Gforth compared with the systems
5172: written in assembly language quite surprising. One important reason for
5173: the disappointing performance of these systems is probably that they are
5174: not written optimally for the 486 (e.g., they use the @code{lods}
5175: instruction). In addition, Win32Forth uses a comfortable, but costly
5176: method for relocating the Forth image: like @code{cforth}, it computes
5177: the actual addresses at run time, resulting in two address computations
1.43 anton 5178: per NEXT (@pxref{Image File Background}).
1.17 anton 5179:
1.26 anton 5180: Only Eforth with the peephole optimizer performs comparable to
5181: Gforth. The speedups achieved with peephole optimization of threaded
5182: code are quite remarkable. Adding a peephole optimizer to Gforth should
5183: cause similar speedups.
5184:
1.30 anton 5185: The speedup of Gforth over PFE, ThisForth and TILE can be easily
1.43 anton 5186: explained with the self-imposed restriction of the latter systems to
5187: standard C, which makes efficient threading impossible (however, the
5188: measured implementation of PFE uses a GNU C extension: @ref{Global Reg
5189: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
5190: Moreover, current C compilers have a hard time optimizing other aspects
5191: of the ThisForth and the TILE source.
1.17 anton 5192:
5193: Note that the performance of Gforth on 386 architecture processors
5194: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
5195: failed to allocate any of the virtual machine registers into real
5196: machine registers by itself and would not work correctly with explicit
5197: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
5198: running the Sieve) than the one measured above.
5199:
1.43 anton 5200: Note also that there have been several releases of Win32Forth since the
5201: release presented here, so the results presented here may have little
5202: predictive value for the performance of Win32Forth today.
5203:
5204: @cindex @file{Benchres}
1.26 anton 5205: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
5206: Maierhofer (presented at EuroForth '95), an indirect threaded version of
5207: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
1.46 anton 5208: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
5209: threaded version used here. The paper available at
1.48 ! anton 5210: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
1.43 anton 5211: it also contains numbers for some native code systems. You can find a
5212: newer version of these measurements at
1.48 ! anton 5213: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
1.43 anton 5214: find numbers for Gforth on various machines in @file{Benchres}.
1.24 anton 5215:
1.43 anton 5216: @node Bugs, Origin, Engine, Top
1.4 anton 5217: @chapter Bugs
1.43 anton 5218: @cindex bug reporting
1.4 anton 5219:
1.17 anton 5220: Known bugs are described in the file BUGS in the Gforth distribution.
5221:
1.24 anton 5222: If you find a bug, please send a bug report to
1.48 ! anton 5223: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should
1.17 anton 5224: describe the Gforth version used (it is announced at the start of an
5225: interactive Gforth session), the machine and operating system (on Unix
5226: systems you can use @code{uname -a} to produce this information), the
1.43 anton 5227: installation options (send the @file{config.status} file), and a
1.24 anton 5228: complete list of changes you (or your installer) have made to the Gforth
5229: sources (if any); it should contain a program (or a sequence of keyboard
5230: commands) that reproduces the bug and a description of what you think
5231: constitutes the buggy behaviour.
1.17 anton 5232:
5233: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
5234: to Report Bugs, gcc.info, GNU C Manual}.
5235:
5236:
1.29 anton 5237: @node Origin, Word Index, Bugs, Top
5238: @chapter Authors and Ancestors of Gforth
5239:
5240: @section Authors and Contributors
1.43 anton 5241: @cindex authors of Gforth
5242: @cindex contributors to Gforth
1.29 anton 5243:
5244: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
1.30 anton 5245: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
5246: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
5247: with their continuous feedback. Lennart Benshop contributed
1.29 anton 5248: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
5249: support for calling C libraries. Helpful comments also came from Paul
1.37 anton 5250: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.39 anton 5251: Wavrik, Barrie Stott and Marc de Groot.
1.29 anton 5252:
1.30 anton 5253: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
5254: and autoconf, among others), and to the creators of the Internet: Gforth
5255: was developed across the Internet, and its authors have not met
5256: physically yet.
5257:
1.29 anton 5258: @section Pedigree
1.43 anton 5259: @cindex Pedigree of Gforth
1.4 anton 5260:
1.17 anton 5261: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24 anton 5262: Dirk Zoller) will cross-fertilize each other. Of course, a significant
5263: part of the design of Gforth was prescribed by ANS Forth.
1.17 anton 5264:
1.23 pazsan 5265: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
5266: 32 bit native code version of VolksForth for the Atari ST, written
5267: mostly by Dietrich Weineck.
5268:
5269: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
5270: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24 anton 5271: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17 anton 5272:
1.34 anton 5273: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17 anton 5274: Forth-83 standard. !! Pedigree? When?
5275:
5276: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24 anton 5277: 1979. Robert Selzer and Bill Ragsdale developed the original
5278: implementation of fig-Forth for the 6502 based on microForth.
5279:
5280: The principal architect of microForth was Dean Sanderson. microForth was
1.41 anton 5281: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
1.24 anton 5282: the 1802, and subsequently implemented on the 8080, the 6800 and the
5283: Z80.
1.17 anton 5284:
1.24 anton 5285: All earlier Forth systems were custom-made, usually by Charles Moore,
1.30 anton 5286: who discovered (as he puts it) Forth during the late 60s. The first full
5287: Forth existed in 1971.
1.17 anton 5288:
5289: A part of the information in this section comes from @cite{The Evolution
5290: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
5291: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
5292: Notices 28(3), 1993. You can find more historical and genealogical
5293: information about Forth there.
5294:
1.43 anton 5295: @node Word Index, Concept Index, Origin, Top
5296: @unnumbered Word Index
1.4 anton 5297:
1.18 anton 5298: This index is as incomplete as the manual. Each word is listed with
5299: stack effect and wordset.
1.17 anton 5300:
5301: @printindex fn
5302:
1.43 anton 5303: @node Concept Index, , Word Index, Top
5304: @unnumbered Concept and Word Index
5305:
5306: This index is as incomplete as the manual. Not all entries listed are
5307: present verbatim in the text. Only the names are listed for the words
5308: here.
1.17 anton 5309:
1.43 anton 5310: @printindex cp
1.1 anton 5311:
5312: @contents
5313: @bye
5314:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>