[gforth] / gforth / Attic / gforth.ds

gforth: gforth/Attic/gforth.ds

1 :	anton	1.1	\input texinfo @c --texinfo--
2 :			@comment The source is gforth.ds, from which gforth.texi is generated
3 :			@comment %**start of header (This is for running Texinfo on a region.)
4 :	anton	1.4	@setfilename gforth.info
5 :	anton	1.17	@settitle Gforth Manual
6 :	anton	1.4	@comment @setchapternewpage odd
7 :	anton	1.1	@comment %**end of header (This is for running Texinfo on a region.)
8 :
9 :			@ifinfo
10 :	anton	1.30	This file documents Gforth 0.2
11 :	anton	1.1
12 :	anton	1.32	Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
13 :	anton	1.1
14 :			Permission is granted to make and distribute verbatim copies of
15 :			this manual provided the copyright notice and this permission notice
16 :			are preserved on all copies.
17 :
18 :	anton	1.4	@ignore
19 :	anton	1.1	Permission is granted to process this file through TeX and print the
20 :			results, provided the printed document carries a copying permission
21 :			notice identical to this one except for the removal of this paragraph
22 :			(this paragraph not being relevant to the printed manual).
23 :
24 :	anton	1.4	@end ignore
25 :	anton	1.1	Permission is granted to copy and distribute modified versions of this
26 :			manual under the conditions for verbatim copying, provided also that the
27 :			sections entitled "Distribution" and "General Public License" are
28 :			included exactly as in the original, and provided that the entire
29 :			resulting derived work is distributed under the terms of a permission
30 :			notice identical to this one.
31 :
32 :			Permission is granted to copy and distribute translations of this manual
33 :			into another language, under the above conditions for modified versions,
34 :			except that the sections entitled "Distribution" and "General Public
35 :			License" may be included in a translation approved by the author instead
36 :			of in the original English.
37 :			@end ifinfo
38 :
39 :	anton	1.24	@finalout
40 :	anton	1.1	@titlepage
41 :			@sp 10
42 :	anton	1.17	@center @titlefont{Gforth Manual}
43 :	anton	1.1	@sp 2
44 :	anton	1.30	@center for version 0.2
45 :	anton	1.1	@sp 2
46 :			@center Anton Ertl
47 :	anton	1.25	@center Bernd Paysan
48 :	anton	1.17	@sp 3
49 :			@center This manual is under construction
50 :	anton	1.1
51 :			@comment The following two commands start the copyright page.
52 :			@page
53 :			@vskip 0pt plus 1filll
54 :	anton	1.32	Copyright @copyright{} 1995,1996 Free Software Foundation, Inc.
55 :	anton	1.1
56 :			@comment !! Published by ... or You can get a copy of this manual ...
57 :
58 :			Permission is granted to make and distribute verbatim copies of
59 :			this manual provided the copyright notice and this permission notice
60 :			are preserved on all copies.
61 :
62 :			Permission is granted to copy and distribute modified versions of this
63 :			manual under the conditions for verbatim copying, provided also that the
64 :			sections entitled "Distribution" and "General Public License" are
65 :			included exactly as in the original, and provided that the entire
66 :			resulting derived work is distributed under the terms of a permission
67 :			notice identical to this one.
68 :
69 :			Permission is granted to copy and distribute translations of this manual
70 :			into another language, under the above conditions for modified versions,
71 :			except that the sections entitled "Distribution" and "General Public
72 :			License" may be included in a translation approved by the author instead
73 :			of in the original English.
74 :			@end titlepage
75 :
76 :
77 :			@node Top, License, (dir), (dir)
78 :			@ifinfo
79 :	anton	1.17	Gforth is a free implementation of ANS Forth available on many
80 :	anton	1.30	personal machines. This manual corresponds to version 0.2.
81 :	anton	1.1	@end ifinfo
82 :
83 :			@menu
84 :	anton	1.4	* License::
85 :	anton	1.17	* Goals:: About the Gforth Project
86 :	anton	1.4	* Other Books:: Things you might want to read
87 :	anton	1.17	* Invocation:: Starting Gforth
88 :			* Words:: Forth words available in Gforth
89 :	anton	1.40	* Tools:: Programming tools
90 :	anton	1.4	* ANS conformance:: Implementation-defined options etc.
91 :	anton	1.17	* Model:: The abstract machine of Gforth
92 :	anton	1.34	* Integrating Gforth:: Forth as scripting language for applications.
93 :	anton	1.17	* Emacs and Gforth:: The Gforth Mode
94 :	anton	1.4	* Internals:: Implementation details
95 :			* Bugs:: How to report them
96 :	anton	1.29	* Origin:: Authors and ancestors of Gforth
97 :	anton	1.4	* Word Index:: An item for each Forth word
98 :			* Node Index:: An item for each node
99 :	anton	1.1	@end menu
100 :
101 :			@node License, Goals, Top, Top
102 :	pazsan	1.20	@unnumbered GNU GENERAL PUBLIC LICENSE
103 :			@center Version 2, June 1991
104 :
105 :			@display
106 :			Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
107 :			675 Mass Ave, Cambridge, MA 02139, USA
108 :
109 :			Everyone is permitted to copy and distribute verbatim copies
110 :			of this license document, but changing it is not allowed.
111 :			@end display
112 :
113 :			@unnumberedsec Preamble
114 :
115 :			The licenses for most software are designed to take away your
116 :			freedom to share and change it. By contrast, the GNU General Public
117 :			License is intended to guarantee your freedom to share and change free
118 :			software---to make sure the software is free for all its users. This
119 :			General Public License applies to most of the Free Software
120 :			Foundation's software and to any other program whose authors commit to
121 :			using it. (Some other Free Software Foundation software is covered by
122 :			the GNU Library General Public License instead.) You can apply it to
123 :			your programs, too.
124 :
125 :			When we speak of free software, we are referring to freedom, not
126 :			price. Our General Public Licenses are designed to make sure that you
127 :			have the freedom to distribute copies of free software (and charge for
128 :			this service if you wish), that you receive source code or can get it
129 :			if you want it, that you can change the software or use pieces of it
130 :			in new free programs; and that you know you can do these things.
131 :
132 :			To protect your rights, we need to make restrictions that forbid
133 :			anyone to deny you these rights or to ask you to surrender the rights.
134 :			These restrictions translate to certain responsibilities for you if you
135 :			distribute copies of the software, or if you modify it.
136 :
137 :			For example, if you distribute copies of such a program, whether
138 :			gratis or for a fee, you must give the recipients all the rights that
139 :			you have. You must make sure that they, too, receive or can get the
140 :			source code. And you must show them these terms so they know their
141 :			rights.
142 :
143 :			We protect your rights with two steps: (1) copyright the software, and
144 :			(2) offer you this license which gives you legal permission to copy,
145 :			distribute and/or modify the software.
146 :
147 :			Also, for each author's protection and ours, we want to make certain
148 :			that everyone understands that there is no warranty for this free
149 :			software. If the software is modified by someone else and passed on, we
150 :			want its recipients to know that what they have is not the original, so
151 :			that any problems introduced by others will not reflect on the original
152 :			authors' reputations.
153 :
154 :			Finally, any free program is threatened constantly by software
155 :			patents. We wish to avoid the danger that redistributors of a free
156 :			program will individually obtain patent licenses, in effect making the
157 :			program proprietary. To prevent this, we have made it clear that any
158 :			patent must be licensed for everyone's free use or not licensed at all.
159 :
160 :			The precise terms and conditions for copying, distribution and
161 :			modification follow.
162 :
163 :			@iftex
164 :			@unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
165 :			@end iftex
166 :			@ifinfo
167 :			@center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
168 :			@end ifinfo
169 :
170 :			@enumerate 0
171 :			@item
172 :			This License applies to any program or other work which contains
173 :			a notice placed by the copyright holder saying it may be distributed
174 :			under the terms of this General Public License. The ``Program'', below,
175 :			refers to any such program or work, and a ``work based on the Program''
176 :			means either the Program or any derivative work under copyright law:
177 :			that is to say, a work containing the Program or a portion of it,
178 :			either verbatim or with modifications and/or translated into another
179 :			language. (Hereinafter, translation is included without limitation in
180 :			the term ``modification''.) Each licensee is addressed as ``you''.
181 :
182 :			Activities other than copying, distribution and modification are not
183 :			covered by this License; they are outside its scope. The act of
184 :			running the Program is not restricted, and the output from the Program
185 :			is covered only if its contents constitute a work based on the
186 :			Program (independent of having been made by running the Program).
187 :			Whether that is true depends on what the Program does.
188 :
189 :			@item
190 :			You may copy and distribute verbatim copies of the Program's
191 :			source code as you receive it, in any medium, provided that you
192 :			conspicuously and appropriately publish on each copy an appropriate
193 :			copyright notice and disclaimer of warranty; keep intact all the
194 :			notices that refer to this License and to the absence of any warranty;
195 :			and give any other recipients of the Program a copy of this License
196 :			along with the Program.
197 :
198 :			You may charge a fee for the physical act of transferring a copy, and
199 :			you may at your option offer warranty protection in exchange for a fee.
200 :
201 :			@item
202 :			You may modify your copy or copies of the Program or any portion
203 :			of it, thus forming a work based on the Program, and copy and
204 :			distribute such modifications or work under the terms of Section 1
205 :			above, provided that you also meet all of these conditions:
206 :
207 :			@enumerate a
208 :			@item
209 :			You must cause the modified files to carry prominent notices
210 :			stating that you changed the files and the date of any change.
211 :
212 :			@item
213 :			You must cause any work that you distribute or publish, that in
214 :			whole or in part contains or is derived from the Program or any
215 :			part thereof, to be licensed as a whole at no charge to all third
216 :			parties under the terms of this License.
217 :
218 :			@item
219 :			If the modified program normally reads commands interactively
220 :			when run, you must cause it, when started running for such
221 :			interactive use in the most ordinary way, to print or display an
222 :			announcement including an appropriate copyright notice and a
223 :			notice that there is no warranty (or else, saying that you provide
224 :			a warranty) and that users may redistribute the program under
225 :			these conditions, and telling the user how to view a copy of this
226 :			License. (Exception: if the Program itself is interactive but
227 :			does not normally print such an announcement, your work based on
228 :			the Program is not required to print an announcement.)
229 :			@end enumerate
230 :
231 :			These requirements apply to the modified work as a whole. If
232 :			identifiable sections of that work are not derived from the Program,
233 :			and can be reasonably considered independent and separate works in
234 :			themselves, then this License, and its terms, do not apply to those
235 :			sections when you distribute them as separate works. But when you
236 :			distribute the same sections as part of a whole which is a work based
237 :			on the Program, the distribution of the whole must be on the terms of
238 :			this License, whose permissions for other licensees extend to the
239 :			entire whole, and thus to each and every part regardless of who wrote it.
240 :
241 :			Thus, it is not the intent of this section to claim rights or contest
242 :			your rights to work written entirely by you; rather, the intent is to
243 :			exercise the right to control the distribution of derivative or
244 :			collective works based on the Program.
245 :
246 :			In addition, mere aggregation of another work not based on the Program
247 :			with the Program (or with a work based on the Program) on a volume of
248 :			a storage or distribution medium does not bring the other work under
249 :			the scope of this License.
250 :
251 :			@item
252 :			You may copy and distribute the Program (or a work based on it,
253 :			under Section 2) in object code or executable form under the terms of
254 :			Sections 1 and 2 above provided that you also do one of the following:
255 :
256 :			@enumerate a
257 :			@item
258 :			Accompany it with the complete corresponding machine-readable
259 :			source code, which must be distributed under the terms of Sections
260 :			1 and 2 above on a medium customarily used for software interchange; or,
261 :
262 :			@item
263 :			Accompany it with a written offer, valid for at least three
264 :			years, to give any third party, for a charge no more than your
265 :			cost of physically performing source distribution, a complete
266 :			machine-readable copy of the corresponding source code, to be
267 :			distributed under the terms of Sections 1 and 2 above on a medium
268 :			customarily used for software interchange; or,
269 :
270 :			@item
271 :			Accompany it with the information you received as to the offer
272 :			to distribute corresponding source code. (This alternative is
273 :			allowed only for noncommercial distribution and only if you
274 :			received the program in object code or executable form with such
275 :			an offer, in accord with Subsection b above.)
276 :			@end enumerate
277 :
278 :			The source code for a work means the preferred form of the work for
279 :			making modifications to it. For an executable work, complete source
280 :			code means all the source code for all modules it contains, plus any
281 :			associated interface definition files, plus the scripts used to
282 :			control compilation and installation of the executable. However, as a
283 :			special exception, the source code distributed need not include
284 :			anything that is normally distributed (in either source or binary
285 :			form) with the major components (compiler, kernel, and so on) of the
286 :			operating system on which the executable runs, unless that component
287 :			itself accompanies the executable.
288 :
289 :			If distribution of executable or object code is made by offering
290 :			access to copy from a designated place, then offering equivalent
291 :			access to copy the source code from the same place counts as
292 :			distribution of the source code, even though third parties are not
293 :			compelled to copy the source along with the object code.
294 :
295 :			@item
296 :			You may not copy, modify, sublicense, or distribute the Program
297 :			except as expressly provided under this License. Any attempt
298 :			otherwise to copy, modify, sublicense or distribute the Program is
299 :			void, and will automatically terminate your rights under this License.
300 :			However, parties who have received copies, or rights, from you under
301 :			this License will not have their licenses terminated so long as such
302 :			parties remain in full compliance.
303 :
304 :			@item
305 :			You are not required to accept this License, since you have not
306 :			signed it. However, nothing else grants you permission to modify or
307 :			distribute the Program or its derivative works. These actions are
308 :			prohibited by law if you do not accept this License. Therefore, by
309 :			modifying or distributing the Program (or any work based on the
310 :			Program), you indicate your acceptance of this License to do so, and
311 :			all its terms and conditions for copying, distributing or modifying
312 :			the Program or works based on it.
313 :
314 :			@item
315 :			Each time you redistribute the Program (or any work based on the
316 :			Program), the recipient automatically receives a license from the
317 :			original licensor to copy, distribute or modify the Program subject to
318 :			these terms and conditions. You may not impose any further
319 :			restrictions on the recipients' exercise of the rights granted herein.
320 :			You are not responsible for enforcing compliance by third parties to
321 :			this License.
322 :
323 :			@item
324 :			If, as a consequence of a court judgment or allegation of patent
325 :			infringement or for any other reason (not limited to patent issues),
326 :			conditions are imposed on you (whether by court order, agreement or
327 :			otherwise) that contradict the conditions of this License, they do not
328 :			excuse you from the conditions of this License. If you cannot
329 :			distribute so as to satisfy simultaneously your obligations under this
330 :			License and any other pertinent obligations, then as a consequence you
331 :			may not distribute the Program at all. For example, if a patent
332 :			license would not permit royalty-free redistribution of the Program by
333 :			all those who receive copies directly or indirectly through you, then
334 :			the only way you could satisfy both it and this License would be to
335 :			refrain entirely from distribution of the Program.
336 :
337 :			If any portion of this section is held invalid or unenforceable under
338 :			any particular circumstance, the balance of the section is intended to
339 :			apply and the section as a whole is intended to apply in other
340 :			circumstances.
341 :
342 :			It is not the purpose of this section to induce you to infringe any
343 :			patents or other property right claims or to contest validity of any
344 :			such claims; this section has the sole purpose of protecting the
345 :			integrity of the free software distribution system, which is
346 :			implemented by public license practices. Many people have made
347 :			generous contributions to the wide range of software distributed
348 :			through that system in reliance on consistent application of that
349 :			system; it is up to the author/donor to decide if he or she is willing
350 :			to distribute software through any other system and a licensee cannot
351 :			impose that choice.
352 :
353 :			This section is intended to make thoroughly clear what is believed to
354 :			be a consequence of the rest of this License.
355 :
356 :			@item
357 :			If the distribution and/or use of the Program is restricted in
358 :			certain countries either by patents or by copyrighted interfaces, the
359 :			original copyright holder who places the Program under this License
360 :			may add an explicit geographical distribution limitation excluding
361 :			those countries, so that distribution is permitted only in or among
362 :			countries not thus excluded. In such case, this License incorporates
363 :			the limitation as if written in the body of this License.
364 :
365 :			@item
366 :			The Free Software Foundation may publish revised and/or new versions
367 :			of the General Public License from time to time. Such new versions will
368 :			be similar in spirit to the present version, but may differ in detail to
369 :			address new problems or concerns.
370 :
371 :			Each version is given a distinguishing version number. If the Program
372 :			specifies a version number of this License which applies to it and ``any
373 :			later version'', you have the option of following the terms and conditions
374 :			either of that version or of any later version published by the Free
375 :			Software Foundation. If the Program does not specify a version number of
376 :			this License, you may choose any version ever published by the Free Software
377 :			Foundation.
378 :
379 :			@item
380 :			If you wish to incorporate parts of the Program into other free
381 :			programs whose distribution conditions are different, write to the author
382 :			to ask for permission. For software which is copyrighted by the Free
383 :			Software Foundation, write to the Free Software Foundation; we sometimes
384 :			make exceptions for this. Our decision will be guided by the two goals
385 :			of preserving the free status of all derivatives of our free software and
386 :			of promoting the sharing and reuse of software generally.
387 :
388 :			@iftex
389 :			@heading NO WARRANTY
390 :			@end iftex
391 :			@ifinfo
392 :			@center NO WARRANTY
393 :			@end ifinfo
394 :
395 :			@item
396 :			BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
397 :			FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
398 :			OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
399 :			PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
400 :			OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
401 :			MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
402 :			TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
403 :			PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
404 :			REPAIR OR CORRECTION.
405 :
406 :			@item
407 :			IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
408 :			WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
409 :			REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
410 :			INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
411 :			OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
412 :			TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
413 :			YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
414 :			PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
415 :			POSSIBILITY OF SUCH DAMAGES.
416 :			@end enumerate
417 :
418 :			@iftex
419 :			@heading END OF TERMS AND CONDITIONS
420 :			@end iftex
421 :			@ifinfo
422 :			@center END OF TERMS AND CONDITIONS
423 :			@end ifinfo
424 :
425 :			@page
426 :			@unnumberedsec How to Apply These Terms to Your New Programs
427 :
428 :			If you develop a new program, and you want it to be of the greatest
429 :			possible use to the public, the best way to achieve this is to make it
430 :			free software which everyone can redistribute and change under these terms.
431 :
432 :			To do so, attach the following notices to the program. It is safest
433 :			to attach them to the start of each source file to most effectively
434 :			convey the exclusion of warranty; and each file should have at least
435 :			the ``copyright'' line and a pointer to where the full notice is found.
436 :
437 :			@smallexample
438 :			@var{one line to give the program's name and a brief idea of what it does.}
439 :			Copyright (C) 19@var{yy} @var{name of author}
440 :
441 :			This program is free software; you can redistribute it and/or modify
442 :			it under the terms of the GNU General Public License as published by
443 :			the Free Software Foundation; either version 2 of the License, or
444 :			(at your option) any later version.
445 :
446 :			This program is distributed in the hope that it will be useful,
447 :			but WITHOUT ANY WARRANTY; without even the implied warranty of
448 :			MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
449 :			GNU General Public License for more details.
450 :
451 :			You should have received a copy of the GNU General Public License
452 :			along with this program; if not, write to the Free Software
453 :			Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
454 :			@end smallexample
455 :
456 :			Also add information on how to contact you by electronic and paper mail.
457 :
458 :			If the program is interactive, make it output a short notice like this
459 :			when it starts in an interactive mode:
460 :
461 :			@smallexample
462 :			Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
463 :			Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
464 :			type `show w'.
465 :			This is free software, and you are welcome to redistribute it
466 :			under certain conditions; type `show c' for details.
467 :			@end smallexample
468 :
469 :			The hypothetical commands @samp{show w} and @samp{show c} should show
470 :			the appropriate parts of the General Public License. Of course, the
471 :			commands you use may be called something other than @samp{show w} and
472 :			@samp{show c}; they could even be mouse-clicks or menu items---whatever
473 :			suits your program.
474 :
475 :			You should also get your employer (if you work as a programmer) or your
476 :			school, if any, to sign a ``copyright disclaimer'' for the program, if
477 :			necessary. Here is a sample; alter the names:
478 :
479 :			@smallexample
480 :			Yoyodyne, Inc., hereby disclaims all copyright interest in the program
481 :			`Gnomovision' (which makes passes at compilers) written by James Hacker.
482 :
483 :			@var{signature of Ty Coon}, 1 April 1989
484 :			Ty Coon, President of Vice
485 :			@end smallexample
486 :
487 :			This General Public License does not permit incorporating your program into
488 :			proprietary programs. If your program is a subroutine library, you may
489 :			consider it more useful to permit linking proprietary applications with the
490 :			library. If this is what you want to do, use the GNU Library General
491 :			Public License instead of this License.
492 :	anton	1.1
493 :			@iftex
494 :	pazsan	1.23	@node Preface
495 :			@comment node-name, next, previous, up
496 :	anton	1.1	@unnumbered Preface
497 :	pazsan	1.23	@cindex Preface
498 :	anton	1.17	This manual documents Gforth. The reader is expected to know
499 :	anton	1.1	Forth. This manual is primarily a reference manual. @xref{Other Books}
500 :			for introductory material.
501 :			@end iftex
502 :
503 :			@node Goals, Other Books, License, Top
504 :			@comment node-name, next, previous, up
505 :	anton	1.17	@chapter Goals of Gforth
506 :	anton	1.1	@cindex Goals
507 :	anton	1.17	The goal of the Gforth Project is to develop a standard model for
508 :	anton	1.1	ANSI Forth. This can be split into several subgoals:
509 :
510 :			@itemize @bullet
511 :			@item
512 :	anton	1.17	Gforth should conform to the ANSI Forth standard.
513 :	anton	1.1	@item
514 :			It should be a model, i.e. it should define all the
515 :			implementation-dependent things.
516 :			@item
517 :			It should become standard, i.e. widely accepted and used. This goal
518 :			is the most difficult one.
519 :			@end itemize
520 :
521 :	anton	1.17	To achieve these goals Gforth should be
522 :	anton	1.1	@itemize @bullet
523 :			@item
524 :			Similar to previous models (fig-Forth, F83)
525 :			@item
526 :			Powerful. It should provide for all the things that are considered
527 :			necessary today and even some that are not yet considered necessary.
528 :			@item
529 :			Efficient. It should not get the reputation of being exceptionally
530 :			slow.
531 :			@item
532 :			Free.
533 :			@item
534 :			Available on many machines/easy to port.
535 :			@end itemize
536 :
537 :	anton	1.17	Have we achieved these goals? Gforth conforms to the ANS Forth
538 :			standard. It may be considered a model, but we have not yet documented
539 :	anton	1.1	which parts of the model are stable and which parts we are likely to
540 :	anton	1.17	change. It certainly has not yet become a de facto standard. It has some
541 :			similarities and some differences to previous models. It has some
542 :			powerful features, but not yet everything that we envisioned. We
543 :			certainly have achieved our execution speed goals (@pxref{Performance}).
544 :			It is free and available on many machines.
545 :	anton	1.1
546 :			@node Other Books, Invocation, Goals, Top
547 :			@chapter Other books on ANS Forth
548 :
549 :			As the standard is relatively new, there are not many books out yet. It
550 :	anton	1.17	is not recommended to learn Forth by using Gforth and a book that is
551 :	anton	1.1	not written for ANS Forth, as you will not know your mistakes from the
552 :			deviations of the book.
553 :
554 :			There is, of course, the standard, the definite reference if you want to
555 :	anton	1.19	write ANS Forth programs. It is available in printed form from the
556 :			National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
557 :			Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
558 :			can also get it from Global Engineering Documents (Tel.: USA (800)
559 :			854-7179; Fax.: (303) 843-9880) for about $300.
560 :
561 :			@cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
562 :			for publication is available electronically and for free in some MS Word
563 :			format, and it has been converted to HTML. Some pointers to these
564 :			versions can be found through
565 :	anton	1.24	@*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
566 :	anton	1.1
567 :	anton	1.21	@cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
568 :	anton	1.1	introductory book based on a draft version of the standard. It does not
569 :			cover the whole standard. It also contains interesting background
570 :			information (Jack Woehr was in the ANS Forth Technical Committe). It is
571 :			not appropriate for complete newbies, but programmers experienced in
572 :			other languages should find it ok.
573 :
574 :			@node Invocation, Words, Other Books, Top
575 :			@chapter Invocation
576 :
577 :			You will usually just say @code{gforth}. In many other cases the default
578 :	anton	1.17	Gforth image will be invoked like this:
579 :	anton	1.1
580 :			@example
581 :			gforth [files] [-e forth-code]
582 :			@end example
583 :
584 :			executing the contents of the files and the Forth code in the order they
585 :			are given.
586 :
587 :			In general, the command line looks like this:
588 :
589 :			@example
590 :			gforth [initialization options] [image-specific options]
591 :			@end example
592 :
593 :			The initialization options must come before the rest of the command
594 :			line. They are:
595 :
596 :			@table @code
597 :			@item --image-file @var{file}
598 :	pazsan	1.20	@item -i @var{file}
599 :	anton	1.1	Loads the Forth image @var{file} instead of the default
600 :			@file{gforth.fi}.
601 :
602 :			@item --path @var{path}
603 :	pazsan	1.20	@item -p @var{path}
604 :	anton	1.39	Uses @var{path} for searching the image file and Forth source code files
605 :			instead of the default in the environment variable @code{GFORTHPATH} or
606 :			the path specified at installation time (e.g.,
607 :			@file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
608 :			directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
609 :	anton	1.1
610 :			@item --dictionary-size @var{size}
611 :			@item -m @var{size}
612 :			Allocate @var{size} space for the Forth dictionary space instead of
613 :			using the default specified in the image (typically 256K). The
614 :			@var{size} specification consists of an integer and a unit (e.g.,
615 :			@code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
616 :			size, in this case Cells), @code{k} (kilobytes), and @code{M}
617 :			(Megabytes). If no unit is specified, @code{e} is used.
618 :
619 :			@item --data-stack-size @var{size}
620 :			@item -d @var{size}
621 :			Allocate @var{size} space for the data stack instead of using the
622 :			default specified in the image (typically 16K).
623 :
624 :			@item --return-stack-size @var{size}
625 :			@item -r @var{size}
626 :			Allocate @var{size} space for the return stack instead of using the
627 :			default specified in the image (typically 16K).
628 :
629 :			@item --fp-stack-size @var{size}
630 :			@item -f @var{size}
631 :			Allocate @var{size} space for the floating point stack instead of
632 :			using the default specified in the image (typically 16K). In this case
633 :			the unit specifier @code{e} refers to floating point numbers.
634 :
635 :			@item --locals-stack-size @var{size}
636 :			@item -l @var{size}
637 :			Allocate @var{size} space for the locals stack instead of using the
638 :			default specified in the image (typically 16K).
639 :
640 :			@end table
641 :
642 :			As explained above, the image-specific command-line arguments for the
643 :			default image @file{gforth.fi} consist of a sequence of filenames and
644 :			@code{-e @var{forth-code}} options that are interpreted in the seqence
645 :			in which they are given. The @code{-e @var{forth-code}} or
646 :			@code{--evaluate @var{forth-code}} option evaluates the forth
647 :			code. This option takes only one argument; if you want to evaluate more
648 :			Forth words, you have to quote them or use several @code{-e}s. To exit
649 :			after processing the command line (instead of entering interactive mode)
650 :			append @code{-e bye} to the command line.
651 :
652 :	anton	1.22	If you have several versions of Gforth installed, @code{gforth} will
653 :			invoke the version that was installed last. @code{gforth-@var{version}}
654 :			invokes a specific version. You may want to use the option
655 :			@code{--path}, if your environment contains the variable
656 :			@code{GFORTHPATH}.
657 :
658 :	anton	1.1	Not yet implemented:
659 :			On startup the system first executes the system initialization file
660 :			(unless the option @code{--no-init-file} is given; note that the system
661 :			resulting from using this option may not be ANS Forth conformant). Then
662 :			the user initialization file @file{.gforth.fs} is executed, unless the
663 :			option @code{--no-rc} is given; this file is first searched in @file{.},
664 :			then in @file{~}, then in the normal path (see above).
665 :
666 :	anton	1.40	@node Words, Tools, Invocation, Top
667 :	anton	1.1	@chapter Forth Words
668 :
669 :			@menu
670 :	anton	1.4	* Notation::
671 :			* Arithmetic::
672 :			* Stack Manipulation::
673 :			* Memory access::
674 :			* Control Structures::
675 :			* Locals::
676 :			* Defining Words::
677 :	anton	1.37	* Tokens for Words::
678 :	anton	1.4	* Wordlists::
679 :			* Files::
680 :			* Blocks::
681 :			* Other I/O::
682 :			* Programming Tools::
683 :	anton	1.18	* Assembler and Code words::
684 :	anton	1.4	* Threading Words::
685 :	anton	1.1	@end menu
686 :
687 :			@node Notation, Arithmetic, Words, Words
688 :			@section Notation
689 :
690 :			The Forth words are described in this section in the glossary notation
691 :			that has become a de-facto standard for Forth texts, i.e.
692 :
693 :	anton	1.4	@format
694 :	anton	1.1	@var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
695 :	anton	1.4	@end format
696 :	anton	1.1	@var{Description}
697 :
698 :			@table @var
699 :			@item word
700 :	anton	1.17	The name of the word. BTW, Gforth is case insensitive, so you can
701 :	anton	1.14	type the words in in lower case (However, @pxref{core-idef}).
702 :	anton	1.1
703 :			@item Stack effect
704 :			The stack effect is written in the notation @code{@var{before} --
705 :			@var{after}}, where @var{before} and @var{after} describe the top of
706 :			stack entries before and after the execution of the word. The rest of
707 :			the stack is not touched by the word. The top of stack is rightmost,
708 :	anton	1.17	i.e., a stack sequence is written as it is typed in. Note that Gforth
709 :	anton	1.1	uses a separate floating point stack, but a unified stack
710 :			notation. Also, return stack effects are not shown in @var{stack
711 :			effect}, but in @var{Description}. The name of a stack item describes
712 :			the type and/or the function of the item. See below for a discussion of
713 :			the types.
714 :
715 :	anton	1.19	All words have two stack effects: A compile-time stack effect and a
716 :			run-time stack effect. The compile-time stack-effect of most words is
717 :			@var{ -- }. If the compile-time stack-effect of a word deviates from
718 :			this standard behaviour, or the word does other unusual things at
719 :			compile time, both stack effects are shown; otherwise only the run-time
720 :			stack effect is shown.
721 :
722 :	anton	1.1	@item pronunciation
723 :			How the word is pronounced
724 :
725 :			@item wordset
726 :			The ANS Forth standard is divided into several wordsets. A standard
727 :			system need not support all of them. So, the fewer wordsets your program
728 :			uses the more portable it will be in theory. However, we suspect that
729 :			most ANS Forth systems on personal machines will feature all
730 :			wordsets. Words that are not defined in the ANS standard have
731 :	anton	1.19	@code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
732 :			describes words that will work in future releases of Gforth;
733 :			@code{gforth-internal} words are more volatile. Environmental query
734 :			strings are also displayed like words; you can recognize them by the
735 :			@code{environment} in the wordset field.
736 :	anton	1.1
737 :			@item Description
738 :			A description of the behaviour of the word.
739 :			@end table
740 :
741 :	anton	1.4	The type of a stack item is specified by the character(s) the name
742 :			starts with:
743 :	anton	1.1
744 :			@table @code
745 :			@item f
746 :			Bool, i.e. @code{false} or @code{true}.
747 :			@item c
748 :			Char
749 :			@item w
750 :			Cell, can contain an integer or an address
751 :			@item n
752 :			signed integer
753 :			@item u
754 :			unsigned integer
755 :			@item d
756 :			double sized signed integer
757 :			@item ud
758 :			double sized unsigned integer
759 :			@item r
760 :	anton	1.36	Float (on the FP stack)
761 :	anton	1.1	@item a_
762 :			Cell-aligned address
763 :			@item c_
764 :	anton	1.36	Char-aligned address (note that a Char may have two bytes in Windows NT)
765 :	anton	1.1	@item f_
766 :			Float-aligned address
767 :			@item df_
768 :			Address aligned for IEEE double precision float
769 :			@item sf_
770 :			Address aligned for IEEE single precision float
771 :			@item xt
772 :			Execution token, same size as Cell
773 :			@item wid
774 :			Wordlist ID, same size as Cell
775 :			@item f83name
776 :			Pointer to a name structure
777 :	anton	1.36	@item "
778 :			string in the input stream (not the stack). The terminating character is
779 :			a blank by default. If it is not a blank, it is shown in @code{<>}
780 :			quotes.
781 :
782 :	anton	1.1	@end table
783 :
784 :	anton	1.4	@node Arithmetic, Stack Manipulation, Notation, Words
785 :	anton	1.1	@section Arithmetic
786 :			Forth arithmetic is not checked, i.e., you will not hear about integer
787 :			overflow on addition or multiplication, you may hear about division by
788 :			zero if you are lucky. The operator is written after the operands, but
789 :			the operands are still in the original order. I.e., the infix @code{2-1}
790 :			corresponds to @code{2 1 -}. Forth offers a variety of division
791 :			operators. If you perform division with potentially negative operands,
792 :			you do not want to use @code{/} or @code{/mod} with its undefined
793 :			behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
794 :	anton	1.4	former, @pxref{Mixed precision}).
795 :
796 :			@menu
797 :			* Single precision::
798 :			* Bitwise operations::
799 :			* Mixed precision:: operations with single and double-cell integers
800 :			* Double precision:: Double-cell integer arithmetic
801 :			* Floating Point::
802 :			@end menu
803 :	anton	1.1
804 :	anton	1.4	@node Single precision, Bitwise operations, Arithmetic, Arithmetic
805 :	anton	1.1	@subsection Single precision
806 :			doc-+
807 :			doc--
808 :			doc-*
809 :			doc-/
810 :			doc-mod
811 :			doc-/mod
812 :			doc-negate
813 :			doc-abs
814 :			doc-min
815 :			doc-max
816 :
817 :	anton	1.4	@node Bitwise operations, Mixed precision, Single precision, Arithmetic
818 :	anton	1.1	@subsection Bitwise operations
819 :			doc-and
820 :			doc-or
821 :			doc-xor
822 :			doc-invert
823 :			doc-2*
824 :			doc-2/
825 :
826 :	anton	1.4	@node Mixed precision, Double precision, Bitwise operations, Arithmetic
827 :	anton	1.1	@subsection Mixed precision
828 :			doc-m+
829 :			doc-*/
830 :			doc-*/mod
831 :			doc-m*
832 :			doc-um*
833 :			doc-m*/
834 :			doc-um/mod
835 :			doc-fm/mod
836 :			doc-sm/rem
837 :
838 :	anton	1.4	@node Double precision, Floating Point, Mixed precision, Arithmetic
839 :	anton	1.1	@subsection Double precision
840 :	anton	1.16
841 :			The outer (aka text) interpreter converts numbers containing a dot into
842 :			a double precision number. Note that only numbers with the dot as last
843 :			character are standard-conforming.
844 :
845 :	anton	1.1	doc-d+
846 :			doc-d-
847 :			doc-dnegate
848 :			doc-dabs
849 :			doc-dmin
850 :			doc-dmax
851 :
852 :	anton	1.4	@node Floating Point, , Double precision, Arithmetic
853 :			@subsection Floating Point
854 :	anton	1.16
855 :			The format of floating point numbers recognized by the outer (aka text)
856 :			interpreter is: a signed decimal number, possibly containing a decimal
857 :			point (@code{.}), followed by @code{E} or @code{e}, optionally followed
858 :			by a signed integer (the exponent). E.g., @code{1e} ist the same as
859 :	anton	1.35	@code{+1.0e+0}. Note that a number without @code{e}
860 :	anton	1.16	is not interpreted as floating-point number, but as double (if the
861 :			number contains a @code{.}) or single precision integer. Also,
862 :			conversions between string and floating point numbers always use base
863 :			10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
864 :			value greater then 14, the @code{E} may be interpreted as digit and the
865 :			number will be interpreted as integer, unless it has a signed exponent
866 :			(both @code{+} and @code{-} are allowed as signs).
867 :	anton	1.4
868 :			Angles in floating point operations are given in radians (a full circle
869 :	anton	1.17	has 2 pi radians). Note, that Gforth has a separate floating point
870 :	anton	1.4	stack, but we use the unified notation.
871 :
872 :			Floating point numbers have a number of unpleasant surprises for the
873 :			unwary (e.g., floating point addition is not associative) and even a few
874 :			for the wary. You should not use them unless you know what you are doing
875 :			or you don't care that the results you get are totally bogus. If you
876 :			want to learn about the problems of floating point numbers (and how to
877 :	anton	1.11	avoid them), you might start with @cite{David Goldberg, What Every
878 :	anton	1.6	Computer Scientist Should Know About Floating-Point Arithmetic, ACM
879 :			Computing Surveys 23(1):5@minus{}48, March 1991}.
880 :	anton	1.4
881 :			doc-f+
882 :			doc-f-
883 :			doc-f*
884 :			doc-f/
885 :			doc-fnegate
886 :			doc-fabs
887 :			doc-fmax
888 :			doc-fmin
889 :			doc-floor
890 :			doc-fround
891 :			doc-f**
892 :			doc-fsqrt
893 :			doc-fexp
894 :			doc-fexpm1
895 :			doc-fln
896 :			doc-flnp1
897 :			doc-flog
898 :	anton	1.6	doc-falog
899 :	anton	1.4	doc-fsin
900 :			doc-fcos
901 :			doc-fsincos
902 :			doc-ftan
903 :			doc-fasin
904 :			doc-facos
905 :			doc-fatan
906 :			doc-fatan2
907 :			doc-fsinh
908 :			doc-fcosh
909 :			doc-ftanh
910 :			doc-fasinh
911 :			doc-facosh
912 :			doc-fatanh
913 :
914 :			@node Stack Manipulation, Memory access, Arithmetic, Words
915 :	anton	1.1	@section Stack Manipulation
916 :
917 :	anton	1.17	Gforth has a data stack (aka parameter stack) for characters, cells,
918 :	anton	1.1	addresses, and double cells, a floating point stack for floating point
919 :			numbers, a return stack for storing the return addresses of colon
920 :			definitions and other data, and a locals stack for storing local
921 :			variables. Note that while every sane Forth has a separate floating
922 :			point stack, this is not strictly required; an ANS Forth system could
923 :			theoretically keep floating point numbers on the data stack. As an
924 :			additional difficulty, you don't know how many cells a floating point
925 :			number takes. It is reportedly possible to write words in a way that
926 :			they work also for a unified stack model, but we do not recommend trying
927 :	anton	1.4	it. Instead, just say that your program has an environmental dependency
928 :			on a separate FP stack.
929 :
930 :			Also, a Forth system is allowed to keep the local variables on the
931 :	anton	1.1	return stack. This is reasonable, as local variables usually eliminate
932 :			the need to use the return stack explicitly. So, if you want to produce
933 :			a standard complying program and if you are using local variables in a
934 :			word, forget about return stack manipulations in that word (see the
935 :			standard document for the exact rules).
936 :
937 :	anton	1.4	@menu
938 :			* Data stack::
939 :			* Floating point stack::
940 :			* Return stack::
941 :			* Locals stack::
942 :			* Stack pointer manipulation::
943 :			@end menu
944 :
945 :			@node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
946 :	anton	1.1	@subsection Data stack
947 :			doc-drop
948 :			doc-nip
949 :			doc-dup
950 :			doc-over
951 :			doc-tuck
952 :			doc-swap
953 :			doc-rot
954 :			doc--rot
955 :			doc-?dup
956 :			doc-pick
957 :			doc-roll
958 :			doc-2drop
959 :			doc-2nip
960 :			doc-2dup
961 :			doc-2over
962 :			doc-2tuck
963 :			doc-2swap
964 :			doc-2rot
965 :
966 :	anton	1.4	@node Floating point stack, Return stack, Data stack, Stack Manipulation
967 :	anton	1.1	@subsection Floating point stack
968 :			doc-fdrop
969 :			doc-fnip
970 :			doc-fdup
971 :			doc-fover
972 :			doc-ftuck
973 :			doc-fswap
974 :			doc-frot
975 :
976 :	anton	1.4	@node Return stack, Locals stack, Floating point stack, Stack Manipulation
977 :	anton	1.1	@subsection Return stack
978 :			doc->r
979 :			doc-r>
980 :			doc-r@
981 :			doc-rdrop
982 :			doc-2>r
983 :			doc-2r>
984 :			doc-2r@
985 :			doc-2rdrop
986 :
987 :	anton	1.4	@node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
988 :	anton	1.1	@subsection Locals stack
989 :
990 :	anton	1.4	@node Stack pointer manipulation, , Locals stack, Stack Manipulation
991 :	anton	1.1	@subsection Stack pointer manipulation
992 :			doc-sp@
993 :			doc-sp!
994 :			doc-fp@
995 :			doc-fp!
996 :			doc-rp@
997 :			doc-rp!
998 :			doc-lp@
999 :			doc-lp!
1000 :
1001 :	anton	1.4	@node Memory access, Control Structures, Stack Manipulation, Words
1002 :	anton	1.1	@section Memory access
1003 :
1004 :	anton	1.4	@menu
1005 :			* Stack-Memory transfers::
1006 :			* Address arithmetic::
1007 :			* Memory block access::
1008 :			@end menu
1009 :
1010 :			@node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1011 :	anton	1.1	@subsection Stack-Memory transfers
1012 :
1013 :			doc-@
1014 :			doc-!
1015 :			doc-+!
1016 :			doc-c@
1017 :			doc-c!
1018 :			doc-2@
1019 :			doc-2!
1020 :			doc-f@
1021 :			doc-f!
1022 :			doc-sf@
1023 :			doc-sf!
1024 :			doc-df@
1025 :			doc-df!
1026 :
1027 :	anton	1.4	@node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1028 :	anton	1.1	@subsection Address arithmetic
1029 :
1030 :			ANS Forth does not specify the sizes of the data types. Instead, it
1031 :			offers a number of words for computing sizes and doing address
1032 :			arithmetic. Basically, address arithmetic is performed in terms of
1033 :			address units (aus); on most systems the address unit is one byte. Note
1034 :			that a character may have more than one au, so @code{chars} is no noop
1035 :			(on systems where it is a noop, it compiles to nothing).
1036 :
1037 :			ANS Forth also defines words for aligning addresses for specific
1038 :			addresses. Many computers require that accesses to specific data types
1039 :			must only occur at specific addresses; e.g., that cells may only be
1040 :			accessed at addresses divisible by 4. Even if a machine allows unaligned
1041 :			accesses, it can usually perform aligned accesses faster.
1042 :
1043 :	anton	1.17	For the performance-conscious: alignment operations are usually only
1044 :	anton	1.1	necessary during the definition of a data structure, not during the
1045 :			(more frequent) accesses to it.
1046 :
1047 :			ANS Forth defines no words for character-aligning addresses. This is not
1048 :			an oversight, but reflects the fact that addresses that are not
1049 :			char-aligned have no use in the standard and therefore will not be
1050 :			created.
1051 :
1052 :			The standard guarantees that addresses returned by @code{CREATE}d words
1053 :	anton	1.17	are cell-aligned; in addition, Gforth guarantees that these addresses
1054 :	anton	1.1	are aligned for all purposes.
1055 :
1056 :	anton	1.9	Note that the standard defines a word @code{char}, which has nothing to
1057 :			do with address arithmetic.
1058 :
1059 :	anton	1.1	doc-chars
1060 :			doc-char+
1061 :			doc-cells
1062 :			doc-cell+
1063 :			doc-align
1064 :			doc-aligned
1065 :			doc-floats
1066 :			doc-float+
1067 :			doc-falign
1068 :			doc-faligned
1069 :			doc-sfloats
1070 :			doc-sfloat+
1071 :			doc-sfalign
1072 :			doc-sfaligned
1073 :			doc-dfloats
1074 :			doc-dfloat+
1075 :			doc-dfalign
1076 :			doc-dfaligned
1077 :	anton	1.10	doc-maxalign
1078 :			doc-maxaligned
1079 :			doc-cfalign
1080 :			doc-cfaligned
1081 :	anton	1.1	doc-address-unit-bits
1082 :
1083 :	anton	1.4	@node Memory block access, , Address arithmetic, Memory access
1084 :	anton	1.1	@subsection Memory block access
1085 :
1086 :			doc-move
1087 :			doc-erase
1088 :
1089 :			While the previous words work on address units, the rest works on
1090 :			characters.
1091 :
1092 :			doc-cmove
1093 :			doc-cmove>
1094 :			doc-fill
1095 :			doc-blank
1096 :
1097 :	anton	1.4	@node Control Structures, Locals, Memory access, Words
1098 :	anton	1.1	@section Control Structures
1099 :
1100 :			Control structures in Forth cannot be used in interpret state, only in
1101 :			compile state, i.e., in a colon definition. We do not like this
1102 :			limitation, but have not seen a satisfying way around it yet, although
1103 :			many schemes have been proposed.
1104 :
1105 :	anton	1.4	@menu
1106 :			* Selection::
1107 :			* Simple Loops::
1108 :			* Counted Loops::
1109 :			* Arbitrary control structures::
1110 :			* Calls and returns::
1111 :			* Exception Handling::
1112 :			@end menu
1113 :
1114 :			@node Selection, Simple Loops, Control Structures, Control Structures
1115 :	anton	1.1	@subsection Selection
1116 :
1117 :			@example
1118 :			@var{flag}
1119 :			IF
1120 :			@var{code}
1121 :			ENDIF
1122 :			@end example
1123 :			or
1124 :			@example
1125 :			@var{flag}
1126 :			IF
1127 :			@var{code1}
1128 :			ELSE
1129 :			@var{code2}
1130 :			ENDIF
1131 :			@end example
1132 :
1133 :	anton	1.4	You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1134 :	anton	1.1	standard, and @code{ENDIF} is not, although it is quite popular. We
1135 :			recommend using @code{ENDIF}, because it is less confusing for people
1136 :			who also know other languages (and is not prone to reinforcing negative
1137 :			prejudices against Forth in these people). Adding @code{ENDIF} to a
1138 :			system that only supplies @code{THEN} is simple:
1139 :			@example
1140 :			: endif POSTPONE then ; immediate
1141 :			@end example
1142 :
1143 :			[According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1144 :			(adv.)} has the following meanings:
1145 :			@quotation
1146 :			... 2b: following next after in order ... 3d: as a necessary consequence
1147 :			(if you were there, then you saw them).
1148 :			@end quotation
1149 :			Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1150 :			and many other programming languages has the meaning 3d.]
1151 :
1152 :	anton	1.31	Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1153 :			you can avoid using @code{?dup}. Using these alternatives is also more
1154 :			efficient than using @code{?dup}. Definitions in plain standard Forth
1155 :			for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1156 :			@file{compat/control.fs}.
1157 :	anton	1.1
1158 :			@example
1159 :			@var{n}
1160 :			CASE
1161 :			@var{n1} OF @var{code1} ENDOF
1162 :			@var{n2} OF @var{code2} ENDOF
1163 :	anton	1.4	@dots{}
1164 :	anton	1.1	ENDCASE
1165 :			@end example
1166 :
1167 :			Executes the first @var{codei}, where the @var{ni} is equal to
1168 :			@var{n}. A default case can be added by simply writing the code after
1169 :			the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1170 :			but must not consume it.
1171 :
1172 :	anton	1.4	@node Simple Loops, Counted Loops, Selection, Control Structures
1173 :	anton	1.1	@subsection Simple Loops
1174 :
1175 :			@example
1176 :			BEGIN
1177 :			@var{code1}
1178 :			@var{flag}
1179 :			WHILE
1180 :			@var{code2}
1181 :			REPEAT
1182 :			@end example
1183 :
1184 :			@var{code1} is executed and @var{flag} is computed. If it is true,
1185 :			@var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1186 :
1187 :			@example
1188 :			BEGIN
1189 :			@var{code}
1190 :			@var{flag}
1191 :			UNTIL
1192 :			@end example
1193 :
1194 :			@var{code} is executed. The loop is restarted if @code{flag} is false.
1195 :
1196 :			@example
1197 :			BEGIN
1198 :			@var{code}
1199 :			AGAIN
1200 :			@end example
1201 :
1202 :			This is an endless loop.
1203 :
1204 :	anton	1.4	@node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1205 :	anton	1.1	@subsection Counted Loops
1206 :
1207 :			The basic counted loop is:
1208 :			@example
1209 :			@var{limit} @var{start}
1210 :			?DO
1211 :			@var{body}
1212 :			LOOP
1213 :			@end example
1214 :
1215 :			This performs one iteration for every integer, starting from @var{start}
1216 :			and up to, but excluding @var{limit}. The counter, aka index, can be
1217 :			accessed with @code{i}. E.g., the loop
1218 :			@example
1219 :			10 0 ?DO
1220 :			i .
1221 :			LOOP
1222 :			@end example
1223 :			prints
1224 :			@example
1225 :			0 1 2 3 4 5 6 7 8 9
1226 :			@end example
1227 :			The index of the innermost loop can be accessed with @code{i}, the index
1228 :			of the next loop with @code{j}, and the index of the third loop with
1229 :			@code{k}.
1230 :
1231 :			The loop control data are kept on the return stack, so there are some
1232 :			restrictions on mixing return stack accesses and counted loop
1233 :			words. E.g., if you put values on the return stack outside the loop, you
1234 :			cannot read them inside the loop. If you put values on the return stack
1235 :			within a loop, you have to remove them before the end of the loop and
1236 :			before accessing the index of the loop.
1237 :
1238 :			There are several variations on the counted loop:
1239 :
1240 :			@code{LEAVE} leaves the innermost counted loop immediately.
1241 :
1242 :	anton	1.18	If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1243 :			(and @code{LOOP} iterates until they become equal by wrap-around
1244 :			arithmetic). This behaviour is usually not what you want. Therefore,
1245 :			Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1246 :			@code{?DO}), which do not enter the loop if @var{start} is greater than
1247 :			@var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1248 :	anton	1.30	unsigned loop parameters.
1249 :	anton	1.18
1250 :	anton	1.1	@code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1251 :			index by @var{n} instead of by 1. The loop is terminated when the border
1252 :			between @var{limit-1} and @var{limit} is crossed. E.g.:
1253 :
1254 :	anton	1.18	@code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1255 :	anton	1.1
1256 :	anton	1.18	@code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1257 :	anton	1.1
1258 :			The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1259 :
1260 :	anton	1.2	@code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1261 :	anton	1.1
1262 :	anton	1.2	@code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1263 :	anton	1.1
1264 :	anton	1.18	Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1265 :			@var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1266 :			index by @var{u} each iteration. The loop is terminated when the border
1267 :			between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1268 :			@code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1269 :	anton	1.1
1270 :	anton	1.18	@code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1271 :	anton	1.1
1272 :	anton	1.18	@code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1273 :	anton	1.1
1274 :	anton	1.18	@code{ 0 0 -DO i . 1 -LOOP} prints nothing
1275 :	anton	1.1
1276 :	anton	1.30	Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1277 :			@code{-LOOP} are not in the ANS Forth standard. However, an
1278 :			implementation for these words that uses only standard words is provided
1279 :			in @file{compat/loops.fs}.
1280 :	anton	1.18
1281 :			@code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1282 :			the loop, independent of the loop parameters. Do not use @code{DO}, even
1283 :			if you know that the loop is entered in any case. Such knowledge tends
1284 :			to become invalid during maintenance of a program, and then the
1285 :			@code{DO} will make trouble.
1286 :	anton	1.1
1287 :			@code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1288 :			@code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1289 :			return stack so @code{EXIT} can get to its return address.
1290 :
1291 :			Another counted loop is
1292 :			@example
1293 :			@var{n}
1294 :			FOR
1295 :			@var{body}
1296 :			NEXT
1297 :			@end example
1298 :			This is the preferred loop of native code compiler writers who are too
1299 :	anton	1.17	lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1300 :	anton	1.1	iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1301 :			and ending with 0. Other Forth systems may behave differently, even if
1302 :	anton	1.30	they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1303 :			loops.
1304 :	anton	1.1
1305 :	anton	1.4	@node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1306 :	anton	1.2	@subsection Arbitrary control structures
1307 :
1308 :			ANS Forth permits and supports using control structures in a non-nested
1309 :			way. Information about incomplete control structures is stored on the
1310 :			control-flow stack. This stack may be implemented on the Forth data
1311 :	anton	1.17	stack, and this is what we have done in Gforth.
1312 :	anton	1.2
1313 :			An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1314 :			entry represents a backward branch target. A few words are the basis for
1315 :			building any control structure possible (except control structures that
1316 :			need storage, like calls, coroutines, and backtracking).
1317 :
1318 :	anton	1.3	doc-if
1319 :			doc-ahead
1320 :			doc-then
1321 :			doc-begin
1322 :			doc-until
1323 :			doc-again
1324 :			doc-cs-pick
1325 :			doc-cs-roll
1326 :	anton	1.2
1327 :	anton	1.17	On many systems control-flow stack items take one word, in Gforth they
1328 :	anton	1.2	currently take three (this may change in the future). Therefore it is a
1329 :			really good idea to manipulate the control flow stack with
1330 :			@code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1331 :			words.
1332 :
1333 :			Some standard control structure words are built from these words:
1334 :
1335 :	anton	1.3	doc-else
1336 :			doc-while
1337 :			doc-repeat
1338 :	anton	1.2
1339 :	anton	1.31	Gforth adds some more control-structure words:
1340 :
1341 :			doc-endif
1342 :			doc-?dup-if
1343 :			doc-?dup-0=-if
1344 :
1345 :	anton	1.2	Counted loop words constitute a separate group of words:
1346 :
1347 :	anton	1.3	doc-?do
1348 :	anton	1.18	doc-+do
1349 :			doc-u+do
1350 :			doc--do
1351 :			doc-u-do
1352 :	anton	1.3	doc-do
1353 :			doc-for
1354 :			doc-loop
1355 :			doc-+loop
1356 :	anton	1.18	doc--loop
1357 :	anton	1.3	doc-next
1358 :			doc-leave
1359 :			doc-?leave
1360 :			doc-unloop
1361 :	anton	1.10	doc-done
1362 :	anton	1.2
1363 :			The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1364 :			@i{do-sys}. Our system allows it, but it's your job to ensure that for
1365 :			every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1366 :	anton	1.3	through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1367 :			fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1368 :	pazsan	1.7	resolved (by using one of the loop-ending words or @code{DONE}).
1369 :	anton	1.2
1370 :			Another group of control structure words are
1371 :
1372 :	anton	1.3	doc-case
1373 :			doc-endcase
1374 :			doc-of
1375 :			doc-endof
1376 :	anton	1.2
1377 :			@i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1378 :			@code{cs-roll}.
1379 :
1380 :	anton	1.3	@subsubsection Programming Style
1381 :
1382 :			In order to ensure readability we recommend that you do not create
1383 :			arbitrary control structures directly, but define new control structure
1384 :			words for the control structure you want and use these words in your
1385 :			program.
1386 :
1387 :			E.g., instead of writing
1388 :
1389 :			@example
1390 :			begin
1391 :			...
1392 :			if [ 1 cs-roll ]
1393 :			...
1394 :			again then
1395 :			@end example
1396 :
1397 :			we recommend defining control structure words, e.g.,
1398 :
1399 :			@example
1400 :			: while ( dest -- orig dest )
1401 :			POSTPONE if
1402 :			1 cs-roll ; immediate
1403 :
1404 :			: repeat ( orig dest -- )
1405 :			POSTPONE again
1406 :			POSTPONE then ; immediate
1407 :			@end example
1408 :
1409 :			and then using these to create the control structure:
1410 :
1411 :			@example
1412 :			begin
1413 :			...
1414 :			while
1415 :			...
1416 :			repeat
1417 :			@end example
1418 :
1419 :	anton	1.30	That's much easier to read, isn't it? Of course, @code{REPEAT} and
1420 :	anton	1.3	@code{WHILE} are predefined, so in this example it would not be
1421 :			necessary to define them.
1422 :
1423 :	anton	1.4	@node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1424 :	anton	1.3	@subsection Calls and returns
1425 :
1426 :			A definition can be called simply be writing the name of the
1427 :	anton	1.17	definition. When the end of the definition is reached, it returns. An
1428 :			earlier return can be forced using
1429 :	anton	1.3
1430 :			doc-exit
1431 :
1432 :			Don't forget to clean up the return stack and @code{UNLOOP} any
1433 :			outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1434 :			primitive compiled by @code{EXIT} is
1435 :
1436 :			doc-;s
1437 :
1438 :	anton	1.4	@node Exception Handling, , Calls and returns, Control Structures
1439 :	anton	1.3	@subsection Exception Handling
1440 :
1441 :			doc-catch
1442 :			doc-throw
1443 :
1444 :	anton	1.4	@node Locals, Defining Words, Control Structures, Words
1445 :	anton	1.1	@section Locals
1446 :
1447 :	anton	1.2	Local variables can make Forth programming more enjoyable and Forth
1448 :			programs easier to read. Unfortunately, the locals of ANS Forth are
1449 :			laden with restrictions. Therefore, we provide not only the ANS Forth
1450 :			locals wordset, but also our own, more powerful locals wordset (we
1451 :			implemented the ANS Forth locals wordset through our locals wordset).
1452 :
1453 :	anton	1.24	The ideas in this section have also been published in the paper
1454 :			@cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1455 :			at EuroForth '94; it is available at
1456 :			@*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1457 :
1458 :	anton	1.2	@menu
1459 :	anton	1.17	* Gforth locals::
1460 :	anton	1.4	* ANS Forth locals::
1461 :	anton	1.2	@end menu
1462 :
1463 :	anton	1.17	@node Gforth locals, ANS Forth locals, Locals, Locals
1464 :			@subsection Gforth locals
1465 :	anton	1.2
1466 :			Locals can be defined with
1467 :
1468 :			@example
1469 :			@{ local1 local2 ... -- comment @}
1470 :			@end example
1471 :			or
1472 :			@example
1473 :			@{ local1 local2 ... @}
1474 :			@end example
1475 :
1476 :			E.g.,
1477 :			@example
1478 :			: max @{ n1 n2 -- n3 @}
1479 :			n1 n2 > if
1480 :			n1
1481 :			else
1482 :			n2
1483 :			endif ;
1484 :			@end example
1485 :
1486 :			The similarity of locals definitions with stack comments is intended. A
1487 :			locals definition often replaces the stack comment of a word. The order
1488 :			of the locals corresponds to the order in a stack comment and everything
1489 :			after the @code{--} is really a comment.
1490 :
1491 :			This similarity has one disadvantage: It is too easy to confuse locals
1492 :			declarations with stack comments, causing bugs and making them hard to
1493 :			find. However, this problem can be avoided by appropriate coding
1494 :			conventions: Do not use both notations in the same program. If you do,
1495 :			they should be distinguished using additional means, e.g. by position.
1496 :
1497 :			The name of the local may be preceded by a type specifier, e.g.,
1498 :			@code{F:} for a floating point value:
1499 :
1500 :			@example
1501 :			: CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1502 :			\ complex multiplication
1503 :			Ar Br f* Ai Bi f* f-
1504 :			Ar Bi f* Ai Br f* f+ ;
1505 :			@end example
1506 :
1507 :	anton	1.17	Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1508 :	anton	1.2	(@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1509 :			(@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1510 :			with @code{W:}, @code{D:} etc.) produces its value and can be changed
1511 :			with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1512 :			produces its address (which becomes invalid when the variable's scope is
1513 :			left). E.g., the standard word @code{emit} can be defined in therms of
1514 :			@code{type} like this:
1515 :
1516 :			@example
1517 :			: emit @{ C^ char* -- @}
1518 :			char* 1 type ;
1519 :			@end example
1520 :
1521 :			A local without type specifier is a @code{W:} local. Both flavours of
1522 :			locals are initialized with values from the data or FP stack.
1523 :
1524 :			Currently there is no way to define locals with user-defined data
1525 :			structures, but we are working on it.
1526 :
1527 :	anton	1.17	Gforth allows defining locals everywhere in a colon definition. This
1528 :	pazsan	1.7	poses the following questions:
1529 :	anton	1.2
1530 :	anton	1.4	@menu
1531 :			* Where are locals visible by name?::
1532 :	anton	1.14	* How long do locals live?::
1533 :	anton	1.4	* Programming Style::
1534 :			* Implementation::
1535 :			@end menu
1536 :
1537 :	anton	1.17	@node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1538 :	anton	1.2	@subsubsection Where are locals visible by name?
1539 :
1540 :			Basically, the answer is that locals are visible where you would expect
1541 :			it in block-structured languages, and sometimes a little longer. If you
1542 :			want to restrict the scope of a local, enclose its definition in
1543 :			@code{SCOPE}...@code{ENDSCOPE}.
1544 :
1545 :			doc-scope
1546 :			doc-endscope
1547 :
1548 :			These words behave like control structure words, so you can use them
1549 :			with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1550 :			arbitrary ways.
1551 :
1552 :			If you want a more exact answer to the visibility question, here's the
1553 :			basic principle: A local is visible in all places that can only be
1554 :			reached through the definition of the local@footnote{In compiler
1555 :			construction terminology, all places dominated by the definition of the
1556 :			local.}. In other words, it is not visible in places that can be reached
1557 :			without going through the definition of the local. E.g., locals defined
1558 :			in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1559 :			defined in @code{BEGIN}...@code{UNTIL} are visible after the
1560 :			@code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1561 :
1562 :			The reasoning behind this solution is: We want to have the locals
1563 :			visible as long as it is meaningful. The user can always make the
1564 :			visibility shorter by using explicit scoping. In a place that can
1565 :			only be reached through the definition of a local, the meaning of a
1566 :			local name is clear. In other places it is not: How is the local
1567 :			initialized at the control flow path that does not contain the
1568 :			definition? Which local is meant, if the same name is defined twice in
1569 :			two independent control flow paths?
1570 :
1571 :			This should be enough detail for nearly all users, so you can skip the
1572 :			rest of this section. If you relly must know all the gory details and
1573 :			options, read on.
1574 :
1575 :			In order to implement this rule, the compiler has to know which places
1576 :			are unreachable. It knows this automatically after @code{AHEAD},
1577 :			@code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1578 :			most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1579 :			compiler that the control flow never reaches that place. If
1580 :			@code{UNREACHABLE} is not used where it could, the only consequence is
1581 :			that the visibility of some locals is more limited than the rule above
1582 :			says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1583 :			lie to the compiler), buggy code will be produced.
1584 :
1585 :			Another problem with this rule is that at @code{BEGIN}, the compiler
1586 :	anton	1.3	does not know which locals will be visible on the incoming
1587 :			back-edge. All problems discussed in the following are due to this
1588 :			ignorance of the compiler (we discuss the problems using @code{BEGIN}
1589 :			loops as examples; the discussion also applies to @code{?DO} and other
1590 :	anton	1.2	loops). Perhaps the most insidious example is:
1591 :			@example
1592 :			AHEAD
1593 :			BEGIN
1594 :			x
1595 :			[ 1 CS-ROLL ] THEN
1596 :	anton	1.4	@{ x @}
1597 :	anton	1.2	...
1598 :			UNTIL
1599 :			@end example
1600 :
1601 :			This should be legal according to the visibility rule. The use of
1602 :			@code{x} can only be reached through the definition; but that appears
1603 :			textually below the use.
1604 :
1605 :			From this example it is clear that the visibility rules cannot be fully
1606 :			implemented without major headaches. Our implementation treats common
1607 :			cases as advertised and the exceptions are treated in a safe way: The
1608 :			compiler makes a reasonable guess about the locals visible after a
1609 :			@code{BEGIN}; if it is too pessimistic, the
1610 :			user will get a spurious error about the local not being defined; if the
1611 :			compiler is too optimistic, it will notice this later and issue a
1612 :			warning. In the case above the compiler would complain about @code{x}
1613 :			being undefined at its use. You can see from the obscure examples in
1614 :			this section that it takes quite unusual control structures to get the
1615 :			compiler into trouble, and even then it will often do fine.
1616 :
1617 :			If the @code{BEGIN} is reachable from above, the most optimistic guess
1618 :			is that all locals visible before the @code{BEGIN} will also be
1619 :			visible after the @code{BEGIN}. This guess is valid for all loops that
1620 :			are entered only through the @code{BEGIN}, in particular, for normal
1621 :			@code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1622 :			@code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1623 :			compiler. When the branch to the @code{BEGIN} is finally generated by
1624 :			@code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1625 :			warns the user if it was too optimisitic:
1626 :			@example
1627 :			IF
1628 :	anton	1.4	@{ x @}
1629 :	anton	1.2	BEGIN
1630 :			\ x ?
1631 :			[ 1 cs-roll ] THEN
1632 :			...
1633 :			UNTIL
1634 :			@end example
1635 :
1636 :			Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1637 :			optimistically assumes that it lives until the @code{THEN}. It notices
1638 :			this difference when it compiles the @code{UNTIL} and issues a
1639 :			warning. The user can avoid the warning, and make sure that @code{x}
1640 :			is not used in the wrong area by using explicit scoping:
1641 :			@example
1642 :			IF
1643 :			SCOPE
1644 :	anton	1.4	@{ x @}
1645 :	anton	1.2	ENDSCOPE
1646 :			BEGIN
1647 :			[ 1 cs-roll ] THEN
1648 :			...
1649 :			UNTIL
1650 :			@end example
1651 :
1652 :			Since the guess is optimistic, there will be no spurious error messages
1653 :			about undefined locals.
1654 :
1655 :			If the @code{BEGIN} is not reachable from above (e.g., after
1656 :			@code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1657 :			optimistic guess, as the locals visible after the @code{BEGIN} may be
1658 :			defined later. Therefore, the compiler assumes that no locals are
1659 :	anton	1.17	visible after the @code{BEGIN}. However, the user can use
1660 :	anton	1.2	@code{ASSUME-LIVE} to make the compiler assume that the same locals are
1661 :	anton	1.17	visible at the BEGIN as at the point where the top control-flow stack
1662 :			item was created.
1663 :	anton	1.2
1664 :			doc-assume-live
1665 :
1666 :			E.g.,
1667 :			@example
1668 :	anton	1.4	@{ x @}
1669 :	anton	1.2	AHEAD
1670 :			ASSUME-LIVE
1671 :			BEGIN
1672 :			x
1673 :			[ 1 CS-ROLL ] THEN
1674 :			...
1675 :			UNTIL
1676 :			@end example
1677 :
1678 :			Other cases where the locals are defined before the @code{BEGIN} can be
1679 :			handled by inserting an appropriate @code{CS-ROLL} before the
1680 :			@code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1681 :			behind the @code{ASSUME-LIVE}).
1682 :
1683 :			Cases where locals are defined after the @code{BEGIN} (but should be
1684 :			visible immediately after the @code{BEGIN}) can only be handled by
1685 :			rearranging the loop. E.g., the ``most insidious'' example above can be
1686 :			arranged into:
1687 :			@example
1688 :			BEGIN
1689 :	anton	1.4	@{ x @}
1690 :	anton	1.2	... 0=
1691 :			WHILE
1692 :			x
1693 :			REPEAT
1694 :			@end example
1695 :
1696 :	anton	1.17	@node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1697 :	anton	1.2	@subsubsection How long do locals live?
1698 :
1699 :			The right answer for the lifetime question would be: A local lives at
1700 :			least as long as it can be accessed. For a value-flavoured local this
1701 :			means: until the end of its visibility. However, a variable-flavoured
1702 :			local could be accessed through its address far beyond its visibility
1703 :			scope. Ultimately, this would mean that such locals would have to be
1704 :			garbage collected. Since this entails un-Forth-like implementation
1705 :			complexities, I adopted the same cowardly solution as some other
1706 :			languages (e.g., C): The local lives only as long as it is visible;
1707 :			afterwards its address is invalid (and programs that access it
1708 :			afterwards are erroneous).
1709 :
1710 :	anton	1.17	@node Programming Style, Implementation, How long do locals live?, Gforth locals
1711 :	anton	1.2	@subsubsection Programming Style
1712 :
1713 :			The freedom to define locals anywhere has the potential to change
1714 :			programming styles dramatically. In particular, the need to use the
1715 :			return stack for intermediate storage vanishes. Moreover, all stack
1716 :			manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1717 :			determined arguments) can be eliminated: If the stack items are in the
1718 :			wrong order, just write a locals definition for all of them; then
1719 :			write the items in the order you want.
1720 :
1721 :			This seems a little far-fetched and eliminating stack manipulations is
1722 :	anton	1.4	unlikely to become a conscious programming objective. Still, the number
1723 :			of stack manipulations will be reduced dramatically if local variables
1724 :	anton	1.17	are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1725 :	anton	1.4	a traditional implementation of @code{max}).
1726 :	anton	1.2
1727 :			This shows one potential benefit of locals: making Forth programs more
1728 :			readable. Of course, this benefit will only be realized if the
1729 :			programmers continue to honour the principle of factoring instead of
1730 :			using the added latitude to make the words longer.
1731 :
1732 :			Using @code{TO} can and should be avoided. Without @code{TO},
1733 :			every value-flavoured local has only a single assignment and many
1734 :			advantages of functional languages apply to Forth. I.e., programs are
1735 :			easier to analyse, to optimize and to read: It is clear from the
1736 :			definition what the local stands for, it does not turn into something
1737 :			different later.
1738 :
1739 :			E.g., a definition using @code{TO} might look like this:
1740 :			@example
1741 :			: strcmp @{ addr1 u1 addr2 u2 -- n @}
1742 :			u1 u2 min 0
1743 :			?do
1744 :	anton	1.36	addr1 c@@ addr2 c@@ -
1745 :	anton	1.31	?dup-if
1746 :	anton	1.2	unloop exit
1747 :			then
1748 :			addr1 char+ TO addr1
1749 :			addr2 char+ TO addr2
1750 :			loop
1751 :			u1 u2 - ;
1752 :			@end example
1753 :			Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1754 :			every loop iteration. @code{strcmp} is a typical example of the
1755 :			readability problems of using @code{TO}. When you start reading
1756 :			@code{strcmp}, you think that @code{addr1} refers to the start of the
1757 :			string. Only near the end of the loop you realize that it is something
1758 :			else.
1759 :
1760 :			This can be avoided by defining two locals at the start of the loop that
1761 :			are initialized with the right value for the current iteration.
1762 :			@example
1763 :			: strcmp @{ addr1 u1 addr2 u2 -- n @}
1764 :			addr1 addr2
1765 :			u1 u2 min 0
1766 :			?do @{ s1 s2 @}
1767 :	anton	1.36	s1 c@@ s2 c@@ -
1768 :	anton	1.31	?dup-if
1769 :	anton	1.2	unloop exit
1770 :			then
1771 :			s1 char+ s2 char+
1772 :			loop
1773 :			2drop
1774 :			u1 u2 - ;
1775 :			@end example
1776 :			Here it is clear from the start that @code{s1} has a different value
1777 :			in every loop iteration.
1778 :
1779 :	anton	1.17	@node Implementation, , Programming Style, Gforth locals
1780 :	anton	1.2	@subsubsection Implementation
1781 :
1782 :	anton	1.17	Gforth uses an extra locals stack. The most compelling reason for
1783 :	anton	1.2	this is that the return stack is not float-aligned; using an extra stack
1784 :			also eliminates the problems and restrictions of using the return stack
1785 :			as locals stack. Like the other stacks, the locals stack grows toward
1786 :			lower addresses. A few primitives allow an efficient implementation:
1787 :
1788 :			doc-@local#
1789 :			doc-f@local#
1790 :			doc-laddr#
1791 :			doc-lp+!#
1792 :			doc-lp!
1793 :			doc->l
1794 :			doc-f>l
1795 :
1796 :			In addition to these primitives, some specializations of these
1797 :			primitives for commonly occurring inline arguments are provided for
1798 :			efficiency reasons, e.g., @code{@@local0} as specialization of
1799 :			@code{@@local#} for the inline argument 0. The following compiling words
1800 :			compile the right specialized version, or the general version, as
1801 :			appropriate:
1802 :
1803 :	anton	1.12	doc-compile-@local
1804 :			doc-compile-f@local
1805 :	anton	1.2	doc-compile-lp+!
1806 :
1807 :			Combinations of conditional branches and @code{lp+!#} like
1808 :			@code{?branch-lp+!#} (the locals pointer is only changed if the branch
1809 :			is taken) are provided for efficiency and correctness in loops.
1810 :
1811 :			A special area in the dictionary space is reserved for keeping the
1812 :			local variable names. @code{@{} switches the dictionary pointer to this
1813 :			area and @code{@}} switches it back and generates the locals
1814 :			initializing code. @code{W:} etc.@ are normal defining words. This
1815 :			special area is cleared at the start of every colon definition.
1816 :
1817 :	anton	1.17	A special feature of Gforth's dictionary is used to implement the
1818 :	anton	1.2	definition of locals without type specifiers: every wordlist (aka
1819 :			vocabulary) has its own methods for searching
1820 :	anton	1.4	etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1821 :	anton	1.2	with a special search method: When it is searched for a word, it
1822 :			actually creates that word using @code{W:}. @code{@{} changes the search
1823 :			order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1824 :			and then the wordlist for defining locals without type specifiers.
1825 :
1826 :			The lifetime rules support a stack discipline within a colon
1827 :			definition: The lifetime of a local is either nested with other locals
1828 :			lifetimes or it does not overlap them.
1829 :
1830 :			At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1831 :			pointer manipulation is generated. Between control structure words
1832 :			locals definitions can push locals onto the locals stack. @code{AGAIN}
1833 :			is the simplest of the other three control flow words. It has to
1834 :			restore the locals stack depth of the corresponding @code{BEGIN}
1835 :			before branching. The code looks like this:
1836 :			@format
1837 :			@code{lp+!#} current-locals-size @minus{} dest-locals-size
1838 :			@code{branch} <begin>
1839 :			@end format
1840 :
1841 :			@code{UNTIL} is a little more complicated: If it branches back, it
1842 :			must adjust the stack just like @code{AGAIN}. But if it falls through,
1843 :			the locals stack must not be changed. The compiler generates the
1844 :			following code:
1845 :			@format
1846 :			@code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1847 :			@end format
1848 :			The locals stack pointer is only adjusted if the branch is taken.
1849 :
1850 :			@code{THEN} can produce somewhat inefficient code:
1851 :			@format
1852 :			@code{lp+!#} current-locals-size @minus{} orig-locals-size
1853 :			<orig target>:
1854 :			@code{lp+!#} orig-locals-size @minus{} new-locals-size
1855 :			@end format
1856 :			The second @code{lp+!#} adjusts the locals stack pointer from the
1857 :	anton	1.4	level at the @var{orig} point to the level after the @code{THEN}. The
1858 :	anton	1.2	first @code{lp+!#} adjusts the locals stack pointer from the current
1859 :			level to the level at the orig point, so the complete effect is an
1860 :			adjustment from the current level to the right level after the
1861 :			@code{THEN}.
1862 :
1863 :			In a conventional Forth implementation a dest control-flow stack entry
1864 :			is just the target address and an orig entry is just the address to be
1865 :			patched. Our locals implementation adds a wordlist to every orig or dest
1866 :			item. It is the list of locals visible (or assumed visible) at the point
1867 :			described by the entry. Our implementation also adds a tag to identify
1868 :			the kind of entry, in particular to differentiate between live and dead
1869 :			(reachable and unreachable) orig entries.
1870 :
1871 :			A few unusual operations have to be performed on locals wordlists:
1872 :
1873 :			doc-common-list
1874 :			doc-sub-list?
1875 :			doc-list-size
1876 :
1877 :			Several features of our locals wordlist implementation make these
1878 :			operations easy to implement: The locals wordlists are organised as
1879 :			linked lists; the tails of these lists are shared, if the lists
1880 :			contain some of the same locals; and the address of a name is greater
1881 :			than the address of the names behind it in the list.
1882 :
1883 :			Another important implementation detail is the variable
1884 :			@code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1885 :			determine if they can be reached directly or only through the branch
1886 :			that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1887 :			@code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1888 :			definition, by @code{BEGIN} and usually by @code{THEN}.
1889 :
1890 :			Counted loops are similar to other loops in most respects, but
1891 :			@code{LEAVE} requires special attention: It performs basically the same
1892 :			service as @code{AHEAD}, but it does not create a control-flow stack
1893 :			entry. Therefore the information has to be stored elsewhere;
1894 :			traditionally, the information was stored in the target fields of the
1895 :			branches created by the @code{LEAVE}s, by organizing these fields into a
1896 :			linked list. Unfortunately, this clever trick does not provide enough
1897 :			space for storing our extended control flow information. Therefore, we
1898 :			introduce another stack, the leave stack. It contains the control-flow
1899 :			stack entries for all unresolved @code{LEAVE}s.
1900 :
1901 :			Local names are kept until the end of the colon definition, even if
1902 :			they are no longer visible in any control-flow path. In a few cases
1903 :			this may lead to increased space needs for the locals name area, but
1904 :			usually less than reclaiming this space would cost in code size.
1905 :
1906 :
1907 :	anton	1.17	@node ANS Forth locals, , Gforth locals, Locals
1908 :	anton	1.2	@subsection ANS Forth locals
1909 :
1910 :			The ANS Forth locals wordset does not define a syntax for locals, but
1911 :			words that make it possible to define various syntaxes. One of the
1912 :	anton	1.17	possible syntaxes is a subset of the syntax we used in the Gforth locals
1913 :	anton	1.2	wordset, i.e.:
1914 :
1915 :			@example
1916 :			@{ local1 local2 ... -- comment @}
1917 :			@end example
1918 :			or
1919 :			@example
1920 :			@{ local1 local2 ... @}
1921 :			@end example
1922 :
1923 :			The order of the locals corresponds to the order in a stack comment. The
1924 :			restrictions are:
1925 :	anton	1.1
1926 :	anton	1.2	@itemize @bullet
1927 :			@item
1928 :	anton	1.17	Locals can only be cell-sized values (no type specifiers are allowed).
1929 :	anton	1.2	@item
1930 :			Locals can be defined only outside control structures.
1931 :			@item
1932 :			Locals can interfere with explicit usage of the return stack. For the
1933 :			exact (and long) rules, see the standard. If you don't use return stack
1934 :	anton	1.17	accessing words in a definition using locals, you will be all right. The
1935 :	anton	1.2	purpose of this rule is to make locals implementation on the return
1936 :			stack easier.
1937 :			@item
1938 :			The whole definition must be in one line.
1939 :			@end itemize
1940 :
1941 :	anton	1.35	Locals defined in this way behave like @code{VALUE}s (@xref{Simple
1942 :			Defining Words}). I.e., they are initialized from the stack. Using their
1943 :	anton	1.2	name produces their value. Their value can be changed using @code{TO}.
1944 :
1945 :	anton	1.17	Since this syntax is supported by Gforth directly, you need not do
1946 :	anton	1.2	anything to use it. If you want to port a program using this syntax to
1947 :	anton	1.30	another ANS Forth system, use @file{compat/anslocal.fs} to implement the
1948 :			syntax on the other system.
1949 :	anton	1.2
1950 :			Note that a syntax shown in the standard, section A.13 looks
1951 :			similar, but is quite different in having the order of locals
1952 :			reversed. Beware!
1953 :
1954 :			The ANS Forth locals wordset itself consists of the following word
1955 :
1956 :			doc-(local)
1957 :
1958 :			The ANS Forth locals extension wordset defines a syntax, but it is so
1959 :			awful that we strongly recommend not to use it. We have implemented this
1960 :	anton	1.17	syntax to make porting to Gforth easy, but do not document it here. The
1961 :	anton	1.2	problem with this syntax is that the locals are defined in an order
1962 :			reversed with respect to the standard stack comment notation, making
1963 :			programs harder to read, and easier to misread and miswrite. The only
1964 :			merit of this syntax is that it is easy to implement using the ANS Forth
1965 :			locals wordset.
1966 :	anton	1.3
1967 :	anton	1.37	@node Defining Words, Tokens for Words, Locals, Words
1968 :	anton	1.4	@section Defining Words
1969 :
1970 :	anton	1.14	@menu
1971 :	anton	1.35	* Simple Defining Words::
1972 :			* Colon Definitions::
1973 :			* User-defined Defining Words::
1974 :			* Supplying names::
1975 :			* Interpretation and Compilation Semantics::
1976 :	anton	1.14	@end menu
1977 :
1978 :	anton	1.35	@node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
1979 :			@subsection Simple Defining Words
1980 :
1981 :			doc-constant
1982 :			doc-2constant
1983 :			doc-fconstant
1984 :			doc-variable
1985 :			doc-2variable
1986 :			doc-fvariable
1987 :			doc-create
1988 :			doc-user
1989 :			doc-value
1990 :			doc-to
1991 :			doc-defer
1992 :			doc-is
1993 :
1994 :			@node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
1995 :			@subsection Colon Definitions
1996 :
1997 :			@example
1998 :			: name ( ... -- ... )
1999 :			word1 word2 word3 ;
2000 :			@end example
2001 :
2002 :			creates a word called @code{name}, that, upon execution, executes
2003 :			@code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2004 :
2005 :			The explanation above is somewhat superficial. @xref{Interpretation and
2006 :			Compilation Semantics} for an in-depth discussion of some of the issues
2007 :			involved.
2008 :
2009 :			doc-:
2010 :			doc-;
2011 :
2012 :			@node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2013 :			@subsection User-defined Defining Words
2014 :
2015 :			You can create new defining words simply by wrapping defining-time code
2016 :			around existing defining words and putting the sequence in a colon
2017 :			definition.
2018 :
2019 :	anton	1.36	If you want the words defined with your defining words to behave
2020 :			differently from words defined with standard defining words, you can
2021 :	anton	1.35	write your defining word like this:
2022 :
2023 :			@example
2024 :			: def-word ( "name" -- )
2025 :			Create @var{code1}
2026 :			DOES> ( ... -- ... )
2027 :			@var{code2} ;
2028 :
2029 :			def-word name
2030 :			@end example
2031 :
2032 :			Technically, this fragment defines a defining word @code{def-word}, and
2033 :			a word @code{name}; when you execute @code{name}, the address of the
2034 :			body of @code{name} is put on the data stack and @var{code2} is executed
2035 :			(the address of the body of @code{name} is the address @code{HERE}
2036 :	anton	1.36	returns immediately after the @code{CREATE}).
2037 :
2038 :			In other words, if you make the following definitions:
2039 :
2040 :			@example
2041 :			: def-word1 ( "name" -- )
2042 :			Create @var{code1} ;
2043 :
2044 :			: action1 ( ... -- ... )
2045 :			@var{code2} ;
2046 :
2047 :			def-word name1
2048 :			@end example
2049 :
2050 :			Using @code{name1 action1} is equivalent to using @code{name}.
2051 :
2052 :			E.g., you can implement @code{Constant} in this way:
2053 :	anton	1.35
2054 :			@example
2055 :			: constant ( w "name" -- )
2056 :			create ,
2057 :			DOES> ( -- w )
2058 :	anton	1.36	@@ ;
2059 :	anton	1.35	@end example
2060 :
2061 :			When you create a constant with @code{5 constant five}, first a new word
2062 :			@code{five} is created, then the value 5 is laid down in the body of
2063 :			@code{five} with @code{,}. When @code{five} is invoked, the address of
2064 :			the body is put on the stack, and @code{@@} retrieves the value 5.
2065 :
2066 :			In the example above the stack comment after the @code{DOES>} specifies
2067 :			the stack effect of the defined words, not the stack effect of the
2068 :			following code (the following code expects the address of the body on
2069 :			the top of stack, which is not reflected in the stack comment). This is
2070 :			the convention that I use and recommend (it clashes a bit with using
2071 :			locals declarations for stack effect specification, though).
2072 :
2073 :			@subsubsection Applications of @code{CREATE..DOES>}
2074 :
2075 :	anton	1.36	You may wonder how to use this feature. Here are some usage patterns:
2076 :	anton	1.35
2077 :			When you see a sequence of code occurring several times, and you can
2078 :			identify a meaning, you will factor it out as a colon definition. When
2079 :			you see similar colon definitions, you can factor them using
2080 :			@code{CREATE..DOES>}. E.g., an assembler usually defines several words
2081 :			that look very similar:
2082 :			@example
2083 :			: ori, ( reg-taget reg-source n -- )
2084 :			0 asm-reg-reg-imm ;
2085 :			: andi, ( reg-taget reg-source n -- )
2086 :			1 asm-reg-reg-imm ;
2087 :			@end example
2088 :
2089 :			This could be factored with:
2090 :			@example
2091 :			: reg-reg-imm ( op-code -- )
2092 :			create ,
2093 :			DOES> ( reg-taget reg-source n -- )
2094 :	anton	1.36	@@ asm-reg-reg-imm ;
2095 :	anton	1.35
2096 :			0 reg-reg-imm ori,
2097 :			1 reg-reg-imm andi,
2098 :			@end example
2099 :
2100 :			Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2101 :			supply a part of the parameters for a word (known as @dfn{currying} in
2102 :			the functional language community). E.g., @code{+} needs two
2103 :			parameters. Creating versions of @code{+} with one parameter fixed can
2104 :			be done like this:
2105 :			@example
2106 :			: curry+ ( n1 -- )
2107 :			create ,
2108 :			DOES> ( n2 -- n1+n2 )
2109 :	anton	1.36	@@ + ;
2110 :	anton	1.35
2111 :			3 curry+ 3+
2112 :			-2 curry+ 2-
2113 :			@end example
2114 :
2115 :			@subsubsection The gory details of @code{CREATE..DOES>}
2116 :
2117 :			doc-does>
2118 :
2119 :			This means that you need not use @code{CREATE} and @code{DOES>} in the
2120 :			same definition; E.g., you can put the @code{DOES>}-part in a separate
2121 :			definition. This allows us to, e.g., select among different DOES>-parts:
2122 :			@example
2123 :			: does1
2124 :			DOES> ( ... -- ... )
2125 :			... ;
2126 :
2127 :			: does2
2128 :			DOES> ( ... -- ... )
2129 :			... ;
2130 :
2131 :			: def-word ( ... -- ... )
2132 :			create ...
2133 :			IF
2134 :			does1
2135 :			ELSE
2136 :			does2
2137 :			ENDIF ;
2138 :			@end example
2139 :
2140 :			In a standard program you can apply a @code{DOES>}-part only if the last
2141 :			word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2142 :			will override the behaviour of the last word defined in any case. In a
2143 :			standard program, you can use @code{DOES>} only in a colon
2144 :			definition. In Gforth, you can also use it in interpretation state, in a
2145 :			kind of one-shot mode:
2146 :			@example
2147 :			CREATE name ( ... -- ... )
2148 :			@var{initialization}
2149 :			DOES>
2150 :			@var{code} ;
2151 :			@end example
2152 :			This is equivalwent to the standard
2153 :			@example
2154 :			:noname
2155 :			DOES>
2156 :			@var{code} ;
2157 :			CREATE name EXECUTE ( ... -- ... )
2158 :			@var{initialization}
2159 :			@end example
2160 :
2161 :			You can get the address of the body of a word with
2162 :
2163 :			doc->body
2164 :
2165 :			@node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2166 :			@subsection Supplying names for the defined words
2167 :
2168 :			By default, defining words take the names for the defined words from the
2169 :			input stream. Sometimes you want to supply the name from a string. You
2170 :			can do this with
2171 :
2172 :			doc-nextname
2173 :
2174 :			E.g.,
2175 :
2176 :			@example
2177 :			s" foo" nextname create
2178 :			@end example
2179 :			is equivalent to
2180 :			@example
2181 :			create foo
2182 :			@end example
2183 :
2184 :			Sometimes you want to define a word without a name. You can do this with
2185 :
2186 :			doc-noname
2187 :
2188 :			To make any use of the newly defined word, you need its execution
2189 :			token. You can get it with
2190 :
2191 :			doc-lastxt
2192 :
2193 :			E.g., you can initialize a deferred word with an anonymous colon
2194 :			definition:
2195 :			@example
2196 :			Defer deferred
2197 :			noname : ( ... -- ... )
2198 :			... ;
2199 :			lastxt IS deferred
2200 :			@end example
2201 :
2202 :			@code{lastxt} also works when the last word was not defined as
2203 :			@code{noname}.
2204 :
2205 :			The standard has also recognized the need for anonymous words and
2206 :			provides
2207 :
2208 :			doc-:noname
2209 :
2210 :			This leaves the execution token for the word on the stack after the
2211 :			closing @code{;}. You can rewrite the last example with @code{:noname}:
2212 :			@example
2213 :			Defer deferred
2214 :			:noname ( ... -- ... )
2215 :			... ;
2216 :			IS deferred
2217 :			@end example
2218 :
2219 :			@node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2220 :			@subsection Interpretation and Compilation Semantics
2221 :
2222 :	anton	1.36	The @dfn{interpretation semantics} of a word are what the text
2223 :			interpreter does when it encounters the word in interpret state. It also
2224 :			appears in some other contexts, e.g., the execution token returned by
2225 :			@code{' @var{word}} identifies the interpretation semantics of
2226 :			@var{word} (in other words, @code{' @var{word} execute} is equivalent to
2227 :			interpret-state text interpretation of @code{@var{word}}).
2228 :
2229 :			The @dfn{compilation semantics} of a word are what the text interpreter
2230 :			does when it encounters the word in compile state. It also appears in
2231 :			other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2232 :			standard terminology, ``appends to the current definition''.} the
2233 :			compilation semantics of @var{word}.
2234 :
2235 :			The standard also talks about @dfn{execution semantics}. They are used
2236 :			only for defining the interpretation and compilation semantics of many
2237 :			words. By default, the interpretation semantics of a word are to
2238 :			@code{execute} its execution semantics, and the compilation semantics of
2239 :			a word are to @code{compile,} its execution semantics.@footnote{In
2240 :			standard terminology: The default interpretation semantics are its
2241 :			execution semantics; the default compilation semantics are to append its
2242 :			execution semantics to the execution semantics of the current
2243 :			definition.}
2244 :
2245 :			You can change the compilation semantics into @code{execute}ing the
2246 :			execution semantics with
2247 :
2248 :	anton	1.35	doc-immediate
2249 :	anton	1.36
2250 :			You can remove the interpretation semantics of a word with
2251 :
2252 :			doc-compile-only
2253 :			doc-restrict
2254 :
2255 :			Note that ticking (@code{'}) compile-only words gives an error
2256 :			(``Interpreting a compile-only word'').
2257 :
2258 :			Gforth also allows you to define words with arbitrary combinations of
2259 :			interpretation and compilation semantics.
2260 :
2261 :	anton	1.35	doc-interpret/compile:
2262 :
2263 :	anton	1.36	This feature was introduced for implementing @code{TO} and @code{S"}. I
2264 :			recommend that you do not define such words, as cute as they may be:
2265 :			they make it hard to get at both parts of the word in some contexts.
2266 :			E.g., assume you want to get an execution token for the compilation
2267 :			part. Instead, define two words, one that embodies the interpretation
2268 :			part, and one that embodies the compilation part.
2269 :
2270 :			There is, however, a potentially useful application of this feature:
2271 :			Providing differing implementations for the default semantics. While
2272 :			this introduces redundancy and is therefore usually a bad idea, a
2273 :			performance improvement may be worth the trouble. E.g., consider the
2274 :			word @code{foobar}:
2275 :
2276 :			@example
2277 :			: foobar
2278 :			foo bar ;
2279 :			@end example
2280 :
2281 :			Let us assume that @code{foobar} is called so frequently that the
2282 :			calling overhead would take a significant amount of the run-time. We can
2283 :			optimize it with @code{interpret/compile:}:
2284 :	anton	1.35
2285 :	anton	1.36	@example
2286 :			:noname
2287 :			foo bar ;
2288 :			:noname
2289 :			POSTPONE foo POSTPONE bar ;
2290 :			interpret/compile: foobar
2291 :			@end example
2292 :
2293 :			This definition has the same interpretation semantics and essentially
2294 :			the same compilation semantics as the simple definition of
2295 :			@code{foobar}, but the implementation of the compilation semantics is
2296 :			more efficient with respect to run-time.
2297 :
2298 :			Some people try to use state-smart words to emulate the feature provided
2299 :			by @code{interpret/compile:} (words are state-smart if they check
2300 :			@code{STATE} during execution). E.g., they would try to code
2301 :			@code{foobar} like this:
2302 :
2303 :			@example
2304 :			: foobar
2305 :			STATE @@
2306 :			IF ( compilation state )
2307 :			POSTPONE foo POSTPONE bar
2308 :			ELSE
2309 :			foo bar
2310 :			ENDIF ; immediate
2311 :			@end example
2312 :
2313 :			While this works if @code{foobar} is processed only by the text
2314 :			interpreter, it does not work in other contexts (like @code{'} or
2315 :			@code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2316 :			for a state-smart word, not for the interpretation semantics of the
2317 :			original @code{foobar}; when you execute this execution token (directly
2318 :			with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2319 :			state, the result will not be what you expected (i.e., it will not
2320 :			perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2321 :			write them!
2322 :
2323 :			It is also possible to write defining words that define words with
2324 :			arbitrary combinations of interpretation and compilation semantics (or,
2325 :			preferably, arbitrary combinations of implementations of the default
2326 :			semantics). In general, this looks like:
2327 :
2328 :			@example
2329 :			: def-word
2330 :			create-interpret/compile
2331 :			@var{code1}
2332 :			interpretation>
2333 :			@var{code2}
2334 :			<interpretation
2335 :			compilation>
2336 :			@var{code3}
2337 :			<compilation ;
2338 :			@end example
2339 :
2340 :			For a @var{word} defined with @code{def-word}, the interpretation
2341 :			semantics are to push the address of the body of @var{word} and perform
2342 :			@var{code2}, and the compilation semantics are to push the address of
2343 :			the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2344 :			can also be defined like this:
2345 :
2346 :			@example
2347 :			: constant ( n "name" -- )
2348 :			create-interpret/compile
2349 :			,
2350 :			interpretation> ( -- n )
2351 :			@@
2352 :			<interpretation
2353 :			compilation> ( compilation. -- ; run-time. -- n )
2354 :			@@ postpone literal
2355 :			<compilation ;
2356 :			@end example
2357 :
2358 :			doc-create-interpret/compile
2359 :			doc-interpretation>
2360 :			doc-<interpretation
2361 :			doc-compilation>
2362 :			doc-<compilation
2363 :
2364 :			Note that words defined with @code{interpret/compile:} and
2365 :			@code{create-interpret/compile} have an extended header structure that
2366 :			differs from other words; however, unless you try to access them with
2367 :			plain address arithmetic, you should not notice this. Words for
2368 :			accessing the header structure usually know how to deal with this; e.g.,
2369 :			@code{' word >body} also gives you the body of a word created with
2370 :			@code{create-interpret/compile}.
2371 :	anton	1.4
2372 :	anton	1.37	@node Tokens for Words, Wordlists, Defining Words, Words
2373 :			@section Tokens for Words
2374 :
2375 :			This chapter describes the creation and use of tokens that represent
2376 :			words on the stack (and in data space).
2377 :
2378 :			Named words have interpretation and compilation semantics. Unnamed words
2379 :			just have execution semantics.
2380 :
2381 :			An @dfn{execution token} represents the execution semantics of an
2382 :			unnamed word. An execution token occupies one cell. As explained in
2383 :			section @ref{Supplying names}, the execution token of the last words
2384 :			defined can be produced with
2385 :
2386 :			short-lastxt
2387 :
2388 :			You can perform the semantics represented by an execution token with
2389 :			doc-execute
2390 :			You can compile the word with
2391 :			doc-compile,
2392 :
2393 :			In Gforth, the abstract data type @emph{execution token} is implemented
2394 :			as CFA (code field address).
2395 :
2396 :			The interpretation semantics of a named word are also represented by an
2397 :			execution token. You can get it with
2398 :
2399 :			doc-[']
2400 :			doc-'
2401 :
2402 :			For literals, you use @code{'} in interpreted code and @code{[']} in
2403 :			compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
2404 :			by complaining about compile-only words. To get an execution token for a
2405 :			compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
2406 :			@var{X} drop}.
2407 :
2408 :			The compilation semantics are represented by a @dfn{compilation token}
2409 :			consisting of two cells: @var{w xt}. The top cell @var{xt} is an
2410 :			execution token. The compilation semantics represented by the
2411 :			compilation token can be performed with @code{execute}, which consumes
2412 :			the whole compilation token, with an additional stack effect determined
2413 :			by the represented compilation semantics.
2414 :
2415 :			doc-[comp']
2416 :			doc-comp'
2417 :
2418 :	anton	1.38	You can compile the compilation semantics with @code{postpone,}. I.e.,
2419 :			@code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
2420 :			@var{word}}.
2421 :
2422 :			doc-postpone,
2423 :
2424 :	anton	1.37	At present, the @var{w} part of a compilation token is an execution
2425 :			token, and the @var{xt} part represents either @code{execute} or
2426 :			@code{compile,}. However, don't rely on that kowledge, unless necessary;
2427 :			we may introduce unusual compilation tokens in the future (e.g.,
2428 :			compilation tokens representing the compilation semantics of literals).
2429 :
2430 :			Named words are also represented by the @dfn{name token}. The abstract
2431 :			data type @emph{name token} is implemented as NFA (name field address).
2432 :
2433 :			doc-find-name
2434 :			doc-name>int
2435 :			doc-name?int
2436 :			doc-name>comp
2437 :			doc-name>string
2438 :
2439 :			@node Wordlists, Files, Tokens for Words, Words
2440 :	anton	1.4	@section Wordlists
2441 :
2442 :			@node Files, Blocks, Wordlists, Words
2443 :			@section Files
2444 :
2445 :			@node Blocks, Other I/O, Files, Words
2446 :			@section Blocks
2447 :
2448 :			@node Other I/O, Programming Tools, Blocks, Words
2449 :			@section Other I/O
2450 :
2451 :	anton	1.18	@node Programming Tools, Assembler and Code words, Other I/O, Words
2452 :	anton	1.4	@section Programming Tools
2453 :
2454 :	anton	1.5	@menu
2455 :			* Debugging:: Simple and quick.
2456 :			* Assertions:: Making your programs self-checking.
2457 :			@end menu
2458 :
2459 :			@node Debugging, Assertions, Programming Tools, Programming Tools
2460 :	anton	1.4	@subsection Debugging
2461 :
2462 :			The simple debugging aids provided in @file{debugging.fs}
2463 :			are meant to support a different style of debugging than the
2464 :			tracing/stepping debuggers used in languages with long turn-around
2465 :			times.
2466 :
2467 :			A much better (faster) way in fast-compilig languages is to add
2468 :			printing code at well-selected places, let the program run, look at
2469 :			the output, see where things went wrong, add more printing code, etc.,
2470 :			until the bug is found.
2471 :
2472 :			The word @code{~~} is easy to insert. It just prints debugging
2473 :			information (by default the source location and the stack contents). It
2474 :			is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2475 :			query-replace them with nothing). The deferred words
2476 :			@code{printdebugdata} and @code{printdebugline} control the output of
2477 :			@code{~~}. The default source location output format works well with
2478 :			Emacs' compilation mode, so you can step through the program at the
2479 :	anton	1.5	source level using @kbd{C-x `} (the advantage over a stepping debugger
2480 :			is that you can step in any direction and you know where the crash has
2481 :			happened or where the strange data has occurred).
2482 :	anton	1.4
2483 :			Note that the default actions clobber the contents of the pictured
2484 :			numeric output string, so you should not use @code{~~}, e.g., between
2485 :			@code{<#} and @code{#>}.
2486 :
2487 :			doc-~~
2488 :			doc-printdebugdata
2489 :			doc-printdebugline
2490 :
2491 :	anton	1.5	@node Assertions, , Debugging, Programming Tools
2492 :	anton	1.4	@subsection Assertions
2493 :
2494 :	anton	1.5	It is a good idea to make your programs self-checking, in particular, if
2495 :			you use an assumption (e.g., that a certain field of a data structure is
2496 :	anton	1.17	never zero) that may become wrong during maintenance. Gforth supports
2497 :	anton	1.5	assertions for this purpose. They are used like this:
2498 :
2499 :			@example
2500 :			assert( @var{flag} )
2501 :			@end example
2502 :
2503 :			The code between @code{assert(} and @code{)} should compute a flag, that
2504 :			should be true if everything is alright and false otherwise. It should
2505 :			not change anything else on the stack. The overall stack effect of the
2506 :			assertion is @code{( -- )}. E.g.
2507 :
2508 :			@example
2509 :			assert( 1 1 + 2 = ) \ what we learn in school
2510 :			assert( dup 0<> ) \ assert that the top of stack is not zero
2511 :			assert( false ) \ this code should not be reached
2512 :			@end example
2513 :
2514 :			The need for assertions is different at different times. During
2515 :			debugging, we want more checking, in production we sometimes care more
2516 :			for speed. Therefore, assertions can be turned off, i.e., the assertion
2517 :			becomes a comment. Depending on the importance of an assertion and the
2518 :			time it takes to check it, you may want to turn off some assertions and
2519 :	anton	1.17	keep others turned on. Gforth provides several levels of assertions for
2520 :	anton	1.5	this purpose:
2521 :
2522 :			doc-assert0(
2523 :			doc-assert1(
2524 :			doc-assert2(
2525 :			doc-assert3(
2526 :			doc-assert(
2527 :			doc-)
2528 :
2529 :			@code{Assert(} is the same as @code{assert1(}. The variable
2530 :			@code{assert-level} specifies the highest assertions that are turned
2531 :			on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2532 :			@code{assert1(} assertions perform checking, while @code{assert2(} and
2533 :			@code{assert3(} assertions are treated as comments.
2534 :
2535 :			Note that the @code{assert-level} is evaluated at compile-time, not at
2536 :			run-time. I.e., you cannot turn assertions on or off at run-time, you
2537 :			have to set the @code{assert-level} appropriately before compiling a
2538 :			piece of code. You can compile several pieces of code at several
2539 :			@code{assert-level}s (e.g., a trusted library at level 1 and newly
2540 :			written code at level 3).
2541 :
2542 :			doc-assert-level
2543 :
2544 :			If an assertion fails, a message compatible with Emacs' compilation mode
2545 :			is produced and the execution is aborted (currently with @code{ABORT"}.
2546 :			If there is interest, we will introduce a special throw code. But if you
2547 :			intend to @code{catch} a specific condition, using @code{throw} is
2548 :			probably more appropriate than an assertion).
2549 :
2550 :	anton	1.18	@node Assembler and Code words, Threading Words, Programming Tools, Words
2551 :			@section Assembler and Code words
2552 :
2553 :			Gforth provides some words for defining primitives (words written in
2554 :			machine code), and for defining the the machine-code equivalent of
2555 :			@code{DOES>}-based defining words. However, the machine-independent
2556 :	anton	1.40	nature of Gforth poses a few problems: First of all, Gforth runs on
2557 :	anton	1.18	several architectures, so it can provide no standard assembler. What's
2558 :			worse is that the register allocation not only depends on the processor,
2559 :	anton	1.25	but also on the @code{gcc} version and options used.
2560 :	anton	1.18
2561 :	anton	1.25	The words that Gforth offers encapsulate some system dependences (e.g., the
2562 :	anton	1.18	header structure), so a system-independent assembler may be used in
2563 :			Gforth. If you do not have an assembler, you can compile machine code
2564 :			directly with @code{,} and @code{c,}.
2565 :
2566 :			doc-assembler
2567 :			doc-code
2568 :			doc-end-code
2569 :			doc-;code
2570 :			doc-flush-icache
2571 :
2572 :			If @code{flush-icache} does not work correctly, @code{code} words
2573 :			etc. will not work (reliably), either.
2574 :
2575 :			These words are rarely used. Therefore they reside in @code{code.fs},
2576 :			which is usually not loaded (except @code{flush-icache}, which is always
2577 :	anton	1.19	present). You can load them with @code{require code.fs}.
2578 :	anton	1.18
2579 :	anton	1.25	In the assembly code you will want to refer to the inner interpreter's
2580 :			registers (e.g., the data stack pointer) and you may want to use other
2581 :			registers for temporary storage. Unfortunately, the register allocation
2582 :			is installation-dependent.
2583 :
2584 :			The easiest solution is to use explicit register declarations
2585 :			(@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2586 :			GNU C Manual}) for all of the inner interpreter's registers: You have to
2587 :			compile Gforth with @code{-DFORCE_REG} (configure option
2588 :			@code{--enable-force-reg}) and the appropriate declarations must be
2589 :			present in the @code{machine.h} file (see @code{mips.h} for an example;
2590 :			you can find a full list of all declarable register symbols with
2591 :			@code{grep register engine.c}). If you give explicit registers to all
2592 :			variables that are declared at the beginning of @code{engine()}, you
2593 :			should be able to use the other caller-saved registers for temporary
2594 :			storage. Alternatively, you can use the @code{gcc} option
2595 :			@code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2596 :			Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2597 :			(however, this restriction on register allocation may slow Gforth
2598 :			significantly).
2599 :
2600 :			If this solution is not viable (e.g., because @code{gcc} does not allow
2601 :			you to explicitly declare all the registers you need), you have to find
2602 :			out by looking at the code where the inner interpreter's registers
2603 :			reside and which registers can be used for temporary storage. You can
2604 :			get an assembly listing of the engine's code with @code{make engine.s}.
2605 :
2606 :			In any case, it is good practice to abstract your assembly code from the
2607 :			actual register allocation. E.g., if the data stack pointer resides in
2608 :			register @code{$17}, create an alias for this register called @code{sp},
2609 :			and use that in your assembly code.
2610 :
2611 :	anton	1.18	Another option for implementing normal and defining words efficiently
2612 :			is: adding the wanted functionality to the source of Gforth. For normal
2613 :	anton	1.35	words you just have to edit @file{primitives} (@pxref{Automatic
2614 :			Generation}), defining words (equivalent to @code{;CODE} words, for fast
2615 :			defined words) may require changes in @file{engine.c}, @file{kernal.fs},
2616 :			@file{prims2x.fs}, and possibly @file{cross.fs}.
2617 :	anton	1.18
2618 :
2619 :			@node Threading Words, , Assembler and Code words, Words
2620 :	anton	1.4	@section Threading Words
2621 :
2622 :			These words provide access to code addresses and other threading stuff
2623 :	anton	1.17	in Gforth (and, possibly, other interpretive Forths). It more or less
2624 :	anton	1.4	abstracts away the differences between direct and indirect threading
2625 :			(and, for direct threading, the machine dependences). However, at
2626 :			present this wordset is still inclomplete. It is also pretty low-level;
2627 :			some day it will hopefully be made unnecessary by an internals words set
2628 :			that abstracts implementation details away completely.
2629 :
2630 :			doc->code-address
2631 :			doc->does-code
2632 :			doc-code-address!
2633 :			doc-does-code!
2634 :			doc-does-handler!
2635 :			doc-/does-handler
2636 :
2637 :	anton	1.18	The code addresses produced by various defining words are produced by
2638 :			the following words:
2639 :	anton	1.14
2640 :	anton	1.18	doc-docol:
2641 :			doc-docon:
2642 :			doc-dovar:
2643 :			doc-douser:
2644 :			doc-dodefer:
2645 :			doc-dofield:
2646 :
2647 :	anton	1.35	You can recognize words defined by a @code{CREATE}...@code{DOES>} word
2648 :			with @code{>DOES-CODE}. If the word was defined in that way, the value
2649 :			returned is different from 0 and identifies the @code{DOES>} used by the
2650 :			defining word.
2651 :	anton	1.14
2652 :	anton	1.40	@node Tools, ANS conformance, Words, Top
2653 :			@chapter Tools
2654 :
2655 :			@menu
2656 :			* ANS Report:: Report the words used, sorted by wordset
2657 :			@end menu
2658 :
2659 :			See also @ref{Emacs and Gforth}.
2660 :
2661 :			@node ANS Report, , Tools, Tools
2662 :			@section @file{ans-report.fs}: Report the words used, sorted by wordset
2663 :
2664 :			If you want to label a Forth program as ANS Forth Program, you must
2665 :			document which wordsets the program uses; for extension wordsets, it is
2666 :			helpful to list the words the program requires from these wordsets
2667 :			(because Forth systems are allowed to provide only some words of them).
2668 :
2669 :			The @file{ans-report.fs} tool makes it easy for you to determine which
2670 :			words from which wordset and which non-ANS words your application
2671 :			uses. You simply have to include @file{ans-report.fs} before loading the
2672 :			program you want to check. After loading your program, you can get the
2673 :			report with @code{print-ans-report}. A typical use is to run this as
2674 :			batch job like this:
2675 :			@example
2676 :			gforth ans-report.fs myprog.fs -e "print-ans-report bye"
2677 :			@end example
2678 :
2679 :			The output looks like this (for @file{compat/control.fs}):
2680 :			@example
2681 :			The program uses the following words
2682 :			from CORE :
2683 :			: POSTPONE THEN ; immediate ?dup IF 0=
2684 :			from BLOCK-EXT :
2685 :			\
2686 :			from FILE :
2687 :			(
2688 :			@end example
2689 :
2690 :			@subsection Caveats
2691 :
2692 :			Note that @file{ans-report.fs} just checks which words are used, not whether
2693 :			they are used in an ANS Forth conforming way!
2694 :
2695 :			Some words are defined in several wordsets in the
2696 :			standard. @file{ans-report.fs} reports them for only one of the
2697 :			wordsets, and not necessarily the one you expect. It depends on usage
2698 :			which wordset is the right one to specify. E.g., if you only use the
2699 :			compilation semantics of @code{S"}, it is a Core word; if you also use
2700 :			its interpretation semantics, it is a File word.
2701 :
2702 :
2703 :			@node ANS conformance, Model, Tools, Top
2704 :	anton	1.4	@chapter ANS conformance
2705 :
2706 :	anton	1.17	To the best of our knowledge, Gforth is an
2707 :	anton	1.14
2708 :	anton	1.15	ANS Forth System
2709 :	anton	1.34	@itemize @bullet
2710 :	anton	1.15	@item providing the Core Extensions word set
2711 :			@item providing the Block word set
2712 :			@item providing the Block Extensions word set
2713 :			@item providing the Double-Number word set
2714 :			@item providing the Double-Number Extensions word set
2715 :			@item providing the Exception word set
2716 :			@item providing the Exception Extensions word set
2717 :			@item providing the Facility word set
2718 :			@item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2719 :			@item providing the File Access word set
2720 :			@item providing the File Access Extensions word set
2721 :			@item providing the Floating-Point word set
2722 :			@item providing the Floating-Point Extensions word set
2723 :			@item providing the Locals word set
2724 :			@item providing the Locals Extensions word set
2725 :			@item providing the Memory-Allocation word set
2726 :			@item providing the Memory-Allocation Extensions word set (that one's easy)
2727 :			@item providing the Programming-Tools word set
2728 :	anton	1.34	@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
2729 :	anton	1.15	@item providing the Search-Order word set
2730 :			@item providing the Search-Order Extensions word set
2731 :			@item providing the String word set
2732 :			@item providing the String Extensions word set (another easy one)
2733 :			@end itemize
2734 :
2735 :			In addition, ANS Forth systems are required to document certain
2736 :			implementation choices. This chapter tries to meet these
2737 :			requirements. In many cases it gives a way to ask the system for the
2738 :			information instead of providing the information directly, in
2739 :			particular, if the information depends on the processor, the operating
2740 :			system or the installation options chosen, or if they are likely to
2741 :	anton	1.17	change during the maintenance of Gforth.
2742 :	anton	1.15
2743 :	anton	1.14	@comment The framework for the rest has been taken from pfe.
2744 :
2745 :			@menu
2746 :			* The Core Words::
2747 :			* The optional Block word set::
2748 :			* The optional Double Number word set::
2749 :			* The optional Exception word set::
2750 :			* The optional Facility word set::
2751 :			* The optional File-Access word set::
2752 :			* The optional Floating-Point word set::
2753 :			* The optional Locals word set::
2754 :			* The optional Memory-Allocation word set::
2755 :			* The optional Programming-Tools word set::
2756 :			* The optional Search-Order word set::
2757 :			@end menu
2758 :
2759 :
2760 :			@c =====================================================================
2761 :			@node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2762 :			@comment node-name, next, previous, up
2763 :			@section The Core Words
2764 :			@c =====================================================================
2765 :
2766 :			@menu
2767 :	anton	1.15	* core-idef:: Implementation Defined Options
2768 :			* core-ambcond:: Ambiguous Conditions
2769 :			* core-other:: Other System Documentation
2770 :	anton	1.14	@end menu
2771 :
2772 :			@c ---------------------------------------------------------------------
2773 :			@node core-idef, core-ambcond, The Core Words, The Core Words
2774 :			@subsection Implementation Defined Options
2775 :			@c ---------------------------------------------------------------------
2776 :
2777 :			@table @i
2778 :
2779 :			@item (Cell) aligned addresses:
2780 :	anton	1.17	processor-dependent. Gforth's alignment words perform natural alignment
2781 :	anton	1.14	(e.g., an address aligned for a datum of size 8 is divisible by
2782 :			8). Unaligned accesses usually result in a @code{-23 THROW}.
2783 :
2784 :			@item @code{EMIT} and non-graphic characters:
2785 :			The character is output using the C library function (actually, macro)
2786 :	anton	1.36	@code{putc}.
2787 :	anton	1.14
2788 :			@item character editing of @code{ACCEPT} and @code{EXPECT}:
2789 :			This is modeled on the GNU readline library (@pxref{Readline
2790 :			Interaction, , Command Line Editing, readline, The GNU Readline
2791 :			Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2792 :			producing a full word completion every time you type it (instead of
2793 :			producing the common prefix of all completions).
2794 :
2795 :			@item character set:
2796 :			The character set of your computer and display device. Gforth is
2797 :			8-bit-clean (but some other component in your system may make trouble).
2798 :
2799 :			@item Character-aligned address requirements:
2800 :			installation-dependent. Currently a character is represented by a C
2801 :			@code{unsigned char}; in the future we might switch to @code{wchar_t}
2802 :			(Comments on that requested).
2803 :
2804 :			@item character-set extensions and matching of names:
2805 :	anton	1.17	Any character except the ASCII NUL charcter can be used in a
2806 :	anton	1.36	name. Matching is case-insensitive (except in @code{TABLE}s. The
2807 :			matching is performed using the C function @code{strncasecmp}, whose
2808 :			function is probably influenced by the locale. E.g., the @code{C} locale
2809 :			does not know about accents and umlauts, so they are matched
2810 :			case-sensitively in that locale. For portability reasons it is best to
2811 :			write programs such that they work in the @code{C} locale. Then one can
2812 :			use libraries written by a Polish programmer (who might use words
2813 :			containing ISO Latin-2 encoded characters) and by a French programmer
2814 :			(ISO Latin-1) in the same program (of course, @code{WORDS} will produce
2815 :			funny results for some of the words (which ones, depends on the font you
2816 :			are using)). Also, the locale you prefer may not be available in other
2817 :			operating systems. Hopefully, Unicode will solve these problems one day.
2818 :	anton	1.14
2819 :			@item conditions under which control characters match a space delimiter:
2820 :			If @code{WORD} is called with the space character as a delimiter, all
2821 :			white-space characters (as identified by the C macro @code{isspace()})
2822 :			are delimiters. @code{PARSE}, on the other hand, treats space like other
2823 :			delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2824 :			like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2825 :			interpreter (aka text interpreter) by default, treats all white-space
2826 :			characters as delimiters.
2827 :
2828 :			@item format of the control flow stack:
2829 :			The data stack is used as control flow stack. The size of a control flow
2830 :			stack item in cells is given by the constant @code{cs-item-size}. At the
2831 :			time of this writing, an item consists of a (pointer to a) locals list
2832 :			(third), an address in the code (second), and a tag for identifying the
2833 :			item (TOS). The following tags are used: @code{defstart},
2834 :			@code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2835 :			@code{scopestart}.
2836 :
2837 :			@item conversion of digits > 35
2838 :			The characters @code{[\]^_'} are the digits with the decimal value
2839 :			36@minus{}41. There is no way to input many of the larger digits.
2840 :
2841 :			@item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2842 :			The cursor is moved to the end of the entered string. If the input is
2843 :			terminated using the @kbd{Return} key, a space is typed.
2844 :
2845 :			@item exception abort sequence of @code{ABORT"}:
2846 :			The error string is stored into the variable @code{"error} and a
2847 :			@code{-2 throw} is performed.
2848 :
2849 :			@item input line terminator:
2850 :	anton	1.36	For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
2851 :			lines. One of these characters is typically produced when you type the
2852 :			@kbd{Enter} or @kbd{Return} key.
2853 :	anton	1.14
2854 :			@item maximum size of a counted string:
2855 :			@code{s" /counted-string" environment? drop .}. Currently 255 characters
2856 :			on all ports, but this may change.
2857 :
2858 :			@item maximum size of a parsed string:
2859 :			Given by the constant @code{/line}. Currently 255 characters.
2860 :
2861 :			@item maximum size of a definition name, in characters:
2862 :			31
2863 :
2864 :			@item maximum string length for @code{ENVIRONMENT?}, in characters:
2865 :			31
2866 :
2867 :			@item method of selecting the user input device:
2868 :	anton	1.17	The user input device is the standard input. There is currently no way to
2869 :			change it from within Gforth. However, the input can typically be
2870 :			redirected in the command line that starts Gforth.
2871 :	anton	1.14
2872 :			@item method of selecting the user output device:
2873 :	anton	1.36	@code{EMIT} and @code{TYPE} output to the file-id stored in the value
2874 :			@code{outfile-id} (@code{stdout} by default). Gforth uses buffered
2875 :			output, so output on a terminal does not become visible before the next
2876 :			newline or buffer overflow. Output on non-terminals is invisible until
2877 :			the buffer overflows.
2878 :	anton	1.14
2879 :			@item methods of dictionary compilation:
2880 :	anton	1.17	What are we expected to document here?
2881 :	anton	1.14
2882 :			@item number of bits in one address unit:
2883 :			@code{s" address-units-bits" environment? drop .}. 8 in all current
2884 :			ports.
2885 :
2886 :			@item number representation and arithmetic:
2887 :			Processor-dependent. Binary two's complement on all current ports.
2888 :
2889 :			@item ranges for integer types:
2890 :			Installation-dependent. Make environmental queries for @code{MAX-N},
2891 :			@code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2892 :			unsigned (and positive) types is 0. The lower bound for signed types on
2893 :			two's complement and one's complement machines machines can be computed
2894 :			by adding 1 to the upper bound.
2895 :
2896 :			@item read-only data space regions:
2897 :			The whole Forth data space is writable.
2898 :
2899 :			@item size of buffer at @code{WORD}:
2900 :			@code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2901 :			shared with the pictured numeric output string. If overwriting
2902 :			@code{PAD} is acceptable, it is as large as the remaining dictionary
2903 :			space, although only as much can be sensibly used as fits in a counted
2904 :			string.
2905 :
2906 :			@item size of one cell in address units:
2907 :			@code{1 cells .}.
2908 :
2909 :			@item size of one character in address units:
2910 :			@code{1 chars .}. 1 on all current ports.
2911 :
2912 :			@item size of the keyboard terminal buffer:
2913 :	anton	1.36	Varies. You can determine the size at a specific time using @code{lp@@
2914 :	anton	1.14	tib - .}. It is shared with the locals stack and TIBs of files that
2915 :			include the current file. You can change the amount of space for TIBs
2916 :	anton	1.17	and locals stack at Gforth startup with the command line option
2917 :	anton	1.14	@code{-l}.
2918 :
2919 :			@item size of the pictured numeric output buffer:
2920 :			@code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2921 :			shared with @code{WORD}.
2922 :
2923 :			@item size of the scratch area returned by @code{PAD}:
2924 :			The remainder of dictionary space. You can even use the unused part of
2925 :	anton	1.36	the data stack space. The current size can be computed with @code{sp@@
2926 :	anton	1.14	pad - .}.
2927 :
2928 :			@item system case-sensitivity characteristics:
2929 :	anton	1.36	Dictionary searches are case insensitive (except in
2930 :			@code{TABLE}s). However, as explained above under @i{character-set
2931 :			extensions}, the matching for non-ASCII characters is determined by the
2932 :			locale you are using. In the default @code{C} locale all non-ASCII
2933 :			characters are matched case-sensitively.
2934 :	anton	1.14
2935 :			@item system prompt:
2936 :			@code{ ok} in interpret state, @code{ compiled} in compile state.
2937 :
2938 :			@item division rounding:
2939 :			installation dependent. @code{s" floored" environment? drop .}. We leave
2940 :	anton	1.25	the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
2941 :	anton	1.14	@code{fm/mod}, @code{sm/rem} or simply @code{/}).
2942 :
2943 :			@item values of @code{STATE} when true:
2944 :			-1.
2945 :
2946 :			@item values returned after arithmetic overflow:
2947 :			On two's complement machines, arithmetic is performed modulo
2948 :			2bits-per-cell for single arithmetic and 4bits-per-cell for double
2949 :			arithmetic (with appropriate mapping for signed types). Division by zero
2950 :	anton	1.36	typically results in a @code{-55 throw} (Floating-point unidentified
2951 :	anton	1.14	fault), although a @code{-10 throw} (divide by zero) would be more
2952 :			appropriate.
2953 :
2954 :			@item whether the current definition can be found after @t{DOES>}:
2955 :			No.
2956 :
2957 :			@end table
2958 :
2959 :			@c ---------------------------------------------------------------------
2960 :			@node core-ambcond, core-other, core-idef, The Core Words
2961 :			@subsection Ambiguous conditions
2962 :			@c ---------------------------------------------------------------------
2963 :
2964 :			@table @i
2965 :
2966 :			@item a name is neither a word nor a number:
2967 :	anton	1.36	@code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
2968 :			preserves the data and FP stack, so you don't lose more work than
2969 :			necessary.
2970 :	anton	1.14
2971 :			@item a definition name exceeds the maximum length allowed:
2972 :			@code{-19 throw} (Word name too long)
2973 :
2974 :			@item addressing a region not inside the various data spaces of the forth system:
2975 :			The stacks, code space and name space are accessible. Machine code space is
2976 :			typically readable. Accessing other addresses gives results dependent on
2977 :			the operating system. On decent systems: @code{-9 throw} (Invalid memory
2978 :			address).
2979 :
2980 :			@item argument type incompatible with parameter:
2981 :			This is usually not caught. Some words perform checks, e.g., the control
2982 :			flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2983 :			mismatch).
2984 :
2985 :			@item attempting to obtain the execution token of a word with undefined execution semantics:
2986 :	anton	1.36	@code{-14 throw} (Interpreting a compile-only word). In some cases, you
2987 :			get an execution token for @code{compile-only-error} (which performs a
2988 :			@code{-14 throw} when executed).
2989 :	anton	1.14
2990 :			@item dividing by zero:
2991 :			typically results in a @code{-55 throw} (floating point unidentified
2992 :			fault), although a @code{-10 throw} (divide by zero) would be more
2993 :			appropriate.
2994 :
2995 :			@item insufficient data stack or return stack space:
2996 :			Not checked. This typically results in mysterious illegal memory
2997 :			accesses, producing @code{-9 throw} (Invalid memory address) or
2998 :			@code{-23 throw} (Address alignment exception).
2999 :
3000 :			@item insufficient space for loop control parameters:
3001 :			like other return stack overflows.
3002 :
3003 :			@item insufficient space in the dictionary:
3004 :			Not checked. Similar results as stack overflows. However, typically the
3005 :			error appears at a different place when one inserts or removes code.
3006 :
3007 :			@item interpreting a word with undefined interpretation semantics:
3008 :			For some words, we defined interpretation semantics. For the others:
3009 :	anton	1.36	@code{-14 throw} (Interpreting a compile-only word).
3010 :	anton	1.14
3011 :			@item modifying the contents of the input buffer or a string literal:
3012 :			These are located in writable memory and can be modified.
3013 :
3014 :			@item overflow of the pictured numeric output string:
3015 :			Not checked.
3016 :
3017 :			@item parsed string overflow:
3018 :			@code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
3019 :
3020 :			@item producing a result out of range:
3021 :			On two's complement machines, arithmetic is performed modulo
3022 :			2bits-per-cell for single arithmetic and 4bits-per-cell for double
3023 :			arithmetic (with appropriate mapping for signed types). Division by zero
3024 :			typically results in a @code{-55 throw} (floatingpoint unidentified
3025 :			fault), although a @code{-10 throw} (divide by zero) would be more
3026 :			appropriate. @code{convert} and @code{>number} currently overflow
3027 :			silently.
3028 :
3029 :			@item reading from an empty data or return stack:
3030 :			The data stack is checked by the outer (aka text) interpreter after
3031 :			every word executed. If it has underflowed, a @code{-4 throw} (Stack
3032 :			underflow) is performed. Apart from that, the stacks are not checked and
3033 :			underflows can result in similar behaviour as overflows (of adjacent
3034 :			stacks).
3035 :
3036 :	anton	1.36	@item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
3037 :	anton	1.14	@code{Create} and its descendants perform a @code{-16 throw} (Attempt to
3038 :			use zero-length string as a name). Words like @code{'} probably will not
3039 :			find what they search. Note that it is possible to create zero-length
3040 :			names with @code{nextname} (should it not?).
3041 :
3042 :			@item @code{>IN} greater than input buffer:
3043 :			The next invocation of a parsing word returns a string wih length 0.
3044 :
3045 :			@item @code{RECURSE} appears after @code{DOES>}:
3046 :	anton	1.36	Compiles a recursive call to the defining word, not to the defined word.
3047 :	anton	1.14
3048 :			@item argument input source different than current input source for @code{RESTORE-INPUT}:
3049 :	anton	1.27	@code{-12 THROW}. Note that, once an input file is closed (e.g., because
3050 :			the end of the file was reached), its source-id may be
3051 :			reused. Therefore, restoring an input source specification referencing a
3052 :			closed file may lead to unpredictable results instead of a @code{-12
3053 :			THROW}.
3054 :
3055 :	anton	1.36	In the future, Gforth may be able to restore input source specifications
3056 :	anton	1.27	from other than the current input soruce.
3057 :	anton	1.14
3058 :			@item data space containing definitions gets de-allocated:
3059 :			Deallocation with @code{allot} is not checked. This typically resuls in
3060 :			memory access faults or execution of illegal instructions.
3061 :
3062 :			@item data space read/write with incorrect alignment:
3063 :			Processor-dependent. Typically results in a @code{-23 throw} (Address
3064 :			alignment exception). Under Linux on a 486 or later processor with
3065 :			alignment turned on, incorrect alignment results in a @code{-9 throw}
3066 :			(Invalid memory address). There are reportedly some processors with
3067 :			alignment restrictions that do not report them.
3068 :
3069 :			@item data space pointer not properly aligned, @code{,}, @code{C,}:
3070 :			Like other alignment errors.
3071 :
3072 :			@item less than u+2 stack items (@code{PICK} and @code{ROLL}):
3073 :			Not checked. May cause an illegal memory access.
3074 :
3075 :			@item loop control parameters not available:
3076 :			Not checked. The counted loop words simply assume that the top of return
3077 :			stack items are loop control parameters and behave accordingly.
3078 :
3079 :			@item most recent definition does not have a name (@code{IMMEDIATE}):
3080 :			@code{abort" last word was headerless"}.
3081 :
3082 :			@item name not defined by @code{VALUE} used by @code{TO}:
3083 :	anton	1.36	@code{-32 throw} (Invalid name argument) (unless name was defined by
3084 :			@code{CONSTANT}; then it just changes the constant).
3085 :	anton	1.14
3086 :	anton	1.15	@item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
3087 :	anton	1.14	@code{-13 throw} (Undefined word)
3088 :
3089 :			@item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
3090 :			Gforth behaves as if they were of the same type. I.e., you can predict
3091 :			the behaviour by interpreting all parameters as, e.g., signed.
3092 :
3093 :			@item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
3094 :	anton	1.36	Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
3095 :			compilation semantics of @code{TO}.
3096 :	anton	1.14
3097 :			@item String longer than a counted string returned by @code{WORD}:
3098 :			Not checked. The string will be ok, but the count will, of course,
3099 :			contain only the least significant bits of the length.
3100 :
3101 :	anton	1.15	@item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
3102 :	anton	1.14	Processor-dependent. Typical behaviours are returning 0 and using only
3103 :			the low bits of the shift count.
3104 :
3105 :			@item word not defined via @code{CREATE}:
3106 :			@code{>BODY} produces the PFA of the word no matter how it was defined.
3107 :
3108 :			@code{DOES>} changes the execution semantics of the last defined word no
3109 :			matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
3110 :			@code{CREATE , DOES>}.
3111 :
3112 :			@item words improperly used outside @code{<#} and @code{#>}:
3113 :			Not checked. As usual, you can expect memory faults.
3114 :
3115 :			@end table
3116 :
3117 :
3118 :			@c ---------------------------------------------------------------------
3119 :			@node core-other, , core-ambcond, The Core Words
3120 :			@subsection Other system documentation
3121 :			@c ---------------------------------------------------------------------
3122 :
3123 :			@table @i
3124 :
3125 :			@item nonstandard words using @code{PAD}:
3126 :			None.
3127 :
3128 :			@item operator's terminal facilities available:
3129 :	anton	1.26	After processing the command line, Gforth goes into interactive mode,
3130 :			and you can give commands to Gforth interactively. The actual facilities
3131 :			available depend on how you invoke Gforth.
3132 :	anton	1.14
3133 :			@item program data space available:
3134 :	anton	1.36	@code{sp@@ here - .} gives the space remaining for dictionary and data
3135 :	anton	1.14	stack together.
3136 :
3137 :			@item return stack space available:
3138 :	anton	1.26	By default 16 KBytes. The default can be overridden with the @code{-r}
3139 :			switch (@pxref{Invocation}) when Gforth starts up.
3140 :	anton	1.14
3141 :			@item stack space available:
3142 :	anton	1.36	@code{sp@@ here - .} gives the space remaining for dictionary and data
3143 :	anton	1.14	stack together.
3144 :
3145 :			@item system dictionary space required, in address units:
3146 :			Type @code{here forthstart - .} after startup. At the time of this
3147 :			writing, this gives 70108 (bytes) on a 32-bit system.
3148 :			@end table
3149 :
3150 :
3151 :			@c =====================================================================
3152 :			@node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
3153 :			@section The optional Block word set
3154 :			@c =====================================================================
3155 :
3156 :			@menu
3157 :	anton	1.15	* block-idef:: Implementation Defined Options
3158 :			* block-ambcond:: Ambiguous Conditions
3159 :			* block-other:: Other System Documentation
3160 :	anton	1.14	@end menu
3161 :
3162 :
3163 :			@c ---------------------------------------------------------------------
3164 :			@node block-idef, block-ambcond, The optional Block word set, The optional Block word set
3165 :			@subsection Implementation Defined Options
3166 :			@c ---------------------------------------------------------------------
3167 :
3168 :			@table @i
3169 :
3170 :			@item the format for display by @code{LIST}:
3171 :			First the screen number is displayed, then 16 lines of 64 characters,
3172 :			each line preceded by the line number.
3173 :
3174 :			@item the length of a line affected by @code{\}:
3175 :			64 characters.
3176 :			@end table
3177 :
3178 :
3179 :			@c ---------------------------------------------------------------------
3180 :			@node block-ambcond, block-other, block-idef, The optional Block word set
3181 :			@subsection Ambiguous conditions
3182 :			@c ---------------------------------------------------------------------
3183 :
3184 :			@table @i
3185 :
3186 :			@item correct block read was not possible:
3187 :			Typically results in a @code{throw} of some OS-derived value (between
3188 :			-512 and -2048). If the blocks file was just not long enough, blanks are
3189 :			supplied for the missing portion.
3190 :
3191 :			@item I/O exception in block transfer:
3192 :			Typically results in a @code{throw} of some OS-derived value (between
3193 :			-512 and -2048).
3194 :
3195 :			@item invalid block number:
3196 :			@code{-35 throw} (Invalid block number)
3197 :
3198 :			@item a program directly alters the contents of @code{BLK}:
3199 :			The input stream is switched to that other block, at the same
3200 :			position. If the storing to @code{BLK} happens when interpreting
3201 :			non-block input, the system will get quite confused when the block ends.
3202 :
3203 :			@item no current block buffer for @code{UPDATE}:
3204 :			@code{UPDATE} has no effect.
3205 :
3206 :			@end table
3207 :
3208 :
3209 :			@c ---------------------------------------------------------------------
3210 :			@node block-other, , block-ambcond, The optional Block word set
3211 :			@subsection Other system documentation
3212 :			@c ---------------------------------------------------------------------
3213 :
3214 :			@table @i
3215 :
3216 :			@item any restrictions a multiprogramming system places on the use of buffer addresses:
3217 :			No restrictions (yet).
3218 :
3219 :			@item the number of blocks available for source and data:
3220 :			depends on your disk space.
3221 :
3222 :			@end table
3223 :
3224 :
3225 :			@c =====================================================================
3226 :			@node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
3227 :			@section The optional Double Number word set
3228 :			@c =====================================================================
3229 :
3230 :			@menu
3231 :	anton	1.15	* double-ambcond:: Ambiguous Conditions
3232 :	anton	1.14	@end menu
3233 :
3234 :
3235 :			@c ---------------------------------------------------------------------
3236 :	anton	1.15	@node double-ambcond, , The optional Double Number word set, The optional Double Number word set
3237 :	anton	1.14	@subsection Ambiguous conditions
3238 :			@c ---------------------------------------------------------------------
3239 :
3240 :			@table @i
3241 :
3242 :	anton	1.15	@item @var{d} outside of range of @var{n} in @code{D>S}:
3243 :	anton	1.14	The least significant cell of @var{d} is produced.
3244 :
3245 :			@end table
3246 :
3247 :
3248 :			@c =====================================================================
3249 :			@node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
3250 :			@section The optional Exception word set
3251 :			@c =====================================================================
3252 :
3253 :			@menu
3254 :	anton	1.15	* exception-idef:: Implementation Defined Options
3255 :	anton	1.14	@end menu
3256 :
3257 :
3258 :			@c ---------------------------------------------------------------------
3259 :	anton	1.15	@node exception-idef, , The optional Exception word set, The optional Exception word set
3260 :	anton	1.14	@subsection Implementation Defined Options
3261 :			@c ---------------------------------------------------------------------
3262 :
3263 :			@table @i
3264 :			@item @code{THROW}-codes used in the system:
3265 :			The codes -256@minus{}-511 are used for reporting signals (see
3266 :			@file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
3267 :			(for file and memory allocation operations). The mapping from OS error
3268 :	anton	1.37	numbers to throw code is -512@minus{}@code{errno}. One side effect of
3269 :	anton	1.14	this mapping is that undefined OS errors produce a message with a
3270 :			strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
3271 :			488} on my system.
3272 :			@end table
3273 :
3274 :			@c =====================================================================
3275 :			@node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
3276 :			@section The optional Facility word set
3277 :			@c =====================================================================
3278 :
3279 :			@menu
3280 :	anton	1.15	* facility-idef:: Implementation Defined Options
3281 :			* facility-ambcond:: Ambiguous Conditions
3282 :	anton	1.14	@end menu
3283 :
3284 :
3285 :			@c ---------------------------------------------------------------------
3286 :			@node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
3287 :			@subsection Implementation Defined Options
3288 :			@c ---------------------------------------------------------------------
3289 :
3290 :			@table @i
3291 :
3292 :			@item encoding of keyboard events (@code{EKEY}):
3293 :			Not yet implemeted.
3294 :
3295 :			@item duration of a system clock tick
3296 :			System dependent. With respect to @code{MS}, the time is specified in
3297 :			microseconds. How well the OS and the hardware implement this, is
3298 :			another question.
3299 :
3300 :			@item repeatability to be expected from the execution of @code{MS}:
3301 :			System dependent. On Unix, a lot depends on load. If the system is
3302 :	anton	1.17	lightly loaded, and the delay is short enough that Gforth does not get
3303 :	anton	1.14	swapped out, the performance should be acceptable. Under MS-DOS and
3304 :			other single-tasking systems, it should be good.
3305 :
3306 :			@end table
3307 :
3308 :
3309 :			@c ---------------------------------------------------------------------
3310 :	anton	1.15	@node facility-ambcond, , facility-idef, The optional Facility word set
3311 :	anton	1.14	@subsection Ambiguous conditions
3312 :			@c ---------------------------------------------------------------------
3313 :
3314 :			@table @i
3315 :
3316 :			@item @code{AT-XY} can't be performed on user output device:
3317 :			Largely terminal dependant. No range checks are done on the arguments.
3318 :			No errors are reported. You may see some garbage appearing, you may see
3319 :			simply nothing happen.
3320 :
3321 :			@end table
3322 :
3323 :
3324 :			@c =====================================================================
3325 :			@node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
3326 :			@section The optional File-Access word set
3327 :			@c =====================================================================
3328 :
3329 :			@menu
3330 :	anton	1.15	* file-idef:: Implementation Defined Options
3331 :			* file-ambcond:: Ambiguous Conditions
3332 :	anton	1.14	@end menu
3333 :
3334 :
3335 :			@c ---------------------------------------------------------------------
3336 :			@node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
3337 :			@subsection Implementation Defined Options
3338 :			@c ---------------------------------------------------------------------
3339 :
3340 :			@table @i
3341 :
3342 :			@item File access methods used:
3343 :			@code{R/O}, @code{R/W} and @code{BIN} work as you would
3344 :			expect. @code{W/O} translates into the C file opening mode @code{w} (or
3345 :			@code{wb}): The file is cleared, if it exists, and created, if it does
3346 :	anton	1.15	not (both with @code{open-file} and @code{create-file}). Under Unix
3347 :	anton	1.14	@code{create-file} creates a file with 666 permissions modified by your
3348 :			umask.
3349 :
3350 :			@item file exceptions:
3351 :			The file words do not raise exceptions (except, perhaps, memory access
3352 :			faults when you pass illegal addresses or file-ids).
3353 :
3354 :			@item file line terminator:
3355 :			System-dependent. Gforth uses C's newline character as line
3356 :			terminator. What the actual character code(s) of this are is
3357 :			system-dependent.
3358 :
3359 :			@item file name format
3360 :			System dependent. Gforth just uses the file name format of your OS.
3361 :
3362 :			@item information returned by @code{FILE-STATUS}:
3363 :			@code{FILE-STATUS} returns the most powerful file access mode allowed
3364 :			for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
3365 :			cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
3366 :			along with the retured mode.
3367 :
3368 :			@item input file state after an exception when including source:
3369 :			All files that are left via the exception are closed.
3370 :
3371 :			@item @var{ior} values and meaning:
3372 :	anton	1.15	The @var{ior}s returned by the file and memory allocation words are
3373 :			intended as throw codes. They typically are in the range
3374 :			-512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3375 :			@var{ior}s is -512@minus{}@var{errno}.
3376 :	anton	1.14
3377 :			@item maximum depth of file input nesting:
3378 :			limited by the amount of return stack, locals/TIB stack, and the number
3379 :			of open files available. This should not give you troubles.
3380 :
3381 :			@item maximum size of input line:
3382 :			@code{/line}. Currently 255.
3383 :
3384 :			@item methods of mapping block ranges to files:
3385 :	anton	1.37	By default, blocks are accessed in the file @file{blocks.fb} in the
3386 :			current working directory. The file can be switched with @code{USE}.
3387 :	anton	1.14
3388 :			@item number of string buffers provided by @code{S"}:
3389 :			1
3390 :
3391 :			@item size of string buffer used by @code{S"}:
3392 :			@code{/line}. currently 255.
3393 :
3394 :			@end table
3395 :
3396 :			@c ---------------------------------------------------------------------
3397 :	anton	1.15	@node file-ambcond, , file-idef, The optional File-Access word set
3398 :	anton	1.14	@subsection Ambiguous conditions
3399 :			@c ---------------------------------------------------------------------
3400 :
3401 :			@table @i
3402 :
3403 :			@item attempting to position a file outside it's boundaries:
3404 :			@code{REPOSITION-FILE} is performed as usual: Afterwards,
3405 :			@code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
3406 :
3407 :			@item attempting to read from file positions not yet written:
3408 :			End-of-file, i.e., zero characters are read and no error is reported.
3409 :
3410 :			@item @var{file-id} is invalid (@code{INCLUDE-FILE}):
3411 :			An appropriate exception may be thrown, but a memory fault or other
3412 :			problem is more probable.
3413 :
3414 :			@item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
3415 :			The @var{ior} produced by the operation, that discovered the problem, is
3416 :			thrown.
3417 :
3418 :			@item named file cannot be opened (@code{included}):
3419 :			The @var{ior} produced by @code{open-file} is thrown.
3420 :
3421 :			@item requesting an unmapped block number:
3422 :			There are no unmapped legal block numbers. On some operating systems,
3423 :			writing a block with a large number may overflow the file system and
3424 :			have an error message as consequence.
3425 :
3426 :			@item using @code{source-id} when @code{blk} is non-zero:
3427 :			@code{source-id} performs its function. Typically it will give the id of
3428 :			the source which loaded the block. (Better ideas?)
3429 :
3430 :			@end table
3431 :
3432 :
3433 :			@c =====================================================================
3434 :			@node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
3435 :	anton	1.15	@section The optional Floating-Point word set
3436 :	anton	1.14	@c =====================================================================
3437 :
3438 :			@menu
3439 :	anton	1.15	* floating-idef:: Implementation Defined Options
3440 :			* floating-ambcond:: Ambiguous Conditions
3441 :	anton	1.14	@end menu
3442 :
3443 :
3444 :			@c ---------------------------------------------------------------------
3445 :			@node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
3446 :			@subsection Implementation Defined Options
3447 :			@c ---------------------------------------------------------------------
3448 :
3449 :			@table @i
3450 :
3451 :	anton	1.15	@item format and range of floating point numbers:
3452 :			System-dependent; the @code{double} type of C.
3453 :	anton	1.14
3454 :	anton	1.15	@item results of @code{REPRESENT} when @var{float} is out of range:
3455 :			System dependent; @code{REPRESENT} is implemented using the C library
3456 :			function @code{ecvt()} and inherits its behaviour in this respect.
3457 :	anton	1.14
3458 :	anton	1.15	@item rounding or truncation of floating-point numbers:
3459 :	anton	1.26	System dependent; the rounding behaviour is inherited from the hosting C
3460 :			compiler. IEEE-FP-based (i.e., most) systems by default round to
3461 :			nearest, and break ties by rounding to even (i.e., such that the last
3462 :			bit of the mantissa is 0).
3463 :	anton	1.14
3464 :	anton	1.15	@item size of floating-point stack:
3465 :			@code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
3466 :			with the command-line option @code{-f}.
3467 :	anton	1.14
3468 :	anton	1.15	@item width of floating-point stack:
3469 :			@code{1 floats}.
3470 :	anton	1.14
3471 :			@end table
3472 :
3473 :
3474 :			@c ---------------------------------------------------------------------
3475 :	anton	1.15	@node floating-ambcond, , floating-idef, The optional Floating-Point word set
3476 :			@subsection Ambiguous conditions
3477 :	anton	1.14	@c ---------------------------------------------------------------------
3478 :
3479 :			@table @i
3480 :
3481 :	anton	1.15	@item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
3482 :	anton	1.37	System-dependent. Typically results in a @code{-23 THROW} like other
3483 :	anton	1.15	alignment violations.
3484 :	anton	1.14
3485 :	anton	1.15	@item @code{f@@} or @code{f!} used with an address that is not float aligned:
3486 :	anton	1.37	System-dependent. Typically results in a @code{-23 THROW} like other
3487 :	anton	1.15	alignment violations.
3488 :	anton	1.14
3489 :	anton	1.15	@item Floating-point result out of range:
3490 :			System-dependent. Can result in a @code{-55 THROW} (Floating-point
3491 :			unidentified fault), or can produce a special value representing, e.g.,
3492 :			Infinity.
3493 :	anton	1.14
3494 :	anton	1.15	@item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
3495 :			System-dependent. Typically results in an alignment fault like other
3496 :			alignment violations.
3497 :	anton	1.14
3498 :	anton	1.15	@item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
3499 :			The floating-point number is converted into decimal nonetheless.
3500 :	anton	1.14
3501 :	anton	1.15	@item Both arguments are equal to zero (@code{FATAN2}):
3502 :			System-dependent. @code{FATAN2} is implemented using the C library
3503 :			function @code{atan2()}.
3504 :	anton	1.14
3505 :	anton	1.15	@item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
3506 :			System-dependent. Anyway, typically the cos of @var{r1} will not be zero
3507 :			because of small errors and the tan will be a very large (or very small)
3508 :			but finite number.
3509 :	anton	1.14
3510 :	anton	1.15	@item @var{d} cannot be presented precisely as a float in @code{D>F}:
3511 :			The result is rounded to the nearest float.
3512 :	anton	1.14
3513 :	anton	1.15	@item dividing by zero:
3514 :			@code{-55 throw} (Floating-point unidentified fault)
3515 :	anton	1.14
3516 :	anton	1.15	@item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
3517 :			System dependent. On IEEE-FP based systems the number is converted into
3518 :			an infinity.
3519 :	anton	1.14
3520 :	anton	1.15	@item @var{float}<1 (@code{facosh}):
3521 :			@code{-55 throw} (Floating-point unidentified fault)
3522 :	anton	1.14
3523 :	anton	1.15	@item @var{float}=<-1 (@code{flnp1}):
3524 :			@code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3525 :			negative infinity is typically produced for @var{float}=-1.
3526 :	anton	1.14
3527 :	anton	1.15	@item @var{float}=<0 (@code{fln}, @code{flog}):
3528 :			@code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3529 :			negative infinity is typically produced for @var{float}=0.
3530 :	anton	1.14
3531 :	anton	1.15	@item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
3532 :			@code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
3533 :			produces values for these inputs on my Linux box (Bug in the C library?)
3534 :	anton	1.14
3535 :	anton	1.15	@item \|@var{float}\|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
3536 :			@code{-55 throw} (Floating-point unidentified fault).
3537 :	anton	1.14
3538 :	anton	1.15	@item integer part of float cannot be represented by @var{d} in @code{f>d}:
3539 :			@code{-55 throw} (Floating-point unidentified fault).
3540 :	anton	1.14
3541 :	anton	1.15	@item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
3542 :			This does not happen.
3543 :			@end table
3544 :	anton	1.14
3545 :
3546 :
3547 :			@c =====================================================================
3548 :	anton	1.15	@node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
3549 :			@section The optional Locals word set
3550 :	anton	1.14	@c =====================================================================
3551 :
3552 :			@menu
3553 :	anton	1.15	* locals-idef:: Implementation Defined Options
3554 :			* locals-ambcond:: Ambiguous Conditions
3555 :	anton	1.14	@end menu
3556 :
3557 :
3558 :			@c ---------------------------------------------------------------------
3559 :	anton	1.15	@node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
3560 :	anton	1.14	@subsection Implementation Defined Options
3561 :			@c ---------------------------------------------------------------------
3562 :
3563 :			@table @i
3564 :
3565 :	anton	1.15	@item maximum number of locals in a definition:
3566 :			@code{s" #locals" environment? drop .}. Currently 15. This is a lower
3567 :			bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
3568 :			characters. The number of locals in a definition is bounded by the size
3569 :			of locals-buffer, which contains the names of the locals.
3570 :	anton	1.14
3571 :			@end table
3572 :
3573 :
3574 :			@c ---------------------------------------------------------------------
3575 :	anton	1.15	@node locals-ambcond, , locals-idef, The optional Locals word set
3576 :	anton	1.14	@subsection Ambiguous conditions
3577 :			@c ---------------------------------------------------------------------
3578 :
3579 :			@table @i
3580 :
3581 :	anton	1.15	@item executing a named local in interpretation state:
3582 :			@code{-14 throw} (Interpreting a compile-only word).
3583 :	anton	1.14
3584 :	anton	1.15	@item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3585 :			@code{-32 throw} (Invalid name argument)
3586 :	anton	1.14
3587 :			@end table
3588 :
3589 :
3590 :			@c =====================================================================
3591 :	anton	1.15	@node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3592 :			@section The optional Memory-Allocation word set
3593 :	anton	1.14	@c =====================================================================
3594 :
3595 :			@menu
3596 :	anton	1.15	* memory-idef:: Implementation Defined Options
3597 :	anton	1.14	@end menu
3598 :
3599 :
3600 :			@c ---------------------------------------------------------------------
3601 :	anton	1.15	@node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
3602 :	anton	1.14	@subsection Implementation Defined Options
3603 :			@c ---------------------------------------------------------------------
3604 :
3605 :			@table @i
3606 :
3607 :	anton	1.15	@item values and meaning of @var{ior}:
3608 :			The @var{ior}s returned by the file and memory allocation words are
3609 :			intended as throw codes. They typically are in the range
3610 :			-512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3611 :			@var{ior}s is -512@minus{}@var{errno}.
3612 :	anton	1.14
3613 :			@end table
3614 :
3615 :			@c =====================================================================
3616 :	anton	1.15	@node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3617 :			@section The optional Programming-Tools word set
3618 :	anton	1.14	@c =====================================================================
3619 :
3620 :			@menu
3621 :	anton	1.15	* programming-idef:: Implementation Defined Options
3622 :			* programming-ambcond:: Ambiguous Conditions
3623 :	anton	1.14	@end menu
3624 :
3625 :
3626 :			@c ---------------------------------------------------------------------
3627 :	anton	1.15	@node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
3628 :	anton	1.14	@subsection Implementation Defined Options
3629 :			@c ---------------------------------------------------------------------
3630 :
3631 :			@table @i
3632 :
3633 :	anton	1.15	@item ending sequence for input following @code{;code} and @code{code}:
3634 :	anton	1.37	@code{end-code}
3635 :	anton	1.14
3636 :	anton	1.15	@item manner of processing input following @code{;code} and @code{code}:
3637 :	anton	1.37	The @code{assembler} vocabulary is pushed on the search order stack, and
3638 :			the input is processed by the text interpreter, (starting) in interpret
3639 :			state.
3640 :	anton	1.15
3641 :			@item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3642 :	anton	1.37	The ANS Forth search order word set.
3643 :	anton	1.15
3644 :			@item source and format of display by @code{SEE}:
3645 :			The source for @code{see} is the intermediate code used by the inner
3646 :			interpreter. The current @code{see} tries to output Forth source code
3647 :			as well as possible.
3648 :
3649 :	anton	1.14	@end table
3650 :
3651 :			@c ---------------------------------------------------------------------
3652 :	anton	1.15	@node programming-ambcond, , programming-idef, The optional Programming-Tools word set
3653 :	anton	1.14	@subsection Ambiguous conditions
3654 :			@c ---------------------------------------------------------------------
3655 :
3656 :			@table @i
3657 :
3658 :	anton	1.15	@item deleting the compilation wordlist (@code{FORGET}):
3659 :			Not implemented (yet).
3660 :	anton	1.14
3661 :	anton	1.15	@item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3662 :			This typically results in an @code{abort"} with a descriptive error
3663 :			message (may change into a @code{-22 throw} (Control structure mismatch)
3664 :			in the future). You may also get a memory access error. If you are
3665 :			unlucky, this ambiguous condition is not caught.
3666 :
3667 :			@item @var{name} can't be found (@code{forget}):
3668 :			Not implemented (yet).
3669 :	anton	1.14
3670 :	anton	1.15	@item @var{name} not defined via @code{CREATE}:
3671 :	anton	1.37	@code{;code} behaves like @code{DOES>} in this respect, i.e., it changes
3672 :			the execution semantics of the last defined word no matter how it was
3673 :			defined.
3674 :	anton	1.14
3675 :	anton	1.15	@item @code{POSTPONE} applied to @code{[IF]}:
3676 :			After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3677 :			equivalent to @code{[IF]}.
3678 :	anton	1.14
3679 :	anton	1.15	@item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3680 :			Continue in the same state of conditional compilation in the next outer
3681 :			input source. Currently there is no warning to the user about this.
3682 :	anton	1.14
3683 :	anton	1.15	@item removing a needed definition (@code{FORGET}):
3684 :			Not implemented (yet).
3685 :	anton	1.14
3686 :			@end table
3687 :
3688 :
3689 :			@c =====================================================================
3690 :	anton	1.15	@node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3691 :			@section The optional Search-Order word set
3692 :	anton	1.14	@c =====================================================================
3693 :
3694 :			@menu
3695 :	anton	1.15	* search-idef:: Implementation Defined Options
3696 :			* search-ambcond:: Ambiguous Conditions
3697 :	anton	1.14	@end menu
3698 :
3699 :
3700 :			@c ---------------------------------------------------------------------
3701 :	anton	1.15	@node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
3702 :	anton	1.14	@subsection Implementation Defined Options
3703 :			@c ---------------------------------------------------------------------
3704 :
3705 :			@table @i
3706 :
3707 :	anton	1.15	@item maximum number of word lists in search order:
3708 :			@code{s" wordlists" environment? drop .}. Currently 16.
3709 :
3710 :			@item minimum search order:
3711 :			@code{root root}.
3712 :	anton	1.14
3713 :			@end table
3714 :
3715 :			@c ---------------------------------------------------------------------
3716 :	anton	1.15	@node search-ambcond, , search-idef, The optional Search-Order word set
3717 :	anton	1.14	@subsection Ambiguous conditions
3718 :			@c ---------------------------------------------------------------------
3719 :
3720 :			@table @i
3721 :
3722 :	anton	1.15	@item changing the compilation wordlist (during compilation):
3723 :	anton	1.33	The word is entered into the wordlist that was the compilation wordlist
3724 :			at the start of the definition. Any changes to the name field (e.g.,
3725 :			@code{immediate}) or the code field (e.g., when executing @code{DOES>})
3726 :			are applied to the latest defined word (as reported by @code{last} or
3727 :			@code{lastxt}), if possible, irrespective of the compilation wordlist.
3728 :	anton	1.14
3729 :	anton	1.15	@item search order empty (@code{previous}):
3730 :			@code{abort" Vocstack empty"}.
3731 :	anton	1.14
3732 :	anton	1.15	@item too many word lists in search order (@code{also}):
3733 :			@code{abort" Vocstack full"}.
3734 :	anton	1.14
3735 :			@end table
3736 :	anton	1.13
3737 :	anton	1.34	@node Model, Integrating Gforth, ANS conformance, Top
3738 :			@chapter Model
3739 :
3740 :			This chapter has yet to be written. It will contain information, on
3741 :			which internal structures you can rely.
3742 :
3743 :			@node Integrating Gforth, Emacs and Gforth, Model, Top
3744 :			@chapter Integrating Gforth into C programs
3745 :
3746 :			This is not yet implemented.
3747 :
3748 :			Several people like to use Forth as scripting language for applications
3749 :			that are otherwise written in C, C++, or some other language.
3750 :
3751 :			The Forth system ATLAST provides facilities for embedding it into
3752 :			applications; unfortunately it has several disadvantages: most
3753 :	anton	1.36	importantly, it is not based on ANS Forth, and it is apparently dead
3754 :	anton	1.34	(i.e., not developed further and not supported). The facilities
3755 :			provided by Gforth in this area are inspired by ATLASTs facilities, so
3756 :			making the switch should not be hard.
3757 :
3758 :			We also tried to design the interface such that it can easily be
3759 :			implemented by other Forth systems, so that we may one day arrive at a
3760 :			standardized interface. Such a standard interface would allow you to
3761 :			replace the Forth system without having to rewrite C code.
3762 :
3763 :			You embed the Gforth interpreter by linking with the library
3764 :			@code{libgforth.a} (give the compiler the option @code{-lgforth}). All
3765 :			global symbols in this library that belong to the interface, have the
3766 :			prefix @code{forth_}. (Global symbols that are used internally have the
3767 :			prefix @code{gforth_}).
3768 :
3769 :			You can include the declarations of Forth types and the functions and
3770 :	anton	1.36	variables of the interface with @code{#include <forth.h>}.
3771 :	anton	1.34
3772 :			Types.
3773 :	anton	1.13
3774 :	anton	1.34	Variables.
3775 :
3776 :			Data and FP Stack pointer. Area sizes.
3777 :
3778 :			functions.
3779 :
3780 :			forth_init(imagefile)
3781 :			forth_evaluate(string) exceptions?
3782 :			forth_goto(address) (or forth_execute(xt)?)
3783 :			forth_continue() (a corountining mechanism)
3784 :
3785 :			Adding primitives.
3786 :
3787 :			No checking.
3788 :
3789 :			Signals?
3790 :
3791 :			Accessing the Stacks
3792 :	anton	1.4
3793 :	anton	1.34	@node Emacs and Gforth, Internals, Integrating Gforth, Top
3794 :	anton	1.17	@chapter Emacs and Gforth
3795 :	anton	1.4
3796 :	anton	1.17	Gforth comes with @file{gforth.el}, an improved version of
3797 :	anton	1.33	@file{forth.el} by Goran Rydqvist (included in the TILE package). The
3798 :	anton	1.4	improvements are a better (but still not perfect) handling of
3799 :			indentation. I have also added comment paragraph filling (@kbd{M-q}),
3800 :	anton	1.8	commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3801 :			removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3802 :			stuff I do not use alone, even though some of it only makes sense for
3803 :			TILE. To get a description of these features, enter Forth mode and type
3804 :			@kbd{C-h m}.
3805 :	anton	1.4
3806 :	anton	1.17	In addition, Gforth supports Emacs quite well: The source code locations
3807 :	anton	1.4	given in error messages, debugging output (from @code{~~}) and failed
3808 :			assertion messages are in the right format for Emacs' compilation mode
3809 :			(@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3810 :			Manual}) so the source location corresponding to an error or other
3811 :			message is only a few keystrokes away (@kbd{C-x `} for the next error,
3812 :			@kbd{C-c C-c} for the error under the cursor).
3813 :
3814 :			Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3815 :			(@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3816 :			contains the definitions of all words defined afterwards. You can then
3817 :			find the source for a word using @kbd{M-.}. Note that emacs can use
3818 :	anton	1.17	several tags files at the same time (e.g., one for the Gforth sources
3819 :	anton	1.28	and one for your program, @pxref{Select Tags Table,,Selecting a Tags
3820 :			Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
3821 :			@file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
3822 :	anton	1.33	@file{/usr/local/share/gforth/0.2.0/TAGS}).
3823 :	anton	1.4
3824 :			To get all these benefits, add the following lines to your @file{.emacs}
3825 :			file:
3826 :
3827 :			@example
3828 :			(autoload 'forth-mode "gforth.el")
3829 :			(setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3830 :			@end example
3831 :
3832 :	anton	1.17	@node Internals, Bugs, Emacs and Gforth, Top
3833 :	anton	1.3	@chapter Internals
3834 :
3835 :	anton	1.17	Reading this section is not necessary for programming with Gforth. It
3836 :			should be helpful for finding your way in the Gforth sources.
3837 :	anton	1.3
3838 :	anton	1.24	The ideas in this section have also been published in the papers
3839 :			@cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
3840 :			the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
3841 :			Ertl, presented at EuroForth '93; the latter is available at
3842 :			@*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
3843 :
3844 :	anton	1.4	@menu
3845 :			* Portability::
3846 :			* Threading::
3847 :			* Primitives::
3848 :			* System Architecture::
3849 :	anton	1.17	* Performance::
3850 :	anton	1.4	@end menu
3851 :
3852 :			@node Portability, Threading, Internals, Internals
3853 :	anton	1.3	@section Portability
3854 :
3855 :			One of the main goals of the effort is availability across a wide range
3856 :			of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3857 :			this goal by manually coding the engine in assembly language for several
3858 :			then-popular processors. This approach is very labor-intensive and the
3859 :			results are short-lived due to progress in computer architecture.
3860 :
3861 :			Others have avoided this problem by coding in C, e.g., Mitch Bradley
3862 :			(cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3863 :			particularly popular for UNIX-based Forths due to the large variety of
3864 :			architectures of UNIX machines. Unfortunately an implementation in C
3865 :			does not mix well with the goals of efficiency and with using
3866 :			traditional techniques: Indirect or direct threading cannot be expressed
3867 :			in C, and switch threading, the fastest technique available in C, is
3868 :			significantly slower. Another problem with C is that it's very
3869 :			cumbersome to express double integer arithmetic.
3870 :
3871 :			Fortunately, there is a portable language that does not have these
3872 :			limitations: GNU C, the version of C processed by the GNU C compiler
3873 :			(@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3874 :			GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3875 :			Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3876 :			threading possible, its @code{long long} type (@pxref{Long Long, ,
3877 :	anton	1.33	Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
3878 :	anton	1.32	double numbers@footnote{Unfortunately, long longs are not implemented
3879 :			properly on all machines (e.g., on alpha-osf1, long longs are only 64
3880 :			bits, the same size as longs (and pointers), but they should be twice as
3881 :			long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
3882 :			C Manual}). So, we had to implement doubles in C after all. Still, on
3883 :			most machines we can use long longs and achieve better performance than
3884 :			with the emulation package.}. GNU C is available for free on all
3885 :			important (and many unimportant) UNIX machines, VMS, 80386s running
3886 :			MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
3887 :			on all these machines.
3888 :	anton	1.3
3889 :			Writing in a portable language has the reputation of producing code that
3890 :			is slower than assembly. For our Forth engine we repeatedly looked at
3891 :			the code produced by the compiler and eliminated most compiler-induced
3892 :			inefficiencies by appropriate changes in the source-code.
3893 :
3894 :			However, register allocation cannot be portably influenced by the
3895 :			programmer, leading to some inefficiencies on register-starved
3896 :			machines. We use explicit register declarations (@pxref{Explicit Reg
3897 :			Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3898 :			improve the speed on some machines. They are turned on by using the
3899 :			@code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3900 :			only depends on the machine, but also on the compiler version: On some
3901 :			machines some compiler versions produce incorrect code when certain
3902 :			explicit register declarations are used. So by default
3903 :			@code{-DFORCE_REG} is not used.
3904 :
3905 :	anton	1.4	@node Threading, Primitives, Portability, Internals
3906 :	anton	1.3	@section Threading
3907 :
3908 :			GNU C's labels as values extension (available since @code{gcc-2.0},
3909 :			@pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3910 :			makes it possible to take the address of @var{label} by writing
3911 :			@code{&&@var{label}}. This address can then be used in a statement like
3912 :			@code{goto @var{address}}. I.e., @code{goto &&x} is the same as
3913 :			@code{goto x}.
3914 :
3915 :			With this feature an indirect threaded NEXT looks like:
3916 :			@example
3917 :			cfa = *ip++;
3918 :			ca = *cfa;
3919 :			goto *ca;
3920 :			@end example
3921 :			For those unfamiliar with the names: @code{ip} is the Forth instruction
3922 :			pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3923 :			execution token and points to the code field of the next word to be
3924 :			executed; The @code{ca} (code address) fetched from there points to some
3925 :			executable code, e.g., a primitive or the colon definition handler
3926 :			@code{docol}.
3927 :
3928 :			Direct threading is even simpler:
3929 :			@example
3930 :			ca = *ip++;
3931 :			goto *ca;
3932 :			@end example
3933 :
3934 :			Of course we have packaged the whole thing neatly in macros called
3935 :			@code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3936 :
3937 :	anton	1.4	@menu
3938 :			* Scheduling::
3939 :			* Direct or Indirect Threaded?::
3940 :			* DOES>::
3941 :			@end menu
3942 :
3943 :			@node Scheduling, Direct or Indirect Threaded?, Threading, Threading
3944 :	anton	1.3	@subsection Scheduling
3945 :
3946 :			There is a little complication: Pipelined and superscalar processors,
3947 :			i.e., RISC and some modern CISC machines can process independent
3948 :			instructions while waiting for the results of an instruction. The
3949 :			compiler usually reorders (schedules) the instructions in a way that
3950 :			achieves good usage of these delay slots. However, on our first tries
3951 :			the compiler did not do well on scheduling primitives. E.g., for
3952 :			@code{+} implemented as
3953 :			@example
3954 :			n=sp[0]+sp[1];
3955 :			sp++;
3956 :			sp[0]=n;
3957 :			NEXT;
3958 :			@end example
3959 :			the NEXT comes strictly after the other code, i.e., there is nearly no
3960 :			scheduling. After a little thought the problem becomes clear: The
3961 :			compiler cannot know that sp and ip point to different addresses (and
3962 :	anton	1.4	the version of @code{gcc} we used would not know it even if it was
3963 :			possible), so it could not move the load of the cfa above the store to
3964 :			the TOS. Indeed the pointers could be the same, if code on or very near
3965 :			the top of stack were executed. In the interest of speed we chose to
3966 :			forbid this probably unused ``feature'' and helped the compiler in
3967 :			scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3968 :			the goto part (@code{NEXT_P2}). @code{+} now looks like:
3969 :	anton	1.3	@example
3970 :			n=sp[0]+sp[1];
3971 :			sp++;
3972 :			NEXT_P1;
3973 :			sp[0]=n;
3974 :			NEXT_P2;
3975 :			@end example
3976 :	anton	1.4	This can be scheduled optimally by the compiler.
3977 :	anton	1.3
3978 :			This division can be turned off with the switch @code{-DCISC_NEXT}. This
3979 :			switch is on by default on machines that do not profit from scheduling
3980 :			(e.g., the 80386), in order to preserve registers.
3981 :
3982 :	anton	1.4	@node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
3983 :	anton	1.3	@subsection Direct or Indirect Threaded?
3984 :
3985 :			Both! After packaging the nasty details in macro definitions we
3986 :			realized that we could switch between direct and indirect threading by
3987 :			simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3988 :			defining a few machine-specific macros for the direct-threading case.
3989 :			On the Forth level we also offer access words that hide the
3990 :			differences between the threading methods (@pxref{Threading Words}).
3991 :
3992 :			Indirect threading is implemented completely
3993 :			machine-independently. Direct threading needs routines for creating
3994 :			jumps to the executable code (e.g. to docol or dodoes). These routines
3995 :			are inherently machine-dependent, but they do not amount to many source
3996 :			lines. I.e., even porting direct threading to a new machine is a small
3997 :			effort.
3998 :
3999 :	anton	1.4	@node DOES>, , Direct or Indirect Threaded?, Threading
4000 :	anton	1.3	@subsection DOES>
4001 :			One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
4002 :			the chunk of code executed by every word defined by a
4003 :			@code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
4004 :			the Forth code to be executed, i.e. the code after the @code{DOES>} (the
4005 :			DOES-code)? There are two solutions:
4006 :
4007 :			In fig-Forth the code field points directly to the dodoes and the
4008 :			DOES-code address is stored in the cell after the code address
4009 :			(i.e. at cfa cell+). It may seem that this solution is illegal in the
4010 :			Forth-79 and all later standards, because in fig-Forth this address
4011 :			lies in the body (which is illegal in these standards). However, by
4012 :			making the code field larger for all words this solution becomes legal
4013 :			again. We use this approach for the indirect threaded version. Leaving
4014 :			a cell unused in most words is a bit wasteful, but on the machines we
4015 :			are targetting this is hardly a problem. The other reason for having a
4016 :			code field size of two cells is to avoid having different image files
4017 :	anton	1.4	for direct and indirect threaded systems (@pxref{System Architecture}).
4018 :	anton	1.3
4019 :			The other approach is that the code field points or jumps to the cell
4020 :			after @code{DOES}. In this variant there is a jump to @code{dodoes} at
4021 :			this address. @code{dodoes} can then get the DOES-code address by
4022 :			computing the code address, i.e., the address of the jump to dodoes,
4023 :			and add the length of that jump field. A variant of this is to have a
4024 :			call to @code{dodoes} after the @code{DOES>}; then the return address
4025 :			(which can be found in the return register on RISCs) is the DOES-code
4026 :			address. Since the two cells available in the code field are usually
4027 :			used up by the jump to the code address in direct threading, we use
4028 :			this approach for direct threading. We did not want to add another
4029 :			cell to the code field.
4030 :
4031 :	anton	1.4	@node Primitives, System Architecture, Threading, Internals
4032 :	anton	1.3	@section Primitives
4033 :
4034 :	anton	1.4	@menu
4035 :			* Automatic Generation::
4036 :			* TOS Optimization::
4037 :			* Produced code::
4038 :			@end menu
4039 :
4040 :			@node Automatic Generation, TOS Optimization, Primitives, Primitives
4041 :	anton	1.3	@subsection Automatic Generation
4042 :
4043 :			Since the primitives are implemented in a portable language, there is no
4044 :			longer any need to minimize the number of primitives. On the contrary,
4045 :			having many primitives is an advantage: speed. In order to reduce the
4046 :			number of errors in primitives and to make programming them easier, we
4047 :			provide a tool, the primitive generator (@file{prims2x.fs}), that
4048 :			automatically generates most (and sometimes all) of the C code for a
4049 :			primitive from the stack effect notation. The source for a primitive
4050 :			has the following form:
4051 :
4052 :			@format
4053 :			@var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
4054 :			[@code{""}@var{glossary entry}@code{""}]
4055 :			@var{C code}
4056 :			[@code{:}
4057 :			@var{Forth code}]
4058 :			@end format
4059 :
4060 :			The items in brackets are optional. The category and glossary fields
4061 :			are there for generating the documentation, the Forth code is there
4062 :			for manual implementations on machines without GNU C. E.g., the source
4063 :			for the primitive @code{+} is:
4064 :			@example
4065 :			+ n1 n2 -- n core plus
4066 :			n = n1+n2;
4067 :			@end example
4068 :
4069 :			This looks like a specification, but in fact @code{n = n1+n2} is C
4070 :			code. Our primitive generation tool extracts a lot of information from
4071 :			the stack effect notations@footnote{We use a one-stack notation, even
4072 :			though we have separate data and floating-point stacks; The separate
4073 :			notation can be generated easily from the unified notation.}: The number
4074 :			of items popped from and pushed on the stack, their type, and by what
4075 :			name they are referred to in the C code. It then generates a C code
4076 :			prelude and postlude for each primitive. The final C code for @code{+}
4077 :			looks like this:
4078 :
4079 :			@example
4080 :			I_plus: /* + ( n1 n2 -- n ) / / label, stack effect */
4081 :			/* / / documentation */
4082 :	anton	1.4	@{
4083 :	anton	1.3	DEF_CA /* definition of variable ca (indirect threading) */
4084 :			Cell n1; /* definitions of variables */
4085 :			Cell n2;
4086 :			Cell n;
4087 :			n1 = (Cell) sp[1]; /* input */
4088 :			n2 = (Cell) TOS;
4089 :			sp += 1; /* stack adjustment */
4090 :			NAME("+") /* debugging output (with -DDEBUG) */
4091 :	anton	1.4	@{
4092 :	anton	1.3	n = n1+n2; /* C code taken from the source */
4093 :	anton	1.4	@}
4094 :	anton	1.3	NEXT_P1; /* NEXT part 1 */
4095 :			TOS = (Cell)n; /* output */
4096 :			NEXT_P2; /* NEXT part 2 */
4097 :	anton	1.4	@}
4098 :	anton	1.3	@end example
4099 :
4100 :			This looks long and inefficient, but the GNU C compiler optimizes quite
4101 :			well and produces optimal code for @code{+} on, e.g., the R3000 and the
4102 :			HP RISC machines: Defining the @code{n}s does not produce any code, and
4103 :			using them as intermediate storage also adds no cost.
4104 :
4105 :			There are also other optimizations, that are not illustrated by this
4106 :			example: Assignments between simple variables are usually for free (copy
4107 :			propagation). If one of the stack items is not used by the primitive
4108 :			(e.g. in @code{drop}), the compiler eliminates the load from the stack
4109 :			(dead code elimination). On the other hand, there are some things that
4110 :			the compiler does not do, therefore they are performed by
4111 :			@file{prims2x.fs}: The compiler does not optimize code away that stores
4112 :			a stack item to the place where it just came from (e.g., @code{over}).
4113 :
4114 :			While programming a primitive is usually easy, there are a few cases
4115 :			where the programmer has to take the actions of the generator into
4116 :			account, most notably @code{?dup}, but also words that do not (always)
4117 :			fall through to NEXT.
4118 :
4119 :	anton	1.4	@node TOS Optimization, Produced code, Automatic Generation, Primitives
4120 :	anton	1.3	@subsection TOS Optimization
4121 :
4122 :			An important optimization for stack machine emulators, e.g., Forth
4123 :			engines, is keeping one or more of the top stack items in
4124 :	anton	1.4	registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
4125 :			@var{out1}...@var{outy}, keeping the top @var{n} items in registers
4126 :	anton	1.34	@itemize @bullet
4127 :	anton	1.3	@item
4128 :			is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
4129 :			due to fewer loads from and stores to the stack.
4130 :			@item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
4131 :			@var{y<n}, due to additional moves between registers.
4132 :			@end itemize
4133 :
4134 :			In particular, keeping one item in a register is never a disadvantage,
4135 :			if there are enough registers. Keeping two items in registers is a
4136 :			disadvantage for frequent words like @code{?branch}, constants,
4137 :			variables, literals and @code{i}. Therefore our generator only produces
4138 :			code that keeps zero or one items in registers. The generated C code
4139 :			covers both cases; the selection between these alternatives is made at
4140 :			C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
4141 :			code for @code{+} is just a simple variable name in the one-item case,
4142 :			otherwise it is a macro that expands into @code{sp[0]}. Note that the
4143 :			GNU C compiler tries to keep simple variables like @code{TOS} in
4144 :			registers, and it usually succeeds, if there are enough registers.
4145 :
4146 :			The primitive generator performs the TOS optimization for the
4147 :			floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
4148 :			operations the benefit of this optimization is even larger:
4149 :			floating-point operations take quite long on most processors, but can be
4150 :			performed in parallel with other operations as long as their results are
4151 :			not used. If the FP-TOS is kept in a register, this works. If
4152 :			it is kept on the stack, i.e., in memory, the store into memory has to
4153 :			wait for the result of the floating-point operation, lengthening the
4154 :			execution time of the primitive considerably.
4155 :
4156 :			The TOS optimization makes the automatic generation of primitives a
4157 :			bit more complicated. Just replacing all occurrences of @code{sp[0]} by
4158 :			@code{TOS} is not sufficient. There are some special cases to
4159 :			consider:
4160 :	anton	1.34	@itemize @bullet
4161 :	anton	1.3	@item In the case of @code{dup ( w -- w w )} the generator must not
4162 :			eliminate the store to the original location of the item on the stack,
4163 :			if the TOS optimization is turned on.
4164 :	anton	1.4	@item Primitives with stack effects of the form @code{--}
4165 :			@var{out1}...@var{outy} must store the TOS to the stack at the start.
4166 :			Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
4167 :	anton	1.3	must load the TOS from the stack at the end. But for the null stack
4168 :			effect @code{--} no stores or loads should be generated.
4169 :			@end itemize
4170 :
4171 :	anton	1.4	@node Produced code, , TOS Optimization, Primitives
4172 :	anton	1.3	@subsection Produced code
4173 :
4174 :			To see what assembly code is produced for the primitives on your machine
4175 :			with your compiler and your flag settings, type @code{make engine.s} and
4176 :	anton	1.4	look at the resulting file @file{engine.s}.
4177 :	anton	1.3
4178 :	anton	1.17	@node System Architecture, Performance, Primitives, Internals
4179 :	anton	1.3	@section System Architecture
4180 :
4181 :			Our Forth system consists not only of primitives, but also of
4182 :			definitions written in Forth. Since the Forth compiler itself belongs
4183 :			to those definitions, it is not possible to start the system with the
4184 :			primitives and the Forth source alone. Therefore we provide the Forth
4185 :			code as an image file in nearly executable form. At the start of the
4186 :			system a C routine loads the image file into memory, sets up the
4187 :			memory (stacks etc.) according to information in the image file, and
4188 :			starts executing Forth code.
4189 :
4190 :			The image file format is a compromise between the goals of making it
4191 :			easy to generate image files and making them portable. The easiest way
4192 :			to generate an image file is to just generate a memory dump. However,
4193 :			this kind of image file cannot be used on a different machine, or on
4194 :			the next version of the engine on the same machine, it even might not
4195 :			work with the same engine compiled by a different version of the C
4196 :			compiler. We would like to have as few versions of the image file as
4197 :			possible, because we do not want to distribute many versions of the
4198 :			same image file, and to make it easy for the users to use their image
4199 :			files on many machines. We currently need to create a different image
4200 :			file for machines with different cell sizes and different byte order
4201 :	anton	1.17	(little- or big-endian)@footnote{We are considering adding information to the
4202 :	anton	1.3	image file that enables the loader to change the byte order.}.
4203 :
4204 :			Forth code that is going to end up in a portable image file has to
4205 :	anton	1.4	comply to some restrictions: addresses have to be stored in memory with
4206 :			special words (@code{A!}, @code{A,}, etc.) in order to make the code
4207 :			relocatable. Cells, floats, etc., have to be stored at the natural
4208 :			alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
4209 :			dividable by~8. This happens automatically in our system when you use
4210 :			the ANS Forth alignment words.}, in order to avoid alignment faults on
4211 :			machines with stricter alignment. The image file is produced by a
4212 :			metacompiler (@file{cross.fs}).
4213 :	anton	1.3
4214 :			So, unlike the image file of Mitch Bradleys @code{cforth}, our image
4215 :			file is not directly executable, but has to undergo some manipulations
4216 :			during loading. Address relocation is performed at image load-time, not
4217 :			at run-time. The loader also has to replace tokens standing for
4218 :			primitive calls with the appropriate code-field addresses (or code
4219 :			addresses in the case of direct threading).
4220 :	anton	1.4
4221 :	anton	1.17	@node Performance, , System Architecture, Internals
4222 :			@section Performance
4223 :
4224 :			On RISCs the Gforth engine is very close to optimal; i.e., it is usually
4225 :			impossible to write a significantly faster engine.
4226 :
4227 :			On register-starved machines like the 386 architecture processors
4228 :			improvements are possible, because @code{gcc} does not utilize the
4229 :			registers as well as a human, even with explicit register declarations;
4230 :			e.g., Bernd Beuster wrote a Forth system fragment in assembly language
4231 :			and hand-tuned it for the 486; this system is 1.19 times faster on the
4232 :			Sieve benchmark on a 486DX2/66 than Gforth compiled with
4233 :			@code{gcc-2.6.3} with @code{-DFORCE_REG}.
4234 :
4235 :			However, this potential advantage of assembly language implementations
4236 :			is not necessarily realized in complete Forth systems: We compared
4237 :	anton	1.26	Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
4238 :			@code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
4239 :			1994) and Eforth (with and without peephole (aka pinhole) optimization
4240 :			of the threaded code); all these systems were written in assembly
4241 :	anton	1.30	language. We also compared Gforth with three systems written in C:
4242 :	anton	1.32	PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
4243 :			configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
4244 :			-DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
4245 :			-fomit-frame-pointer; ThisForth employs peephole optimization of the
4246 :			threaded code) and TILE (compiled with @code{make opt}). We benchmarked
4247 :			Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
4248 :			O'Heskin kindly provided the results for Win32Forth and NT Forth on a
4249 :			486DX2/66 with similar memory performance under Windows NT. Marcel
4250 :			Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
4251 :			added the peephole optimizer, ran the benchmarks and reported the
4252 :			results.
4253 :	anton	1.17
4254 :			We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
4255 :			matrix multiplication come from the Stanford integer benchmarks and have
4256 :			been translated into Forth by Martin Fraeman; we used the versions
4257 :	anton	1.30	included in the TILE Forth package, but with bigger data set sizes; and
4258 :			a recursive Fibonacci number computation for benchmarking calling
4259 :			performance. The following table shows the time taken for the benchmarks
4260 :			scaled by the time taken by Gforth (in other words, it shows the speedup
4261 :			factor that Gforth achieved over the other systems).
4262 :	anton	1.17
4263 :			@example
4264 :	anton	1.30	relative Win32- NT eforth This-
4265 :			time Gforth Forth Forth eforth +opt PFE Forth TILE
4266 :	anton	1.32	sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
4267 :			bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
4268 :	anton	1.38	matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
4269 :			fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
4270 :	anton	1.17	@end example
4271 :
4272 :			You may find the good performance of Gforth compared with the systems
4273 :			written in assembly language quite surprising. One important reason for
4274 :			the disappointing performance of these systems is probably that they are
4275 :			not written optimally for the 486 (e.g., they use the @code{lods}
4276 :			instruction). In addition, Win32Forth uses a comfortable, but costly
4277 :			method for relocating the Forth image: like @code{cforth}, it computes
4278 :			the actual addresses at run time, resulting in two address computations
4279 :			per NEXT (@pxref{System Architecture}).
4280 :
4281 :	anton	1.26	Only Eforth with the peephole optimizer performs comparable to
4282 :			Gforth. The speedups achieved with peephole optimization of threaded
4283 :			code are quite remarkable. Adding a peephole optimizer to Gforth should
4284 :			cause similar speedups.
4285 :
4286 :	anton	1.30	The speedup of Gforth over PFE, ThisForth and TILE can be easily
4287 :			explained with the self-imposed restriction to standard C, which makes
4288 :			efficient threading impossible (however, the measured implementation of
4289 :			PFE uses a GNU C extension: @ref{Global Reg Vars, , Defining Global
4290 :			Register Variables, gcc.info, GNU C Manual}). Moreover, current C
4291 :			compilers have a hard time optimizing other aspects of the ThisForth
4292 :			and the TILE source.
4293 :	anton	1.17
4294 :			Note that the performance of Gforth on 386 architecture processors
4295 :			varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
4296 :			failed to allocate any of the virtual machine registers into real
4297 :			machine registers by itself and would not work correctly with explicit
4298 :			register declarations, giving a 1.3 times slower engine (on a 486DX2/66
4299 :			running the Sieve) than the one measured above.
4300 :
4301 :	anton	1.26	In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
4302 :			Maierhofer (presented at EuroForth '95), an indirect threaded version of
4303 :			Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
4304 :			version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
4305 :			used here. The paper available at
4306 :	anton	1.24	@*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
4307 :			it also contains numbers for some native code systems. You can find
4308 :			numbers for Gforth on various machines in @file{Benchres}.
4309 :
4310 :	anton	1.29	@node Bugs, Origin, Internals, Top
4311 :	anton	1.4	@chapter Bugs
4312 :
4313 :	anton	1.17	Known bugs are described in the file BUGS in the Gforth distribution.
4314 :
4315 :	anton	1.24	If you find a bug, please send a bug report to
4316 :	anton	1.32	@code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
4317 :	anton	1.17	describe the Gforth version used (it is announced at the start of an
4318 :			interactive Gforth session), the machine and operating system (on Unix
4319 :			systems you can use @code{uname -a} to produce this information), the
4320 :	anton	1.24	installation options (send the @code{config.status} file), and a
4321 :			complete list of changes you (or your installer) have made to the Gforth
4322 :			sources (if any); it should contain a program (or a sequence of keyboard
4323 :			commands) that reproduces the bug and a description of what you think
4324 :			constitutes the buggy behaviour.
4325 :	anton	1.17
4326 :			For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
4327 :			to Report Bugs, gcc.info, GNU C Manual}.
4328 :
4329 :
4330 :	anton	1.29	@node Origin, Word Index, Bugs, Top
4331 :			@chapter Authors and Ancestors of Gforth
4332 :
4333 :			@section Authors and Contributors
4334 :
4335 :			The Gforth project was started in mid-1992 by Bernd Paysan and Anton
4336 :	anton	1.30	Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
4337 :			one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
4338 :			with their continuous feedback. Lennart Benshop contributed
4339 :	anton	1.29	@file{glosgen.fs}, while Stuart Ramsden has been working on automatic
4340 :			support for calling C libraries. Helpful comments also came from Paul
4341 :	anton	1.37	Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
4342 :	anton	1.39	Wavrik, Barrie Stott and Marc de Groot.
4343 :	anton	1.29
4344 :	anton	1.30	Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
4345 :			and autoconf, among others), and to the creators of the Internet: Gforth
4346 :			was developed across the Internet, and its authors have not met
4347 :			physically yet.
4348 :
4349 :	anton	1.29	@section Pedigree
4350 :	anton	1.4
4351 :	anton	1.17	Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
4352 :	anton	1.24	Dirk Zoller) will cross-fertilize each other. Of course, a significant
4353 :			part of the design of Gforth was prescribed by ANS Forth.
4354 :	anton	1.17
4355 :	pazsan	1.23	Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
4356 :			32 bit native code version of VolksForth for the Atari ST, written
4357 :			mostly by Dietrich Weineck.
4358 :
4359 :			VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
4360 :			Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
4361 :	anton	1.24	UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
4362 :	anton	1.17
4363 :	anton	1.34	Henry Laxen and Mike Perry wrote F83 as a model implementation of the
4364 :	anton	1.17	Forth-83 standard. !! Pedigree? When?
4365 :
4366 :			A team led by Bill Ragsdale implemented fig-Forth on many processors in
4367 :	anton	1.24	1979. Robert Selzer and Bill Ragsdale developed the original
4368 :			implementation of fig-Forth for the 6502 based on microForth.
4369 :
4370 :			The principal architect of microForth was Dean Sanderson. microForth was
4371 :			FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
4372 :			the 1802, and subsequently implemented on the 8080, the 6800 and the
4373 :			Z80.
4374 :	anton	1.17
4375 :	anton	1.24	All earlier Forth systems were custom-made, usually by Charles Moore,
4376 :	anton	1.30	who discovered (as he puts it) Forth during the late 60s. The first full
4377 :			Forth existed in 1971.
4378 :	anton	1.17
4379 :			A part of the information in this section comes from @cite{The Evolution
4380 :			of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
4381 :			H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
4382 :			Notices 28(3), 1993. You can find more historical and genealogical
4383 :			information about Forth there.
4384 :
4385 :	anton	1.29	@node Word Index, Node Index, Origin, Top
4386 :	anton	1.4	@chapter Word Index
4387 :
4388 :	anton	1.18	This index is as incomplete as the manual. Each word is listed with
4389 :			stack effect and wordset.
4390 :	anton	1.17
4391 :			@printindex fn
4392 :
4393 :	anton	1.4	@node Node Index, , Word Index, Top
4394 :			@chapter Node Index
4395 :	anton	1.17
4396 :			This index is even less complete than the manual.
4397 :	anton	1.1
4398 :			@contents
4399 :			@bye
4400 :

CVS Admin

gforth: gforth/Attic/gforth.ds

ViewCVS and CVS Help