|
428
|
1 \input texinfo @c -*-texinfo-*-
|
|
|
2 @c %**start of header
|
|
|
3 @setfilename ../../info/internals.info
|
|
|
4 @settitle XEmacs Internals Manual
|
|
|
5 @c %**end of header
|
|
|
6
|
|
|
7 @ifinfo
|
|
|
8 @dircategory XEmacs Editor
|
|
|
9 @direntry
|
|
440
|
10 * Internals: (internals). XEmacs Internals Manual.
|
|
428
|
11 @end direntry
|
|
|
12
|
|
|
13 Copyright @copyright{} 1992 - 1996 Ben Wing.
|
|
|
14 Copyright @copyright{} 1996, 1997 Sun Microsystems.
|
|
|
15 Copyright @copyright{} 1994 - 1998 Free Software Foundation.
|
|
|
16 Copyright @copyright{} 1994, 1995 Board of Trustees, University of Illinois.
|
|
|
17
|
|
|
18
|
|
|
19 Permission is granted to make and distribute verbatim copies of this
|
|
|
20 manual provided the copyright notice and this permission notice are
|
|
|
21 preserved on all copies.
|
|
|
22
|
|
|
23 @ignore
|
|
|
24 Permission is granted to process this file through TeX and print the
|
|
|
25 results, provided the printed document carries copying permission notice
|
|
|
26 identical to this one except for the removal of this paragraph (this
|
|
|
27 paragraph not being relevant to the printed manual).
|
|
|
28
|
|
|
29 @end ignore
|
|
|
30 Permission is granted to copy and distribute modified versions of this
|
|
|
31 manual under the conditions for verbatim copying, provided that the
|
|
|
32 entire resulting derived work is distributed under the terms of a
|
|
|
33 permission notice identical to this one.
|
|
|
34
|
|
|
35 Permission is granted to copy and distribute translations of this manual
|
|
|
36 into another language, under the above conditions for modified versions,
|
|
|
37 except that this permission notice may be stated in a translation
|
|
|
38 approved by the Foundation.
|
|
|
39
|
|
|
40 Permission is granted to copy and distribute modified versions of this
|
|
|
41 manual under the conditions for verbatim copying, provided also that the
|
|
|
42 section entitled ``GNU General Public License'' is included exactly as
|
|
|
43 in the original, and provided that the entire resulting derived work is
|
|
|
44 distributed under the terms of a permission notice identical to this
|
|
|
45 one.
|
|
|
46
|
|
|
47 Permission is granted to copy and distribute translations of this manual
|
|
|
48 into another language, under the above conditions for modified versions,
|
|
|
49 except that the section entitled ``GNU General Public License'' may be
|
|
|
50 included in a translation approved by the Free Software Foundation
|
|
|
51 instead of in the original English.
|
|
|
52 @end ifinfo
|
|
|
53
|
|
|
54 @c Combine indices.
|
|
|
55 @synindex cp fn
|
|
|
56 @syncodeindex vr fn
|
|
|
57 @syncodeindex ky fn
|
|
|
58 @syncodeindex pg fn
|
|
|
59 @syncodeindex tp fn
|
|
|
60
|
|
|
61 @setchapternewpage odd
|
|
|
62 @finalout
|
|
|
63
|
|
|
64 @titlepage
|
|
|
65 @title XEmacs Internals Manual
|
|
|
66 @subtitle Version 1.3, August 1999
|
|
|
67
|
|
|
68 @author Ben Wing
|
|
|
69 @author Martin Buchholz
|
|
|
70 @author Hrvoje Niksic
|
|
|
71 @author Matthias Neubauer
|
|
|
72 @page
|
|
|
73 @vskip 0pt plus 1fill
|
|
|
74
|
|
|
75 @noindent
|
|
|
76 Copyright @copyright{} 1992 - 1996 Ben Wing. @*
|
|
|
77 Copyright @copyright{} 1996, 1997 Sun Microsystems, Inc. @*
|
|
|
78 Copyright @copyright{} 1994 - 1998 Free Software Foundation. @*
|
|
|
79 Copyright @copyright{} 1994, 1995 Board of Trustees, University of Illinois.
|
|
|
80
|
|
|
81 @sp 2
|
|
|
82 Version 1.3 @*
|
|
|
83 August 1999.@*
|
|
|
84
|
|
|
85 Permission is granted to make and distribute verbatim copies of this
|
|
|
86 manual provided the copyright notice and this permission notice are
|
|
|
87 preserved on all copies.
|
|
|
88
|
|
|
89 Permission is granted to copy and distribute modified versions of this
|
|
|
90 manual under the conditions for verbatim copying, provided also that the
|
|
|
91 section entitled ``GNU General Public License'' is included
|
|
|
92 exactly as in the original, and provided that the entire resulting
|
|
|
93 derived work is distributed under the terms of a permission notice
|
|
|
94 identical to this one.
|
|
|
95
|
|
|
96 Permission is granted to copy and distribute translations of this manual
|
|
|
97 into another language, under the above conditions for modified versions,
|
|
|
98 except that the section entitled ``GNU General Public License'' may be
|
|
|
99 included in a translation approved by the Free Software Foundation
|
|
|
100 instead of in the original English.
|
|
|
101 @end titlepage
|
|
|
102 @page
|
|
|
103
|
|
|
104 @node Top, A History of Emacs, (dir), (dir)
|
|
|
105
|
|
|
106 @ifinfo
|
|
|
107 This Info file contains v1.0 of the XEmacs Internals Manual.
|
|
|
108 @end ifinfo
|
|
|
109
|
|
|
110 @menu
|
|
|
111 * A History of Emacs:: Times, dates, important events.
|
|
|
112 * XEmacs From the Outside:: A broad conceptual overview.
|
|
|
113 * The Lisp Language:: An overview.
|
|
|
114 * XEmacs From the Perspective of Building::
|
|
|
115 * XEmacs From the Inside::
|
|
|
116 * The XEmacs Object System (Abstractly Speaking)::
|
|
|
117 * How Lisp Objects Are Represented in C::
|
|
|
118 * Rules When Writing New C Code::
|
|
|
119 * A Summary of the Various XEmacs Modules::
|
|
|
120 * Allocation of Objects in XEmacs Lisp::
|
|
|
121 * Events and the Event Loop::
|
|
|
122 * Evaluation; Stack Frames; Bindings::
|
|
|
123 * Symbols and Variables::
|
|
|
124 * Buffers and Textual Representation::
|
|
|
125 * MULE Character Sets and Encodings::
|
|
|
126 * The Lisp Reader and Compiler::
|
|
|
127 * Lstreams::
|
|
|
128 * Consoles; Devices; Frames; Windows::
|
|
|
129 * The Redisplay Mechanism::
|
|
|
130 * Extents::
|
|
|
131 * Faces::
|
|
|
132 * Glyphs::
|
|
|
133 * Specifiers::
|
|
|
134 * Menus::
|
|
|
135 * Subprocesses::
|
|
|
136 * Interface to X Windows::
|
|
|
137 * Index:: Index including concepts, functions, variables,
|
|
|
138 and other terms.
|
|
|
139
|
|
|
140 --- The Detailed Node Listing ---
|
|
|
141
|
|
|
142 Here are other nodes that are inferiors of those already listed,
|
|
|
143 mentioned here so you can get to them in one step:
|
|
|
144
|
|
|
145 A History of Emacs
|
|
|
146
|
|
|
147 * Through Version 18:: Unification prevails.
|
|
|
148 * Lucid Emacs:: One version 19 Emacs.
|
|
|
149 * GNU Emacs 19:: The other version 19 Emacs.
|
|
|
150 * XEmacs:: The continuation of Lucid Emacs.
|
|
|
151
|
|
|
152 Rules When Writing New C Code
|
|
|
153
|
|
|
154 * General Coding Rules::
|
|
|
155 * Writing Lisp Primitives::
|
|
|
156 * Adding Global Lisp Variables::
|
|
|
157 * Techniques for XEmacs Developers::
|
|
|
158
|
|
|
159 A Summary of the Various XEmacs Modules
|
|
|
160
|
|
|
161 * Low-Level Modules::
|
|
|
162 * Basic Lisp Modules::
|
|
|
163 * Modules for Standard Editing Operations::
|
|
|
164 * Editor-Level Control Flow Modules::
|
|
|
165 * Modules for the Basic Displayable Lisp Objects::
|
|
|
166 * Modules for other Display-Related Lisp Objects::
|
|
|
167 * Modules for the Redisplay Mechanism::
|
|
|
168 * Modules for Interfacing with the File System::
|
|
|
169 * Modules for Other Aspects of the Lisp Interpreter and Object System::
|
|
|
170 * Modules for Interfacing with the Operating System::
|
|
|
171 * Modules for Interfacing with X Windows::
|
|
|
172 * Modules for Internationalization::
|
|
|
173
|
|
|
174 Allocation of Objects in XEmacs Lisp
|
|
|
175
|
|
|
176 * Introduction to Allocation::
|
|
|
177 * Garbage Collection::
|
|
|
178 * GCPROing::
|
|
|
179 * Garbage Collection - Step by Step::
|
|
|
180 * Integers and Characters::
|
|
|
181 * Allocation from Frob Blocks::
|
|
|
182 * lrecords::
|
|
|
183 * Low-level allocation::
|
|
|
184 * Pure Space::
|
|
|
185 * Cons::
|
|
|
186 * Vector::
|
|
|
187 * Bit Vector::
|
|
|
188 * Symbol::
|
|
|
189 * Marker::
|
|
|
190 * String::
|
|
|
191 * Compiled Function::
|
|
|
192
|
|
|
193 Events and the Event Loop
|
|
|
194
|
|
|
195 * Introduction to Events::
|
|
|
196 * Main Loop::
|
|
|
197 * Specifics of the Event Gathering Mechanism::
|
|
|
198 * Specifics About the Emacs Event::
|
|
|
199 * The Event Stream Callback Routines::
|
|
|
200 * Other Event Loop Functions::
|
|
|
201 * Converting Events::
|
|
|
202 * Dispatching Events; The Command Builder::
|
|
|
203
|
|
|
204 Evaluation; Stack Frames; Bindings
|
|
|
205
|
|
|
206 * Evaluation::
|
|
|
207 * Dynamic Binding; The specbinding Stack; Unwind-Protects::
|
|
|
208 * Simple Special Forms::
|
|
|
209 * Catch and Throw::
|
|
|
210
|
|
|
211 Symbols and Variables
|
|
|
212
|
|
|
213 * Introduction to Symbols::
|
|
|
214 * Obarrays::
|
|
|
215 * Symbol Values::
|
|
|
216
|
|
|
217 Buffers and Textual Representation
|
|
|
218
|
|
|
219 * Introduction to Buffers:: A buffer holds a block of text such as a file.
|
|
|
220 * The Text in a Buffer:: Representation of the text in a buffer.
|
|
|
221 * Buffer Lists:: Keeping track of all buffers.
|
|
|
222 * Markers and Extents:: Tagging locations within a buffer.
|
|
|
223 * Bufbytes and Emchars:: Representation of individual characters.
|
|
|
224 * The Buffer Object:: The Lisp object corresponding to a buffer.
|
|
|
225
|
|
|
226 MULE Character Sets and Encodings
|
|
|
227
|
|
|
228 * Character Sets::
|
|
|
229 * Encodings::
|
|
|
230 * Internal Mule Encodings::
|
|
|
231
|
|
|
232 Encodings
|
|
|
233
|
|
|
234 * Japanese EUC (Extended Unix Code)::
|
|
|
235 * JIS7::
|
|
|
236
|
|
|
237 Internal Mule Encodings
|
|
|
238
|
|
|
239 * Internal String Encoding::
|
|
|
240 * Internal Character Encoding::
|
|
|
241
|
|
|
242 The Lisp Reader and Compiler
|
|
|
243
|
|
|
244 Lstreams
|
|
|
245
|
|
|
246 Consoles; Devices; Frames; Windows
|
|
|
247
|
|
|
248 * Introduction to Consoles; Devices; Frames; Windows::
|
|
|
249 * Point::
|
|
|
250 * Window Hierarchy::
|
|
|
251
|
|
|
252 The Redisplay Mechanism
|
|
|
253
|
|
|
254 * Critical Redisplay Sections::
|
|
|
255 * Line Start Cache::
|
|
|
256
|
|
|
257 Extents
|
|
|
258
|
|
|
259 * Introduction to Extents:: Extents are ranges over text, with properties.
|
|
|
260 * Extent Ordering:: How extents are ordered internally.
|
|
|
261 * Format of the Extent Info:: The extent information in a buffer or string.
|
|
|
262 * Zero-Length Extents:: A weird special case.
|
|
|
263 * Mathematics of Extent Ordering:: A rigorous foundation.
|
|
|
264 * Extent Fragments:: Cached information useful for redisplay.
|
|
|
265
|
|
|
266 Faces
|
|
|
267
|
|
|
268 Glyphs
|
|
|
269
|
|
|
270 Specifiers
|
|
|
271
|
|
|
272 Menus
|
|
|
273
|
|
|
274 Subprocesses
|
|
|
275
|
|
|
276 Interface to X Windows
|
|
|
277
|
|
|
278 @end menu
|
|
|
279
|
|
|
280 @node A History of Emacs, XEmacs From the Outside, Top, Top
|
|
|
281 @chapter A History of Emacs
|
|
|
282 @cindex history of Emacs
|
|
|
283 @cindex Hackers (Steven Levy)
|
|
|
284 @cindex Levy, Steven
|
|
|
285 @cindex ITS (Incompatible Timesharing System)
|
|
|
286 @cindex Stallman, Richard
|
|
|
287 @cindex RMS
|
|
|
288 @cindex MIT
|
|
|
289 @cindex TECO
|
|
|
290 @cindex FSF
|
|
|
291 @cindex Free Software Foundation
|
|
|
292
|
|
|
293 XEmacs is a powerful, customizable text editor and development
|
|
|
294 environment. It began as Lucid Emacs, which was in turn derived from
|
|
|
295 GNU Emacs, a program written by Richard Stallman of the Free Software
|
|
|
296 Foundation. GNU Emacs dates back to the 1970's, and was modelled
|
|
|
297 after a package called ``Emacs'', written in 1976, that was a set of
|
|
|
298 macros on top of TECO, an old, old text editor written at MIT on the
|
|
|
299 DEC PDP 10 under one of the earliest time-sharing operating systems,
|
|
|
300 ITS (Incompatible Timesharing System). (ITS dates back well before
|
|
|
301 Unix.) ITS, TECO, and Emacs were products of a group of people at MIT
|
|
|
302 who called themselves ``hackers'', who shared an idealistic belief
|
|
|
303 system about the free exchange of information and were fanatical in
|
|
|
304 their devotion to and time spent with computers. (The hacker
|
|
|
305 subculture dates back to the late 1950's at MIT and is described in
|
|
|
306 detail in Steven Levy's book @cite{Hackers}. This book also includes
|
|
|
307 a lot of information about Stallman himself and the development of
|
|
|
308 Lisp, a programming language developed at MIT that underlies Emacs.)
|
|
|
309
|
|
|
310 @menu
|
|
|
311 * Through Version 18:: Unification prevails.
|
|
|
312 * Lucid Emacs:: One version 19 Emacs.
|
|
|
313 * GNU Emacs 19:: The other version 19 Emacs.
|
|
|
314 * GNU Emacs 20:: The other version 20 Emacs.
|
|
|
315 * XEmacs:: The continuation of Lucid Emacs.
|
|
|
316 @end menu
|
|
|
317
|
|
|
318 @node Through Version 18
|
|
|
319 @section Through Version 18
|
|
|
320 @cindex Gosling, James
|
|
|
321 @cindex Great Usenet Renaming
|
|
|
322
|
|
|
323 Although the history of the early versions of GNU Emacs is unclear,
|
|
|
324 the history is well-known from the middle of 1985. A time line is:
|
|
|
325
|
|
|
326 @itemize @bullet
|
|
|
327 @item
|
|
|
328 GNU Emacs version 15 (15.34) was released sometime in 1984 or 1985 and
|
|
|
329 shared some code with a version of Emacs written by James Gosling (the
|
|
|
330 same James Gosling who later created the Java language).
|
|
|
331 @item
|
|
|
332 GNU Emacs version 16 (first released version was 16.56) was released on
|
|
|
333 July 15, 1985. All Gosling code was removed due to potential copyright
|
|
|
334 problems with the code.
|
|
|
335 @item
|
|
|
336 version 16.57: released on September 16, 1985.
|
|
|
337 @item
|
|
|
338 versions 16.58, 16.59: released on September 17, 1985.
|
|
|
339 @item
|
|
|
340 version 16.60: released on September 19, 1985. These later version 16's
|
|
|
341 incorporated patches from the net, esp. for getting Emacs to work under
|
|
|
342 System V.
|
|
|
343 @item
|
|
|
344 version 17.36 (first official v17 release) released on December 20,
|
|
|
345 1985. Included a TeX-able user manual. First official unpatched
|
|
|
346 version that worked on vanilla System V machines.
|
|
|
347 @item
|
|
|
348 version 17.43 (second official v17 release) released on January 25,
|
|
|
349 1986.
|
|
|
350 @item
|
|
|
351 version 17.45 released on January 30, 1986.
|
|
|
352 @item
|
|
|
353 version 17.46 released on February 4, 1986.
|
|
|
354 @item
|
|
|
355 version 17.48 released on February 10, 1986.
|
|
|
356 @item
|
|
|
357 version 17.49 released on February 12, 1986.
|
|
|
358 @item
|
|
|
359 version 17.55 released on March 18, 1986.
|
|
|
360 @item
|
|
|
361 version 17.57 released on March 27, 1986.
|
|
|
362 @item
|
|
|
363 version 17.58 released on April 4, 1986.
|
|
|
364 @item
|
|
|
365 version 17.61 released on April 12, 1986.
|
|
|
366 @item
|
|
|
367 version 17.63 released on May 7, 1986.
|
|
|
368 @item
|
|
|
369 version 17.64 released on May 12, 1986.
|
|
|
370 @item
|
|
|
371 version 18.24 (a beta version) released on October 2, 1986.
|
|
|
372 @item
|
|
|
373 version 18.30 (a beta version) released on November 15, 1986.
|
|
|
374 @item
|
|
|
375 version 18.31 (a beta version) released on November 23, 1986.
|
|
|
376 @item
|
|
|
377 version 18.32 (a beta version) released on December 7, 1986.
|
|
|
378 @item
|
|
|
379 version 18.33 (a beta version) released on December 12, 1986.
|
|
|
380 @item
|
|
|
381 version 18.35 (a beta version) released on January 5, 1987.
|
|
|
382 @item
|
|
|
383 version 18.36 (a beta version) released on January 21, 1987.
|
|
|
384 @item
|
|
|
385 January 27, 1987: The Great Usenet Renaming. net.emacs is now
|
|
|
386 comp.emacs.
|
|
|
387 @item
|
|
|
388 version 18.37 (a beta version) released on February 12, 1987.
|
|
|
389 @item
|
|
|
390 version 18.38 (a beta version) released on March 3, 1987.
|
|
|
391 @item
|
|
|
392 version 18.39 (a beta version) released on March 14, 1987.
|
|
|
393 @item
|
|
|
394 version 18.40 (a beta version) released on March 18, 1987.
|
|
|
395 @item
|
|
|
396 version 18.41 (the first ``official'' release) released on March 22,
|
|
|
397 1987.
|
|
|
398 @item
|
|
|
399 version 18.45 released on June 2, 1987.
|
|
|
400 @item
|
|
|
401 version 18.46 released on June 9, 1987.
|
|
|
402 @item
|
|
|
403 version 18.47 released on June 18, 1987.
|
|
|
404 @item
|
|
|
405 version 18.48 released on September 3, 1987.
|
|
|
406 @item
|
|
|
407 version 18.49 released on September 18, 1987.
|
|
|
408 @item
|
|
|
409 version 18.50 released on February 13, 1988.
|
|
|
410 @item
|
|
|
411 version 18.51 released on May 7, 1988.
|
|
|
412 @item
|
|
|
413 version 18.52 released on September 1, 1988.
|
|
|
414 @item
|
|
|
415 version 18.53 released on February 24, 1989.
|
|
|
416 @item
|
|
|
417 version 18.54 released on April 26, 1989.
|
|
|
418 @item
|
|
|
419 version 18.55 released on August 23, 1989. This is the earliest version
|
|
|
420 that is still available by FTP.
|
|
|
421 @item
|
|
|
422 version 18.56 released on January 17, 1991.
|
|
|
423 @item
|
|
|
424 version 18.57 released late January, 1991.
|
|
|
425 @item
|
|
|
426 version 18.58 released ?????.
|
|
|
427 @item
|
|
|
428 version 18.59 released October 31, 1992.
|
|
|
429 @end itemize
|
|
|
430
|
|
|
431 @node Lucid Emacs
|
|
|
432 @section Lucid Emacs
|
|
|
433 @cindex Lucid Emacs
|
|
|
434 @cindex Lucid Inc.
|
|
|
435 @cindex Energize
|
|
|
436 @cindex Epoch
|
|
|
437
|
|
|
438 Lucid Emacs was developed by the (now-defunct) Lucid Inc., a maker of
|
|
|
439 C++ and Lisp development environments. It began when Lucid decided they
|
|
|
440 wanted to use Emacs as the editor and cornerstone of their C++
|
|
|
441 development environment (called ``Energize''). They needed many features
|
|
|
442 that were not available in the existing version of GNU Emacs (version
|
|
|
443 18.5something), in particular good and integrated support for GUI
|
|
|
444 elements such as mouse support, multiple fonts, multiple window-system
|
|
|
445 windows, etc. A branch of GNU Emacs called Epoch, written at the
|
|
|
446 University of Illinois, existed that supplied many of these features;
|
|
|
447 however, Lucid needed more than what existed in Epoch. At the time, the
|
|
|
448 Free Software Foundation was working on version 19 of Emacs (this was
|
|
|
449 sometime around 1991), which was planned to have similar features, and
|
|
|
450 so Lucid decided to work with the Free Software Foundation. Their plan
|
|
|
451 was to add features that they needed, and coordinate with the FSF so
|
|
|
452 that the features would get included back into Emacs version 19.
|
|
|
453
|
|
|
454 Delays in the release of version 19 occurred, however (resulting in it
|
|
|
455 finally being released more than a year after what was initially
|
|
|
456 planned), and Lucid encountered unexpected technical resistance in
|
|
|
457 getting their changes merged back into version 19, so they decided to
|
|
|
458 release their own version of Emacs, which became Lucid Emacs 19.0.
|
|
|
459
|
|
|
460 @cindex Zawinski, Jamie
|
|
|
461 @cindex Sexton, Harlan
|
|
|
462 @cindex Benson, Eric
|
|
|
463 @cindex Devin, Matthieu
|
|
|
464 The initial authors of Lucid Emacs were Matthieu Devin, Harlan Sexton,
|
|
|
465 and Eric Benson, and the work was later taken over by Jamie Zawinski,
|
|
|
466 who became ``Mr. Lucid Emacs'' for many releases.
|
|
|
467
|
|
|
468 A time line for Lucid Emacs/XEmacs is
|
|
|
469
|
|
|
470 @itemize @bullet
|
|
|
471 @item
|
|
|
472 version 19.0 shipped with Energize 1.0, April 1992.
|
|
|
473 @item
|
|
|
474 version 19.1 released June 4, 1992.
|
|
|
475 @item
|
|
|
476 version 19.2 released June 19, 1992.
|
|
|
477 @item
|
|
|
478 version 19.3 released September 9, 1992.
|
|
|
479 @item
|
|
|
480 version 19.4 released January 21, 1993.
|
|
|
481 @item
|
|
|
482 version 19.5 was a repackaging of 19.4 with a few bug fixes and
|
|
|
483 shipped with Energize 2.0. Never released to the net.
|
|
|
484 @item
|
|
|
485 version 19.6 released April 9, 1993.
|
|
|
486 @item
|
|
|
487 version 19.7 was a repackaging of 19.6 with a few bug fixes and
|
|
|
488 shipped with Energize 2.1. Never released to the net.
|
|
|
489 @item
|
|
|
490 version 19.8 released September 6, 1993.
|
|
|
491 @item
|
|
|
492 version 19.9 released January 12, 1994.
|
|
|
493 @item
|
|
|
494 version 19.10 released May 27, 1994.
|
|
|
495 @item
|
|
|
496 version 19.11 (first XEmacs) released September 13, 1994.
|
|
|
497 @item
|
|
|
498 version 19.12 released June 23, 1995.
|
|
|
499 @item
|
|
|
500 version 19.13 released September 1, 1995.
|
|
|
501 @item
|
|
|
502 version 19.14 released June 23, 1996.
|
|
|
503 @item
|
|
|
504 version 20.0 released February 9, 1997.
|
|
|
505 @item
|
|
|
506 version 19.15 released March 28, 1997.
|
|
|
507 @item
|
|
|
508 version 20.1 (not released to the net) April 15, 1997.
|
|
|
509 @item
|
|
|
510 version 20.2 released May 16, 1997.
|
|
|
511 @item
|
|
|
512 version 19.16 released October 31, 1997.
|
|
|
513 @item
|
|
|
514 version 20.3 (the first stable version of XEmacs 20.x) released November 30,
|
|
|
515 1997.
|
|
|
516 version 20.4 released February 28, 1998.
|
|
|
517 @end itemize
|
|
|
518
|
|
|
519 @node GNU Emacs 19
|
|
|
520 @section GNU Emacs 19
|
|
|
521 @cindex GNU Emacs 19
|
|
|
522 @cindex FSF Emacs
|
|
|
523
|
|
|
524 About a year after the initial release of Lucid Emacs, the FSF
|
|
|
525 released a beta of their version of Emacs 19 (referred to here as ``GNU
|
|
|
526 Emacs''). By this time, the current version of Lucid Emacs was
|
|
|
527 19.6. (Strangely, the first released beta from the FSF was GNU Emacs
|
|
|
528 19.7.) A time line for GNU Emacs version 19 is
|
|
|
529
|
|
|
530 @itemize @bullet
|
|
|
531 @item
|
|
|
532 version 19.8 (beta) released May 27, 1993.
|
|
|
533 @item
|
|
|
534 version 19.9 (beta) released May 27, 1993.
|
|
|
535 @item
|
|
|
536 version 19.10 (beta) released May 30, 1993.
|
|
|
537 @item
|
|
|
538 version 19.11 (beta) released June 1, 1993.
|
|
|
539 @item
|
|
|
540 version 19.12 (beta) released June 2, 1993.
|
|
|
541 @item
|
|
|
542 version 19.13 (beta) released June 8, 1993.
|
|
|
543 @item
|
|
|
544 version 19.14 (beta) released June 17, 1993.
|
|
|
545 @item
|
|
|
546 version 19.15 (beta) released June 19, 1993.
|
|
|
547 @item
|
|
|
548 version 19.16 (beta) released July 6, 1993.
|
|
|
549 @item
|
|
|
550 version 19.17 (beta) released late July, 1993.
|
|
|
551 @item
|
|
|
552 version 19.18 (beta) released August 9, 1993.
|
|
|
553 @item
|
|
|
554 version 19.19 (beta) released August 15, 1993.
|
|
|
555 @item
|
|
|
556 version 19.20 (beta) released November 17, 1993.
|
|
|
557 @item
|
|
|
558 version 19.21 (beta) released November 17, 1993.
|
|
|
559 @item
|
|
|
560 version 19.22 (beta) released November 28, 1993.
|
|
|
561 @item
|
|
|
562 version 19.23 (beta) released May 17, 1994.
|
|
|
563 @item
|
|
|
564 version 19.24 (beta) released May 16, 1994.
|
|
|
565 @item
|
|
|
566 version 19.25 (beta) released June 3, 1994.
|
|
|
567 @item
|
|
|
568 version 19.26 (beta) released September 11, 1994.
|
|
|
569 @item
|
|
|
570 version 19.27 (beta) released September 14, 1994.
|
|
|
571 @item
|
|
|
572 version 19.28 (first ``official'' release) released November 1, 1994.
|
|
|
573 @item
|
|
|
574 version 19.29 released June 21, 1995.
|
|
|
575 @item
|
|
|
576 version 19.30 released November 24, 1995.
|
|
|
577 @item
|
|
|
578 version 19.31 released May 25, 1996.
|
|
|
579 @item
|
|
|
580 version 19.32 released July 31, 1996.
|
|
|
581 @item
|
|
|
582 version 19.33 released August 11, 1996.
|
|
|
583 @item
|
|
|
584 version 19.34 released August 21, 1996.
|
|
|
585 @item
|
|
|
586 version 19.34b released September 6, 1996.
|
|
|
587 @end itemize
|
|
|
588
|
|
|
589 @cindex Mlynarik, Richard
|
|
|
590 In some ways, GNU Emacs 19 was better than Lucid Emacs; in some ways,
|
|
|
591 worse. Lucid soon began incorporating features from GNU Emacs 19 into
|
|
|
592 Lucid Emacs; the work was mostly done by Richard Mlynarik, who had been
|
|
|
593 working on and using GNU Emacs for a long time (back as far as version
|
|
|
594 16 or 17).
|
|
|
595
|
|
|
596 @node GNU Emacs 20
|
|
|
597 @section GNU Emacs 20
|
|
|
598 @cindex GNU Emacs 20
|
|
|
599 @cindex FSF Emacs
|
|
|
600
|
|
|
601 On February 2, 1997 work began on GNU Emacs to integrate Mule. The first
|
|
|
602 release was made in September of that year.
|
|
|
603
|
|
|
604 A timeline for Emacs 20 is
|
|
|
605
|
|
|
606 @itemize @bullet
|
|
|
607 @item
|
|
|
608 version 20.1 released September 17, 1997.
|
|
|
609 @item
|
|
|
610 version 20.2 released September 20, 1997.
|
|
|
611 @item
|
|
|
612 version 20.3 released August 19, 1998.
|
|
|
613 @end itemize
|
|
|
614
|
|
|
615 @node XEmacs
|
|
|
616 @section XEmacs
|
|
|
617 @cindex XEmacs
|
|
|
618
|
|
|
619 @cindex Sun Microsystems
|
|
|
620 @cindex University of Illinois
|
|
|
621 @cindex Illinois, University of
|
|
|
622 @cindex SPARCWorks
|
|
|
623 @cindex Andreessen, Marc
|
|
|
624 @cindex Baur, Steve
|
|
|
625 @cindex Buchholz, Martin
|
|
|
626 @cindex Kaplan, Simon
|
|
|
627 @cindex Wing, Ben
|
|
|
628 @cindex Thompson, Chuck
|
|
|
629 @cindex Win-Emacs
|
|
|
630 @cindex Epoch
|
|
|
631 @cindex Amdahl Corporation
|
|
|
632 Around the time that Lucid was developing Energize, Sun Microsystems
|
|
|
633 was developing their own development environment (called ``SPARCWorks'')
|
|
|
634 and also decided to use Emacs. They joined forces with the Epoch team
|
|
|
635 at the University of Illinois and later with Lucid. The maintainer of
|
|
|
636 the last-released version of Epoch was Marc Andreessen, but he dropped
|
|
|
637 out and the Epoch project, headed by Simon Kaplan, lured Chuck Thompson
|
|
|
638 away from a system administration job to become the primary Lucid Emacs
|
|
|
639 author for Epoch and Sun. Chuck's area of specialty became the
|
|
|
640 redisplay engine (he replaced the old Lucid Emacs redisplay engine with
|
|
|
641 a ported version from Epoch and then later rewrote it from scratch).
|
|
|
642 Sun also hired Ben Wing (the author of Win-Emacs, a port of Lucid Emacs
|
|
|
643 to Microsoft Windows 3.1) in 1993, for what was initially a one-month
|
|
|
644 contract to fix some event problems but later became a many-year
|
|
|
645 involvement, punctuated by a six-month contract with Amdahl Corporation.
|
|
|
646
|
|
|
647 @cindex rename to XEmacs
|
|
|
648 In 1994, Sun and Lucid agreed to rename Lucid Emacs to XEmacs (a name
|
|
|
649 not favorable to either company); the first release called XEmacs was
|
|
|
650 version 19.11. In June 1994, Lucid folded and Jamie quit to work for
|
|
|
651 the newly formed Mosaic Communications Corp., later Netscape
|
|
|
652 Communications Corp. (co-founded by the same Marc Andreessen, who had
|
|
|
653 quit his Epoch job to work on a graphical browser for the World Wide
|
|
|
654 Web). Chuck then become the primary maintainer of XEmacs, and put out
|
|
|
655 versions 19.11 through 19.14 in conjunction with Ben. For 19.12 and
|
|
|
656 19.13, Chuck added the new redisplay and many other display improvements
|
|
|
657 and Ben added MULE support (support for Asian and other languages) and
|
|
|
658 redesigned most of the internal Lisp subsystems to better support the
|
|
|
659 MULE work and the various other features being added to XEmacs. After
|
|
|
660 19.14 Chuck retired as primary maintainer and Steve Baur stepped in.
|
|
|
661
|
|
|
662 @cindex MULE merged XEmacs appears
|
|
|
663 Soon after 19.13 was released, work began in earnest on the MULE
|
|
|
664 internationalization code and the source tree was divided into two
|
|
|
665 development paths. The MULE version was initially called 19.20, but was
|
|
|
666 soon renamed to 20.0. In 1996 Martin Buchholz of Sun Microsystems took
|
|
|
667 over the care and feeding of it and worked on it in parallel with the
|
|
|
668 19.14 development that was occurring at the same time. After much work
|
|
|
669 by Martin, it was decided to release 20.0 ahead of 19.15 in February
|
|
|
670 1997. The source tree remained divided until 20.2 when the version 19
|
|
|
671 source was finally retired at version 19.16.
|
|
|
672
|
|
|
673 @cindex Baur, Steve
|
|
|
674 @cindex Buchholz, Martin
|
|
|
675 @cindex Jones, Kyle
|
|
|
676 @cindex Niksic, Hrvoje
|
|
|
677 @cindex XEmacs goes it alone
|
|
|
678 In 1997, Sun finally dropped all pretense of support for XEmacs and
|
|
|
679 Martin Buchholz left the company in November. Since then, and mostly
|
|
|
680 for the previous year, because Steve Baur was never paid to work on
|
|
|
681 XEmacs, XEmacs has existed solely on the contributions of volunteers
|
|
|
682 from the Free Software Community. Starting from 1997, Hrvoje Niksic and
|
|
|
683 Kyle Jones have figured prominently in XEmacs development.
|
|
|
684
|
|
|
685 @cindex merging attempts
|
|
|
686 Many attempts have been made to merge XEmacs and GNU Emacs, but they
|
|
|
687 have consistently failed.
|
|
|
688
|
|
|
689 A more detailed history is contained in the XEmacs About page.
|
|
|
690
|
|
|
691 @node XEmacs From the Outside, The Lisp Language, A History of Emacs, Top
|
|
|
692 @chapter XEmacs From the Outside
|
|
|
693 @cindex read-eval-print
|
|
|
694
|
|
|
695 XEmacs appears to the outside world as an editor, but it is really a
|
|
|
696 Lisp environment. At its heart is a Lisp interpreter; it also
|
|
|
697 ``happens'' to contain many specialized object types (e.g. buffers,
|
|
|
698 windows, frames, events) that are useful for implementing an editor.
|
|
|
699 Some of these objects (in particular windows and frames) have
|
|
|
700 displayable representations, and XEmacs provides a function
|
|
|
701 @code{redisplay()} that ensures that the display of all such objects
|
|
|
702 matches their internal state. Most of the time, a standard Lisp
|
|
440
|
703 environment is in a @dfn{read-eval-print} loop---i.e. ``read some Lisp
|
|
428
|
704 code, execute it, and print the results''. XEmacs has a similar loop:
|
|
|
705
|
|
|
706 @itemize @bullet
|
|
|
707 @item
|
|
|
708 read an event
|
|
|
709 @item
|
|
|
710 dispatch the event (i.e. ``do it'')
|
|
|
711 @item
|
|
|
712 redisplay
|
|
|
713 @end itemize
|
|
|
714
|
|
|
715 Reading an event is done using the Lisp function @code{next-event},
|
|
|
716 which waits for something to happen (typically, the user presses a key
|
|
|
717 or moves the mouse) and returns an event object describing this.
|
|
|
718 Dispatching an event is done using the Lisp function
|
|
|
719 @code{dispatch-event}, which looks up the event in a keymap object (a
|
|
|
720 particular kind of object that associates an event with a Lisp function)
|
|
|
721 and calls that function. The function ``does'' what the user has
|
|
|
722 requested by changing the state of particular frame objects, buffer
|
|
|
723 objects, etc. Finally, @code{redisplay()} is called, which updates the
|
|
|
724 display to reflect those changes just made. Thus is an ``editor'' born.
|
|
|
725
|
|
|
726 @cindex bridge, playing
|
|
|
727 @cindex taxes, doing
|
|
|
728 @cindex pi, calculating
|
|
|
729 Note that you do not have to use XEmacs as an editor; you could just
|
|
|
730 as well make it do your taxes, compute pi, play bridge, etc. You'd just
|
|
|
731 have to write functions to do those operations in Lisp.
|
|
|
732
|
|
|
733 @node The Lisp Language, XEmacs From the Perspective of Building, XEmacs From the Outside, Top
|
|
|
734 @chapter The Lisp Language
|
|
|
735 @cindex Lisp vs. C
|
|
|
736 @cindex C vs. Lisp
|
|
|
737 @cindex Lisp vs. Java
|
|
|
738 @cindex Java vs. Lisp
|
|
|
739 @cindex dynamic scoping
|
|
|
740 @cindex scoping, dynamic
|
|
|
741 @cindex dynamic types
|
|
|
742 @cindex types, dynamic
|
|
|
743 @cindex Java
|
|
|
744 @cindex Common Lisp
|
|
|
745 @cindex Gosling, James
|
|
|
746
|
|
|
747 Lisp is a general-purpose language that is higher-level than C and in
|
|
|
748 many ways more powerful than C. Powerful dialects of Lisp such as
|
|
|
749 Common Lisp are probably much better languages for writing very large
|
|
|
750 applications than is C. (Unfortunately, for many non-technical
|
|
|
751 reasons C and its successor C++ have become the dominant languages for
|
|
|
752 application development. These languages are both inadequate for
|
|
|
753 extremely large applications, which is evidenced by the fact that newer,
|
|
|
754 larger programs are becoming ever harder to write and are requiring ever
|
|
|
755 more programmers despite great increases in C development environments;
|
|
|
756 and by the fact that, although hardware speeds and reliability have been
|
|
|
757 growing at an exponential rate, most software is still generally
|
|
|
758 considered to be slow and buggy.)
|
|
|
759
|
|
|
760 The new Java language holds promise as a better general-purpose
|
|
|
761 development language than C. Java has many features in common with
|
|
|
762 Lisp that are not shared by C (this is not a coincidence, since
|
|
|
763 Java was designed by James Gosling, a former Lisp hacker). This
|
|
|
764 will be discussed more later.
|
|
|
765
|
|
|
766 For those used to C, here is a summary of the basic differences between
|
|
|
767 C and Lisp:
|
|
|
768
|
|
|
769 @enumerate
|
|
|
770 @item
|
|
|
771 Lisp has an extremely regular syntax. Every function, expression,
|
|
|
772 and control statement is written in the form
|
|
|
773
|
|
|
774 @example
|
|
|
775 (@var{func} @var{arg1} @var{arg2} ...)
|
|
|
776 @end example
|
|
|
777
|
|
|
778 This is as opposed to C, which writes functions as
|
|
|
779
|
|
|
780 @example
|
|
|
781 func(@var{arg1}, @var{arg2}, ...)
|
|
|
782 @end example
|
|
|
783
|
|
|
784 but writes expressions involving operators as (e.g.)
|
|
|
785
|
|
|
786 @example
|
|
|
787 @var{arg1} + @var{arg2}
|
|
|
788 @end example
|
|
|
789
|
|
|
790 and writes control statements as (e.g.)
|
|
|
791
|
|
|
792 @example
|
|
|
793 while (@var{expr}) @{ @var{statement1}; @var{statement2}; ... @}
|
|
|
794 @end example
|
|
|
795
|
|
|
796 Lisp equivalents of the latter two would be
|
|
|
797
|
|
|
798 @example
|
|
|
799 (+ @var{arg1} @var{arg2} ...)
|
|
|
800 @end example
|
|
|
801
|
|
|
802 and
|
|
|
803
|
|
|
804 @example
|
|
|
805 (while @var{expr} @var{statement1} @var{statement2} ...)
|
|
|
806 @end example
|
|
|
807
|
|
|
808 @item
|
|
|
809 Lisp is a safe language. Assuming there are no bugs in the Lisp
|
|
|
810 interpreter/compiler, it is impossible to write a program that ``core
|
|
|
811 dumps'' or otherwise causes the machine to execute an illegal
|
|
|
812 instruction. This is very different from C, where perhaps the most
|
|
|
813 common outcome of a bug is exactly such a crash. A corollary of this is that
|
|
|
814 the C operation of casting a pointer is impossible (and unnecessary) in
|
|
|
815 Lisp, and that it is impossible to access memory outside the bounds of
|
|
|
816 an array.
|
|
|
817
|
|
|
818 @item
|
|
|
819 Programs and data are written in the same form. The
|
|
|
820 parenthesis-enclosing form described above for statements is the same
|
|
|
821 form used for the most common data type in Lisp, the list. Thus, it is
|
|
|
822 possible to represent any Lisp program using Lisp data types, and for
|
|
|
823 one program to construct Lisp statements and then dynamically
|
|
|
824 @dfn{evaluate} them, or cause them to execute.
|
|
|
825
|
|
|
826 @item
|
|
|
827 All objects are @dfn{dynamically typed}. This means that part of every
|
|
|
828 object is an indication of what type it is. A Lisp program can
|
|
|
829 manipulate an object without knowing what type it is, and can query an
|
|
|
830 object to determine its type. This means that, correspondingly,
|
|
|
831 variables and function parameters can hold objects of any type and are
|
|
|
832 not normally declared as being of any particular type. This is opposed
|
|
|
833 to the @dfn{static typing} of C, where variables can hold exactly one
|
|
|
834 type of object and must be declared as such, and objects do not contain
|
|
|
835 an indication of their type because it's implicit in the variables they
|
|
|
836 are stored in. It is possible in C to have a variable hold different
|
|
|
837 types of objects (e.g. through the use of @code{void *} pointers or
|
|
|
838 variable-argument functions), but the type information must then be
|
|
|
839 passed explicitly in some other fashion, leading to additional program
|
|
|
840 complexity.
|
|
|
841
|
|
|
842 @item
|
|
|
843 Allocated memory is automatically reclaimed when it is no longer in use.
|
|
|
844 This operation is called @dfn{garbage collection} and involves looking
|
|
|
845 through all variables to see what memory is being pointed to, and
|
|
|
846 reclaiming any memory that is not pointed to and is thus
|
|
|
847 ``inaccessible'' and out of use. This is as opposed to C, in which
|
|
|
848 allocated memory must be explicitly reclaimed using @code{free()}. If
|
|
|
849 you simply drop all pointers to memory without freeing it, it becomes
|
|
|
850 ``leaked'' memory that still takes up space. Over a long period of
|
|
|
851 time, this can cause your program to grow and grow until it runs out of
|
|
|
852 memory.
|
|
|
853
|
|
|
854 @item
|
|
|
855 Lisp has built-in facilities for handling errors and exceptions. In C,
|
|
|
856 when an error occurs, usually either the program exits entirely or the
|
|
|
857 routine in which the error occurs returns a value indicating this. If
|
|
|
858 an error occurs in a deeply-nested routine, then every routine currently
|
|
|
859 called must unwind itself normally and return an error value back up to
|
|
|
860 the next routine. This means that every routine must explicitly check
|
|
|
861 for an error in all the routines it calls; if it does not do so,
|
|
|
862 unexpected and often random behavior results. This is an extremely
|
|
|
863 common source of bugs in C programs. An alternative would be to do a
|
|
|
864 non-local exit using @code{longjmp()}, but that is often very dangerous
|
|
|
865 because the routines that were exited past had no opportunity to clean
|
|
|
866 up after themselves and may leave things in an inconsistent state,
|
|
|
867 causing a crash shortly afterwards.
|
|
|
868
|
|
|
869 Lisp provides mechanisms to make such non-local exits safe. When an
|
|
|
870 error occurs, a routine simply signals that an error of a particular
|
|
|
871 class has occurred, and a non-local exit takes place. Any routine can
|
|
|
872 trap errors occurring in routines it calls by registering an error
|
|
|
873 handler for some or all classes of errors. (If no handler is registered,
|
|
|
874 a default handler, generally installed by the top-level event loop, is
|
|
|
875 executed; this prints out the error and continues.) Routines can also
|
|
|
876 specify cleanup code (called an @dfn{unwind-protect}) that will be
|
|
|
877 called when control exits from a block of code, no matter how that exit
|
|
440
|
878 occurs---i.e. even if a function deeply nested below it causes a
|
|
428
|
879 non-local exit back to the top level.
|
|
|
880
|
|
|
881 Note that this facility has appeared in some recent vintages of C, in
|
|
|
882 particular Visual C++ and other PC compilers written for the Microsoft
|
|
|
883 Win32 API.
|
|
|
884
|
|
|
885 @item
|
|
|
886 In Emacs Lisp, local variables are @dfn{dynamically scoped}. This means
|
|
|
887 that if you declare a local variable in a particular function, and then
|
|
|
888 call another function, that subfunction can ``see'' the local variable
|
|
|
889 you declared. This is actually considered a bug in Emacs Lisp and in
|
|
|
890 all other early dialects of Lisp, and was corrected in Common Lisp. (In
|
|
|
891 Common Lisp, you can still declare dynamically scoped variables if you
|
|
440
|
892 want to---they are sometimes useful---but variables by default are
|
|
428
|
893 @dfn{lexically scoped} as in C.)
|
|
|
894 @end enumerate
|
|
|
895
|
|
|
896 For those familiar with Lisp, Emacs Lisp is modelled after MacLisp, an
|
|
|
897 early dialect of Lisp developed at MIT (no relation to the Macintosh
|
|
|
898 computer). There is a Common Lisp compatibility package available for
|
|
|
899 Emacs that provides many of the features of Common Lisp.
|
|
|
900
|
|
|
901 The Java language is derived in many ways from C, and shares a similar
|
|
|
902 syntax, but has the following features in common with Lisp (and different
|
|
|
903 from C):
|
|
|
904
|
|
|
905 @enumerate
|
|
|
906 @item
|
|
|
907 Java is a safe language, like Lisp.
|
|
|
908 @item
|
|
|
909 Java provides garbage collection, like Lisp.
|
|
|
910 @item
|
|
|
911 Java has built-in facilities for handling errors and exceptions, like
|
|
|
912 Lisp.
|
|
|
913 @item
|
|
|
914 Java has a type system that combines the best advantages of both static
|
|
|
915 and dynamic typing. Objects (except very simple types) are explicitly
|
|
|
916 marked with their type, as in dynamic typing; but there is a hierarchy
|
|
|
917 of types and functions are declared to accept only certain types, thus
|
|
|
918 providing the increased compile-time error-checking of static typing.
|
|
|
919 @end enumerate
|
|
|
920
|
|
|
921 The Java language also has some negative attributes:
|
|
|
922
|
|
|
923 @enumerate
|
|
|
924 @item
|
|
|
925 Java uses the edit/compile/run model of software development. This
|
|
|
926 makes it hard to use interactively. For example, to use Java like
|
|
|
927 @code{bc} it is necessary to write a special purpose, albeit tiny,
|
|
|
928 application. In Emacs Lisp, a calculator comes built-in without any
|
|
|
929 effort - one can always just type an expression in the @code{*scratch*}
|
|
|
930 buffer.
|
|
|
931 @item
|
|
|
932 Java tries too hard to enforce, not merely enable, portability, making
|
|
|
933 ordinary access to standard OS facilities painful. Java has an
|
|
|
934 @dfn{agenda}. I think this is why @code{chdir} is not part of standard
|
|
|
935 Java, which is inexcusable.
|
|
|
936 @end enumerate
|
|
|
937
|
|
|
938 Unfortunately, there is no perfect language. Static typing allows a
|
|
|
939 compiler to catch programmer errors and produce more efficient code, but
|
|
|
940 makes programming more tedious and less fun. For the forseeable future,
|
|
|
941 an Ideal Editing and Programming Environment (and that is what XEmacs
|
|
|
942 aspires to) will be programmable in multiple languages: high level ones
|
|
|
943 like Lisp for user customization and prototyping, and lower level ones
|
|
|
944 for infrastructure and industrial strength applications. If I had my
|
|
|
945 way, XEmacs would be friendly towards the Python, Scheme, C++, ML,
|
|
|
946 etc... communities. But there are serious technical difficulties to
|
|
|
947 achieving that goal.
|
|
|
948
|
|
|
949 The word @dfn{application} in the previous paragraph was used
|
|
|
950 intentionally. XEmacs implements an API for programs written in Lisp
|
|
|
951 that makes it a full-fledged application platform, very much like an OS
|
|
|
952 inside the real OS.
|
|
|
953
|
|
|
954 @node XEmacs From the Perspective of Building, XEmacs From the Inside, The Lisp Language, Top
|
|
|
955 @chapter XEmacs From the Perspective of Building
|
|
|
956
|
|
|
957 The heart of XEmacs is the Lisp environment, which is written in C.
|
|
|
958 This is contained in the @file{src/} subdirectory. Underneath
|
|
|
959 @file{src/} are two subdirectories of header files: @file{s/} (header
|
|
|
960 files for particular operating systems) and @file{m/} (header files for
|
|
|
961 particular machine types). In practice the distinction between the two
|
|
|
962 types of header files is blurred. These header files define or undefine
|
|
|
963 certain preprocessor constants and macros to indicate particular
|
|
|
964 characteristics of the associated machine or operating system. As part
|
|
|
965 of the configure process, one @file{s/} file and one @file{m/} file is
|
|
|
966 identified for the particular environment in which XEmacs is being
|
|
|
967 built.
|
|
|
968
|
|
|
969 XEmacs also contains a great deal of Lisp code. This implements the
|
|
|
970 operations that make XEmacs useful as an editor as well as just a Lisp
|
|
|
971 environment, and also contains many add-on packages that allow XEmacs to
|
|
|
972 browse directories, act as a mail and Usenet news reader, compile Lisp
|
|
|
973 code, etc. There is actually more Lisp code than C code associated with
|
|
|
974 XEmacs, but much of the Lisp code is peripheral to the actual operation
|
|
|
975 of the editor. The Lisp code all lies in subdirectories underneath the
|
|
|
976 @file{lisp/} directory.
|
|
|
977
|
|
|
978 The @file{lwlib/} directory contains C code that implements a
|
|
|
979 generalized interface onto different X widget toolkits and also
|
|
|
980 implements some widgets of its own that behave like Motif widgets but
|
|
|
981 are faster, free, and in some cases more powerful. The code in this
|
|
|
982 directory compiles into a library and is mostly independent from XEmacs.
|
|
|
983
|
|
|
984 The @file{etc/} directory contains various data files associated with
|
|
|
985 XEmacs. Some of them are actually read by XEmacs at startup; others
|
|
|
986 merely contain useful information of various sorts.
|
|
|
987
|
|
|
988 The @file{lib-src/} directory contains C code for various auxiliary
|
|
|
989 programs that are used in connection with XEmacs. Some of them are used
|
|
|
990 during the build process; others are used to perform certain functions
|
|
|
991 that cannot conveniently be placed in the XEmacs executable (e.g. the
|
|
|
992 @file{movemail} program for fetching mail out of @file{/var/spool/mail},
|
|
|
993 which must be setgid to @file{mail} on many systems; and the
|
|
|
994 @file{gnuclient} program, which allows an external script to communicate
|
|
|
995 with a running XEmacs process).
|
|
|
996
|
|
|
997 The @file{man/} directory contains the sources for the XEmacs
|
|
|
998 documentation. It is mostly in a form called Texinfo, which can be
|
|
|
999 converted into either a printed document (by passing it through @TeX{})
|
|
|
1000 or into on-line documentation called @dfn{info files}.
|
|
|
1001
|
|
|
1002 The @file{info/} directory contains the results of formatting the XEmacs
|
|
|
1003 documentation as @dfn{info files}, for on-line use. These files are
|
|
|
1004 used when you enter the Info system using @kbd{C-h i} or through the
|
|
|
1005 Help menu.
|
|
|
1006
|
|
|
1007 The @file{dynodump/} directory contains auxiliary code used to build
|
|
|
1008 XEmacs on Solaris platforms.
|
|
|
1009
|
|
|
1010 The other directories contain various miscellaneous code and information
|
|
|
1011 that is not normally used or needed.
|
|
|
1012
|
|
|
1013 The first step of building involves running the @file{configure} program
|
|
|
1014 and passing it various parameters to specify any optional features you
|
|
|
1015 want and compiler arguments and such, as described in the @file{INSTALL}
|
|
|
1016 file. This determines what the build environment is, chooses the
|
|
|
1017 appropriate @file{s/} and @file{m/} file, and runs a series of tests to
|
|
|
1018 determine many details about your environment, such as which library
|
|
|
1019 functions are available and exactly how they work. The reason for
|
|
|
1020 running these tests is that it allows XEmacs to be compiled on a much
|
|
|
1021 wider variety of platforms than those that the XEmacs developers happen
|
|
|
1022 to be familiar with, including various sorts of hybrid platforms. This
|
|
|
1023 is especially important now that many operating systems give you a great
|
|
|
1024 deal of control over exactly what features you want installed, and allow
|
|
|
1025 for easy upgrading of parts of a system without upgrading the rest. It
|
|
|
1026 would be impossible to pre-determine and pre-specify the information for
|
|
|
1027 all possible configurations.
|
|
|
1028
|
|
|
1029 In fact, the @file{s/} and @file{m/} files are basically @emph{evil},
|
|
|
1030 since they contain unmaintainable platform-specific hard-coded
|
|
|
1031 information. XEmacs has been moving in the direction of having all
|
|
|
1032 system-specific information be determined dynamically by
|
|
|
1033 @file{configure}. Perhaps someday we can @code{rm -rf src/s src/m}.
|
|
|
1034
|
|
|
1035 When configure is done running, it generates @file{Makefile}s and
|
|
|
1036 @file{GNUmakefile}s and the file @file{src/config.h} (which describes
|
|
|
1037 the features of your system) from template files. You then run
|
|
|
1038 @file{make}, which compiles the auxiliary code and programs in
|
|
|
1039 @file{lib-src/} and @file{lwlib/} and the main XEmacs executable in
|
|
|
1040 @file{src/}. The result of compiling and linking is an executable
|
|
|
1041 called @file{temacs}, which is @emph{not} the final XEmacs executable.
|
|
|
1042 @file{temacs} by itself is not intended to function as an editor or even
|
|
|
1043 display any windows on the screen, and if you simply run it, it will
|
|
|
1044 exit immediately. The @file{Makefile} runs @file{temacs} with certain
|
|
|
1045 options that cause it to initialize itself, read in a number of basic
|
|
|
1046 Lisp files, and then dump itself out into a new executable called
|
|
|
1047 @file{xemacs}. This new executable has been pre-initialized and
|
|
|
1048 contains pre-digested Lisp code that is necessary for the editor to
|
|
|
1049 function (this includes most basic editing functions,
|
|
|
1050 e.g. @code{kill-line}, that can be defined in terms of other Lisp
|
|
|
1051 primitives; some initialization code that is called when certain
|
|
|
1052 objects, such as frames, are created; and all of the standard
|
|
|
1053 keybindings and code for the actions they result in). This executable,
|
|
|
1054 @file{xemacs}, is the executable that you run to use the XEmacs editor.
|
|
|
1055
|
|
|
1056 Although @file{temacs} is not intended to be run as an editor, it can,
|
|
|
1057 by using the incantation @code{temacs -batch -l loadup.el run-temacs}.
|
|
|
1058 This is useful when the dumping procedure described above is broken, or
|
|
|
1059 when using certain program debugging tools such as Purify. These tools
|
|
|
1060 get mighty confused by the tricks played by the XEmacs build process,
|
|
|
1061 such as allocation memory in one process, and freeing it in the next.
|
|
|
1062
|
|
|
1063 @node XEmacs From the Inside, The XEmacs Object System (Abstractly Speaking), XEmacs From the Perspective of Building, Top
|
|
|
1064 @chapter XEmacs From the Inside
|
|
|
1065
|
|
|
1066 Internally, XEmacs is quite complex, and can be very confusing. To
|
|
|
1067 simplify things, it can be useful to think of XEmacs as containing an
|
|
|
1068 event loop that ``drives'' everything, and a number of other subsystems,
|
|
|
1069 such as a Lisp engine and a redisplay mechanism. Each of these other
|
|
|
1070 subsystems exists simultaneously in XEmacs, and each has a certain
|
|
|
1071 state. The flow of control continually passes in and out of these
|
|
|
1072 different subsystems in the course of normal operation of the editor.
|
|
|
1073
|
|
|
1074 It is important to keep in mind that, most of the time, the editor is
|
|
|
1075 ``driven'' by the event loop. Except during initialization and batch
|
|
|
1076 mode, all subsystems are entered directly or indirectly through the
|
|
|
1077 event loop, and ultimately, control exits out of all subsystems back up
|
|
|
1078 to the event loop. This cycle of entering a subsystem, exiting back out
|
|
|
1079 to the event loop, and starting another iteration of the event loop
|
|
|
1080 occurs once each keystroke, mouse motion, etc.
|
|
|
1081
|
|
|
1082 If you're trying to understand a particular subsystem (other than the
|
|
|
1083 event loop), think of it as a ``daemon'' process or ``servant'' that is
|
|
|
1084 responsible for one particular aspect of a larger system, and
|
|
|
1085 periodically receives commands or environment changes that cause it to
|
|
|
1086 do something. Ultimately, these commands and environment changes are
|
|
|
1087 always triggered by the event loop. For example:
|
|
|
1088
|
|
|
1089 @itemize @bullet
|
|
|
1090 @item
|
|
|
1091 The window and frame mechanism is responsible for keeping track of what
|
|
|
1092 windows and frames exist, what buffers are in them, etc. It is
|
|
|
1093 periodically given commands (usually from the user) to make a change to
|
|
|
1094 the current window/frame state: i.e. create a new frame, delete a
|
|
|
1095 window, etc.
|
|
|
1096
|
|
|
1097 @item
|
|
|
1098 The buffer mechanism is responsible for keeping track of what buffers
|
|
|
1099 exist and what text is in them. It is periodically given commands
|
|
|
1100 (usually from the user) to insert or delete text, create a buffer, etc.
|
|
|
1101 When it receives a text-change command, it notifies the redisplay
|
|
|
1102 mechanism.
|
|
|
1103
|
|
|
1104 @item
|
|
|
1105 The redisplay mechanism is responsible for making sure that windows and
|
|
|
1106 frames are displayed correctly. It is periodically told (by the event
|
|
|
1107 loop) to actually ``do its job'', i.e. snoop around and see what the
|
|
|
1108 current state of the environment (mostly of the currently-existing
|
|
|
1109 windows, frames, and buffers) is, and make sure that that state matches
|
|
|
1110 what's actually displayed. It keeps lots and lots of information around
|
|
|
1111 (such as what is actually being displayed currently, and what the
|
|
|
1112 environment was last time it checked) so that it can minimize the work
|
|
|
1113 it has to do. It is also helped along in that whenever a relevant
|
|
|
1114 change to the environment occurs, the redisplay mechanism is told about
|
|
|
1115 this, so it has a pretty good idea of where it has to look to find
|
|
|
1116 possible changes and doesn't have to look everywhere.
|
|
|
1117
|
|
|
1118 @item
|
|
|
1119 The Lisp engine is responsible for executing the Lisp code in which most
|
|
|
1120 user commands are written. It is entered through a call to @code{eval}
|
|
|
1121 or @code{funcall}, which occurs as a result of dispatching an event from
|
|
|
1122 the event loop. The functions it calls issue commands to the buffer
|
|
|
1123 mechanism, the window/frame subsystem, etc.
|
|
|
1124
|
|
|
1125 @item
|
|
|
1126 The Lisp allocation subsystem is responsible for keeping track of Lisp
|
|
|
1127 objects. It is given commands from the Lisp engine to allocate objects,
|
|
|
1128 garbage collect, etc.
|
|
|
1129 @end itemize
|
|
|
1130
|
|
|
1131 etc.
|
|
|
1132
|
|
|
1133 The important idea here is that there are a number of independent
|
|
|
1134 subsystems each with its own responsibility and persistent state, just
|
|
|
1135 like different employees in a company, and each subsystem is
|
|
|
1136 periodically given commands from other subsystems. Commands can flow
|
|
|
1137 from any one subsystem to any other, but there is usually some sort of
|
|
|
1138 hierarchy, with all commands originating from the event subsystem.
|
|
|
1139
|
|
|
1140 XEmacs is entered in @code{main()}, which is in @file{emacs.c}. When
|
|
|
1141 this is called the first time (in a properly-invoked @file{temacs}), it
|
|
|
1142 does the following:
|
|
|
1143
|
|
|
1144 @enumerate
|
|
|
1145 @item
|
|
|
1146 It does some very basic environment initializations, such as determining
|
|
|
1147 where it and its directories (e.g. @file{lisp/} and @file{etc/}) reside
|
|
|
1148 and setting up signal handlers.
|
|
|
1149 @item
|
|
|
1150 It initializes the entire Lisp interpreter.
|
|
|
1151 @item
|
|
|
1152 It sets the initial values of many built-in variables (including many
|
|
|
1153 variables that are visible to Lisp programs), such as the global keymap
|
|
|
1154 object and the built-in faces (a face is an object that describes the
|
|
|
1155 display characteristics of text). This involves creating Lisp objects
|
|
|
1156 and thus is dependent on step (2).
|
|
|
1157 @item
|
|
|
1158 It performs various other initializations that are relevant to the
|
|
|
1159 particular environment it is running in, such as retrieving environment
|
|
|
1160 variables, determining the current date and the user who is running the
|
|
|
1161 program, examining its standard input, creating any necessary file
|
|
|
1162 descriptors, etc.
|
|
|
1163 @item
|
|
|
1164 At this point, the C initialization is complete. A Lisp program that
|
|
|
1165 was specified on the command line (usually @file{loadup.el}) is called
|
|
|
1166 (temacs is normally invoked as @code{temacs -batch -l loadup.el dump}).
|
|
|
1167 @file{loadup.el} loads all of the other Lisp files that are needed for
|
|
|
1168 the operation of the editor, calls the @code{dump-emacs} function to
|
|
|
1169 write out @file{xemacs}, and then kills the temacs process.
|
|
|
1170 @end enumerate
|
|
|
1171
|
|
|
1172 When @file{xemacs} is then run, it only redoes steps (1) and (4)
|
|
|
1173 above; all variables already contain the values they were set to when
|
|
|
1174 the executable was dumped, and all memory that was allocated with
|
|
|
1175 @code{malloc()} is still around. (XEmacs knows whether it is being run
|
|
|
1176 as @file{xemacs} or @file{temacs} because it sets the global variable
|
|
|
1177 @code{initialized} to 1 after step (4) above.) At this point,
|
|
|
1178 @file{xemacs} calls a Lisp function to do any further initialization,
|
|
|
1179 which includes parsing the command-line (the C code can only do limited
|
|
|
1180 command-line parsing, which includes looking for the @samp{-batch} and
|
|
|
1181 @samp{-l} flags and a few other flags that it needs to know about before
|
|
|
1182 initialization is complete), creating the first frame (or @dfn{window}
|
|
|
1183 in standard window-system parlance), running the user's init file
|
|
|
1184 (usually the file @file{.emacs} in the user's home directory), etc. The
|
|
|
1185 function to do this is usually called @code{normal-top-level};
|
|
|
1186 @file{loadup.el} tells the C code about this function by setting its
|
|
|
1187 name as the value of the Lisp variable @code{top-level}.
|
|
|
1188
|
|
|
1189 When the Lisp initialization code is done, the C code enters the event
|
|
|
1190 loop, and stays there for the duration of the XEmacs process. The code
|
|
|
1191 for the event loop is contained in @file{keyboard.c}, and is called
|
|
|
1192 @code{Fcommand_loop_1()}. Note that this event loop could very well be
|
|
|
1193 written in Lisp, and in fact a Lisp version exists; but apparently,
|
|
|
1194 doing this makes XEmacs run noticeably slower.
|
|
|
1195
|
|
|
1196 Notice how much of the initialization is done in Lisp, not in C.
|
|
|
1197 In general, XEmacs tries to move as much code as is possible
|
|
|
1198 into Lisp. Code that remains in C is code that implements the
|
|
|
1199 Lisp interpreter itself, or code that needs to be very fast, or
|
|
|
1200 code that needs to do system calls or other such stuff that
|
|
|
1201 needs to be done in C, or code that needs to have access to
|
|
|
1202 ``forbidden'' structures. (One conscious aspect of the design of
|
|
|
1203 Lisp under XEmacs is a clean separation between the external
|
|
|
1204 interface to a Lisp object's functionality and its internal
|
|
|
1205 implementation. Part of this design is that Lisp programs
|
|
|
1206 are forbidden from accessing the contents of the object other
|
|
|
1207 than through using a standard API. In this respect, XEmacs Lisp
|
|
|
1208 is similar to modern Lisp dialects but differs from GNU Emacs,
|
|
|
1209 which tends to expose the implementation and allow Lisp
|
|
|
1210 programs to look at it directly. The major advantage of
|
|
|
1211 hiding the implementation is that it allows the implementation
|
|
|
1212 to be redesigned without affecting any Lisp programs, including
|
|
|
1213 those that might want to be ``clever'' by looking directly at
|
|
|
1214 the object's contents and possibly manipulating them.)
|
|
|
1215
|
|
|
1216 Moving code into Lisp makes the code easier to debug and maintain and
|
|
|
1217 makes it much easier for people who are not XEmacs developers to
|
|
|
1218 customize XEmacs, because they can make a change with much less chance
|
|
|
1219 of obscure and unwanted interactions occurring than if they were to
|
|
|
1220 change the C code.
|
|
|
1221
|
|
|
1222 @node The XEmacs Object System (Abstractly Speaking), How Lisp Objects Are Represented in C, XEmacs From the Inside, Top
|
|
|
1223 @chapter The XEmacs Object System (Abstractly Speaking)
|
|
|
1224
|
|
|
1225 At the heart of the Lisp interpreter is its management of objects.
|
|
|
1226 XEmacs Lisp contains many built-in objects, some of which are
|
|
|
1227 simple and others of which can be very complex; and some of which
|
|
|
1228 are very common, and others of which are rarely used or are only
|
|
|
1229 used internally. (Since the Lisp allocation system, with its
|
|
|
1230 automatic reclamation of unused storage, is so much more convenient
|
|
|
1231 than @code{malloc()} and @code{free()}, the C code makes extensive use of it
|
|
|
1232 in its internal operations.)
|
|
|
1233
|
|
|
1234 The basic Lisp objects are
|
|
|
1235
|
|
|
1236 @table @code
|
|
|
1237 @item integer
|
|
|
1238 28 or 31 bits of precision, or 60 or 63 bits on 64-bit machines; the
|
|
|
1239 reason for this is described below when the internal Lisp object
|
|
|
1240 representation is described.
|
|
|
1241 @item float
|
|
|
1242 Same precision as a double in C.
|
|
|
1243 @item cons
|
|
|
1244 A simple container for two Lisp objects, used to implement lists and
|
|
|
1245 most other data structures in Lisp.
|
|
|
1246 @item char
|
|
|
1247 An object representing a single character of text; chars behave like
|
|
|
1248 integers in many ways but are logically considered text rather than
|
|
|
1249 numbers and have a different read syntax. (the read syntax for a char
|
|
440
|
1250 contains the char itself or some textual encoding of it---for example,
|
|
428
|
1251 a Japanese Kanji character might be encoded as @samp{^[$(B#&^[(B} using the
|
|
440
|
1252 ISO-2022 encoding standard---rather than the numerical representation
|
|
428
|
1253 of the char; this way, if the mapping between chars and integers
|
|
|
1254 changes, which is quite possible for Kanji characters and other extended
|
|
|
1255 characters, the same character will still be created. Note that some
|
|
|
1256 primitives confuse chars and integers. The worst culprit is @code{eq},
|
|
|
1257 which makes a special exception and considers a char to be @code{eq} to
|
|
|
1258 its integer equivalent, even though in no other case are objects of two
|
|
|
1259 different types @code{eq}. The reason for this monstrosity is
|
|
|
1260 compatibility with existing code; the separation of char from integer
|
|
|
1261 came fairly recently.)
|
|
|
1262 @item symbol
|
|
|
1263 An object that contains Lisp objects and is referred to by name;
|
|
|
1264 symbols are used to implement variables and named functions
|
|
|
1265 and to provide the equivalent of preprocessor constants in C.
|
|
|
1266 @item vector
|
|
|
1267 A one-dimensional array of Lisp objects providing constant-time access
|
|
|
1268 to any of the objects; access to an arbitrary object in a vector is
|
|
|
1269 faster than for lists, but the operations that can be done on a vector
|
|
|
1270 are more limited.
|
|
|
1271 @item string
|
|
|
1272 Self-explanatory; behaves much like a vector of chars
|
|
|
1273 but has a different read syntax and is stored and manipulated
|
|
|
1274 more compactly.
|
|
|
1275 @item bit-vector
|
|
|
1276 A vector of bits; similar to a string in spirit.
|
|
|
1277 @item compiled-function
|
|
|
1278 An object containing compiled Lisp code, known as @dfn{byte code}.
|
|
|
1279 @item subr
|
|
|
1280 A Lisp primitive, i.e. a Lisp-callable function implemented in C.
|
|
|
1281 @end table
|
|
|
1282
|
|
|
1283 @cindex closure
|
|
|
1284 Note that there is no basic ``function'' type, as in more powerful
|
|
|
1285 versions of Lisp (where it's called a @dfn{closure}). XEmacs Lisp does
|
|
|
1286 not provide the closure semantics implemented by Common Lisp and Scheme.
|
|
|
1287 The guts of a function in XEmacs Lisp are represented in one of four
|
|
|
1288 ways: a symbol specifying another function (when one function is an
|
|
|
1289 alias for another), a list (whose first element must be the symbol
|
|
|
1290 @code{lambda}) containing the function's source code, a
|
|
|
1291 compiled-function object, or a subr object. (In other words, given a
|
|
|
1292 symbol specifying the name of a function, calling @code{symbol-function}
|
|
|
1293 to retrieve the contents of the symbol's function cell will return one
|
|
|
1294 of these types of objects.)
|
|
|
1295
|
|
|
1296 XEmacs Lisp also contains numerous specialized objects used to implement
|
|
|
1297 the editor:
|
|
|
1298
|
|
|
1299 @table @code
|
|
|
1300 @item buffer
|
|
|
1301 Stores text like a string, but is optimized for insertion and deletion
|
|
|
1302 and has certain other properties that can be set.
|
|
|
1303 @item frame
|
|
|
1304 An object with various properties whose displayable representation is a
|
|
|
1305 @dfn{window} in window-system parlance.
|
|
|
1306 @item window
|
|
|
1307 A section of a frame that displays the contents of a buffer;
|
|
|
1308 often called a @dfn{pane} in window-system parlance.
|
|
|
1309 @item window-configuration
|
|
|
1310 An object that represents a saved configuration of windows in a frame.
|
|
|
1311 @item device
|
|
|
1312 An object representing a screen on which frames can be displayed;
|
|
|
1313 equivalent to a @dfn{display} in the X Window System and a @dfn{TTY} in
|
|
|
1314 character mode.
|
|
|
1315 @item face
|
|
|
1316 An object specifying the appearance of text or graphics; it has
|
|
|
1317 properties such as font, foreground color, and background color.
|
|
|
1318 @item marker
|
|
|
1319 An object that refers to a particular position in a buffer and moves
|
|
|
1320 around as text is inserted and deleted to stay in the same relative
|
|
|
1321 position to the text around it.
|
|
|
1322 @item extent
|
|
|
1323 Similar to a marker but covers a range of text in a buffer; can also
|
|
|
1324 specify properties of the text, such as a face in which the text is to
|
|
|
1325 be displayed, whether the text is invisible or unmodifiable, etc.
|
|
|
1326 @item event
|
|
|
1327 Generated by calling @code{next-event} and contains information
|
|
|
1328 describing a particular event happening in the system, such as the user
|
|
|
1329 pressing a key or a process terminating.
|
|
|
1330 @item keymap
|
|
|
1331 An object that maps from events (described using lists, vectors, and
|
|
|
1332 symbols rather than with an event object because the mapping is for
|
|
|
1333 classes of events, rather than individual events) to functions to
|
|
|
1334 execute or other events to recursively look up; the functions are
|
|
|
1335 described by name, using a symbol, or using lists to specify the
|
|
|
1336 function's code.
|
|
|
1337 @item glyph
|
|
|
1338 An object that describes the appearance of an image (e.g. pixmap) on
|
|
|
1339 the screen; glyphs can be attached to the beginning or end of extents
|
|
|
1340 and in some future version of XEmacs will be able to be inserted
|
|
|
1341 directly into a buffer.
|
|
|
1342 @item process
|
|
|
1343 An object that describes a connection to an externally-running process.
|
|
|
1344 @end table
|
|
|
1345
|
|
|
1346 There are some other, less-commonly-encountered general objects:
|
|
|
1347
|
|
|
1348 @table @code
|
|
|
1349 @item hash-table
|
|
|
1350 An object that maps from an arbitrary Lisp object to another arbitrary
|
|
|
1351 Lisp object, using hashing for fast lookup.
|
|
|
1352 @item obarray
|
|
|
1353 A limited form of hash-table that maps from strings to symbols; obarrays
|
|
|
1354 are used to look up a symbol given its name and are not actually their
|
|
|
1355 own object type but are kludgily represented using vectors with hidden
|
|
|
1356 fields (this representation derives from GNU Emacs).
|
|
|
1357 @item specifier
|
|
|
1358 A complex object used to specify the value of a display property; a
|
|
|
1359 default value is given and different values can be specified for
|
|
|
1360 particular frames, buffers, windows, devices, or classes of device.
|
|
|
1361 @item char-table
|
|
|
1362 An object that maps from chars or classes of chars to arbitrary Lisp
|
|
|
1363 objects; internally char tables use a complex nested-vector
|
|
|
1364 representation that is optimized to the way characters are represented
|
|
|
1365 as integers.
|
|
|
1366 @item range-table
|
|
|
1367 An object that maps from ranges of integers to arbitrary Lisp objects.
|
|
|
1368 @end table
|
|
|
1369
|
|
|
1370 And some strange special-purpose objects:
|
|
|
1371
|
|
|
1372 @table @code
|
|
|
1373 @item charset
|
|
|
1374 @itemx coding-system
|
|
|
1375 Objects used when MULE, or multi-lingual/Asian-language, support is
|
|
|
1376 enabled.
|
|
|
1377 @item color-instance
|
|
|
1378 @itemx font-instance
|
|
|
1379 @itemx image-instance
|
|
|
1380 An object that encapsulates a window-system resource; instances are
|
|
|
1381 mostly used internally but are exposed on the Lisp level for cleanness
|
|
|
1382 of the specifier model and because it's occasionally useful for Lisp
|
|
|
1383 program to create or query the properties of instances.
|
|
|
1384 @item subwindow
|
|
|
1385 An object that encapsulate a @dfn{subwindow} resource, i.e. a
|
|
|
1386 window-system child window that is drawn into by an external process;
|
|
|
1387 this object should be integrated into the glyph system but isn't yet,
|
|
|
1388 and may change form when this is done.
|
|
|
1389 @item tooltalk-message
|
|
|
1390 @itemx tooltalk-pattern
|
|
|
1391 Objects that represent resources used in the ToolTalk interprocess
|
|
|
1392 communication protocol.
|
|
|
1393 @item toolbar-button
|
|
|
1394 An object used in conjunction with the toolbar.
|
|
|
1395 @end table
|
|
|
1396
|
|
|
1397 And objects that are only used internally:
|
|
|
1398
|
|
|
1399 @table @code
|
|
|
1400 @item opaque
|
|
|
1401 A generic object for encapsulating arbitrary memory; this allows you the
|
|
|
1402 generality of @code{malloc()} and the convenience of the Lisp object
|
|
|
1403 system.
|
|
|
1404 @item lstream
|
|
|
1405 A buffering I/O stream, used to provide a unified interface to anything
|
|
|
1406 that can accept output or provide input, such as a file descriptor, a
|
|
|
1407 stdio stream, a chunk of memory, a Lisp buffer, a Lisp string, etc.;
|
|
|
1408 it's a Lisp object to make its memory management more convenient.
|
|
|
1409 @item char-table-entry
|
|
|
1410 Subsidiary objects in the internal char-table representation.
|
|
|
1411 @item extent-auxiliary
|
|
|
1412 @itemx menubar-data
|
|
|
1413 @itemx toolbar-data
|
|
|
1414 Various special-purpose objects that are basically just used to
|
|
|
1415 encapsulate memory for particular subsystems, similar to the more
|
|
|
1416 general ``opaque'' object.
|
|
|
1417 @item symbol-value-forward
|
|
|
1418 @itemx symbol-value-buffer-local
|
|
|
1419 @itemx symbol-value-varalias
|
|
|
1420 @itemx symbol-value-lisp-magic
|
|
|
1421 Special internal-only objects that are placed in the value cell of a
|
|
|
1422 symbol to indicate that there is something special with this variable --
|
|
|
1423 e.g. it has no value, it mirrors another variable, or it mirrors some C
|
|
|
1424 variable; there is really only one kind of object, called a
|
|
|
1425 @dfn{symbol-value-magic}, but it is sort-of halfway kludged into
|
|
|
1426 semi-different object types.
|
|
|
1427 @end table
|
|
|
1428
|
|
|
1429 @cindex permanent objects
|
|
|
1430 @cindex temporary objects
|
|
|
1431 Some types of objects are @dfn{permanent}, meaning that once created,
|
|
|
1432 they do not disappear until explicitly destroyed, using a function such
|
|
|
1433 as @code{delete-buffer}, @code{delete-window}, @code{delete-frame}, etc.
|
|
|
1434 Others will disappear once they are not longer used, through the garbage
|
|
|
1435 collection mechanism. Buffers, frames, windows, devices, and processes
|
|
|
1436 are among the objects that are permanent. Note that some objects can go
|
|
|
1437 both ways: Faces can be created either way; extents are normally
|
|
|
1438 permanent, but detached extents (extents not referring to any text, as
|
|
|
1439 happens to some extents when the text they are referring to is deleted)
|
|
|
1440 are temporary. Note that some permanent objects, such as faces and
|
|
|
1441 coding systems, cannot be deleted. Note also that windows are unique in
|
|
|
1442 that they can be @emph{undeleted} after having previously been
|
|
|
1443 deleted. (This happens as a result of restoring a window configuration.)
|
|
|
1444
|
|
|
1445 @cindex read syntax
|
|
|
1446 Note that many types of objects have a @dfn{read syntax}, i.e. a way of
|
|
|
1447 specifying an object of that type in Lisp code. When you load a Lisp
|
|
|
1448 file, or type in code to be evaluated, what really happens is that the
|
|
|
1449 function @code{read} is called, which reads some text and creates an object
|
|
|
1450 based on the syntax of that text; then @code{eval} is called, which
|
|
|
1451 possibly does something special; then this loop repeats until there's
|
|
|
1452 no more text to read. (@code{eval} only actually does something special
|
|
|
1453 with symbols, which causes the symbol's value to be returned,
|
|
|
1454 similar to referencing a variable; and with conses [i.e. lists],
|
|
|
1455 which cause a function invocation. All other values are returned
|
|
|
1456 unchanged.)
|
|
|
1457
|
|
|
1458 The read syntax
|
|
|
1459
|
|
|
1460 @example
|
|
|
1461 17297
|
|
|
1462 @end example
|
|
|
1463
|
|
|
1464 converts to an integer whose value is 17297.
|
|
|
1465
|
|
|
1466 @example
|
|
|
1467 1.983e-4
|
|
|
1468 @end example
|
|
|
1469
|
|
|
1470 converts to a float whose value is 1.983e-4, or .0001983.
|
|
|
1471
|
|
|
1472 @example
|
|
|
1473 ?b
|
|
|
1474 @end example
|
|
|
1475
|
|
|
1476 converts to a char that represents the lowercase letter b.
|
|
|
1477
|
|
|
1478 @example
|
|
|
1479 ?^[$(B#&^[(B
|
|
|
1480 @end example
|
|
|
1481
|
|
|
1482 (where @samp{^[} actually is an @samp{ESC} character) converts to a
|
|
|
1483 particular Kanji character when using an ISO2022-based coding system for
|
|
|
1484 input. (To decode this goo: @samp{ESC} begins an escape sequence;
|
|
|
1485 @samp{ESC $ (} is a class of escape sequences meaning ``switch to a
|
|
|
1486 94x94 character set''; @samp{ESC $ ( B} means ``switch to Japanese
|
|
|
1487 Kanji''; @samp{#} and @samp{&} collectively index into a 94-by-94 array
|
|
|
1488 of characters [subtract 33 from the ASCII value of each character to get
|
|
|
1489 the corresponding index]; @samp{ESC (} is a class of escape sequences
|
|
|
1490 meaning ``switch to a 94 character set''; @samp{ESC (B} means ``switch
|
|
|
1491 to US ASCII''. It is a coincidence that the letter @samp{B} is used to
|
|
|
1492 denote both Japanese Kanji and US ASCII. If the first @samp{B} were
|
|
|
1493 replaced with an @samp{A}, you'd be requesting a Chinese Hanzi character
|
|
|
1494 from the GB2312 character set.)
|
|
|
1495
|
|
|
1496 @example
|
|
|
1497 "foobar"
|
|
|
1498 @end example
|
|
|
1499
|
|
|
1500 converts to a string.
|
|
|
1501
|
|
|
1502 @example
|
|
|
1503 foobar
|
|
|
1504 @end example
|
|
|
1505
|
|
|
1506 converts to a symbol whose name is @code{"foobar"}. This is done by
|
|
|
1507 looking up the string equivalent in the global variable
|
|
|
1508 @code{obarray}, whose contents should be an obarray. If no symbol
|
|
|
1509 is found, a new symbol with the name @code{"foobar"} is automatically
|
|
|
1510 created and added to @code{obarray}; this process is called
|
|
|
1511 @dfn{interning} the symbol.
|
|
|
1512 @cindex interning
|
|
|
1513
|
|
|
1514 @example
|
|
|
1515 (foo . bar)
|
|
|
1516 @end example
|
|
|
1517
|
|
|
1518 converts to a cons cell containing the symbols @code{foo} and @code{bar}.
|
|
|
1519
|
|
|
1520 @example
|
|
|
1521 (1 a 2.5)
|
|
|
1522 @end example
|
|
|
1523
|
|
|
1524 converts to a three-element list containing the specified objects
|
|
|
1525 (note that a list is actually a set of nested conses; see the
|
|
|
1526 XEmacs Lisp Reference).
|
|
|
1527
|
|
|
1528 @example
|
|
|
1529 [1 a 2.5]
|
|
|
1530 @end example
|
|
|
1531
|
|
|
1532 converts to a three-element vector containing the specified objects.
|
|
|
1533
|
|
|
1534 @example
|
|
|
1535 #[... ... ... ...]
|
|
|
1536 @end example
|
|
|
1537
|
|
|
1538 converts to a compiled-function object (the actual contents are not
|
|
|
1539 shown since they are not relevant here; look at a file that ends with
|
|
|
1540 @file{.elc} for examples).
|
|
|
1541
|
|
|
1542 @example
|
|
|
1543 #*01110110
|
|
|
1544 @end example
|
|
|
1545
|
|
|
1546 converts to a bit-vector.
|
|
|
1547
|
|
|
1548 @example
|
|
|
1549 #s(hash-table ... ...)
|
|
|
1550 @end example
|
|
|
1551
|
|
|
1552 converts to a hash table (the actual contents are not shown).
|
|
|
1553
|
|
|
1554 @example
|
|
|
1555 #s(range-table ... ...)
|
|
|
1556 @end example
|
|
|
1557
|
|
|
1558 converts to a range table (the actual contents are not shown).
|
|
|
1559
|
|
|
1560 @example
|
|
|
1561 #s(char-table ... ...)
|
|
|
1562 @end example
|
|
|
1563
|
|
|
1564 converts to a char table (the actual contents are not shown).
|
|
|
1565
|
|
|
1566 Note that the @code{#s()} syntax is the general syntax for structures,
|
|
|
1567 which are not really implemented in XEmacs Lisp but should be.
|
|
|
1568
|
|
|
1569 When an object is printed out (using @code{print} or a related
|
|
|
1570 function), the read syntax is used, so that the same object can be read
|
|
|
1571 in again.
|
|
|
1572
|
|
|
1573 The other objects do not have read syntaxes, usually because it does not
|
|
|
1574 really make sense to create them in this fashion (i.e. processes, where
|
|
|
1575 it doesn't make sense to have a subprocess created as a side effect of
|
|
|
1576 reading some Lisp code), or because they can't be created at all
|
|
|
1577 (e.g. subrs). Permanent objects, as a rule, do not have a read syntax;
|
|
|
1578 nor do most complex objects, which contain too much state to be easily
|
|
|
1579 initialized through a read syntax.
|
|
|
1580
|
|
|
1581 @node How Lisp Objects Are Represented in C, Rules When Writing New C Code, The XEmacs Object System (Abstractly Speaking), Top
|
|
|
1582 @chapter How Lisp Objects Are Represented in C
|
|
|
1583
|
|
|
1584 Lisp objects are represented in C using a 32-bit or 64-bit machine word
|
|
|
1585 (depending on the processor; i.e. DEC Alphas use 64-bit Lisp objects and
|
|
|
1586 most other processors use 32-bit Lisp objects). The representation
|
|
|
1587 stuffs a pointer together with a tag, as follows:
|
|
|
1588
|
|
|
1589 @example
|
|
|
1590 [ 3 3 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 ]
|
|
|
1591 [ 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 ]
|
|
|
1592
|
|
|
1593 <---> ^ <------------------------------------------------------>
|
|
|
1594 tag | a pointer to a structure, or an integer
|
|
|
1595 |
|
|
|
1596 mark bit
|
|
|
1597 @end example
|
|
|
1598
|
|
|
1599 The tag describes the type of the Lisp object. For integers and chars,
|
|
|
1600 the lower 28 bits contain the value of the integer or char; for all
|
|
|
1601 others, the lower 28 bits contain a pointer. The mark bit is used
|
|
|
1602 during garbage-collection, and is always 0 when garbage collection is
|
|
|
1603 not happening. (The way that garbage collection works, basically, is that it
|
|
440
|
1604 loops over all places where Lisp objects could exist---this includes
|
|
428
|
1605 all global variables in C that contain Lisp objects [including
|
|
|
1606 @code{Vobarray}, the C equivalent of @code{obarray}; through this, all
|
|
440
|
1607 Lisp variables will get marked], plus various other places---and
|
|
428
|
1608 recursively scans through the Lisp objects, marking each object it finds
|
|
|
1609 by setting the mark bit. Then it goes through the lists of all objects
|
|
|
1610 allocated, freeing the ones that are not marked and turning off the mark
|
|
|
1611 bit of the ones that are marked.)
|
|
|
1612
|
|
|
1613 Lisp objects use the typedef @code{Lisp_Object}, but the actual C type
|
|
|
1614 used for the Lisp object can vary. It can be either a simple type
|
|
|
1615 (@code{long} on the DEC Alpha, @code{int} on other machines) or a
|
|
|
1616 structure whose fields are bit fields that line up properly (actually, a
|
|
|
1617 union of structures is used). Generally the simple integral type is
|
|
|
1618 preferable because it ensures that the compiler will actually use a
|
|
|
1619 machine word to represent the object (some compilers will use more
|
|
|
1620 general and less efficient code for unions and structs even if they can
|
|
|
1621 fit in a machine word). The union type, however, has the advantage of
|
|
|
1622 stricter type checking (if you accidentally pass an integer where a Lisp
|
|
|
1623 object is desired, you get a compile error), and it makes it easier to
|
|
|
1624 decode Lisp objects when debugging. The choice of which type to use is
|
|
|
1625 determined by the preprocessor constant @code{USE_UNION_TYPE} which is
|
|
|
1626 defined via the @code{--use-union-type} option to @code{configure}.
|
|
|
1627
|
|
|
1628 @cindex record type
|
|
|
1629
|
|
|
1630 Note that there are only eight types that the tag can represent, but
|
|
|
1631 many more actual types than this. This is handled by having one of the
|
|
|
1632 tag types specify a meta-type called a @dfn{record}; for all such
|
|
|
1633 objects, the first four bytes of the pointed-to structure indicate what
|
|
|
1634 the actual type is.
|
|
|
1635
|
|
|
1636 Note also that having 28 bits for pointers and integers restricts a lot
|
|
|
1637 of things to 256 megabytes of memory. (Basically, enough pointers and
|
|
|
1638 indices and whatnot get stuffed into Lisp objects that the total amount
|
|
|
1639 of memory used by XEmacs can't grow above 256 megabytes. In older
|
|
|
1640 versions of XEmacs and GNU Emacs, the tag was 5 bits wide, allowing for
|
|
|
1641 32 types, which was more than the actual number of types that existed at
|
|
|
1642 the time, and no ``record'' type was necessary. However, this limited
|
|
|
1643 the editor to 64 megabytes total, which some users who edited large
|
|
|
1644 files might conceivably exceed.)
|
|
|
1645
|
|
|
1646 Also, note that there is an implicit assumption here that all pointers
|
|
|
1647 are low enough that the top bits are all zero and can just be chopped
|
|
|
1648 off. On standard machines that allocate memory from the bottom up (and
|
|
|
1649 give each process its own address space), this works fine. Some
|
|
|
1650 machines, however, put the data space somewhere else in memory
|
|
|
1651 (e.g. beginning at 0x80000000). Those machines cope by defining
|
|
|
1652 @code{DATA_SEG_BITS} in the corresponding @file{m/} or @file{s/} file to
|
|
|
1653 the proper mask. Then, pointers retrieved from Lisp objects are
|
|
|
1654 automatically OR'ed with this value prior to being used.
|
|
|
1655
|
|
|
1656 A corollary of the previous paragraph is that @strong{(pointers to)
|
|
|
1657 stack-allocated structures cannot be put into Lisp objects}. The stack
|
|
|
1658 is generally located near the top of memory; if you put such a pointer
|
|
|
1659 into a Lisp object, it will get its top bits chopped off, and you will
|
|
|
1660 lose.
|
|
|
1661
|
|
|
1662 Actually, there's an alternative representation of a @code{Lisp_Object},
|
|
|
1663 invented by Kyle Jones, that is used when the
|
|
|
1664 @code{--use-minimal-tagbits} option to @code{configure} is used. In
|
|
|
1665 this case the 2 lower bits are used for the tag bits. This
|
|
|
1666 representation assumes that pointers to structs are always aligned to
|
|
|
1667 multiples of 4, so the lower 2 bits are always zero.
|
|
|
1668
|
|
|
1669 @example
|
|
|
1670 [ 3 3 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 ]
|
|
|
1671 [ 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 ]
|
|
|
1672
|
|
|
1673 <---------------------------------------------------------> <->
|
|
|
1674 a pointer to a structure, or an integer tag
|
|
|
1675 @end example
|
|
|
1676
|
|
|
1677 A tag of 00 is used for all pointer object types, a tag of 10 is used
|
|
|
1678 for characters, and the other two tags 01 and 11 are joined together to
|
|
|
1679 form the integer object type. The markbit is moved to part of the
|
|
|
1680 structure being pointed at (integers and chars do not need to be marked,
|
|
|
1681 since no memory is allocated). This representation has these
|
|
|
1682 advantages:
|
|
|
1683
|
|
|
1684 @enumerate
|
|
|
1685 @item
|
|
|
1686 31 bits can be used for Lisp Integers.
|
|
|
1687 @item
|
|
|
1688 @emph{Any} pointer can be represented directly, and no bit masking
|
|
|
1689 operations are necessary.
|
|
|
1690 @end enumerate
|
|
|
1691
|
|
|
1692 The disadvantages are:
|
|
|
1693
|
|
|
1694 @enumerate
|
|
|
1695 @item
|
|
|
1696 An extra level of indirection is needed when accessing the object types
|
|
|
1697 that were not record types. So checking whether a Lisp object is a cons
|
|
|
1698 cell becomes a slower operation.
|
|
|
1699 @item
|
|
|
1700 Mark bits can no longer be stored directly in Lisp objects, so another
|
|
|
1701 place for them must be found. This means that a cons cell requires more
|
|
|
1702 memory than merely room for 2 lisp objects, leading to extra memory use.
|
|
|
1703 @end enumerate
|
|
|
1704
|
|
|
1705 Various macros are used to construct Lisp objects and extract the
|
|
|
1706 components. Macros of the form @code{XINT()}, @code{XCHAR()},
|
|
|
1707 @code{XSTRING()}, @code{XSYMBOL()}, etc. mask out the pointer/integer
|
|
|
1708 field and cast it to the appropriate type. All of the macros that
|
|
|
1709 construct pointers will @code{OR} with @code{DATA_SEG_BITS} if
|
|
|
1710 necessary. @code{XINT()} needs to be a bit tricky so that negative
|
|
|
1711 numbers are properly sign-extended: Usually it does this by shifting the
|
|
|
1712 number four bits to the left and then four bits to the right. This
|
|
|
1713 assumes that the right-shift operator does an arithmetic shift (i.e. it
|
|
|
1714 leaves the most-significant bit as-is rather than shifting in a zero, so
|
|
|
1715 that it mimics a divide-by-two even for negative numbers). Not all
|
|
|
1716 machines/compilers do this, and on the ones that don't, a more
|
|
|
1717 complicated definition is selected by defining
|
|
|
1718 @code{EXPLICIT_SIGN_EXTEND}.
|
|
|
1719
|
|
|
1720 Note that when @code{ERROR_CHECK_TYPECHECK} is defined, the extractor
|
|
440
|
1721 macros become more complicated---they check the tag bits and/or the
|
|
428
|
1722 type field in the first four bytes of a record type to ensure that the
|
|
|
1723 object is really of the correct type. This is great for catching places
|
|
440
|
1724 where an incorrect type is being dereferenced---this typically results
|
|
428
|
1725 in a pointer being dereferenced as the wrong type of structure, with
|
|
|
1726 unpredictable (and sometimes not easily traceable) results.
|
|
|
1727
|
|
|
1728 There are similar @code{XSET@var{TYPE}()} macros that construct a Lisp
|
|
|
1729 object. These macros are of the form @code{XSET@var{TYPE}
|
|
|
1730 (@var{lvalue}, @var{result})},
|
|
|
1731 i.e. they have to be a statement rather than just used in an expression.
|
|
|
1732 The reason for this is that standard C doesn't let you ``construct'' a
|
|
|
1733 structure (but GCC does). Granted, this sometimes isn't too convenient;
|
|
|
1734 for the case of integers, at least, you can use the function
|
|
|
1735 @code{make_int()}, which constructs and @emph{returns} an integer
|
|
|
1736 Lisp object. Note that the @code{XSET@var{TYPE}()} macros are also
|
|
|
1737 affected by @code{ERROR_CHECK_TYPECHECK} and make sure that the
|
|
|
1738 structure is of the right type in the case of record types, where the
|
|
|
1739 type is contained in the structure.
|
|
|
1740
|
|
|
1741 The C programmer is responsible for @strong{guaranteeing} that a
|
|
|
1742 Lisp_Object is is the correct type before using the @code{X@var{TYPE}}
|
|
|
1743 macros. This is especially important in the case of lists. Use
|
|
|
1744 @code{XCAR} and @code{XCDR} if a Lisp_Object is certainly a cons cell,
|
|
|
1745 else use @code{Fcar()} and @code{Fcdr()}. Trust other C code, but not
|
|
|
1746 Lisp code. On the other hand, if XEmacs has an internal logic error,
|
|
|
1747 it's better to crash immediately, so sprinkle ``unreachable''
|
|
|
1748 @code{abort()}s liberally about the source code.
|
|
|
1749
|
|
|
1750 @node Rules When Writing New C Code, A Summary of the Various XEmacs Modules, How Lisp Objects Are Represented in C, Top
|
|
|
1751 @chapter Rules When Writing New C Code
|
|
|
1752
|
|
|
1753 The XEmacs C Code is extremely complex and intricate, and there are many
|
|
|
1754 rules that are more or less consistently followed throughout the code.
|
|
|
1755 Many of these rules are not obvious, so they are explained here. It is
|
|
|
1756 of the utmost importance that you follow them. If you don't, you may
|
|
|
1757 get something that appears to work, but which will crash in odd
|
|
|
1758 situations, often in code far away from where the actual breakage is.
|
|
|
1759
|
|
|
1760 @menu
|
|
|
1761 * General Coding Rules::
|
|
|
1762 * Writing Lisp Primitives::
|
|
|
1763 * Adding Global Lisp Variables::
|
|
|
1764 * Coding for Mule::
|
|
|
1765 * Techniques for XEmacs Developers::
|
|
|
1766 @end menu
|
|
|
1767
|
|
|
1768 @node General Coding Rules
|
|
|
1769 @section General Coding Rules
|
|
|
1770
|
|
|
1771 The C code is actually written in a dialect of C called @dfn{Clean C},
|
|
|
1772 meaning that it can be compiled, mostly warning-free, with either a C or
|
|
|
1773 C++ compiler. Coding in Clean C has several advantages over plain C.
|
|
|
1774 C++ compilers are more nit-picking, and a number of coding errors have
|
|
|
1775 been found by compiling with C++. The ability to use both C and C++
|
|
|
1776 tools means that a greater variety of development tools are available to
|
|
|
1777 the developer.
|
|
|
1778
|
|
|
1779 Almost every module contains a @code{syms_of_*()} function and a
|
|
|
1780 @code{vars_of_*()} function. The former declares any Lisp primitives
|
|
|
1781 you have defined and defines any symbols you will be using. The latter
|
|
|
1782 declares any global Lisp variables you have added and initializes global
|
|
|
1783 C variables in the module. For each such function, declare it in
|
|
|
1784 @file{symsinit.h} and make sure it's called in the appropriate place in
|
|
|
1785 @file{emacs.c}. @strong{Important}: There are stringent requirements on
|
|
|
1786 exactly what can go into these functions. See the comment in
|
|
|
1787 @file{emacs.c}. The reason for this is to avoid obscure unwanted
|
|
|
1788 interactions during initialization. If you don't follow these rules,
|
|
|
1789 you'll be sorry! If you want to do anything that isn't allowed, create
|
|
|
1790 a @code{complex_vars_of_*()} function for it. Doing this is tricky,
|
|
|
1791 though: You have to make sure your function is called at the right time
|
|
|
1792 so that all the initialization dependencies work out.
|
|
|
1793
|
|
|
1794 Every module includes @file{<config.h>} (angle brackets so that
|
|
|
1795 @samp{--srcdir} works correctly; @file{config.h} may or may not be in
|
|
|
1796 the same directory as the C sources) and @file{lisp.h}. @file{config.h}
|
|
|
1797 must always be included before any other header files (including
|
|
|
1798 system header files) to ensure that certain tricks played by various
|
|
|
1799 @file{s/} and @file{m/} files work out correctly.
|
|
|
1800
|
|
440
|
1801 When including header files, always use angle brackets, not double
|
|
|
1802 quotes, except when the file to be included is in the same directory as
|
|
|
1803 the including file. If either file is a generated file, then that is
|
|
|
1804 not likely to be the case. In order to understand why we have this
|
|
|
1805 rule, imagine what happens when you do a build in the source directory
|
|
|
1806 using @samp{./configure} and another build in another directory using
|
|
|
1807 @samp{../work/configure}. There will be two different @file{config.h}
|
|
|
1808 files. Which one will be used if you @samp{#include "config.h"}?
|
|
|
1809
|
|
428
|
1810 @strong{All global and static variables that are to be modifiable must
|
|
|
1811 be declared uninitialized.} This means that you may not use the
|
|
|
1812 ``declare with initializer'' form for these variables, such as @code{int
|
|
|
1813 some_variable = 0;}. The reason for this has to do with some kludges
|
|
|
1814 done during the dumping process: If possible, the initialized data
|
|
|
1815 segment is re-mapped so that it becomes part of the (unmodifiable) code
|
|
|
1816 segment in the dumped executable. This allows this memory to be shared
|
|
|
1817 among multiple running XEmacs processes. XEmacs is careful to place as
|
|
|
1818 much constant data as possible into initialized variables (in
|
|
440
|
1819 particular, into what's called the @dfn{pure space}---see below) during
|
|
428
|
1820 the @file{temacs} phase.
|
|
|
1821
|
|
|
1822 @cindex copy-on-write
|
|
|
1823 @strong{Please note:} This kludge only works on a few systems nowadays,
|
|
|
1824 and is rapidly becoming irrelevant because most modern operating systems
|
|
|
1825 provide @dfn{copy-on-write} semantics. All data is initially shared
|
|
|
1826 between processes, and a private copy is automatically made (on a
|
|
|
1827 page-by-page basis) when a process first attempts to write to a page of
|
|
|
1828 memory.
|
|
|
1829
|
|
|
1830 Formerly, there was a requirement that static variables not be declared
|
|
|
1831 inside of functions. This had to do with another hack along the same
|
|
|
1832 vein as what was just described: old USG systems put statically-declared
|
|
|
1833 variables in the initialized data space, so those header files had a
|
|
|
1834 @code{#define static} declaration. (That way, the data-segment remapping
|
|
|
1835 described above could still work.) This fails badly on static variables
|
|
|
1836 inside of functions, which suddenly become automatic variables;
|
|
|
1837 therefore, you weren't supposed to have any of them. This awful kludge
|
|
|
1838 has been removed in XEmacs because
|
|
|
1839
|
|
|
1840 @enumerate
|
|
|
1841 @item
|
|
|
1842 almost all of the systems that used this kludge ended up having
|
|
|
1843 to disable the data-segment remapping anyway;
|
|
|
1844 @item
|
|
|
1845 the only systems that didn't were extremely outdated ones;
|
|
|
1846 @item
|
|
|
1847 this hack completely messed up inline functions.
|
|
|
1848 @end enumerate
|
|
|
1849
|
|
|
1850 The C source code makes heavy use of C preprocessor macros. One popular
|
|
|
1851 macro style is:
|
|
|
1852
|
|
|
1853 @example
|
|
440
|
1854 #define FOO(var, value) do @{ \
|
|
|
1855 Lisp_Object FOO_value = (value); \
|
|
|
1856 ... /* compute using FOO_value */ \
|
|
|
1857 (var) = bar; \
|
|
428
|
1858 @} while (0)
|
|
|
1859 @end example
|
|
|
1860
|
|
|
1861 The @code{do @{...@} while (0)} is a standard trick to allow FOO to have
|
|
|
1862 statement semantics, so that it can safely be used within an @code{if}
|
|
|
1863 statement in C, for example. Multiple evaluation is prevented by
|
|
|
1864 copying a supplied argument into a local variable, so that
|
|
|
1865 @code{FOO(var,fun(1))} only calls @code{fun} once.
|
|
|
1866
|
|
|
1867 Lisp lists are popular data structures in the C code as well as in
|
|
|
1868 Elisp. There are two sets of macros that iterate over lists.
|
|
|
1869 @code{EXTERNAL_LIST_LOOP_@var{n}} should be used when the list has been
|
|
|
1870 supplied by the user, and cannot be trusted to be acyclic and
|
|
|
1871 nil-terminated. A @code{malformed-list} or @code{circular-list} error
|
|
|
1872 will be generated if the list being iterated over is not entirely
|
|
|
1873 kosher. @code{LIST_LOOP_@var{n}}, on the other hand, is faster and less
|
|
|
1874 safe, and can be used only on trusted lists.
|
|
|
1875
|
|
|
1876 Related macros are @code{GET_EXTERNAL_LIST_LENGTH} and
|
|
|
1877 @code{GET_LIST_LENGTH}, which calculate the length of a list, and in the
|
|
|
1878 case of @code{GET_EXTERNAL_LIST_LENGTH}, validating the properness of
|
|
|
1879 the list. The macros @code{EXTERNAL_LIST_LOOP_DELETE_IF} and
|
|
|
1880 @code{LIST_LOOP_DELETE_IF} delete elements from a lisp list satisfying some
|
|
|
1881 predicate.
|
|
|
1882
|
|
|
1883 @node Writing Lisp Primitives
|
|
|
1884 @section Writing Lisp Primitives
|
|
|
1885
|
|
|
1886 Lisp primitives are Lisp functions implemented in C. The details of
|
|
|
1887 interfacing the C function so that Lisp can call it are handled by a few
|
|
|
1888 C macros. The only way to really understand how to write new C code is
|
|
|
1889 to read the source, but we can explain some things here.
|
|
|
1890
|
|
|
1891 An example of a special form is the definition of @code{prog1}, from
|
|
|
1892 @file{eval.c}. (An ordinary function would have the same general
|
|
|
1893 appearance.)
|
|
|
1894
|
|
|
1895 @cindex garbage collection protection
|
|
|
1896 @smallexample
|
|
|
1897 @group
|
|
|
1898 DEFUN ("prog1", Fprog1, 1, UNEVALLED, 0, /*
|
|
|
1899 Similar to `progn', but the value of the first form is returned.
|
|
|
1900 \(prog1 FIRST BODY...): All the arguments are evaluated sequentially.
|
|
|
1901 The value of FIRST is saved during evaluation of the remaining args,
|
|
|
1902 whose values are discarded.
|
|
|
1903 */
|
|
|
1904 (args))
|
|
|
1905 @{
|
|
|
1906 /* This function can GC */
|
|
|
1907 REGISTER Lisp_Object val, form, tail;
|
|
|
1908 struct gcpro gcpro1;
|
|
|
1909
|
|
|
1910 val = Feval (XCAR (args));
|
|
|
1911
|
|
|
1912 GCPRO1 (val);
|
|
|
1913
|
|
|
1914 LIST_LOOP_3 (form, XCDR (args), tail)
|
|
|
1915 Feval (form);
|
|
|
1916
|
|
|
1917 UNGCPRO;
|
|
|
1918 return val;
|
|
|
1919 @}
|
|
|
1920 @end group
|
|
|
1921 @end smallexample
|
|
|
1922
|
|
|
1923 Let's start with a precise explanation of the arguments to the
|
|
|
1924 @code{DEFUN} macro. Here is a template for them:
|
|
|
1925
|
|
|
1926 @example
|
|
|
1927 @group
|
|
|
1928 DEFUN (@var{lname}, @var{fname}, @var{min_args}, @var{max_args}, @var{interactive}, /*
|
|
|
1929 @var{docstring}
|
|
|
1930 */
|
|
|
1931 (@var{arglist}))
|
|
|
1932 @end group
|
|
|
1933 @end example
|
|
|
1934
|
|
|
1935 @table @var
|
|
|
1936 @item lname
|
|
|
1937 This string is the name of the Lisp symbol to define as the function
|
|
|
1938 name; in the example above, it is @code{"prog1"}.
|
|
|
1939
|
|
|
1940 @item fname
|
|
|
1941 This is the C function name for this function. This is the name that is
|
|
|
1942 used in C code for calling the function. The name is, by convention,
|
|
|
1943 @samp{F} prepended to the Lisp name, with all dashes (@samp{-}) in the
|
|
|
1944 Lisp name changed to underscores. Thus, to call this function from C
|
|
|
1945 code, call @code{Fprog1}. Remember that the arguments are of type
|
|
|
1946 @code{Lisp_Object}; various macros and functions for creating values of
|
|
|
1947 type @code{Lisp_Object} are declared in the file @file{lisp.h}.
|
|
|
1948
|
|
|
1949 Primitives whose names are special characters (e.g. @code{+} or
|
|
|
1950 @code{<}) are named by spelling out, in some fashion, the special
|
|
|
1951 character: e.g. @code{Fplus()} or @code{Flss()}. Primitives whose names
|
|
|
1952 begin with normal alphanumeric characters but also contain special
|
|
|
1953 characters are spelled out in some creative way, e.g. @code{let*}
|
|
|
1954 becomes @code{FletX()}.
|
|
|
1955
|
|
|
1956 Each function also has an associated structure that holds the data for
|
|
|
1957 the subr object that represents the function in Lisp. This structure
|
|
|
1958 conveys the Lisp symbol name to the initialization routine that will
|
|
|
1959 create the symbol and store the subr object as its definition. The C
|
|
|
1960 variable name of this structure is always @samp{S} prepended to the
|
|
|
1961 @var{fname}. You hardly ever need to be aware of the existence of this
|
|
|
1962 structure, since @code{DEFUN} plus @code{DEFSUBR} takes care of all the
|
|
|
1963 details.
|
|
|
1964
|
|
|
1965 @item min_args
|
|
|
1966 This is the minimum number of arguments that the function requires. The
|
|
|
1967 function @code{prog1} allows a minimum of one argument.
|
|
|
1968
|
|
|
1969 @item max_args
|
|
|
1970 This is the maximum number of arguments that the function accepts, if
|
|
|
1971 there is a fixed maximum. Alternatively, it can be @code{UNEVALLED},
|
|
|
1972 indicating a special form that receives unevaluated arguments, or
|
|
|
1973 @code{MANY}, indicating an unlimited number of evaluated arguments (the
|
|
|
1974 C equivalent of @code{&rest}). Both @code{UNEVALLED} and @code{MANY}
|
|
|
1975 are macros. If @var{max_args} is a number, it may not be less than
|
|
|
1976 @var{min_args} and it may not be greater than 8. (If you need to add a
|
|
|
1977 function with more than 8 arguments, use the @code{MANY} form. Resist
|
|
|
1978 the urge to edit the definition of @code{DEFUN} in @file{lisp.h}. If
|
|
|
1979 you do it anyways, make sure to also add another clause to the switch
|
|
|
1980 statement in @code{primitive_funcall().})
|
|
|
1981
|
|
|
1982 @item interactive
|
|
|
1983 This is an interactive specification, a string such as might be used as
|
|
|
1984 the argument of @code{interactive} in a Lisp function. In the case of
|
|
|
1985 @code{prog1}, it is 0 (a null pointer), indicating that @code{prog1}
|
|
|
1986 cannot be called interactively. A value of @code{""} indicates a
|
|
|
1987 function that should receive no arguments when called interactively.
|
|
|
1988
|
|
|
1989 @item docstring
|
|
|
1990 This is the documentation string. It is written just like a
|
|
|
1991 documentation string for a function defined in Lisp; in particular, the
|
|
|
1992 first line should be a single sentence. Note how the documentation
|
|
|
1993 string is enclosed in a comment, none of the documentation is placed on
|
|
|
1994 the same lines as the comment-start and comment-end characters, and the
|
|
|
1995 comment-start characters are on the same line as the interactive
|
|
|
1996 specification. @file{make-docfile}, which scans the C files for
|
|
|
1997 documentation strings, is very particular about what it looks for, and
|
|
|
1998 will not properly extract the doc string if it's not in this exact format.
|
|
|
1999
|
|
|
2000 In order to make both @file{etags} and @file{make-docfile} happy, make
|
|
|
2001 sure that the @code{DEFUN} line contains the @var{lname} and
|
|
|
2002 @var{fname}, and that the comment-start characters for the doc string
|
|
|
2003 are on the same line as the interactive specification, and put a newline
|
|
|
2004 directly after them (and before the comment-end characters).
|
|
|
2005
|
|
|
2006 @item arglist
|
|
|
2007 This is the comma-separated list of arguments to the C function. For a
|
|
|
2008 function with a fixed maximum number of arguments, provide a C argument
|
|
|
2009 for each Lisp argument. In this case, unlike regular C functions, the
|
|
|
2010 types of the arguments are not declared; they are simply always of type
|
|
|
2011 @code{Lisp_Object}.
|
|
|
2012
|
|
|
2013 The names of the C arguments will be used as the names of the arguments
|
|
|
2014 to the Lisp primitive as displayed in its documentation, modulo the same
|
|
|
2015 concerns described above for @code{F...} names (in particular,
|
|
|
2016 underscores in the C arguments become dashes in the Lisp arguments).
|
|
|
2017
|
|
|
2018 There is one additional kludge: A trailing `_' on the C argument is
|
|
|
2019 discarded when forming the Lisp argument. This allows C language
|
|
|
2020 reserved words (like @code{default}) or global symbols (like
|
|
|
2021 @code{dirname}) to be used as argument names without compiler warnings
|
|
|
2022 or errors.
|
|
|
2023
|
|
|
2024 A Lisp function with @w{@var{max_args} = @code{UNEVALLED}} is a
|
|
|
2025 @w{@dfn{special form}}; its arguments are not evaluated. Instead it
|
|
|
2026 receives one argument of type @code{Lisp_Object}, a (Lisp) list of the
|
|
|
2027 unevaluated arguments, conventionally named @code{(args)}.
|
|
|
2028
|
|
|
2029 When a Lisp function has no upper limit on the number of arguments,
|
|
|
2030 specify @w{@var{max_args} = @code{MANY}}. In this case its implementation in
|
|
|
2031 C actually receives exactly two arguments: the number of Lisp arguments
|
|
|
2032 (an @code{int}) and the address of a block containing their values (a
|
|
|
2033 @w{@code{Lisp_Object *}}). In this case only are the C types specified
|
|
|
2034 in the @var{arglist}: @w{@code{(int nargs, Lisp_Object *args)}}.
|
|
|
2035
|
|
|
2036 @end table
|
|
|
2037
|
|
|
2038 Within the function @code{Fprog1} itself, note the use of the macros
|
|
|
2039 @code{GCPRO1} and @code{UNGCPRO}. @code{GCPRO1} is used to ``protect''
|
|
|
2040 a variable from garbage collection---to inform the garbage collector
|
|
|
2041 that it must look in that variable and regard the object pointed at by
|
|
|
2042 its contents as an accessible object. This is necessary whenever you
|
|
|
2043 call @code{Feval} or anything that can directly or indirectly call
|
|
|
2044 @code{Feval} (this includes the @code{QUIT} macro!). At such a time,
|
|
|
2045 any Lisp object that you intend to refer to again must be protected
|
|
|
2046 somehow. @code{UNGCPRO} cancels the protection of the variables that
|
|
|
2047 are protected in the current function. It is necessary to do this
|
|
|
2048 explicitly.
|
|
|
2049
|
|
|
2050 The macro @code{GCPRO1} protects just one local variable. If you want
|
|
|
2051 to protect two, use @code{GCPRO2} instead; repeating @code{GCPRO1} will
|
|
|
2052 not work. Macros @code{GCPRO3} and @code{GCPRO4} also exist.
|
|
|
2053
|
|
|
2054 These macros implicitly use local variables such as @code{gcpro1}; you
|
|
|
2055 must declare these explicitly, with type @code{struct gcpro}. Thus, if
|
|
|
2056 you use @code{GCPRO2}, you must declare @code{gcpro1} and @code{gcpro2}.
|
|
|
2057
|
|
|
2058 @cindex caller-protects (@code{GCPRO} rule)
|
|
|
2059 Note also that the general rule is @dfn{caller-protects}; i.e. you are
|
|
|
2060 only responsible for protecting those Lisp objects that you create. Any
|
|
|
2061 objects passed to you as arguments should have been protected by whoever
|
|
|
2062 created them, so you don't in general have to protect them.
|
|
|
2063
|
|
|
2064 In particular, the arguments to any Lisp primitive are always
|
|
|
2065 automatically @code{GCPRO}ed, when called ``normally'' from Lisp code or
|
|
|
2066 bytecode. So only a few Lisp primitives that are called frequently from
|
|
|
2067 C code, such as @code{Fprogn} protect their arguments as a service to
|
|
|
2068 their caller. You don't need to protect your arguments when writing a
|
|
|
2069 new @code{DEFUN}.
|
|
|
2070
|
|
|
2071 @code{GCPRO}ing is perhaps the trickiest and most error-prone part of
|
|
|
2072 XEmacs coding. It is @strong{extremely} important that you get this
|
|
|
2073 right and use a great deal of discipline when writing this code.
|
|
|
2074 @xref{GCPROing, ,@code{GCPRO}ing}, for full details on how to do this.
|
|
|
2075
|
|
|
2076 What @code{DEFUN} actually does is declare a global structure of type
|
|
|
2077 @code{Lisp_Subr} whose name begins with capital @samp{SF} and which
|
|
|
2078 contains information about the primitive (e.g. a pointer to the
|
|
|
2079 function, its minimum and maximum allowed arguments, a string describing
|
|
|
2080 its Lisp name); @code{DEFUN} then begins a normal C function declaration
|
|
|
2081 using the @code{F...} name. The Lisp subr object that is the function
|
|
|
2082 definition of a primitive (i.e. the object in the function slot of the
|
|
|
2083 symbol that names the primitive) actually points to this @samp{SF}
|
|
|
2084 structure; when @code{Feval} encounters a subr, it looks in the
|
|
|
2085 structure to find out how to call the C function.
|
|
|
2086
|
|
|
2087 Defining the C function is not enough to make a Lisp primitive
|
|
|
2088 available; you must also create the Lisp symbol for the primitive (the
|
|
|
2089 symbol is @dfn{interned}; @pxref{Obarrays}) and store a suitable subr
|
|
|
2090 object in its function cell. (If you don't do this, the primitive won't
|
|
|
2091 be seen by Lisp code.) The code looks like this:
|
|
|
2092
|
|
|
2093 @example
|
|
|
2094 DEFSUBR (@var{fname});
|
|
|
2095 @end example
|
|
|
2096
|
|
|
2097 @noindent
|
|
|
2098 Here @var{fname} is the same name you used as the second argument to
|
|
|
2099 @code{DEFUN}.
|
|
|
2100
|
|
|
2101 This call to @code{DEFSUBR} should go in the @code{syms_of_*()} function
|
|
|
2102 at the end of the module. If no such function exists, create it and
|
|
|
2103 make sure to also declare it in @file{symsinit.h} and call it from the
|
|
|
2104 appropriate spot in @code{main()}. @xref{General Coding Rules}.
|
|
|
2105
|
|
|
2106 Note that C code cannot call functions by name unless they are defined
|
|
|
2107 in C. The way to call a function written in Lisp from C is to use
|
|
|
2108 @code{Ffuncall}, which embodies the Lisp function @code{funcall}. Since
|
|
|
2109 the Lisp function @code{funcall} accepts an unlimited number of
|
|
|
2110 arguments, in C it takes two: the number of Lisp-level arguments, and a
|
|
|
2111 one-dimensional array containing their values. The first Lisp-level
|
|
|
2112 argument is the Lisp function to call, and the rest are the arguments to
|
|
|
2113 pass to it. Since @code{Ffuncall} can call the evaluator, you must
|
|
|
2114 protect pointers from garbage collection around the call to
|
|
|
2115 @code{Ffuncall}. (However, @code{Ffuncall} explicitly protects all of
|
|
|
2116 its parameters, so you don't have to protect any pointers passed as
|
|
|
2117 parameters to it.)
|
|
|
2118
|
|
|
2119 The C functions @code{call0}, @code{call1}, @code{call2}, and so on,
|
|
|
2120 provide handy ways to call a Lisp function conveniently with a fixed
|
|
|
2121 number of arguments. They work by calling @code{Ffuncall}.
|
|
|
2122
|
|
|
2123 @file{eval.c} is a very good file to look through for examples;
|
|
|
2124 @file{lisp.h} contains the definitions for important macros and
|
|
|
2125 functions.
|
|
|
2126
|
|
|
2127 @node Adding Global Lisp Variables
|
|
|
2128 @section Adding Global Lisp Variables
|
|
|
2129
|
|
|
2130 Global variables whose names begin with @samp{Q} are constants whose
|
|
|
2131 value is a symbol of a particular name. The name of the variable should
|
|
|
2132 be derived from the name of the symbol using the same rules as for Lisp
|
|
|
2133 primitives. These variables are initialized using a call to
|
|
|
2134 @code{defsymbol()} in the @code{syms_of_*()} function. (This call
|
|
|
2135 interns a symbol, sets the C variable to the resulting Lisp object, and
|
|
|
2136 calls @code{staticpro()} on the C variable to tell the
|
|
|
2137 garbage-collection mechanism about this variable. What
|
|
|
2138 @code{staticpro()} does is add a pointer to the variable to a large
|
|
|
2139 global array; when garbage-collection happens, all pointers listed in
|
|
|
2140 the array are used as starting points for marking Lisp objects. This is
|
|
|
2141 important because it's quite possible that the only current reference to
|
|
|
2142 the object is the C variable. In the case of symbols, the
|
|
|
2143 @code{staticpro()} doesn't matter all that much because the symbol is
|
|
|
2144 contained in @code{obarray}, which is itself @code{staticpro()}ed.
|
|
|
2145 However, it's possible that a naughty user could do something like
|
|
|
2146 uninterning the symbol out of @code{obarray} or even setting
|
|
|
2147 @code{obarray} to a different value [although this is likely to make
|
|
|
2148 XEmacs crash!].)
|
|
|
2149
|
|
|
2150 @strong{Please note:} It is potentially deadly if you declare a
|
|
|
2151 @samp{Q...} variable in two different modules. The two calls to
|
|
|
2152 @code{defsymbol()} are no problem, but some linkers will complain about
|
|
|
2153 multiply-defined symbols. The most insidious aspect of this is that
|
|
|
2154 often the link will succeed anyway, but then the resulting executable
|
|
|
2155 will sometimes crash in obscure ways during certain operations! To
|
|
|
2156 avoid this problem, declare any symbols with common names (such as
|
|
|
2157 @code{text}) that are not obviously associated with this particular
|
|
|
2158 module in the module @file{general.c}.
|
|
|
2159
|
|
|
2160 Global variables whose names begin with @samp{V} are variables that
|
|
|
2161 contain Lisp objects. The convention here is that all global variables
|
|
|
2162 of type @code{Lisp_Object} begin with @samp{V}, and all others don't
|
|
|
2163 (including integer and boolean variables that have Lisp
|
|
|
2164 equivalents). Most of the time, these variables have equivalents in
|
|
|
2165 Lisp, but some don't. Those that do are declared this way by a call to
|
|
|
2166 @code{DEFVAR_LISP()} in the @code{vars_of_*()} initializer for the
|
|
|
2167 module. What this does is create a special @dfn{symbol-value-forward}
|
|
|
2168 Lisp object that contains a pointer to the C variable, intern a symbol
|
|
|
2169 whose name is as specified in the call to @code{DEFVAR_LISP()}, and set
|
|
|
2170 its value to the symbol-value-forward Lisp object; it also calls
|
|
|
2171 @code{staticpro()} on the C variable to tell the garbage-collection
|
|
|
2172 mechanism about the variable. When @code{eval} (or actually
|
|
|
2173 @code{symbol-value}) encounters this special object in the process of
|
|
|
2174 retrieving a variable's value, it follows the indirection to the C
|
|
|
2175 variable and gets its value. @code{setq} does similar things so that
|
|
|
2176 the C variable gets changed.
|
|
|
2177
|
|
|
2178 Whether or not you @code{DEFVAR_LISP()} a variable, you need to
|
|
|
2179 initialize it in the @code{vars_of_*()} function; otherwise it will end
|
|
|
2180 up as all zeroes, which is the integer 0 (@emph{not} @code{nil}), and
|
|
|
2181 this is probably not what you want. Also, if the variable is not
|
|
|
2182 @code{DEFVAR_LISP()}ed, @strong{you must call} @code{staticpro()} on the
|
|
|
2183 C variable in the @code{vars_of_*()} function. Otherwise, the
|
|
|
2184 garbage-collection mechanism won't know that the object in this variable
|
|
|
2185 is in use, and will happily collect it and reuse its storage for another
|
|
|
2186 Lisp object, and you will be the one who's unhappy when you can't figure
|
|
|
2187 out how your variable got overwritten.
|
|
|
2188
|
|
|
2189 @node Coding for Mule
|
|
|
2190 @section Coding for Mule
|
|
|
2191 @cindex Coding for Mule
|
|
|
2192
|
|
|
2193 Although Mule support is not compiled by default in XEmacs, many people
|
|
|
2194 are using it, and we consider it crucial that new code works correctly
|
|
|
2195 with multibyte characters. This is not hard; it is only a matter of
|
|
|
2196 following several simple user-interface guidelines. Even if you never
|
|
|
2197 compile with Mule, with a little practice you will find it quite easy
|
|
|
2198 to code Mule-correctly.
|
|
|
2199
|
|
|
2200 Note that these guidelines are not necessarily tied to the current Mule
|
|
|
2201 implementation; they are also a good idea to follow on the grounds of
|
|
|
2202 code generalization for future I18N work.
|
|
|
2203
|
|
|
2204 @menu
|
|
|
2205 * Character-Related Data Types::
|
|
|
2206 * Working With Character and Byte Positions::
|
|
|
2207 * Conversion to and from External Data::
|
|
|
2208 * General Guidelines for Writing Mule-Aware Code::
|
|
|
2209 * An Example of Mule-Aware Code::
|
|
|
2210 @end menu
|
|
|
2211
|
|
|
2212 @node Character-Related Data Types
|
|
|
2213 @subsection Character-Related Data Types
|
|
|
2214
|
|
|
2215 First, let's review the basic character-related datatypes used by
|
|
|
2216 XEmacs. Note that the separate @code{typedef}s are not mandatory in the
|
|
|
2217 current implementation (all of them boil down to @code{unsigned char} or
|
|
|
2218 @code{int}), but they improve clarity of code a great deal, because one
|
|
|
2219 glance at the declaration can tell the intended use of the variable.
|
|
|
2220
|
|
|
2221 @table @code
|
|
|
2222 @item Emchar
|
|
|
2223 @cindex Emchar
|
|
|
2224 An @code{Emchar} holds a single Emacs character.
|
|
|
2225
|
|
|
2226 Obviously, the equality between characters and bytes is lost in the Mule
|
|
|
2227 world. Characters can be represented by one or more bytes in the
|
|
|
2228 buffer, and @code{Emchar} is the C type large enough to hold any
|
|
|
2229 character.
|
|
|
2230
|
|
|
2231 Without Mule support, an @code{Emchar} is equivalent to an
|
|
|
2232 @code{unsigned char}.
|
|
|
2233
|
|
|
2234 @item Bufbyte
|
|
|
2235 @cindex Bufbyte
|
|
|
2236 The data representing the text in a buffer or string is logically a set
|
|
|
2237 of @code{Bufbyte}s.
|
|
|
2238
|
|
|
2239 XEmacs does not work with character formats all the time; when reading
|
|
|
2240 characters from the outside, it decodes them to an internal format, and
|
|
|
2241 likewise encodes them when writing. @code{Bufbyte} (in fact
|
|
|
2242 @code{unsigned char}) is the basic unit of XEmacs internal buffers and
|
|
|
2243 strings format.
|
|
|
2244
|
|
|
2245 One character can correspond to one or more @code{Bufbyte}s. In the
|
|
|
2246 current implementation, an ASCII character is represented by the same
|
|
|
2247 @code{Bufbyte}, and extended characters are represented by a sequence of
|
|
|
2248 @code{Bufbyte}s.
|
|
|
2249
|
|
|
2250 Without Mule support, a @code{Bufbyte} is equivalent to an
|
|
|
2251 @code{Emchar}.
|
|
|
2252
|
|
|
2253 @item Bufpos
|
|
|
2254 @itemx Charcount
|
|
|
2255 @cindex Bufpos
|
|
|
2256 @cindex Charcount
|
|
|
2257 A @code{Bufpos} represents a character position in a buffer or string.
|
|
|
2258 A @code{Charcount} represents a number (count) of characters.
|
|
|
2259 Logically, subtracting two @code{Bufpos} values yields a
|
|
|
2260 @code{Charcount} value. Although all of these are @code{typedef}ed to
|
|
|
2261 @code{int}, we use them in preference to @code{int} to make it clear
|
|
|
2262 what sort of position is being used.
|
|
|
2263
|
|
|
2264 @code{Bufpos} and @code{Charcount} values are the only ones that are
|
|
|
2265 ever visible to Lisp.
|
|
|
2266
|
|
|
2267 @item Bytind
|
|
|
2268 @itemx Bytecount
|
|
|
2269 @cindex Bytind
|
|
|
2270 @cindex Bytecount
|
|
|
2271 A @code{Bytind} represents a byte position in a buffer or string. A
|
|
|
2272 @code{Bytecount} represents the distance between two positions in bytes.
|
|
|
2273 The relationship between @code{Bytind} and @code{Bytecount} is the same
|
|
|
2274 as the relationship between @code{Bufpos} and @code{Charcount}.
|
|
|
2275
|
|
|
2276 @item Extbyte
|
|
|
2277 @itemx Extcount
|
|
|
2278 @cindex Extbyte
|
|
|
2279 @cindex Extcount
|
|
|
2280 When dealing with the outside world, XEmacs works with @code{Extbyte}s,
|
|
|
2281 which are equivalent to @code{unsigned char}. Obviously, an
|
|
|
2282 @code{Extcount} is the distance between two @code{Extbyte}s. Extbytes
|
|
|
2283 and Extcounts are not all that frequent in XEmacs code.
|
|
|
2284 @end table
|
|
|
2285
|
|
|
2286 @node Working With Character and Byte Positions
|
|
|
2287 @subsection Working With Character and Byte Positions
|
|
|
2288
|
|
|
2289 Now that we have defined the basic character-related types, we can look
|
|
|
2290 at the macros and functions designed for work with them and for
|
|
|
2291 conversion between them. Most of these macros are defined in
|
|
|
2292 @file{buffer.h}, and we don't discuss all of them here, but only the
|
|
|
2293 most important ones. Examining the existing code is the best way to
|
|
|
2294 learn about them.
|
|
|
2295
|
|
|
2296 @table @code
|
|
|
2297 @item MAX_EMCHAR_LEN
|
|
|
2298 @cindex MAX_EMCHAR_LEN
|
|
|
2299 This preprocessor constant is the maximum number of buffer bytes per
|
|
|
2300 Emacs character, i.e. the byte length of an @code{Emchar}. It is useful
|
|
|
2301 when allocating temporary strings to keep a known number of characters.
|
|
|
2302 For instance:
|
|
|
2303
|
|
|
2304 @example
|
|
|
2305 @group
|
|
|
2306 @{
|
|
|
2307 Charcount cclen;
|
|
|
2308 ...
|
|
|
2309 @{
|
|
|
2310 /* Allocate place for @var{cclen} characters. */
|
|
|
2311 Bufbyte *buf = (Bufbyte *)alloca (cclen * MAX_EMCHAR_LEN);
|
|
|
2312 ...
|
|
|
2313 @end group
|
|
|
2314 @end example
|
|
|
2315
|
|
|
2316 If you followed the previous section, you can guess that, logically,
|
|
|
2317 multiplying a @code{Charcount} value with @code{MAX_EMCHAR_LEN} produces
|
|
|
2318 a @code{Bytecount} value.
|
|
|
2319
|
|
|
2320 In the current Mule implementation, @code{MAX_EMCHAR_LEN} equals 4.
|
|
|
2321 Without Mule, it is 1.
|
|
|
2322
|
|
|
2323 @item charptr_emchar
|
|
|
2324 @itemx set_charptr_emchar
|
|
|
2325 @cindex charptr_emchar
|
|
|
2326 @cindex set_charptr_emchar
|
|
|
2327 The @code{charptr_emchar} macro takes a @code{Bufbyte} pointer and
|
|
|
2328 returns the @code{Emchar} stored at that position. If it were a
|
|
|
2329 function, its prototype would be:
|
|
|
2330
|
|
|
2331 @example
|
|
|
2332 Emchar charptr_emchar (Bufbyte *p);
|
|
|
2333 @end example
|
|
|
2334
|
|
|
2335 @code{set_charptr_emchar} stores an @code{Emchar} to the specified byte
|
|
|
2336 position. It returns the number of bytes stored:
|
|
|
2337
|
|
|
2338 @example
|
|
|
2339 Bytecount set_charptr_emchar (Bufbyte *p, Emchar c);
|
|
|
2340 @end example
|
|
|
2341
|
|
|
2342 It is important to note that @code{set_charptr_emchar} is safe only for
|
|
|
2343 appending a character at the end of a buffer, not for overwriting a
|
|
|
2344 character in the middle. This is because the width of characters
|
|
|
2345 varies, and @code{set_charptr_emchar} cannot resize the string if it
|
|
|
2346 writes, say, a two-byte character where a single-byte character used to
|
|
|
2347 reside.
|
|
|
2348
|
|
|
2349 A typical use of @code{set_charptr_emchar} can be demonstrated by this
|
|
|
2350 example, which copies characters from buffer @var{buf} to a temporary
|
|
|
2351 string of Bufbytes.
|
|
|
2352
|
|
|
2353 @example
|
|
|
2354 @group
|
|
|
2355 @{
|
|
|
2356 Bufpos pos;
|
|
|
2357 for (pos = beg; pos < end; pos++)
|
|
|
2358 @{
|
|
|
2359 Emchar c = BUF_FETCH_CHAR (buf, pos);
|
|
|
2360 p += set_charptr_emchar (buf, c);
|
|
|
2361 @}
|
|
|
2362 @}
|
|
|
2363 @end group
|
|
|
2364 @end example
|
|
|
2365
|
|
|
2366 Note how @code{set_charptr_emchar} is used to store the @code{Emchar}
|
|
|
2367 and increment the counter, at the same time.
|
|
|
2368
|
|
|
2369 @item INC_CHARPTR
|
|
|
2370 @itemx DEC_CHARPTR
|
|
|
2371 @cindex INC_CHARPTR
|
|
|
2372 @cindex DEC_CHARPTR
|
|
|
2373 These two macros increment and decrement a @code{Bufbyte} pointer,
|
|
|
2374 respectively. They will adjust the pointer by the appropriate number of
|
|
|
2375 bytes according to the byte length of the character stored there. Both
|
|
|
2376 macros assume that the memory address is located at the beginning of a
|
|
|
2377 valid character.
|
|
|
2378
|
|
|
2379 Without Mule support, @code{INC_CHARPTR (p)} and @code{DEC_CHARPTR (p)}
|
|
|
2380 simply expand to @code{p++} and @code{p--}, respectively.
|
|
|
2381
|
|
|
2382 @item bytecount_to_charcount
|
|
|
2383 @cindex bytecount_to_charcount
|
|
|
2384 Given a pointer to a text string and a length in bytes, return the
|
|
|
2385 equivalent length in characters.
|
|
|
2386
|
|
|
2387 @example
|
|
|
2388 Charcount bytecount_to_charcount (Bufbyte *p, Bytecount bc);
|
|
|
2389 @end example
|
|
|
2390
|
|
|
2391 @item charcount_to_bytecount
|
|
|
2392 @cindex charcount_to_bytecount
|
|
|
2393 Given a pointer to a text string and a length in characters, return the
|
|
|
2394 equivalent length in bytes.
|
|
|
2395
|
|
|
2396 @example
|
|
|
2397 Bytecount charcount_to_bytecount (Bufbyte *p, Charcount cc);
|
|
|
2398 @end example
|
|
|
2399
|
|
|
2400 @item charptr_n_addr
|
|
|
2401 @cindex charptr_n_addr
|
|
|
2402 Return a pointer to the beginning of the character offset @var{cc} (in
|
|
|
2403 characters) from @var{p}.
|
|
|
2404
|
|
|
2405 @example
|
|
|
2406 Bufbyte *charptr_n_addr (Bufbyte *p, Charcount cc);
|
|
|
2407 @end example
|
|
|
2408 @end table
|
|
|
2409
|
|
|
2410 @node Conversion to and from External Data
|
|
|
2411 @subsection Conversion to and from External Data
|
|
|
2412
|
|
|
2413 When an external function, such as a C library function, returns a
|
|
|
2414 @code{char} pointer, you should almost never treat it as @code{Bufbyte}.
|
|
|
2415 This is because these returned strings may contain 8bit characters which
|
|
|
2416 can be misinterpreted by XEmacs, and cause a crash. Likewise, when
|
|
|
2417 exporting a piece of internal text to the outside world, you should
|
|
|
2418 always convert it to an appropriate external encoding, lest the internal
|
|
|
2419 stuff (such as the infamous \201 characters) leak out.
|
|
|
2420
|
|
|
2421 The interface to conversion between the internal and external
|
|
|
2422 representations of text are the numerous conversion macros defined in
|
|
|
2423 @file{buffer.h}. Before looking at them, we'll look at the external
|
|
|
2424 formats supported by these macros.
|
|
|
2425
|
|
|
2426 Currently meaningful formats are @code{FORMAT_BINARY},
|
|
|
2427 @code{FORMAT_FILENAME}, @code{FORMAT_OS}, and @code{FORMAT_CTEXT}. Here
|
|
|
2428 is a description of these.
|
|
|
2429
|
|
|
2430 @table @code
|
|
|
2431 @item FORMAT_BINARY
|
|
|
2432 Binary format. This is the simplest format and is what we use in the
|
|
|
2433 absence of a more appropriate format. This converts according to the
|
|
|
2434 @code{binary} coding system:
|
|
|
2435
|
|
|
2436 @enumerate a
|
|
|
2437 @item
|
|
|
2438 On input, bytes 0--255 are converted into characters 0--255.
|
|
|
2439 @item
|
|
|
2440 On output, characters 0--255 are converted into bytes 0--255 and other
|
|
|
2441 characters are converted into `X'.
|
|
|
2442 @end enumerate
|
|
|
2443
|
|
|
2444 @item FORMAT_FILENAME
|
|
|
2445 Format used for filenames. In the original Mule, this is user-definable
|
|
|
2446 with the @code{pathname-coding-system} variable. For the moment, we
|
|
|
2447 just use the @code{binary} coding system.
|
|
|
2448
|
|
|
2449 @item FORMAT_OS
|
|
|
2450 Format used for the external Unix environment---@code{argv[]}, stuff
|
|
|
2451 from @code{getenv()}, stuff from the @file{/etc/passwd} file, etc.
|
|
|
2452
|
|
|
2453 Perhaps should be the same as FORMAT_FILENAME.
|
|
|
2454
|
|
|
2455 @item FORMAT_CTEXT
|
|
|
2456 Compound--text format. This is the standard X format used for data
|
|
|
2457 stored in properties, selections, and the like. This is an 8-bit
|
|
|
2458 no-lock-shift ISO2022 coding system.
|
|
|
2459 @end table
|
|
|
2460
|
|
|
2461 The macros to convert between these formats and the internal format, and
|
|
|
2462 vice versa, follow.
|
|
|
2463
|
|
|
2464 @table @code
|
|
|
2465 @item GET_CHARPTR_INT_DATA_ALLOCA
|
|
|
2466 @itemx GET_CHARPTR_EXT_DATA_ALLOCA
|
|
|
2467 These two are the most basic conversion macros.
|
|
|
2468 @code{GET_CHARPTR_INT_DATA_ALLOCA} converts external data to internal
|
|
|
2469 format, and @code{GET_CHARPTR_EXT_DATA_ALLOCA} converts the other way
|
|
|
2470 around. The arguments each of these receives are @var{ptr} (pointer to
|
|
|
2471 the text in external format), @var{len} (length of texts in bytes),
|
|
|
2472 @var{fmt} (format of the external text), @var{ptr_out} (lvalue to which
|
|
|
2473 new text should be copied), and @var{len_out} (lvalue which will be
|
|
|
2474 assigned the length of the internal text in bytes). The resulting text
|
|
|
2475 is stored to a stack-allocated buffer. If the text doesn't need
|
|
|
2476 changing, these macros will do nothing, except for setting
|
|
|
2477 @var{len_out}.
|
|
|
2478
|
|
|
2479 The macros above take many arguments which makes them unwieldy. For
|
|
|
2480 this reason, a number of convenience macros are defined with obvious
|
|
|
2481 functionality, but accepting less arguments. The general rule is that
|
|
|
2482 macros with @samp{INT} in their name convert text to internal Emacs
|
|
|
2483 representation, whereas the @samp{EXT} macros convert to external
|
|
|
2484 representation.
|
|
|
2485
|
|
|
2486 @item GET_C_CHARPTR_INT_DATA_ALLOCA
|
|
|
2487 @itemx GET_C_CHARPTR_EXT_DATA_ALLOCA
|
|
|
2488 As their names imply, these macros work on C char pointers, which are
|
|
|
2489 zero-terminated, and thus do not need @var{len} or @var{len_out}
|
|
|
2490 parameters.
|
|
|
2491
|
|
|
2492 @item GET_STRING_EXT_DATA_ALLOCA
|
|
|
2493 @itemx GET_C_STRING_EXT_DATA_ALLOCA
|
|
|
2494 These two macros convert a Lisp string into an external representation.
|
|
|
2495 The difference between them is that @code{GET_STRING_EXT_DATA_ALLOCA}
|
|
|
2496 stores its output to a generic string, providing @var{len_out}, the
|
|
|
2497 length of the resulting external string. On the other hand,
|
|
|
2498 @code{GET_C_STRING_EXT_DATA_ALLOCA} assumes that the caller will be
|
|
|
2499 satisfied with output string being zero-terminated.
|
|
|
2500
|
|
|
2501 Note that for Lisp strings only one conversion direction makes sense.
|
|
|
2502
|
|
|
2503 @item GET_C_CHARPTR_EXT_BINARY_DATA_ALLOCA
|
|
|
2504 @itemx GET_CHARPTR_EXT_BINARY_DATA_ALLOCA
|
|
|
2505 @itemx GET_STRING_BINARY_DATA_ALLOCA
|
|
|
2506 @itemx GET_C_STRING_BINARY_DATA_ALLOCA
|
|
|
2507 @itemx GET_C_CHARPTR_EXT_FILENAME_DATA_ALLOCA
|
|
|
2508 @itemx ...
|
|
|
2509 These macros convert internal text to a specific external
|
|
|
2510 representation, with the external format being encoded into the name of
|
|
|
2511 the macro. Note that the @code{GET_STRING_...} and
|
|
|
2512 @code{GET_C_STRING...} macros lack the @samp{EXT} tag, because they
|
|
|
2513 only make sense in that direction.
|
|
|
2514
|
|
|
2515 @item GET_C_CHARPTR_INT_BINARY_DATA_ALLOCA
|
|
|
2516 @itemx GET_CHARPTR_INT_BINARY_DATA_ALLOCA
|
|
|
2517 @itemx GET_C_CHARPTR_INT_FILENAME_DATA_ALLOCA
|
|
|
2518 @itemx ...
|
|
|
2519 These macros convert external text of a specific format to its internal
|
|
|
2520 representation, with the external format being incoded into the name of
|
|
|
2521 the macro.
|
|
|
2522 @end table
|
|
|
2523
|
|
|
2524 @node General Guidelines for Writing Mule-Aware Code
|
|
|
2525 @subsection General Guidelines for Writing Mule-Aware Code
|
|
|
2526
|
|
|
2527 This section contains some general guidance on how to write Mule-aware
|
|
|
2528 code, as well as some pitfalls you should avoid.
|
|
|
2529
|
|
|
2530 @table @emph
|
|
|
2531 @item Never use @code{char} and @code{char *}.
|
|
|
2532 In XEmacs, the use of @code{char} and @code{char *} is almost always a
|
|
|
2533 mistake. If you want to manipulate an Emacs character from ``C'', use
|
|
|
2534 @code{Emchar}. If you want to examine a specific octet in the internal
|
|
|
2535 format, use @code{Bufbyte}. If you want a Lisp-visible character, use a
|
|
|
2536 @code{Lisp_Object} and @code{make_char}. If you want a pointer to move
|
|
|
2537 through the internal text, use @code{Bufbyte *}. Also note that you
|
|
|
2538 almost certainly do not need @code{Emchar *}.
|
|
|
2539
|
|
|
2540 @item Be careful not to confuse @code{Charcount}, @code{Bytecount}, and @code{Bufpos}.
|
|
|
2541 The whole point of using different types is to avoid confusion about the
|
|
|
2542 use of certain variables. Lest this effect be nullified, you need to be
|
|
|
2543 careful about using the right types.
|
|
|
2544
|
|
|
2545 @item Always convert external data
|
|
|
2546 It is extremely important to always convert external data, because
|
|
|
2547 XEmacs can crash if unexpected 8bit sequences are copied to its internal
|
|
|
2548 buffers literally.
|
|
|
2549
|
|
|
2550 This means that when a system function, such as @code{readdir}, returns
|
|
|
2551 a string, you need to convert it using one of the conversion macros
|
|
|
2552 described in the previous chapter, before passing it further to Lisp.
|
|
|
2553 In the case of @code{readdir}, you would use the
|
|
|
2554 @code{GET_C_CHARPTR_INT_FILENAME_DATA_ALLOCA} macro.
|
|
|
2555
|
|
|
2556 Also note that many internal functions, such as @code{make_string},
|
|
|
2557 accept Bufbytes, which removes the need for them to convert the data
|
|
|
2558 they receive. This increases efficiency because that way external data
|
|
|
2559 needs to be decoded only once, when it is read. After that, it is
|
|
|
2560 passed around in internal format.
|
|
|
2561 @end table
|
|
|
2562
|
|
|
2563 @node An Example of Mule-Aware Code
|
|
|
2564 @subsection An Example of Mule-Aware Code
|
|
|
2565
|
|
|
2566 As an example of Mule-aware code, we shall will analyze the
|
|
|
2567 @code{string} function, which conses up a Lisp string from the character
|
|
|
2568 arguments it receives. Here is the definition, pasted from
|
|
|
2569 @code{alloc.c}:
|
|
|
2570
|
|
|
2571 @example
|
|
|
2572 @group
|
|
|
2573 DEFUN ("string", Fstring, 0, MANY, 0, /*
|
|
|
2574 Concatenate all the argument characters and make the result a string.
|
|
|
2575 */
|
|
|
2576 (int nargs, Lisp_Object *args))
|
|
|
2577 @{
|
|
|
2578 Bufbyte *storage = alloca_array (Bufbyte, nargs * MAX_EMCHAR_LEN);
|
|
|
2579 Bufbyte *p = storage;
|
|
|
2580
|
|
|
2581 for (; nargs; nargs--, args++)
|
|
|
2582 @{
|
|
|
2583 Lisp_Object lisp_char = *args;
|
|
|
2584 CHECK_CHAR_COERCE_INT (lisp_char);
|
|
|
2585 p += set_charptr_emchar (p, XCHAR (lisp_char));
|
|
|
2586 @}
|
|
|
2587 return make_string (storage, p - storage);
|
|
|
2588 @}
|
|
|
2589 @end group
|
|
|
2590 @end example
|
|
|
2591
|
|
|
2592 Now we can analyze the source line by line.
|
|
|
2593
|
|
|
2594 Obviously, string will be as long as there are arguments to the
|
|
|
2595 function. This is why we allocate @code{MAX_EMCHAR_LEN} * @var{nargs}
|
|
|
2596 bytes on the stack, i.e. the worst-case number of bytes for @var{nargs}
|
|
|
2597 @code{Emchar}s to fit in the string.
|
|
|
2598
|
|
|
2599 Then, the loop checks that each element is a character, converting
|
|
|
2600 integers in the process. Like many other functions in XEmacs, this
|
|
|
2601 function silently accepts integers where characters are expected, for
|
|
|
2602 historical and compatibility reasons. Unless you know what you are
|
|
|
2603 doing, @code{CHECK_CHAR} will also suffice. @code{XCHAR (lisp_char)}
|
|
|
2604 extracts the @code{Emchar} from the @code{Lisp_Object}, and
|
|
|
2605 @code{set_charptr_emchar} stores it to storage, increasing @code{p} in
|
|
|
2606 the process.
|
|
|
2607
|
|
|
2608 Other instructive examples of correct coding under Mule can be found all
|
|
|
2609 over the XEmacs code. For starters, I recommend
|
|
|
2610 @code{Fnormalize_menu_item_name} in @file{menubar.c}. After you have
|
|
|
2611 understood this section of the manual and studied the examples, you can
|
|
|
2612 proceed writing new Mule-aware code.
|
|
|
2613
|
|
|
2614 @node Techniques for XEmacs Developers
|
|
|
2615 @section Techniques for XEmacs Developers
|
|
|
2616
|
|
|
2617 To make a quantified XEmacs, do: @code{make quantmacs}.
|
|
|
2618
|
|
|
2619 You simply can't dump Quantified and Purified images. Run the image
|
|
|
2620 like so: @code{quantmacs -batch -l loadup.el run-temacs @var{xemacs-args...}}.
|
|
|
2621
|
|
|
2622 Before you go through the trouble, are you compiling with all
|
|
|
2623 debugging and error-checking off? If not try that first. Be warned
|
|
|
2624 that while Quantify is directly responsible for quite a few
|
|
|
2625 optimizations which have been made to XEmacs, doing a run which
|
|
|
2626 generates results which can be acted upon is not necessarily a trivial
|
|
|
2627 task.
|
|
|
2628
|
|
|
2629 Also, if you're still willing to do some runs make sure you configure
|
|
|
2630 with the @samp{--quantify} flag. That will keep Quantify from starting
|
|
|
2631 to record data until after the loadup is completed and will shut off
|
|
|
2632 recording right before it shuts down (which generates enough bogus data
|
|
|
2633 to throw most results off). It also enables three additional elisp
|
|
|
2634 commands: @code{quantify-start-recording-data},
|
|
|
2635 @code{quantify-stop-recording-data} and @code{quantify-clear-data}.
|
|
|
2636
|
|
|
2637 If you want to make XEmacs faster, target your favorite slow benchmark,
|
|
|
2638 run a profiler like Quantify, @code{gprof}, or @code{tcov}, and figure
|
|
|
2639 out where the cycles are going. Specific projects:
|
|
|
2640
|
|
|
2641 @itemize @bullet
|
|
|
2642 @item
|
|
|
2643 Make the garbage collector faster. Figure out how to write an
|
|
|
2644 incremental garbage collector.
|
|
|
2645 @item
|
|
|
2646 Write a compiler that takes bytecode and spits out C code.
|
|
|
2647 Unfortunately, you will then need a C compiler and a more fully
|
|
|
2648 developed module system.
|
|
|
2649 @item
|
|
|
2650 Speed up redisplay.
|
|
|
2651 @item
|
|
|
2652 Speed up syntax highlighting. Maybe moving some of the syntax
|
|
|
2653 highlighting capabilities into C would make a difference.
|
|
|
2654 @item
|
|
|
2655 Implement tail recursion in Emacs Lisp (hard!).
|
|
|
2656 @end itemize
|
|
|
2657
|
|
|
2658 Unfortunately, Emacs Lisp is slow, and is going to stay slow. Function
|
|
|
2659 calls in elisp are especially expensive. Iterating over a long list is
|
|
|
2660 going to be 30 times faster implemented in C than in Elisp.
|
|
|
2661
|
|
438
|
2662 To get started debugging XEmacs, take a look at the @file{.gdbinit} and
|
|
|
2663 @file{.dbxrc} files in the @file{src} directory.
|
|
428
|
2664 @xref{Q2.1.15 - How to Debug an XEmacs problem with a debugger,,,
|
|
|
2665 xemacs-faq, XEmacs FAQ}.
|
|
|
2666
|
|
|
2667 After making source code changes, run @code{make check} to ensure that
|
|
|
2668 you haven't introduced any regressions. If you're feeling ambitious,
|
|
|
2669 you can try to improve the test suite in @file{tests/automated}.
|
|
|
2670
|
|
|
2671 Here are things to know when you create a new source file:
|
|
|
2672
|
|
|
2673 @itemize @bullet
|
|
|
2674 @item
|
|
|
2675 All @file{.c} files should @code{#include <config.h>} first. Almost all
|
|
|
2676 @file{.c} files should @code{#include "lisp.h"} second.
|
|
|
2677
|
|
|
2678 @item
|
|
|
2679 Generated header files should be included using the @code{#include <...>} syntax,
|
|
|
2680 not the @code{#include "..."} syntax. The generated headers are:
|
|
|
2681
|
|
|
2682 @file{config.h puresize-adjust.h sheap-adjust.h paths.h Emacs.ad.h}
|
|
|
2683
|
|
|
2684 The basic rule is that you should assume builds using @code{--srcdir}
|
|
|
2685 and the @code{#include <...>} syntax needs to be used when the
|
|
|
2686 to-be-included generated file is in a potentially different directory
|
|
|
2687 @emph{at compile time}. The non-obvious C rule is that @code{#include "..."}
|
|
|
2688 means to search for the included file in the same directory as the
|
|
|
2689 including file, @emph{not} in the current directory.
|
|
|
2690
|
|
|
2691 @item
|
|
|
2692 Header files should @emph{not} include @code{<config.h>} and
|
|
|
2693 @code{"lisp.h"}. It is the responsibility of the @file{.c} files that
|
|
|
2694 use it to do so.
|
|
|
2695
|
|
|
2696 @item
|
|
|
2697 If the header uses @code{INLINE}, either directly or through
|
|
|
2698 @code{DECLARE_LRECORD}, then it must be added to @file{inline.c}'s
|
|
|
2699 includes.
|
|
|
2700
|
|
|
2701 @item
|
|
|
2702 Try compiling at least once with
|
|
|
2703
|
|
|
2704 @example
|
|
|
2705 gcc --with-mule --with-union-type --error-checking=all
|
|
|
2706 @end example
|
|
|
2707
|
|
|
2708 @item
|
|
|
2709 Did I mention that you should run the test suite?
|
|
|
2710 @example
|
|
|
2711 make check
|
|
|
2712 @end example
|
|
|
2713 @end itemize
|
|
|
2714
|
|
|
2715
|
|
|
2716 @node A Summary of the Various XEmacs Modules, Allocation of Objects in XEmacs Lisp, Rules When Writing New C Code, Top
|
|
|
2717 @chapter A Summary of the Various XEmacs Modules
|
|
|
2718
|
|
|
2719 This is accurate as of XEmacs 20.0.
|
|
|
2720
|
|
|
2721 @menu
|
|
|
2722 * Low-Level Modules::
|
|
|
2723 * Basic Lisp Modules::
|
|
|
2724 * Modules for Standard Editing Operations::
|
|
|
2725 * Editor-Level Control Flow Modules::
|
|
|
2726 * Modules for the Basic Displayable Lisp Objects::
|
|
|
2727 * Modules for other Display-Related Lisp Objects::
|
|
|
2728 * Modules for the Redisplay Mechanism::
|
|
|
2729 * Modules for Interfacing with the File System::
|
|
|
2730 * Modules for Other Aspects of the Lisp Interpreter and Object System::
|
|
|
2731 * Modules for Interfacing with the Operating System::
|
|
|
2732 * Modules for Interfacing with X Windows::
|
|
|
2733 * Modules for Internationalization::
|
|
|
2734 @end menu
|
|
|
2735
|
|
|
2736 @node Low-Level Modules
|
|
|
2737 @section Low-Level Modules
|
|
|
2738
|
|
|
2739 @example
|
|
|
2740 config.h
|
|
|
2741 @end example
|
|
|
2742
|
|
|
2743 This is automatically generated from @file{config.h.in} based on the
|
|
|
2744 results of configure tests and user-selected optional features and
|
|
|
2745 contains preprocessor definitions specifying the nature of the
|
|
|
2746 environment in which XEmacs is being compiled.
|
|
|
2747
|
|
|
2748
|
|
|
2749
|
|
|
2750 @example
|
|
|
2751 paths.h
|
|
|
2752 @end example
|
|
|
2753
|
|
|
2754 This is automatically generated from @file{paths.h.in} based on supplied
|
|
|
2755 configure values, and allows for non-standard installed configurations
|
|
|
2756 of the XEmacs directories. It's currently broken, though.
|
|
|
2757
|
|
|
2758
|
|
|
2759
|
|
|
2760 @example
|
|
|
2761 emacs.c
|
|
|
2762 signal.c
|
|
|
2763 @end example
|
|
|
2764
|
|
|
2765 @file{emacs.c} contains @code{main()} and other code that performs the most
|
|
|
2766 basic environment initializations and handles shutting down the XEmacs
|
|
|
2767 process (this includes @code{kill-emacs}, the normal way that XEmacs is
|
|
|
2768 exited; @code{dump-emacs}, which is used during the build process to
|
|
|
2769 write out the XEmacs executable; @code{run-emacs-from-temacs}, which can
|
|
|
2770 be used to start XEmacs directly when temacs has finished loading all
|
|
|
2771 the Lisp code; and emergency code to handle crashes [XEmacs tries to
|
|
|
2772 auto-save all files before it crashes]).
|
|
|
2773
|
|
|
2774 Low-level code that directly interacts with the Unix signal mechanism,
|
|
|
2775 however, is in @file{signal.c}. Note that this code does not handle system
|
|
|
2776 dependencies in interfacing to signals; that is handled using the
|
|
|
2777 @file{syssignal.h} header file, described in section J below.
|
|
|
2778
|
|
|
2779
|
|
|
2780
|
|
|
2781 @example
|
|
|
2782 unexaix.c
|
|
|
2783 unexalpha.c
|
|
|
2784 unexapollo.c
|
|
|
2785 unexconvex.c
|
|
|
2786 unexec.c
|
|
|
2787 unexelf.c
|
|
|
2788 unexelfsgi.c
|
|
|
2789 unexencap.c
|
|
|
2790 unexenix.c
|
|
|
2791 unexfreebsd.c
|
|
|
2792 unexfx2800.c
|
|
|
2793 unexhp9k3.c
|
|
|
2794 unexhp9k800.c
|
|
|
2795 unexmips.c
|
|
|
2796 unexnext.c
|
|
|
2797 unexsol2.c
|
|
|
2798 unexsunos4.c
|
|
|
2799 @end example
|
|
|
2800
|
|
|
2801 These modules contain code dumping out the XEmacs executable on various
|
|
|
2802 different systems. (This process is highly machine-specific and
|
|
|
2803 requires intimate knowledge of the executable format and the memory map
|
|
|
2804 of the process.) Only one of these modules is actually used; this is
|
|
|
2805 chosen by @file{configure}.
|
|
|
2806
|
|
|
2807
|
|
|
2808
|
|
|
2809 @example
|
|
|
2810 crt0.c
|
|
|
2811 lastfile.c
|
|
|
2812 pre-crt0.c
|
|
|
2813 @end example
|
|
|
2814
|
|
|
2815 These modules are used in conjunction with the dump mechanism. On some
|
|
|
2816 systems, an alternative version of the C startup code (the actual code
|
|
|
2817 that receives control from the operating system when the process is
|
|
|
2818 started, and which calls @code{main()}) is required so that the dumping
|
|
|
2819 process works properly; @file{crt0.c} provides this.
|
|
|
2820
|
|
|
2821 @file{pre-crt0.c} and @file{lastfile.c} should be the very first and
|
|
|
2822 very last file linked, respectively. (Actually, this is not really true.
|
|
|
2823 @file{lastfile.c} should be after all Emacs modules whose initialized
|
|
|
2824 data should be made constant, and before all other Emacs files and all
|
|
|
2825 libraries. In particular, the allocation modules @file{gmalloc.c},
|
|
|
2826 @file{alloca.c}, etc. are normally placed past @file{lastfile.c}, and
|
|
|
2827 all of the files that implement Xt widget classes @emph{must} be placed
|
|
|
2828 after @file{lastfile.c} because they contain various structures that
|
|
|
2829 must be statically initialized and into which Xt writes at various
|
|
|
2830 times.) @file{pre-crt0.c} and @file{lastfile.c} contain exported symbols
|
|
|
2831 that are used to determine the start and end of XEmacs' initialized
|
|
|
2832 data space when dumping.
|
|
|
2833
|
|
|
2834
|
|
|
2835
|
|
|
2836 @example
|
|
|
2837 alloca.c
|
|
|
2838 free-hook.c
|
|
|
2839 getpagesize.h
|
|
|
2840 gmalloc.c
|
|
|
2841 malloc.c
|
|
|
2842 mem-limits.h
|
|
|
2843 ralloc.c
|
|
|
2844 vm-limit.c
|
|
|
2845 @end example
|
|
|
2846
|
|
|
2847 These handle basic C allocation of memory. @file{alloca.c} is an emulation of
|
|
|
2848 the stack allocation function @code{alloca()} on machines that lack
|
|
|
2849 this. (XEmacs makes extensive use of @code{alloca()} in its code.)
|
|
|
2850
|
|
|
2851 @file{gmalloc.c} and @file{malloc.c} are two implementations of the standard C
|
|
|
2852 functions @code{malloc()}, @code{realloc()} and @code{free()}. They are
|
|
|
2853 often used in place of the standard system-provided @code{malloc()}
|
|
|
2854 because they usually provide a much faster implementation, at the
|
|
|
2855 expense of additional memory use. @file{gmalloc.c} is a newer implementation
|
|
|
2856 that is much more memory-efficient for large allocations than @file{malloc.c},
|
|
|
2857 and should always be preferred if it works. (At one point, @file{gmalloc.c}
|
|
|
2858 didn't work on some systems where @file{malloc.c} worked; but this should be
|
|
|
2859 fixed now.)
|
|
|
2860
|
|
|
2861 @cindex relocating allocator
|
|
|
2862 @file{ralloc.c} is the @dfn{relocating allocator}. It provides
|
|
|
2863 functions similar to @code{malloc()}, @code{realloc()} and @code{free()}
|
|
|
2864 that allocate memory that can be dynamically relocated in memory. The
|
|
|
2865 advantage of this is that allocated memory can be shuffled around to
|
|
|
2866 place all the free memory at the end of the heap, and the heap can then
|
|
|
2867 be shrunk, releasing the memory back to the operating system. The use
|
|
|
2868 of this can be controlled with the configure option @code{--rel-alloc};
|
|
|
2869 if enabled, memory allocated for buffers will be relocatable, so that if
|
|
|
2870 a very large file is visited and the buffer is later killed, the memory
|
|
|
2871 can be released to the operating system. (The disadvantage of this
|
|
|
2872 mechanism is that it can be very slow. On systems with the
|
|
|
2873 @code{mmap()} system call, the XEmacs version of @file{ralloc.c} uses
|
|
|
2874 this to move memory around without actually having to block-copy it,
|
|
|
2875 which can speed things up; but it can still cause noticeable performance
|
|
|
2876 degradation.)
|
|
|
2877
|
|
|
2878 @file{free-hook.c} contains some debugging functions for checking for invalid
|
|
|
2879 arguments to @code{free()}.
|
|
|
2880
|
|
|
2881 @file{vm-limit.c} contains some functions that warn the user when memory is
|
|
|
2882 getting low. These are callback functions that are called by @file{gmalloc.c}
|
|
|
2883 and @file{malloc.c} at appropriate times.
|
|
|
2884
|
|
|
2885 @file{getpagesize.h} provides a uniform interface for retrieving the size of a
|
|
|
2886 page in virtual memory. @file{mem-limits.h} provides a uniform interface for
|
|
|
2887 retrieving the total amount of available virtual memory. Both are
|
|
|
2888 similar in spirit to the @file{sys*.h} files described in section J, below.
|
|
|
2889
|
|
|
2890
|
|
|
2891
|
|
|
2892 @example
|
|
|
2893 blocktype.c
|
|
|
2894 blocktype.h
|
|
|
2895 dynarr.c
|
|
|
2896 @end example
|
|
|
2897
|
|
|
2898 These implement a couple of basic C data types to facilitate memory
|
|
|
2899 allocation. The @code{Blocktype} type efficiently manages the
|
|
|
2900 allocation of fixed-size blocks by minimizing the number of times that
|
|
|
2901 @code{malloc()} and @code{free()} are called. It allocates memory in
|
|
|
2902 large chunks, subdivides the chunks into blocks of the proper size, and
|
|
|
2903 returns the blocks as requested. When blocks are freed, they are placed
|
|
|
2904 onto a linked list, so they can be efficiently reused. This data type
|
|
|
2905 is not much used in XEmacs currently, because it's a fairly new
|
|
|
2906 addition.
|
|
|
2907
|
|
|
2908 @cindex dynamic array
|
|
|
2909 The @code{Dynarr} type implements a @dfn{dynamic array}, which is
|
|
|
2910 similar to a standard C array but has no fixed limit on the number of
|
|
|
2911 elements it can contain. Dynamic arrays can hold elements of any type,
|
|
|
2912 and when you add a new element, the array automatically resizes itself
|
|
|
2913 if it isn't big enough. Dynarrs are extensively used in the redisplay
|
|
|
2914 mechanism.
|
|
|
2915
|
|
|
2916
|
|
|
2917
|
|
|
2918 @example
|
|
|
2919 inline.c
|
|
|
2920 @end example
|
|
|
2921
|
|
|
2922 This module is used in connection with inline functions (available in
|
|
|
2923 some compilers). Often, inline functions need to have a corresponding
|
|
|
2924 non-inline function that does the same thing. This module is where they
|
|
|
2925 reside. It contains no actual code, but defines some special flags that
|
|
|
2926 cause inline functions defined in header files to be rendered as actual
|
|
|
2927 functions. It then includes all header files that contain any inline
|
|
|
2928 function definitions, so that each one gets a real function equivalent.
|
|
|
2929
|
|
|
2930
|
|
|
2931
|
|
|
2932 @example
|
|
|
2933 debug.c
|
|
|
2934 debug.h
|
|
|
2935 @end example
|
|
|
2936
|
|
|
2937 These functions provide a system for doing internal consistency checks
|
|
|
2938 during code development. This system is not currently used; instead the
|
|
|
2939 simpler @code{assert()} macro is used along with the various checks
|
|
|
2940 provided by the @samp{--error-check-*} configuration options.
|
|
|
2941
|
|
|
2942
|
|
|
2943
|
|
|
2944 @example
|
|
|
2945 prefix-args.c
|
|
|
2946 @end example
|
|
|
2947
|
|
|
2948 This is actually the source for a small, self-contained program
|
|
|
2949 used during building.
|
|
|
2950
|
|
|
2951
|
|
|
2952 @example
|
|
|
2953 universe.h
|
|
|
2954 @end example
|
|
|
2955
|
|
|
2956 This is not currently used.
|
|
|
2957
|
|
|
2958
|
|
|
2959
|
|
|
2960 @node Basic Lisp Modules
|
|
|
2961 @section Basic Lisp Modules
|
|
|
2962
|
|
|
2963 @example
|
|
|
2964 emacsfns.h
|
|
|
2965 lisp-disunion.h
|
|
|
2966 lisp-union.h
|
|
|
2967 lisp.h
|
|
|
2968 lrecord.h
|
|
|
2969 symsinit.h
|
|
|
2970 @end example
|
|
|
2971
|
|
|
2972 These are the basic header files for all XEmacs modules. Each module
|
|
|
2973 includes @file{lisp.h}, which brings the other header files in.
|
|
|
2974 @file{lisp.h} contains the definitions of the structures and extractor
|
|
|
2975 and constructor macros for the basic Lisp objects and various other
|
|
|
2976 basic definitions for the Lisp environment, as well as some
|
|
|
2977 general-purpose definitions (e.g. @code{min()} and @code{max()}).
|
|
|
2978 @file{lisp.h} includes either @file{lisp-disunion.h} or
|
|
|
2979 @file{lisp-union.h}, depending on whether @code{USE_UNION_TYPE} is
|
|
|
2980 defined. These files define the typedef of the Lisp object itself (as
|
|
|
2981 described above) and the low-level macros that hide the actual
|
|
|
2982 implementation of the Lisp object. All extractor and constructor macros
|
|
|
2983 for particular types of Lisp objects are defined in terms of these
|
|
|
2984 low-level macros.
|
|
|
2985
|
|
|
2986 As a general rule, all typedefs should go into the typedefs section of
|
|
|
2987 @file{lisp.h} rather than into a module-specific header file even if the
|
|
|
2988 structure is defined elsewhere. This allows function prototypes that
|
|
|
2989 use the typedef to be placed into other header files. Forward structure
|
|
|
2990 declarations (i.e. a simple declaration like @code{struct foo;} where
|
|
|
2991 the structure itself is defined elsewhere) should be placed into the
|
|
|
2992 typedefs section as necessary.
|
|
|
2993
|
|
|
2994 @file{lrecord.h} contains the basic structures and macros that implement
|
|
440
|
2995 all record-type Lisp objects---i.e. all objects whose type is a field
|
|
428
|
2996 in their C structure, which includes all objects except the few most
|
|
|
2997 basic ones.
|
|
|
2998
|
|
|
2999 @file{lisp.h} contains prototypes for most of the exported functions in
|
|
|
3000 the various modules. Lisp primitives defined using @code{DEFUN} that
|
|
|
3001 need to be called by C code should be declared using @code{EXFUN}.
|
|
|
3002 Other function prototypes should be placed either into the appropriate
|
|
|
3003 section of @code{lisp.h}, or into a module-specific header file,
|
|
|
3004 depending on how general-purpose the function is and whether it has
|
|
|
3005 special-purpose argument types requiring definitions not in
|
|
|
3006 @file{lisp.h}.) All initialization functions are prototyped in
|
|
|
3007 @file{symsinit.h}.
|
|
|
3008
|
|
|
3009
|
|
|
3010
|
|
|
3011 @example
|
|
|
3012 alloc.c
|
|
|
3013 pure.c
|
|
|
3014 puresize.h
|
|
|
3015 @end example
|
|
|
3016
|
|
|
3017 The large module @file{alloc.c} implements all of the basic allocation and
|
|
|
3018 garbage collection for Lisp objects. The most commonly used Lisp
|
|
|
3019 objects are allocated in chunks, similar to the Blocktype data type
|
|
|
3020 described above; others are allocated in individually @code{malloc()}ed
|
|
|
3021 blocks. This module provides the foundation on which all other aspects
|
|
|
3022 of the Lisp environment sit, and is the first module initialized at
|
|
|
3023 startup.
|
|
|
3024
|
|
|
3025 Note that @file{alloc.c} provides a series of generic functions that are
|
|
|
3026 not dependent on any particular object type, and interfaces to
|
|
|
3027 particular types of objects using a standardized interface of
|
|
|
3028 type-specific methods. This scheme is a fundamental principle of
|
|
|
3029 object-oriented programming and is heavily used throughout XEmacs. The
|
|
|
3030 great advantage of this is that it allows for a clean separation of
|
|
440
|
3031 functionality into different modules---new classes of Lisp objects, new
|
|
428
|
3032 event interfaces, new device types, new stream interfaces, etc. can be
|
|
|
3033 added transparently without affecting code anywhere else in XEmacs.
|
|
|
3034 Because the different subsystems are divided into general and specific
|
|
|
3035 code, adding a new subtype within a subsystem will in general not
|
|
|
3036 require changes to the generic subsystem code or affect any of the other
|
|
|
3037 subtypes in the subsystem; this provides a great deal of robustness to
|
|
|
3038 the XEmacs code.
|
|
|
3039
|
|
|
3040 @cindex pure space
|
|
|
3041 @file{pure.c} contains the declaration of the @dfn{purespace} array.
|
|
|
3042 Pure space is a hack used to place some constant Lisp data into the code
|
|
|
3043 segment of the XEmacs executable, even though the data needs to be
|
|
|
3044 initialized through function calls. (See above in section VIII for more
|
|
|
3045 info about this.) During startup, certain sorts of data is
|
|
|
3046 automatically copied into pure space, and other data is copied manually
|
|
|
3047 in some of the basic Lisp files by calling the function @code{purecopy},
|
|
|
3048 which copies the object if possible (this only works in temacs, of
|
|
|
3049 course) and returns the new object. In particular, while temacs is
|
|
|
3050 executing, the Lisp reader automatically copies all compiled-function
|
|
|
3051 objects that it reads into pure space. Since compiled-function objects
|
|
|
3052 are large, are never modified, and typically comprise the majority of
|
|
|
3053 the contents of a compiled-Lisp file, this works well. While XEmacs is
|
|
|
3054 running, any attempt to modify an object that resides in pure space
|
|
|
3055 causes an error. Objects in pure space are never garbage collected --
|
|
|
3056 almost all of the time, they're intended to be permanent, and in any
|
|
|
3057 case you can't write into pure space to set the mark bits.
|
|
|
3058
|
|
|
3059 @file{puresize.h} contains the declaration of the size of the pure space
|
|
|
3060 array. This depends on the optional features that are compiled in, any
|
|
|
3061 extra purespace requested by the user at compile time, and certain other
|
|
|
3062 factors (e.g. 64-bit machines need more pure space because their Lisp
|
|
|
3063 objects are larger). The smallest size that suffices should be used, so
|
|
|
3064 that there's no wasted space. If there's not enough pure space, you
|
|
|
3065 will get an error during the build process, specifying how much more
|
|
|
3066 pure space is needed.
|
|
|
3067
|
|
|
3068
|
|
|
3069
|
|
|
3070 @example
|
|
|
3071 eval.c
|
|
|
3072 backtrace.h
|
|
|
3073 @end example
|
|
|
3074
|
|
|
3075 This module contains all of the functions to handle the flow of control.
|
|
|
3076 This includes the mechanisms of defining functions, calling functions,
|
|
|
3077 traversing stack frames, and binding variables; the control primitives
|
|
|
3078 and other special forms such as @code{while}, @code{if}, @code{eval},
|
|
|
3079 @code{let}, @code{and}, @code{or}, @code{progn}, etc.; handling of
|
|
|
3080 non-local exits, unwind-protects, and exception handlers; entering the
|
|
|
3081 debugger; methods for the subr Lisp object type; etc. It does
|
|
|
3082 @emph{not} include the @code{read} function, the @code{print} function,
|
|
|
3083 or the handling of symbols and obarrays.
|
|
|
3084
|
|
|
3085 @file{backtrace.h} contains some structures related to stack frames and the
|
|
|
3086 flow of control.
|
|
|
3087
|
|
|
3088
|
|
|
3089
|
|
|
3090 @example
|
|
|
3091 lread.c
|
|
|
3092 @end example
|
|
|
3093
|
|
|
3094 This module implements the Lisp reader and the @code{read} function,
|
|
|
3095 which converts text into Lisp objects, according to the read syntax of
|
|
|
3096 the objects, as described above. This is similar to the parser that is
|
|
|
3097 a part of all compilers.
|
|
|
3098
|
|
|
3099
|
|
|
3100
|
|
|
3101 @example
|
|
|
3102 print.c
|
|
|
3103 @end example
|
|
|
3104
|
|
|
3105 This module implements the Lisp print mechanism and the @code{print}
|
|
|
3106 function and related functions. This is the inverse of the Lisp reader
|
|
|
3107 -- it converts Lisp objects to a printed, textual representation.
|
|
|
3108 (Hopefully something that can be read back in using @code{read} to get
|
|
|
3109 an equivalent object.)
|
|
|
3110
|
|
|
3111
|
|
|
3112
|
|
|
3113 @example
|
|
|
3114 general.c
|
|
|
3115 symbols.c
|
|
|
3116 symeval.h
|
|
|
3117 @end example
|
|
|
3118
|
|
|
3119 @file{symbols.c} implements the handling of symbols, obarrays, and
|
|
|
3120 retrieving the values of symbols. Much of the code is devoted to
|
|
|
3121 handling the special @dfn{symbol-value-magic} objects that define
|
|
440
|
3122 special types of variables---this includes buffer-local variables,
|
|
428
|
3123 variable aliases, variables that forward into C variables, etc. This
|
|
|
3124 module is initialized extremely early (right after @file{alloc.c}),
|
|
|
3125 because it is here that the basic symbols @code{t} and @code{nil} are
|
|
|
3126 created, and those symbols are used everywhere throughout XEmacs.
|
|
|
3127
|
|
|
3128 @file{symeval.h} contains the definitions of symbol structures and the
|
|
|
3129 @code{DEFVAR_LISP()} and related macros for declaring variables.
|
|
|
3130
|
|
|
3131
|
|
|
3132
|
|
|
3133 @example
|
|
|
3134 data.c
|
|
|
3135 floatfns.c
|
|
|
3136 fns.c
|
|
|
3137 @end example
|
|
|
3138
|
|
|
3139 These modules implement the methods and standard Lisp primitives for all
|
|
|
3140 the basic Lisp object types other than symbols (which are described
|
|
|
3141 above). @file{data.c} contains all the predicates (primitives that return
|
|
|
3142 whether an object is of a particular type); the integer arithmetic
|
|
|
3143 functions; and the basic accessor and mutator primitives for the various
|
|
|
3144 object types. @file{fns.c} contains all the standard predicates for working
|
|
|
3145 with sequences (where, abstractly speaking, a sequence is an ordered set
|
|
|
3146 of objects, and can be represented by a list, string, vector, or
|
|
|
3147 bit-vector); it also contains @code{equal}, perhaps on the grounds that
|
|
|
3148 bulk of the operation of @code{equal} is comparing sequences.
|
|
|
3149 @file{floatfns.c} contains methods and primitives for floats and floating-point
|
|
|
3150 arithmetic.
|
|
|
3151
|
|
|
3152
|
|
|
3153
|
|
|
3154 @example
|
|
|
3155 bytecode.c
|
|
|
3156 bytecode.h
|
|
|
3157 @end example
|
|
|
3158
|
|
|
3159 @file{bytecode.c} implements the byte-code interpreter and
|
|
|
3160 compiled-function objects, and @file{bytecode.h} contains associated
|
|
|
3161 structures. Note that the byte-code @emph{compiler} is written in Lisp.
|
|
|
3162
|
|
|
3163
|
|
|
3164
|
|
|
3165
|
|
|
3166 @node Modules for Standard Editing Operations
|
|
|
3167 @section Modules for Standard Editing Operations
|
|
|
3168
|
|
|
3169 @example
|
|
|
3170 buffer.c
|
|
|
3171 buffer.h
|
|
|
3172 bufslots.h
|
|
|
3173 @end example
|
|
|
3174
|
|
|
3175 @file{buffer.c} implements the @dfn{buffer} Lisp object type. This
|
|
|
3176 includes functions that create and destroy buffers; retrieve buffers by
|
|
|
3177 name or by other properties; manipulate lists of buffers (remember that
|
|
|
3178 buffers are permanent objects and stored in various ordered lists);
|
|
|
3179 retrieve or change buffer properties; etc. It also contains the
|
|
|
3180 definitions of all the built-in buffer-local variables (which can be
|
|
|
3181 viewed as buffer properties). It does @emph{not} contain code to
|
|
|
3182 manipulate buffer-local variables (that's in @file{symbols.c}, described
|
|
|
3183 above); or code to manipulate the text in a buffer.
|
|
|
3184
|
|
|
3185 @file{buffer.h} defines the structures associated with a buffer and the various
|
|
|
3186 macros for retrieving text from a buffer and special buffer positions
|
|
|
3187 (e.g. @code{point}, the default location for text insertion). It also
|
|
|
3188 contains macros for working with buffer positions and converting between
|
|
|
3189 their representations as character offsets and as byte offsets (under
|
|
|
3190 MULE, they are different, because characters can be multi-byte). It is
|
|
|
3191 one of the largest header files.
|
|
|
3192
|
|
|
3193 @file{bufslots.h} defines the fields in the buffer structure that correspond to
|
|
|
3194 the built-in buffer-local variables. It is its own header file because
|
|
|
3195 it is included many times in @file{buffer.c}, as a way of iterating over all
|
|
|
3196 the built-in buffer-local variables.
|
|
|
3197
|
|
|
3198
|
|
|
3199
|
|
|
3200 @example
|
|
|
3201 insdel.c
|
|
|
3202 insdel.h
|
|
|
3203 @end example
|
|
|
3204
|
|
|
3205 @file{insdel.c} contains low-level functions for inserting and deleting text in
|
|
|
3206 a buffer, keeping track of changed regions for use by redisplay, and
|
|
|
3207 calling any before-change and after-change functions that may have been
|
|
|
3208 registered for the buffer. It also contains the actual functions that
|
|
|
3209 convert between byte offsets and character offsets.
|
|
|
3210
|
|
|
3211 @file{insdel.h} contains associated headers.
|
|
|
3212
|
|
|
3213
|
|
|
3214
|
|
|
3215 @example
|
|
|
3216 marker.c
|
|
|
3217 @end example
|
|
|
3218
|
|
|
3219 This module implements the @dfn{marker} Lisp object type, which
|
|
|
3220 conceptually is a pointer to a text position in a buffer that moves
|
|
|
3221 around as text is inserted and deleted, so as to remain in the same
|
|
|
3222 relative position. This module doesn't actually move the markers around
|
|
|
3223 -- that's handled in @file{insdel.c}. This module just creates them and
|
|
|
3224 implements the primitives for working with them. As markers are simple
|
|
|
3225 objects, this does not entail much.
|
|
|
3226
|
|
|
3227 Note that the standard arithmetic primitives (e.g. @code{+}) accept
|
|
|
3228 markers in place of integers and automatically substitute the value of
|
|
|
3229 @code{marker-position} for the marker, i.e. an integer describing the
|
|
|
3230 current buffer position of the marker.
|
|
|
3231
|
|
|
3232
|
|
|
3233
|
|
|
3234 @example
|
|
|
3235 extents.c
|
|
|
3236 extents.h
|
|
|
3237 @end example
|
|
|
3238
|
|
|
3239 This module implements the @dfn{extent} Lisp object type, which is like
|
|
|
3240 a marker that works over a range of text rather than a single position.
|
|
|
3241 Extents are also much more complex and powerful than markers and have a
|
|
|
3242 more efficient (and more algorithmically complex) implementation. The
|
|
|
3243 implementation is described in detail in comments in @file{extents.c}.
|
|
|
3244
|
|
|
3245 The code in @file{extents.c} works closely with @file{insdel.c} so that
|
|
|
3246 extents are properly moved around as text is inserted and deleted.
|
|
|
3247 There is also code in @file{extents.c} that provides information needed
|
|
|
3248 by the redisplay mechanism for efficient operation. (Remember that
|
|
|
3249 extents can have display properties that affect [sometimes drastically,
|
|
|
3250 as in the @code{invisible} property] the display of the text they
|
|
|
3251 cover.)
|
|
|
3252
|
|
|
3253
|
|
|
3254
|
|
|
3255 @example
|
|
|
3256 editfns.c
|
|
|
3257 @end example
|
|
|
3258
|
|
|
3259 @file{editfns.c} contains the standard Lisp primitives for working with
|
|
|
3260 a buffer's text, and calls the low-level functions in @file{insdel.c}.
|
|
|
3261 It also contains primitives for working with @code{point} (the default
|
|
|
3262 buffer insertion location).
|
|
|
3263
|
|
|
3264 @file{editfns.c} also contains functions for retrieving various
|
|
|
3265 characteristics from the external environment: the current time, the
|
|
|
3266 process ID of the running XEmacs process, the name of the user who ran
|
|
|
3267 this XEmacs process, etc. It's not clear why this code is in
|
|
|
3268 @file{editfns.c}.
|
|
|
3269
|
|
|
3270
|
|
|
3271
|
|
|
3272 @example
|
|
|
3273 callint.c
|
|
|
3274 cmds.c
|
|
|
3275 commands.h
|
|
|
3276 @end example
|
|
|
3277
|
|
|
3278 @cindex interactive
|
|
|
3279 These modules implement the basic @dfn{interactive} commands,
|
|
|
3280 i.e. user-callable functions. Commands, as opposed to other functions,
|
|
|
3281 have special ways of getting their parameters interactively (by querying
|
|
|
3282 the user), as opposed to having them passed in a normal function
|
|
|
3283 invocation. Many commands are not really meant to be called from other
|
|
|
3284 Lisp functions, because they modify global state in a way that's often
|
|
|
3285 undesired as part of other Lisp functions.
|
|
|
3286
|
|
|
3287 @file{callint.c} implements the mechanism for querying the user for
|
|
|
3288 parameters and calling interactive commands. The bulk of this module is
|
|
|
3289 code that parses the interactive spec that is supplied with an
|
|
|
3290 interactive command.
|
|
|
3291
|
|
|
3292 @file{cmds.c} implements the basic, most commonly used editing commands:
|
|
|
3293 commands to move around the current buffer and insert and delete
|
|
|
3294 characters. These commands are implemented using the Lisp primitives
|
|
|
3295 defined in @file{editfns.c}.
|
|
|
3296
|
|
|
3297 @file{commands.h} contains associated structure definitions and prototypes.
|
|
|
3298
|
|
|
3299
|
|
|
3300
|
|
|
3301 @example
|
|
|
3302 regex.c
|
|
|
3303 regex.h
|
|
|
3304 search.c
|
|
|
3305 @end example
|
|
|
3306
|
|
|
3307 @file{search.c} implements the Lisp primitives for searching for text in
|
|
|
3308 a buffer, and some of the low-level algorithms for doing this. In
|
|
|
3309 particular, the fast fixed-string Boyer-Moore search algorithm is
|
|
|
3310 implemented in @file{search.c}. The low-level algorithms for doing
|
|
|
3311 regular-expression searching, however, are implemented in @file{regex.c}
|
|
|
3312 and @file{regex.h}. These two modules are largely independent of
|
|
|
3313 XEmacs, and are similar to (and based upon) the regular-expression
|
|
|
3314 routines used in @file{grep} and other GNU utilities.
|
|
|
3315
|
|
|
3316
|
|
|
3317
|
|
|
3318 @example
|
|
|
3319 doprnt.c
|
|
|
3320 @end example
|
|
|
3321
|
|
|
3322 @file{doprnt.c} implements formatted-string processing, similar to
|
|
|
3323 @code{printf()} command in C.
|
|
|
3324
|
|
|
3325
|
|
|
3326
|
|
|
3327 @example
|
|
|
3328 undo.c
|
|
|
3329 @end example
|
|
|
3330
|
|
|
3331 This module implements the undo mechanism for tracking buffer changes.
|
|
|
3332 Most of this could be implemented in Lisp.
|
|
|
3333
|
|
|
3334
|
|
|
3335
|
|
|
3336 @node Editor-Level Control Flow Modules
|
|
|
3337 @section Editor-Level Control Flow Modules
|
|
|
3338
|
|
|
3339 @example
|
|
|
3340 event-Xt.c
|
|
|
3341 event-stream.c
|
|
|
3342 event-tty.c
|
|
|
3343 events.c
|
|
|
3344 events.h
|
|
|
3345 @end example
|
|
|
3346
|
|
|
3347 These implement the handling of events (user input and other system
|
|
|
3348 notifications).
|
|
|
3349
|
|
|
3350 @file{events.c} and @file{events.h} define the @dfn{event} Lisp object
|
|
|
3351 type and primitives for manipulating it.
|
|
|
3352
|
|
|
3353 @file{event-stream.c} implements the basic functions for working with
|
|
|
3354 event queues, dispatching an event by looking it up in relevant keymaps
|
|
|
3355 and such, and handling timeouts; this includes the primitives
|
|
|
3356 @code{next-event} and @code{dispatch-event}, as well as related
|
|
|
3357 primitives such as @code{sit-for}, @code{sleep-for}, and
|
|
|
3358 @code{accept-process-output}. (@file{event-stream.c} is one of the
|
|
|
3359 hairiest and trickiest modules in XEmacs. Beware! You can easily mess
|
|
|
3360 things up here.)
|
|
|
3361
|
|
|
3362 @file{event-Xt.c} and @file{event-tty.c} implement the low-level
|
|
|
3363 interfaces onto retrieving events from Xt (the X toolkit) and from TTY's
|
|
|
3364 (using @code{read()} and @code{select()}), respectively. The event
|
|
|
3365 interface enforces a clean separation between the specific code for
|
|
|
3366 interfacing with the operating system and the generic code for working
|
|
|
3367 with events, by defining an API of basic, low-level event methods;
|
|
|
3368 @file{event-Xt.c} and @file{event-tty.c} are two different
|
|
|
3369 implementations of this API. To add support for a new operating system
|
|
|
3370 (e.g. NeXTstep), one merely needs to provide another implementation of
|
|
|
3371 those API functions.
|
|
|
3372
|
|
|
3373 Note that the choice of whether to use @file{event-Xt.c} or
|
|
|
3374 @file{event-tty.c} is made at compile time! Or at the very latest, it
|
|
|
3375 is made at startup time. @file{event-Xt.c} handles events for
|
|
|
3376 @emph{both} X and TTY frames; @file{event-tty.c} is only used when X
|
|
|
3377 support is not compiled into XEmacs. The reason for this is that there
|
|
|
3378 is only one event loop in XEmacs: thus, it needs to be able to receive
|
|
|
3379 events from all different kinds of frames.
|
|
|
3380
|
|
|
3381
|
|
|
3382
|
|
|
3383 @example
|
|
|
3384 keymap.c
|
|
|
3385 keymap.h
|
|
|
3386 @end example
|
|
|
3387
|
|
|
3388 @file{keymap.c} and @file{keymap.h} define the @dfn{keymap} Lisp object
|
|
|
3389 type and associated methods and primitives. (Remember that keymaps are
|
|
|
3390 objects that associate event descriptions with functions to be called to
|
|
|
3391 ``execute'' those events; @code{dispatch-event} looks up events in the
|
|
|
3392 relevant keymaps.)
|
|
|
3393
|
|
|
3394
|
|
|
3395
|
|
|
3396 @example
|
|
|
3397 keyboard.c
|
|
|
3398 @end example
|
|
|
3399
|
|
|
3400 @file{keyboard.c} contains functions that implement the actual editor
|
|
440
|
3401 command loop---i.e. the event loop that cyclically retrieves and
|
|
428
|
3402 dispatches events. This code is also rather tricky, just like
|
|
|
3403 @file{event-stream.c}.
|
|
|
3404
|
|
|
3405
|
|
|
3406
|
|
|
3407 @example
|
|
|
3408 macros.c
|
|
|
3409 macros.h
|
|
|
3410 @end example
|
|
|
3411
|
|
|
3412 These two modules contain the basic code for defining keyboard macros.
|
|
|
3413 These functions don't actually do much; most of the code that handles keyboard
|
|
|
3414 macros is mixed in with the event-handling code in @file{event-stream.c}.
|
|
|
3415
|
|
|
3416
|
|
|
3417
|
|
|
3418 @example
|
|
|
3419 minibuf.c
|
|
|
3420 @end example
|
|
|
3421
|
|
|
3422 This contains some miscellaneous code related to the minibuffer (most of
|
|
|
3423 the minibuffer code was moved into Lisp by Richard Mlynarik). This
|
|
|
3424 includes the primitives for completion (although filename completion is
|
|
|
3425 in @file{dired.c}), the lowest-level interface to the minibuffer (if the
|
|
|
3426 command loop were cleaned up, this too could be in Lisp), and code for
|
|
|
3427 dealing with the echo area (this, too, was mostly moved into Lisp, and
|
|
|
3428 the only code remaining is code to call out to Lisp or provide simple
|
|
|
3429 bootstrapping implementations early in temacs, before the echo-area Lisp
|
|
|
3430 code is loaded).
|
|
|
3431
|
|
|
3432
|
|
|
3433
|
|
|
3434 @node Modules for the Basic Displayable Lisp Objects
|
|
|
3435 @section Modules for the Basic Displayable Lisp Objects
|
|
|
3436
|
|
|
3437 @example
|
|
|
3438 device-ns.h
|
|
|
3439 device-stream.c
|
|
|
3440 device-stream.h
|
|
|
3441 device-tty.c
|
|
|
3442 device-tty.h
|
|
|
3443 device-x.c
|
|
|
3444 device-x.h
|
|
|
3445 device.c
|
|
|
3446 device.h
|
|
|
3447 @end example
|
|
|
3448
|
|
|
3449 These modules implement the @dfn{device} Lisp object type. This
|
|
|
3450 abstracts a particular screen or connection on which frames are
|
|
|
3451 displayed. As with Lisp objects, event interfaces, and other
|
|
|
3452 subsystems, the device code is separated into a generic component that
|
|
|
3453 contains a standardized interface (in the form of a set of methods) onto
|
|
|
3454 particular device types.
|
|
|
3455
|
|
|
3456 The device subsystem defines all the methods and provides method
|
|
|
3457 services for not only device operations but also for the frame, window,
|
|
|
3458 menubar, scrollbar, toolbar, and other displayable-object subsystems.
|
|
|
3459 The reason for this is that all of these subsystems have the same
|
|
|
3460 subtypes (X, TTY, NeXTstep, Microsoft Windows, etc.) as devices do.
|
|
|
3461
|
|
|
3462
|
|
|
3463
|
|
|
3464 @example
|
|
|
3465 frame-ns.h
|
|
|
3466 frame-tty.c
|
|
|
3467 frame-x.c
|
|
|
3468 frame-x.h
|
|
|
3469 frame.c
|
|
|
3470 frame.h
|
|
|
3471 @end example
|
|
|
3472
|
|
|
3473 Each device contains one or more frames in which objects (e.g. text) are
|
|
|
3474 displayed. A frame corresponds to a window in the window system;
|
|
|
3475 usually this is a top-level window but it could potentially be one of a
|
|
|
3476 number of overlapping child windows within a top-level window, using the
|
|
|
3477 MDI (Multiple Document Interface) protocol in Microsoft Windows or a
|
|
|
3478 similar scheme.
|
|
|
3479
|
|
|
3480 The @file{frame-*} files implement the @dfn{frame} Lisp object type and
|
|
|
3481 provide the generic and device-type-specific operations on frames
|
|
|
3482 (e.g. raising, lowering, resizing, moving, etc.).
|
|
|
3483
|
|
|
3484
|
|
|
3485
|
|
|
3486 @example
|
|
|
3487 window.c
|
|
|
3488 window.h
|
|
|
3489 @end example
|
|
|
3490
|
|
|
3491 @cindex window (in Emacs)
|
|
|
3492 @cindex pane
|
|
|
3493 Each frame consists of one or more non-overlapping @dfn{windows} (better
|
|
|
3494 known as @dfn{panes} in standard window-system terminology) in which a
|
|
|
3495 buffer's text can be displayed. Windows can also have scrollbars
|
|
|
3496 displayed around their edges.
|
|
|
3497
|
|
|
3498 @file{window.c} and @file{window.h} implement the @dfn{window} Lisp
|
|
|
3499 object type and provide code to manage windows. Since windows have no
|
|
|
3500 associated resources in the window system (the window system knows only
|
|
|
3501 about the frame; no child windows or anything are used for XEmacs
|
|
|
3502 windows), there is no device-type-specific code here; all of that code
|
|
|
3503 is part of the redisplay mechanism or the code for particular object
|
|
|
3504 types such as scrollbars.
|
|
|
3505
|
|
|
3506
|
|
|
3507
|
|
|
3508 @node Modules for other Display-Related Lisp Objects
|
|
|
3509 @section Modules for other Display-Related Lisp Objects
|
|
|
3510
|
|
|
3511 @example
|
|
|
3512 faces.c
|
|
|
3513 faces.h
|
|
|
3514 @end example
|
|
|
3515
|
|
|
3516
|
|
|
3517
|
|
|
3518 @example
|
|
|
3519 bitmaps.h
|
|
|
3520 glyphs-ns.h
|
|
|
3521 glyphs-x.c
|
|
|
3522 glyphs-x.h
|
|
|
3523 glyphs.c
|
|
|
3524 glyphs.h
|
|
|
3525 @end example
|
|
|
3526
|
|
|
3527
|
|
|
3528
|
|
|
3529 @example
|
|
|
3530 objects-ns.h
|
|
|
3531 objects-tty.c
|
|
|
3532 objects-tty.h
|
|
|
3533 objects-x.c
|
|
|
3534 objects-x.h
|
|
|
3535 objects.c
|
|
|
3536 objects.h
|
|
|
3537 @end example
|
|
|
3538
|
|
|
3539
|
|
|
3540
|
|
|
3541 @example
|
|
|
3542 menubar-x.c
|
|
|
3543 menubar.c
|
|
|
3544 @end example
|
|
|
3545
|
|
|
3546
|
|
|
3547
|
|
|
3548 @example
|
|
|
3549 scrollbar-x.c
|
|
|
3550 scrollbar-x.h
|
|
|
3551 scrollbar.c
|
|
|
3552 scrollbar.h
|
|
|
3553 @end example
|
|
|
3554
|
|
|
3555
|
|
|
3556
|
|
|
3557 @example
|
|
|
3558 toolbar-x.c
|
|
|
3559 toolbar.c
|
|
|
3560 toolbar.h
|
|
|
3561 @end example
|
|
|
3562
|
|
|
3563
|
|
|
3564
|
|
|
3565 @example
|
|
|
3566 font-lock.c
|
|
|
3567 @end example
|
|
|
3568
|
|
440
|
3569 This file provides C support for syntax highlighting---i.e.
|
|
428
|
3570 highlighting different syntactic constructs of a source file in
|
|
|
3571 different colors, for easy reading. The C support is provided so that
|
|
|
3572 this is fast.
|
|
|
3573
|
|
|
3574
|
|
|
3575
|
|
|
3576 @example
|
|
|
3577 dgif_lib.c
|
|
|
3578 gif_err.c
|
|
|
3579 gif_lib.h
|
|
|
3580 gifalloc.c
|
|
|
3581 @end example
|
|
|
3582
|
|
|
3583 These modules decode GIF-format image files, for use with glyphs.
|
|
|
3584
|
|
|
3585
|
|
|
3586
|
|
|
3587 @node Modules for the Redisplay Mechanism
|
|
|
3588 @section Modules for the Redisplay Mechanism
|
|
|
3589
|
|
|
3590 @example
|
|
|
3591 redisplay-output.c
|
|
|
3592 redisplay-tty.c
|
|
|
3593 redisplay-x.c
|
|
|
3594 redisplay.c
|
|
|
3595 redisplay.h
|
|
|
3596 @end example
|
|
|
3597
|
|
|
3598 These files provide the redisplay mechanism. As with many other
|
|
|
3599 subsystems in XEmacs, there is a clean separation between the general
|
|
|
3600 and device-specific support.
|
|
|
3601
|
|
|
3602 @file{redisplay.c} contains the bulk of the redisplay engine. These
|
|
|
3603 functions update the redisplay structures (which describe how the screen
|
|
|
3604 is to appear) to reflect any changes made to the state of any
|
|
|
3605 displayable objects (buffer, frame, window, etc.) since the last time
|
|
|
3606 that redisplay was called. These functions are highly optimized to
|
|
|
3607 avoid doing more work than necessary (since redisplay is called
|
|
|
3608 extremely often and is potentially a huge time sink), and depend heavily
|
|
|
3609 on notifications from the objects themselves that changes have occurred,
|
|
|
3610 so that redisplay doesn't explicitly have to check each possible object.
|
|
|
3611 The redisplay mechanism also contains a great deal of caching to further
|
|
|
3612 speed things up; some of this caching is contained within the various
|
|
|
3613 displayable objects.
|
|
|
3614
|
|
|
3615 @file{redisplay-output.c} goes through the redisplay structures and converts
|
|
|
3616 them into calls to device-specific methods to actually output the screen
|
|
|
3617 changes.
|
|
|
3618
|
|
|
3619 @file{redisplay-x.c} and @file{redisplay-tty.c} are two implementations
|
|
|
3620 of these redisplay output methods, for X frames and TTY frames,
|
|
|
3621 respectively.
|
|
|
3622
|
|
|
3623
|
|
|
3624
|
|
|
3625 @example
|
|
|
3626 indent.c
|
|
|
3627 @end example
|
|
|
3628
|
|
|
3629 This module contains various functions and Lisp primitives for
|
|
|
3630 converting between buffer positions and screen positions. These
|
|
|
3631 functions call the redisplay mechanism to do most of the work, and then
|
|
|
3632 examine the redisplay structures to get the necessary information. This
|
|
|
3633 module needs work.
|
|
|
3634
|
|
|
3635
|
|
|
3636
|
|
|
3637 @example
|
|
|
3638 termcap.c
|
|
|
3639 terminfo.c
|
|
|
3640 tparam.c
|
|
|
3641 @end example
|
|
|
3642
|
|
|
3643 These files contain functions for working with the termcap (BSD-style)
|
|
|
3644 and terminfo (System V style) databases of terminal capabilities and
|
|
|
3645 escape sequences, used when XEmacs is displaying in a TTY.
|
|
|
3646
|
|
|
3647
|
|
|
3648
|
|
|
3649 @example
|
|
|
3650 cm.c
|
|
|
3651 cm.h
|
|
|
3652 @end example
|
|
|
3653
|
|
|
3654 These files provide some miscellaneous TTY-output functions and should
|
|
|
3655 probably be merged into @file{redisplay-tty.c}.
|
|
|
3656
|
|
|
3657
|
|
|
3658
|
|
|
3659 @node Modules for Interfacing with the File System
|
|
|
3660 @section Modules for Interfacing with the File System
|
|
|
3661
|
|
|
3662 @example
|
|
|
3663 lstream.c
|
|
|
3664 lstream.h
|
|
|
3665 @end example
|
|
|
3666
|
|
|
3667 These modules implement the @dfn{stream} Lisp object type. This is an
|
|
|
3668 internal-only Lisp object that implements a generic buffering stream.
|
|
|
3669 The idea is to provide a uniform interface onto all sources and sinks of
|
|
|
3670 data, including file descriptors, stdio streams, chunks of memory, Lisp
|
|
|
3671 buffers, Lisp strings, etc. That way, I/O functions can be written to
|
|
|
3672 the stream interface and can transparently handle all possible sources
|
|
|
3673 and sinks. (For example, the @code{read} function can read data from a
|
|
|
3674 file, a string, a buffer, or even a function that is called repeatedly
|
|
|
3675 to return data, without worrying about where the data is coming from or
|
|
|
3676 what-size chunks it is returned in.)
|
|
|
3677
|
|
|
3678 @cindex lstream
|
|
|
3679 Note that in the C code, streams are called @dfn{lstreams} (for ``Lisp
|
|
|
3680 streams'') to distinguish them from other kinds of streams, e.g. stdio
|
|
|
3681 streams and C++ I/O streams.
|
|
|
3682
|
|
|
3683 Similar to other subsystems in XEmacs, lstreams are separated into
|
|
|
3684 generic functions and a set of methods for the different types of
|
|
|
3685 lstreams. @file{lstream.c} provides implementations of many different
|
|
|
3686 types of streams; others are provided, e.g., in @file{mule-coding.c}.
|
|
|
3687
|
|
|
3688
|
|
|
3689
|
|
|
3690 @example
|
|
|
3691 fileio.c
|
|
|
3692 @end example
|
|
|
3693
|
|
|
3694 This implements the basic primitives for interfacing with the file
|
|
|
3695 system. This includes primitives for reading files into buffers,
|
|
|
3696 writing buffers into files, checking for the presence or accessibility
|
|
|
3697 of files, canonicalizing file names, etc. Note that these primitives
|
|
|
3698 are usually not invoked directly by the user: There is a great deal of
|
|
|
3699 higher-level Lisp code that implements the user commands such as
|
|
|
3700 @code{find-file} and @code{save-buffer}. This is similar to the
|
|
|
3701 distinction between the lower-level primitives in @file{editfns.c} and
|
|
|
3702 the higher-level user commands in @file{commands.c} and
|
|
|
3703 @file{simple.el}.
|
|
|
3704
|
|
|
3705
|
|
|
3706
|
|
|
3707 @example
|
|
|
3708 filelock.c
|
|
|
3709 @end example
|
|
|
3710
|
|
|
3711 This file provides functions for detecting clashes between different
|
|
|
3712 processes (e.g. XEmacs and some external process, or two different
|
|
|
3713 XEmacs processes) modifying the same file. (XEmacs can optionally use
|
|
|
3714 the @file{lock/} subdirectory to provide a form of ``locking'' between
|
|
|
3715 different XEmacs processes.) This module is also used by the low-level
|
|
|
3716 functions in @file{insdel.c} to ensure that, if the first modification
|
|
|
3717 is being made to a buffer whose corresponding file has been externally
|
|
|
3718 modified, the user is made aware of this so that the buffer can be
|
|
|
3719 synched up with the external changes if necessary.
|
|
|
3720
|
|
|
3721
|
|
|
3722 @example
|
|
|
3723 filemode.c
|
|
|
3724 @end example
|
|
|
3725
|
|
|
3726 This file provides some miscellaneous functions that construct a
|
|
|
3727 @samp{rwxr-xr-x}-type permissions string (as might appear in an
|
|
|
3728 @file{ls}-style directory listing) given the information returned by the
|
|
|
3729 @code{stat()} system call.
|
|
|
3730
|
|
|
3731
|
|
|
3732
|
|
|
3733 @example
|
|
|
3734 dired.c
|
|
|
3735 ndir.h
|
|
|
3736 @end example
|
|
|
3737
|
|
|
3738 These files implement the XEmacs interface to directory searching. This
|
|
|
3739 includes a number of primitives for determining the files in a directory
|
|
|
3740 and for doing filename completion. (Remember that generic completion is
|
|
|
3741 handled by a different mechanism, in @file{minibuf.c}.)
|
|
|
3742
|
|
|
3743 @file{ndir.h} is a header file used for the directory-searching
|
|
|
3744 emulation functions provided in @file{sysdep.c} (see section J below),
|
|
|
3745 for systems that don't provide any directory-searching functions. (On
|
|
|
3746 those systems, directories can be read directly as files, and parsed.)
|
|
|
3747
|
|
|
3748
|
|
|
3749
|
|
|
3750 @example
|
|
|
3751 realpath.c
|
|
|
3752 @end example
|
|
|
3753
|
|
|
3754 This file provides an implementation of the @code{realpath()} function
|
|
|
3755 for expanding symbolic links, on systems that don't implement it or have
|
|
|
3756 a broken implementation.
|
|
|
3757
|
|
|
3758
|
|
|
3759
|
|
|
3760 @node Modules for Other Aspects of the Lisp Interpreter and Object System
|
|
|
3761 @section Modules for Other Aspects of the Lisp Interpreter and Object System
|
|
|
3762
|
|
|
3763 @example
|
|
|
3764 elhash.c
|
|
|
3765 elhash.h
|
|
|
3766 hash.c
|
|
|
3767 hash.h
|
|
|
3768 @end example
|
|
|
3769
|
|
|
3770 These files provide two implementations of hash tables. Files
|
|
|
3771 @file{hash.c} and @file{hash.h} provide a generic C implementation of
|
|
|
3772 hash tables which can stand independently of XEmacs. Files
|
|
|
3773 @file{elhash.c} and @file{elhash.h} provide a separate implementation of
|
|
|
3774 hash tables that can store only Lisp objects, and knows about Lispy
|
|
|
3775 things like garbage collection, and implement the @dfn{hash-table} Lisp
|
|
|
3776 object type.
|
|
|
3777
|
|
|
3778
|
|
|
3779 @example
|
|
|
3780 specifier.c
|
|
|
3781 specifier.h
|
|
|
3782 @end example
|
|
|
3783
|
|
|
3784 This module implements the @dfn{specifier} Lisp object type. This is
|
|
|
3785 primarily used for displayable properties, and allows for values that
|
|
|
3786 are specific to a particular buffer, window, frame, device, or device
|
|
|
3787 class, as well as a default value existing. This is used, for example,
|
|
|
3788 to control the height of the horizontal scrollbar or the appearance of
|
|
|
3789 the @code{default}, @code{bold}, or other faces. The specifier object
|
|
|
3790 consists of a number of specifications, each of which maps from a
|
|
|
3791 buffer, window, etc. to a value. The function @code{specifier-instance}
|
|
|
3792 looks up a value given a window (from which a buffer, frame, and device
|
|
|
3793 can be derived).
|
|
|
3794
|
|
|
3795
|
|
|
3796 @example
|
|
|
3797 chartab.c
|
|
|
3798 chartab.h
|
|
|
3799 casetab.c
|
|
|
3800 @end example
|
|
|
3801
|
|
|
3802 @file{chartab.c} and @file{chartab.h} implement the @dfn{char table}
|
|
|
3803 Lisp object type, which maps from characters or certain sorts of
|
|
|
3804 character ranges to Lisp objects. The implementation of this object
|
|
|
3805 type is optimized for the internal representation of characters. Char
|
|
|
3806 tables come in different types, which affect the allowed object types to
|
|
|
3807 which a character can be mapped and also dictate certain other
|
|
|
3808 properties of the char table.
|
|
|
3809
|
|
|
3810 @cindex case table
|
|
|
3811 @file{casetab.c} implements one sort of char table, the @dfn{case
|
|
|
3812 table}, which maps characters to other characters of possibly different
|
|
|
3813 case. These are used by XEmacs to implement case-changing primitives
|
|
|
3814 and to do case-insensitive searching.
|
|
|
3815
|
|
|
3816
|
|
|
3817
|
|
|
3818 @example
|
|
|
3819 syntax.c
|
|
|
3820 syntax.h
|
|
|
3821 @end example
|
|
|
3822
|
|
|
3823 @cindex scanner
|
|
|
3824 This module implements @dfn{syntax tables}, another sort of char table
|
|
|
3825 that maps characters into syntax classes that define the syntax of these
|
|
|
3826 characters (e.g. a parenthesis belongs to a class of @samp{open}
|
|
|
3827 characters that have corresponding @samp{close} characters and can be
|
|
|
3828 nested). This module also implements the Lisp @dfn{scanner}, a set of
|
|
|
3829 primitives for scanning over text based on syntax tables. This is used,
|
|
|
3830 for example, to find the matching parenthesis in a command such as
|
|
|
3831 @code{forward-sexp}, and by @file{font-lock.c} to locate quoted strings,
|
|
|
3832 comments, etc.
|
|
|
3833
|
|
|
3834
|
|
|
3835
|
|
|
3836 @example
|
|
|
3837 casefiddle.c
|
|
|
3838 @end example
|
|
|
3839
|
|
|
3840 This module implements various Lisp primitives for upcasing, downcasing
|
|
|
3841 and capitalizing strings or regions of buffers.
|
|
|
3842
|
|
|
3843
|
|
|
3844
|
|
|
3845 @example
|
|
|
3846 rangetab.c
|
|
|
3847 @end example
|
|
|
3848
|
|
|
3849 This module implements the @dfn{range table} Lisp object type, which
|
|
|
3850 provides for a mapping from ranges of integers to arbitrary Lisp
|
|
|
3851 objects.
|
|
|
3852
|
|
|
3853
|
|
|
3854
|
|
|
3855 @example
|
|
|
3856 opaque.c
|
|
|
3857 opaque.h
|
|
|
3858 @end example
|
|
|
3859
|
|
|
3860 This module implements the @dfn{opaque} Lisp object type, an
|
|
|
3861 internal-only Lisp object that encapsulates an arbitrary block of memory
|
|
|
3862 so that it can be managed by the Lisp allocation system. To create an
|
|
|
3863 opaque object, you call @code{make_opaque()}, passing a pointer to a
|
|
|
3864 block of memory. An object is created that is big enough to hold the
|
|
|
3865 memory, which is copied into the object's storage. The object will then
|
|
|
3866 stick around as long as you keep pointers to it, after which it will be
|
|
|
3867 automatically reclaimed.
|
|
|
3868
|
|
|
3869 @cindex mark method
|
|
|
3870 Opaque objects can also have an arbitrary @dfn{mark method} associated
|
|
|
3871 with them, in case the block of memory contains other Lisp objects that
|
|
|
3872 need to be marked for garbage-collection purposes. (If you need other
|
|
|
3873 object methods, such as a finalize method, you should just go ahead and
|
|
440
|
3874 create a new Lisp object type---it's not hard.)
|
|
428
|
3875
|
|
|
3876
|
|
|
3877
|
|
|
3878 @example
|
|
|
3879 abbrev.c
|
|
|
3880 @end example
|
|
|
3881
|
|
|
3882 This function provides a few primitives for doing dynamic abbreviation
|
|
|
3883 expansion. In XEmacs, most of the code for this has been moved into
|
|
|
3884 Lisp. Some C code remains for speed and because the primitive
|
|
|
3885 @code{self-insert-command} (which is executed for all self-inserting
|
|
|
3886 characters) hooks into the abbrev mechanism. (@code{self-insert-command}
|
|
|
3887 is itself in C only for speed.)
|
|
|
3888
|
|
|
3889
|
|
|
3890
|
|
|
3891 @example
|
|
|
3892 doc.c
|
|
|
3893 @end example
|
|
|
3894
|
|
|
3895 This function provides primitives for retrieving the documentation
|
|
|
3896 strings of functions and variables. These documentation strings contain
|
|
|
3897 certain special markers that get dynamically expanded (e.g. a
|
|
|
3898 reverse-lookup is performed on some named functions to retrieve their
|
|
|
3899 current key bindings). Some documentation strings (in particular, for
|
|
|
3900 the built-in primitives and pre-loaded Lisp functions) are stored
|
|
|
3901 externally in a file @file{DOC} in the @file{lib-src/} directory and
|
|
|
3902 need to be fetched from that file. (Part of the build stage involves
|
|
|
3903 building this file, and another part involves constructing an index for
|
|
|
3904 this file and embedding it into the executable, so that the functions in
|
|
|
3905 @file{doc.c} do not have to search the entire @file{DOC} file to find
|
|
|
3906 the appropriate documentation string.)
|
|
|
3907
|
|
|
3908
|
|
|
3909
|
|
|
3910 @example
|
|
|
3911 md5.c
|
|
|
3912 @end example
|
|
|
3913
|
|
|
3914 This function provides a Lisp primitive that implements the MD5 secure
|
|
|
3915 hashing scheme, used to create a large hash value of a string of data such that
|
|
|
3916 the data cannot be derived from the hash value. This is used for
|
|
|
3917 various security applications on the Internet.
|
|
|
3918
|
|
|
3919
|
|
|
3920
|
|
|
3921
|
|
|
3922 @node Modules for Interfacing with the Operating System
|
|
|
3923 @section Modules for Interfacing with the Operating System
|
|
|
3924
|
|
|
3925 @example
|
|
|
3926 callproc.c
|
|
|
3927 process.c
|
|
|
3928 process.h
|
|
|
3929 @end example
|
|
|
3930
|
|
|
3931 These modules allow XEmacs to spawn and communicate with subprocesses
|
|
|
3932 and network connections.
|
|
|
3933
|
|
|
3934 @cindex synchronous subprocesses
|
|
|
3935 @cindex subprocesses, synchronous
|
|
|
3936 @file{callproc.c} implements (through the @code{call-process}
|
|
|
3937 primitive) what are called @dfn{synchronous subprocesses}. This means
|
|
|
3938 that XEmacs runs a program, waits till it's done, and retrieves its
|
|
|
3939 output. A typical example might be calling the @file{ls} program to get
|
|
|
3940 a directory listing.
|
|
|
3941
|
|
|
3942 @cindex asynchronous subprocesses
|
|
|
3943 @cindex subprocesses, asynchronous
|
|
|
3944 @file{process.c} and @file{process.h} implement @dfn{asynchronous
|
|
|
3945 subprocesses}. This means that XEmacs starts a program and then
|
|
|
3946 continues normally, not waiting for the process to finish. Data can be
|
|
|
3947 sent to the process or retrieved from it as it's running. This is used
|
|
|
3948 for the @code{shell} command (which provides a front end onto a shell
|
|
|
3949 program such as @file{csh}), the mail and news readers implemented in
|
|
|
3950 XEmacs, etc. The result of calling @code{start-process} to start a
|
|
|
3951 subprocess is a process object, a particular kind of object used to
|
|
|
3952 communicate with the subprocess. You can send data to the process by
|
|
|
3953 passing the process object and the data to @code{send-process}, and you
|
|
|
3954 can specify what happens to data retrieved from the process by setting
|
|
|
3955 properties of the process object. (When the process sends data, XEmacs
|
|
|
3956 receives a process event, which says that there is data ready. When
|
|
|
3957 @code{dispatch-event} is called on this event, it reads the data from
|
|
|
3958 the process and does something with it, as specified by the process
|
|
|
3959 object's properties. Typically, this means inserting the data into a
|
|
|
3960 buffer or calling a function.) Another property of the process object is
|
|
|
3961 called the @dfn{sentinel}, which is a function that is called when the
|
|
|
3962 process terminates.
|
|
|
3963
|
|
|
3964 @cindex network connections
|
|
|
3965 Process objects are also used for network connections (connections to a
|
|
|
3966 process running on another machine). Network connections are started
|
|
|
3967 with @code{open-network-stream} but otherwise work just like
|
|
|
3968 subprocesses.
|
|
|
3969
|
|
|
3970
|
|
|
3971
|
|
|
3972 @example
|
|
|
3973 sysdep.c
|
|
|
3974 sysdep.h
|
|
|
3975 @end example
|
|
|
3976
|
|
|
3977 These modules implement most of the low-level, messy operating-system
|
|
|
3978 interface code. This includes various device control (ioctl) operations
|
|
|
3979 for file descriptors, TTY's, pseudo-terminals, etc. (usually this stuff
|
|
|
3980 is fairly system-dependent; thus the name of this module), and emulation
|
|
|
3981 of standard library functions and system calls on systems that don't
|
|
|
3982 provide them or have broken versions.
|
|
|
3983
|
|
|
3984
|
|
|
3985
|
|
|
3986 @example
|
|
|
3987 sysdir.h
|
|
|
3988 sysfile.h
|
|
|
3989 sysfloat.h
|
|
|
3990 sysproc.h
|
|
|
3991 syspwd.h
|
|
|
3992 syssignal.h
|
|
|
3993 systime.h
|
|
|
3994 systty.h
|
|
|
3995 syswait.h
|
|
|
3996 @end example
|
|
|
3997
|
|
|
3998 These header files provide consistent interfaces onto system-dependent
|
|
|
3999 header files and system calls. The idea is that, instead of including a
|
|
|
4000 standard header file like @file{<sys/param.h>} (which may or may not
|
|
|
4001 exist on various systems) or having to worry about whether all system
|
|
|
4002 provide a particular preprocessor constant, or having to deal with the
|
|
|
4003 four different paradigms for manipulating signals, you just include the
|
|
|
4004 appropriate @file{sys*.h} header file, which includes all the right
|
|
|
4005 system header files, defines and missing preprocessor constants,
|
|
|
4006 provides a uniform interface onto system calls, etc.
|
|
|
4007
|
|
|
4008 @file{sysdir.h} provides a uniform interface onto directory-querying
|
|
|
4009 functions. (In some cases, this is in conjunction with emulation
|
|
|
4010 functions in @file{sysdep.c}.)
|
|
|
4011
|
|
|
4012 @file{sysfile.h} includes all the necessary header files for standard
|
|
|
4013 system calls (e.g. @code{read()}), ensures that all necessary
|
|
|
4014 @code{open()} and @code{stat()} preprocessor constants are defined, and
|
|
|
4015 possibly (usually) substitutes sugared versions of @code{read()},
|
|
|
4016 @code{write()}, etc. that automatically restart interrupted I/O
|
|
|
4017 operations.
|
|
|
4018
|
|
|
4019 @file{sysfloat.h} includes the necessary header files for floating-point
|
|
|
4020 operations.
|
|
|
4021
|
|
|
4022 @file{sysproc.h} includes the necessary header files for calling
|
|
|
4023 @code{select()}, @code{fork()}, @code{execve()}, socket operations, and
|
|
|
4024 the like, and ensures that the @code{FD_*()} macros for descriptor-set
|
|
|
4025 manipulations are available.
|
|
|
4026
|
|
|
4027 @file{syspwd.h} includes the necessary header files for obtaining
|
|
|
4028 information from @file{/etc/passwd} (the functions are emulated under
|
|
|
4029 VMS).
|
|
|
4030
|
|
|
4031 @file{syssignal.h} includes the necessary header files for
|
|
|
4032 signal-handling and provides a uniform interface onto the different
|
|
|
4033 signal-handling and signal-blocking paradigms.
|
|
|
4034
|
|
|
4035 @file{systime.h} includes the necessary header files and provides
|
|
|
4036 uniform interfaces for retrieving the time of day, setting file
|
|
|
4037 access/modification times, getting the amount of time used by the XEmacs
|
|
|
4038 process, etc.
|
|
|
4039
|
|
|
4040 @file{systty.h} buffers against the infinitude of different ways of
|
|
|
4041 controlling TTY's.
|
|
|
4042
|
|
|
4043 @file{syswait.h} provides a uniform way of retrieving the exit status
|
|
|
4044 from a @code{wait()}ed-on process (some systems use a union, others use
|
|
|
4045 an int).
|
|
|
4046
|
|
|
4047
|
|
|
4048
|
|
|
4049 @example
|
|
|
4050 hpplay.c
|
|
|
4051 libsst.c
|
|
|
4052 libsst.h
|
|
|
4053 libst.h
|
|
|
4054 linuxplay.c
|
|
|
4055 nas.c
|
|
|
4056 sgiplay.c
|
|
|
4057 sound.c
|
|
|
4058 sunplay.c
|
|
|
4059 @end example
|
|
|
4060
|
|
|
4061 These files implement the ability to play various sounds on some types
|
|
|
4062 of computers. You have to configure your XEmacs with sound support in
|
|
|
4063 order to get this capability.
|
|
|
4064
|
|
|
4065 @file{sound.c} provides the generic interface. It implements various
|
|
|
4066 Lisp primitives and variables that let you specify which sounds should
|
|
|
4067 be played in certain conditions. (The conditions are identified by
|
|
|
4068 symbols, which are passed to @code{ding} to make a sound. Various
|
|
|
4069 standard functions call this function at certain times; if sound support
|
|
|
4070 does not exist, a simple beep results.
|
|
|
4071
|
|
|
4072 @cindex native sound
|
|
|
4073 @cindex sound, native
|
|
|
4074 @file{sgiplay.c}, @file{sunplay.c}, @file{hpplay.c}, and
|
|
|
4075 @file{linuxplay.c} interface to the machine's speaker for various
|
|
|
4076 different kind of machines. This is called @dfn{native} sound.
|
|
|
4077
|
|
|
4078 @cindex sound, network
|
|
|
4079 @cindex network sound
|
|
|
4080 @cindex NAS
|
|
|
4081 @file{nas.c} interfaces to a computer somewhere else on the network
|
|
|
4082 using the NAS (Network Audio Server) protocol, playing sounds on that
|
|
|
4083 machine. This allows you to run XEmacs on a remote machine, with its
|
|
|
4084 display set to your local machine, and have the sounds be made on your
|
|
|
4085 local machine, provided that you have a NAS server running on your local
|
|
|
4086 machine.
|
|
|
4087
|
|
|
4088 @file{libsst.c}, @file{libsst.h}, and @file{libst.h} provide some
|
|
|
4089 additional functions for playing sound on a Sun SPARC but are not
|
|
|
4090 currently in use.
|
|
|
4091
|
|
|
4092
|
|
|
4093
|
|
|
4094 @example
|
|
|
4095 tooltalk.c
|
|
|
4096 tooltalk.h
|
|
|
4097 @end example
|
|
|
4098
|
|
|
4099 These two modules implement an interface to the ToolTalk protocol, which
|
|
|
4100 is an interprocess communication protocol implemented on some versions
|
|
|
4101 of Unix. ToolTalk is a high-level protocol that allows processes to
|
|
|
4102 register themselves as providers of particular services; other processes
|
|
|
4103 can then request a service without knowing or caring exactly who is
|
|
|
4104 providing the service. It is similar in spirit to the DDE protocol
|
|
|
4105 provided under Microsoft Windows. ToolTalk is a part of the new CDE
|
|
|
4106 (Common Desktop Environment) specification and is used to connect the
|
|
|
4107 parts of the SPARCWorks development environment.
|
|
|
4108
|
|
|
4109
|
|
|
4110
|
|
|
4111 @example
|
|
|
4112 getloadavg.c
|
|
|
4113 @end example
|
|
|
4114
|
|
|
4115 This module provides the ability to retrieve the system's current load
|
|
|
4116 average. (The way to do this is highly system-specific, unfortunately,
|
|
|
4117 and requires a lot of special-case code.)
|
|
|
4118
|
|
|
4119
|
|
|
4120
|
|
|
4121 @example
|
|
|
4122 sunpro.c
|
|
|
4123 @end example
|
|
|
4124
|
|
|
4125 This module provides a small amount of code used internally at Sun to
|
|
|
4126 keep statistics on the usage of XEmacs.
|
|
|
4127
|
|
|
4128
|
|
|
4129
|
|
|
4130 @example
|
|
|
4131 broken-sun.h
|
|
|
4132 strcmp.c
|
|
|
4133 strcpy.c
|
|
|
4134 sunOS-fix.c
|
|
|
4135 @end example
|
|
|
4136
|
|
|
4137 These files provide replacement functions and prototypes to fix numerous
|
|
|
4138 bugs in early releases of SunOS 4.1.
|
|
|
4139
|
|
|
4140
|
|
|
4141
|
|
|
4142 @example
|
|
|
4143 hftctl.c
|
|
|
4144 @end example
|
|
|
4145
|
|
|
4146 This module provides some terminal-control code necessary on versions of
|
|
|
4147 AIX prior to 4.1.
|
|
|
4148
|
|
|
4149
|
|
|
4150
|
|
|
4151 @example
|
|
|
4152 msdos.c
|
|
|
4153 msdos.h
|
|
|
4154 @end example
|
|
|
4155
|
|
|
4156 These modules are used for MS-DOS support, which does not work in
|
|
|
4157 XEmacs.
|
|
|
4158
|
|
|
4159
|
|
|
4160
|
|
|
4161 @node Modules for Interfacing with X Windows
|
|
|
4162 @section Modules for Interfacing with X Windows
|
|
|
4163
|
|
|
4164 @example
|
|
|
4165 Emacs.ad.h
|
|
|
4166 @end example
|
|
|
4167
|
|
|
4168 A file generated from @file{Emacs.ad}, which contains XEmacs-supplied
|
|
|
4169 fallback resources (so that XEmacs has pretty defaults).
|
|
|
4170
|
|
|
4171
|
|
|
4172
|
|
|
4173 @example
|
|
|
4174 EmacsFrame.c
|
|
|
4175 EmacsFrame.h
|
|
|
4176 EmacsFrameP.h
|
|
|
4177 @end example
|
|
|
4178
|
|
|
4179 These modules implement an Xt widget class that encapsulates a frame.
|
|
|
4180 This is for ease in integrating with Xt. The EmacsFrame widget covers
|
|
|
4181 the entire X window except for the menubar; the scrollbars are
|
|
|
4182 positioned on top of the EmacsFrame widget.
|
|
|
4183
|
|
|
4184 @strong{Warning:} Abandon hope, all ye who enter here. This code took
|
|
|
4185 an ungodly amount of time to get right, and is likely to fall apart
|
|
|
4186 mercilessly at the slightest change. Such is life under Xt.
|
|
|
4187
|
|
|
4188
|
|
|
4189
|
|
|
4190 @example
|
|
|
4191 EmacsManager.c
|
|
|
4192 EmacsManager.h
|
|
|
4193 EmacsManagerP.h
|
|
|
4194 @end example
|
|
|
4195
|
|
|
4196 These modules implement a simple Xt manager (i.e. composite) widget
|
|
|
4197 class that simply lets its children set whatever geometry they want.
|
|
|
4198 It's amazing that Xt doesn't provide this standardly, but on second
|
|
|
4199 thought, it makes sense, considering how amazingly broken Xt is.
|
|
|
4200
|
|
|
4201
|
|
|
4202 @example
|
|
|
4203 EmacsShell-sub.c
|
|
|
4204 EmacsShell.c
|
|
|
4205 EmacsShell.h
|
|
|
4206 EmacsShellP.h
|
|
|
4207 @end example
|
|
|
4208
|
|
|
4209 These modules implement two Xt widget classes that are subclasses of
|
|
|
4210 the TopLevelShell and TransientShell classes. This is necessary to deal
|
|
|
4211 with more brokenness that Xt has sadistically thrust onto the backs of
|
|
|
4212 developers.
|
|
|
4213
|
|
|
4214
|
|
|
4215
|
|
|
4216 @example
|
|
|
4217 xgccache.c
|
|
|
4218 xgccache.h
|
|
|
4219 @end example
|
|
|
4220
|
|
|
4221 These modules provide functions for maintenance and caching of GC's
|
|
|
4222 (graphics contexts) under the X Window System. This code is junky and
|
|
|
4223 needs to be rewritten.
|
|
|
4224
|
|
|
4225
|
|
|
4226
|
|
|
4227 @example
|
|
|
4228 xselect.c
|
|
|
4229 @end example
|
|
|
4230
|
|
|
4231 @cindex selections
|
|
|
4232 This module provides an interface to the X Window System's concept of
|
|
|
4233 @dfn{selections}, the standard way for X applications to communicate
|
|
|
4234 with each other.
|
|
|
4235
|
|
|
4236
|
|
|
4237
|
|
|
4238 @example
|
|
|
4239 xintrinsic.h
|
|
|
4240 xintrinsicp.h
|
|
|
4241 xmmanagerp.h
|
|
|
4242 xmprimitivep.h
|
|
|
4243 @end example
|
|
|
4244
|
|
|
4245 These header files are similar in spirit to the @file{sys*.h} files and buffer
|
|
|
4246 against different implementations of Xt and Motif.
|
|
|
4247
|
|
|
4248 @itemize @bullet
|
|
|
4249 @item
|
|
|
4250 @file{xintrinsic.h} should be included in place of @file{<Intrinsic.h>}.
|
|
|
4251 @item
|
|
|
4252 @file{xintrinsicp.h} should be included in place of @file{<IntrinsicP.h>}.
|
|
|
4253 @item
|
|
|
4254 @file{xmmanagerp.h} should be included in place of @file{<XmManagerP.h>}.
|
|
|
4255 @item
|
|
|
4256 @file{xmprimitivep.h} should be included in place of @file{<XmPrimitiveP.h>}.
|
|
|
4257 @end itemize
|
|
|
4258
|
|
|
4259
|
|
|
4260
|
|
|
4261 @example
|
|
|
4262 xmu.c
|
|
|
4263 xmu.h
|
|
|
4264 @end example
|
|
|
4265
|
|
|
4266 These files provide an emulation of the Xmu library for those systems
|
|
|
4267 (i.e. HPUX) that don't provide it as a standard part of X.
|
|
|
4268
|
|
|
4269
|
|
|
4270
|
|
|
4271 @example
|
|
|
4272 ExternalClient-Xlib.c
|
|
|
4273 ExternalClient.c
|
|
|
4274 ExternalClient.h
|
|
|
4275 ExternalClientP.h
|
|
|
4276 ExternalShell.c
|
|
|
4277 ExternalShell.h
|
|
|
4278 ExternalShellP.h
|
|
|
4279 extw-Xlib.c
|
|
|
4280 extw-Xlib.h
|
|
|
4281 extw-Xt.c
|
|
|
4282 extw-Xt.h
|
|
|
4283 @end example
|
|
|
4284
|
|
|
4285 @cindex external widget
|
|
|
4286 These files provide the @dfn{external widget} interface, which allows an
|
|
|
4287 XEmacs frame to appear as a widget in another application. To do this,
|
|
|
4288 you have to configure with @samp{--external-widget}.
|
|
|
4289
|
|
|
4290 @file{ExternalShell*} provides the server (XEmacs) side of the
|
|
|
4291 connection.
|
|
|
4292
|
|
|
4293 @file{ExternalClient*} provides the client (other application) side of
|
|
|
4294 the connection. These files are not compiled into XEmacs but are
|
|
|
4295 compiled into libraries that are then linked into your application.
|
|
|
4296
|
|
|
4297 @file{extw-*} is common code that is used for both the client and server.
|
|
|
4298
|
|
|
4299 Don't touch this code; something is liable to break if you do.
|
|
|
4300
|
|
|
4301
|
|
|
4302
|
|
|
4303 @node Modules for Internationalization
|
|
|
4304 @section Modules for Internationalization
|
|
|
4305
|
|
|
4306 @example
|
|
|
4307 mule-canna.c
|
|
|
4308 mule-ccl.c
|
|
|
4309 mule-charset.c
|
|
|
4310 mule-charset.h
|
|
|
4311 mule-coding.c
|
|
|
4312 mule-coding.h
|
|
|
4313 mule-mcpath.c
|
|
|
4314 mule-mcpath.h
|
|
|
4315 mule-wnnfns.c
|
|
|
4316 mule.c
|
|
|
4317 @end example
|
|
|
4318
|
|
|
4319 These files implement the MULE (Asian-language) support. Note that MULE
|
|
|
4320 actually provides a general interface for all sorts of languages, not
|
|
|
4321 just Asian languages (although they are generally the most complicated
|
|
|
4322 to support). This code is still in beta.
|
|
|
4323
|
|
|
4324 @file{mule-charset.*} and @file{mule-coding.*} provide the heart of the
|
|
|
4325 XEmacs MULE support. @file{mule-charset.*} implements the @dfn{charset}
|
|
|
4326 Lisp object type, which encapsulates a character set (an ordered one- or
|
|
|
4327 two-dimensional set of characters, such as US ASCII or JISX0208 Japanese
|
|
|
4328 Kanji).
|
|
|
4329
|
|
|
4330 @file{mule-coding.*} implements the @dfn{coding-system} Lisp object
|
|
|
4331 type, which encapsulates a method of converting between different
|
|
|
4332 encodings. An encoding is a representation of a stream of characters,
|
|
|
4333 possibly from multiple character sets, using a stream of bytes or words,
|
|
|
4334 and defines (e.g.) which escape sequences are used to specify particular
|
|
|
4335 character sets, how the indices for a character are converted into bytes
|
|
|
4336 (sometimes this involves setting the high bit; sometimes complicated
|
|
|
4337 rearranging of the values takes place, as in the Shift-JIS encoding),
|
|
|
4338 etc.
|
|
|
4339
|
|
|
4340 @file{mule-ccl.c} provides the CCL (Code Conversion Language)
|
|
|
4341 interpreter. CCL is similar in spirit to Lisp byte code and is used to
|
|
|
4342 implement converters for custom encodings.
|
|
|
4343
|
|
|
4344 @file{mule-canna.c} and @file{mule-wnnfns.c} implement interfaces to
|
|
|
4345 external programs used to implement the Canna and WNN input methods,
|
|
|
4346 respectively. This is currently in beta.
|
|
|
4347
|
|
|
4348 @file{mule-mcpath.c} provides some functions to allow for pathnames
|
|
|
4349 containing extended characters. This code is fragmentary, obsolete, and
|
|
|
4350 completely non-working. Instead, @var{pathname-coding-system} is used
|
|
|
4351 to specify conversions of names of files and directories. The standard
|
|
|
4352 C I/O functions like @samp{open()} are wrapped so that conversion occurs
|
|
|
4353 automatically.
|
|
|
4354
|
|
|
4355 @file{mule.c} provides a few miscellaneous things that should probably
|
|
|
4356 be elsewhere.
|
|
|
4357
|
|
|
4358
|
|
|
4359
|
|
|
4360 @example
|
|
|
4361 intl.c
|
|
|
4362 @end example
|
|
|
4363
|
|
|
4364 This provides some miscellaneous internationalization code for
|
|
|
4365 implementing message translation and interfacing to the Ximp input
|
|
|
4366 method. None of this code is currently working.
|
|
|
4367
|
|
|
4368
|
|
|
4369
|
|
|
4370 @example
|
|
|
4371 iso-wide.h
|
|
|
4372 @end example
|
|
|
4373
|
|
|
4374 This contains leftover code from an earlier implementation of
|
|
|
4375 Asian-language support, and is not currently used.
|
|
|
4376
|
|
|
4377
|
|
|
4378
|
|
|
4379
|
|
|
4380 @node Allocation of Objects in XEmacs Lisp, Events and the Event Loop, A Summary of the Various XEmacs Modules, Top
|
|
|
4381 @chapter Allocation of Objects in XEmacs Lisp
|
|
|
4382
|
|
|
4383 @menu
|
|
|
4384 * Introduction to Allocation::
|
|
|
4385 * Garbage Collection::
|
|
|
4386 * GCPROing::
|
|
|
4387 * Garbage Collection - Step by Step::
|
|
|
4388 * Integers and Characters::
|
|
|
4389 * Allocation from Frob Blocks::
|
|
|
4390 * lrecords::
|
|
|
4391 * Low-level allocation::
|
|
|
4392 * Pure Space::
|
|
|
4393 * Cons::
|
|
|
4394 * Vector::
|
|
|
4395 * Bit Vector::
|
|
|
4396 * Symbol::
|
|
|
4397 * Marker::
|
|
|
4398 * String::
|
|
|
4399 * Compiled Function::
|
|
|
4400 @end menu
|
|
|
4401
|
|
|
4402 @node Introduction to Allocation
|
|
|
4403 @section Introduction to Allocation
|
|
|
4404
|
|
|
4405 Emacs Lisp, like all Lisps, has garbage collection. This means that
|
|
|
4406 the programmer never has to explicitly free (destroy) an object; it
|
|
|
4407 happens automatically when the object becomes inaccessible. Most
|
|
|
4408 experts agree that garbage collection is a necessity in a modern,
|
|
|
4409 high-level language. Its omission from C stems from the fact that C was
|
|
|
4410 originally designed to be a nice abstract layer on top of assembly
|
|
|
4411 language, for writing kernels and basic system utilities rather than
|
|
|
4412 large applications.
|
|
|
4413
|
|
|
4414 Lisp objects can be created by any of a number of Lisp primitives.
|
|
|
4415 Most object types have one or a small number of basic primitives
|
|
|
4416 for creating objects. For conses, the basic primitive is @code{cons};
|
|
|
4417 for vectors, the primitives are @code{make-vector} and @code{vector}; for
|
|
|
4418 symbols, the primitives are @code{make-symbol} and @code{intern}; etc.
|
|
|
4419 Some Lisp objects, especially those that are primarily used internally,
|
|
|
4420 have no corresponding Lisp primitives. Every Lisp object, though,
|
|
|
4421 has at least one C primitive for creating it.
|
|
|
4422
|
|
|
4423 Recall from section (VII) that a Lisp object, as stored in a 32-bit
|
|
|
4424 or 64-bit word, has a mark bit, a few tag bits, and a ``value'' that
|
|
|
4425 occupies the remainder of the bits. We can separate the different
|
|
|
4426 Lisp object types into four broad categories:
|
|
|
4427
|
|
|
4428 @itemize @bullet
|
|
|
4429 @item
|
|
|
4430 (a) Those for whom the value directly represents the contents of the
|
|
|
4431 Lisp object. Only two types are in this category: integers and
|
|
|
4432 characters. No special allocation or garbage collection is necessary
|
|
|
4433 for such objects. Lisp objects of these types do not need to be
|
|
|
4434 @code{GCPRO}ed.
|
|
|
4435 @end itemize
|
|
|
4436
|
|
|
4437 In the remaining three categories, the value is a pointer to a
|
|
|
4438 structure.
|
|
|
4439
|
|
|
4440 @itemize @bullet
|
|
|
4441 @item
|
|
|
4442 @cindex frob block
|
|
|
4443 (b) Those for whom the tag directly specifies the type. Recall that
|
|
|
4444 there are only three tag bits; this means that at most five types can be
|
|
|
4445 specified this way. The most commonly-used types are stored in this
|
|
|
4446 format; this includes conses, strings, vectors, and sometimes symbols.
|
|
|
4447 With the exception of vectors, objects in this category are allocated in
|
|
|
4448 @dfn{frob blocks}, i.e. large blocks of memory that are subdivided into
|
|
|
4449 individual objects. This saves a lot on malloc overhead, since there
|
|
|
4450 are typically quite a lot of these objects around, and the objects are
|
|
440
|
4451 small. (A cons, for example, occupies 8 bytes on 32-bit machines---4
|
|
428
|
4452 bytes for each of the two objects it contains.) Vectors are individually
|
|
|
4453 @code{malloc()}ed since they are of variable size. (It would be
|
|
|
4454 possible, and desirable, to allocate vectors of certain small sizes out
|
|
|
4455 of frob blocks, but it isn't currently done.) Strings are handled
|
|
|
4456 specially: Each string is allocated in two parts, a fixed size structure
|
|
|
4457 containing a length and a data pointer, and the actual data of the
|
|
|
4458 string. The former structure is allocated in frob blocks as usual, and
|
|
|
4459 the latter data is stored in @dfn{string chars blocks} and is relocated
|
|
|
4460 during garbage collection to eliminate holes.
|
|
|
4461 @end itemize
|
|
|
4462
|
|
|
4463 In the remaining two categories, the type is stored in the object
|
|
|
4464 itself. The tag for all such objects is the generic @dfn{lrecord}
|
|
|
4465 (Lisp_Record) tag. The first four bytes (or eight, for 64-bit machines)
|
|
|
4466 of the object's structure are a pointer to a structure that describes
|
|
|
4467 the object's type, which includes method pointers and a pointer to a
|
|
|
4468 string naming the type. Note that it's possible to save some space by
|
|
|
4469 using a one- or two-byte tag, rather than a four- or eight-byte pointer
|
|
|
4470 to store the type, but it's not clear it's worth making the change.
|
|
|
4471
|
|
|
4472 @itemize @bullet
|
|
|
4473 @item
|
|
|
4474 (c) Those lrecords that are allocated in frob blocks (see above). This
|
|
|
4475 includes the objects that are most common and relatively small, and
|
|
|
4476 includes floats, compiled functions, symbols (when not in category (b)),
|
|
|
4477 extents, events, and markers. With the cleanup of frob blocks done in
|
|
|
4478 19.12, it's not terribly hard to add more objects to this category, but
|
|
|
4479 it's a bit trickier than adding an object type to type (d) (esp. if the
|
|
|
4480 object needs a finalization method), and is not likely to save much
|
|
|
4481 space unless the object is small and there are many of them. (In fact,
|
|
|
4482 if there are very few of them, it might actually waste space.)
|
|
|
4483 @item
|
|
|
4484 (d) Those lrecords that are individually @code{malloc()}ed. These are
|
|
|
4485 called @dfn{lcrecords}. All other types are in this category. Adding a
|
|
|
4486 new type to this category is comparatively easy, and all types added
|
|
|
4487 since 19.8 (when the current allocation scheme was devised, by Richard
|
|
|
4488 Mlynarik), with the exception of the character type, have been in this
|
|
|
4489 category.
|
|
|
4490 @end itemize
|
|
|
4491
|
|
|
4492 Note that bit vectors are a bit of a special case. They are
|
|
|
4493 simple lrecords as in category (c), but are individually @code{malloc()}ed
|
|
|
4494 like vectors. You can basically view them as exactly like vectors
|
|
|
4495 except that their type is stored in lrecord fashion rather than
|
|
|
4496 in directly-tagged fashion.
|
|
|
4497
|
|
|
4498 Note that FSF Emacs redesigned their object system in 19.29 to follow
|
|
|
4499 a similar scheme. However, given RMS's expressed dislike for data
|
|
|
4500 abstraction, the FSF scheme is not nearly as clean or as easy to
|
|
|
4501 extend. (FSF calls items of type (c) @code{Lisp_Misc} and items of type
|
|
|
4502 (d) @code{Lisp_Vectorlike}, with separate tags for each, although
|
|
|
4503 @code{Lisp_Vectorlike} is also used for vectors.)
|
|
|
4504
|
|
|
4505 @node Garbage Collection
|
|
|
4506 @section Garbage Collection
|
|
|
4507 @cindex garbage collection
|
|
|
4508
|
|
|
4509 @cindex mark and sweep
|
|
|
4510 Garbage collection is simple in theory but tricky to implement.
|
|
|
4511 Emacs Lisp uses the oldest garbage collection method, called
|
|
|
4512 @dfn{mark and sweep}. Garbage collection begins by starting with
|
|
|
4513 all accessible locations (i.e. all variables and other slots where
|
|
|
4514 Lisp objects might occur) and recursively traversing all objects
|
|
|
4515 accessible from those slots, marking each one that is found.
|
|
|
4516 We then go through all of memory and free each object that is
|
|
|
4517 not marked, and unmarking each object that is marked. Note
|
|
|
4518 that ``all of memory'' means all currently allocated objects.
|
|
|
4519 Traversing all these objects means traversing all frob blocks,
|
|
|
4520 all vectors (which are chained in one big list), and all
|
|
|
4521 lcrecords (which are likewise chained).
|
|
|
4522
|
|
|
4523 Note that, when an object is marked, the mark has to occur
|
|
|
4524 inside of the object's structure, rather than in the 32-bit
|
|
|
4525 @code{Lisp_Object} holding the object's pointer; i.e. you can't just
|
|
|
4526 set the pointer's mark bit. This is because there may be many
|
|
|
4527 pointers to the same object. This means that the method of
|
|
|
4528 marking an object can differ depending on the type. The
|
|
|
4529 different marking methods are approximately as follows:
|
|
|
4530
|
|
|
4531 @enumerate
|
|
|
4532 @item
|
|
|
4533 For conses, the mark bit of the car is set.
|
|
|
4534 @item
|
|
|
4535 For strings, the mark bit of the string's plist is set.
|
|
|
4536 @item
|
|
|
4537 For symbols when not lrecords, the mark bit of the
|
|
|
4538 symbol's plist is set.
|
|
|
4539 @item
|
|
|
4540 For vectors, the length is negated after adding 1.
|
|
|
4541 @item
|
|
|
4542 For lrecords, the pointer to the structure describing
|
|
|
4543 the type is changed (see below).
|
|
|
4544 @item
|
|
|
4545 Integers and characters do not need to be marked, since
|
|
|
4546 no allocation occurs for them.
|
|
|
4547 @end enumerate
|
|
|
4548
|
|
|
4549 The details of this are in the @code{mark_object()} function.
|
|
|
4550
|
|
|
4551 Note that any code that operates during garbage collection has
|
|
|
4552 to be especially careful because of the fact that some objects
|
|
|
4553 may be marked and as such may not look like they normally do.
|
|
|
4554 In particular:
|
|
|
4555
|
|
|
4556 @itemize @bullet
|
|
|
4557 Some object pointers may have their mark bit set. This will make
|
|
|
4558 @code{FOOBARP()} predicates fail. Use @code{GC_FOOBARP()} to deal with
|
|
|
4559 this.
|
|
|
4560 @item
|
|
|
4561 Even if you clear the mark bit, @code{FOOBARP()} will still fail
|
|
|
4562 for lrecords because the implementation pointer has been
|
|
|
4563 changed (see below). @code{GC_FOOBARP()} will correctly deal with
|
|
|
4564 this.
|
|
|
4565 @item
|
|
|
4566 Vectors have their size field munged, so anything that
|
|
|
4567 looks at this field will fail.
|
|
|
4568 @item
|
|
|
4569 Note that @code{XFOOBAR()} macros @emph{will} work correctly on object
|
|
|
4570 pointers with their mark bit set, because the logical shift operations
|
|
|
4571 that remove the tag also remove the mark bit.
|
|
|
4572 @end itemize
|
|
|
4573
|
|
|
4574 Finally, note that garbage collection can be invoked explicitly
|
|
|
4575 by calling @code{garbage-collect} but is also called automatically
|
|
|
4576 by @code{eval}, once a certain amount of memory has been allocated
|
|
|
4577 since the last garbage collection (according to @code{gc-cons-threshold}).
|
|
|
4578
|
|
|
4579 @node GCPROing
|
|
|
4580 @section @code{GCPRO}ing
|
|
|
4581
|
|
|
4582 @code{GCPRO}ing is one of the ugliest and trickiest parts of Emacs
|
|
|
4583 internals. The basic idea is that whenever garbage collection
|
|
|
4584 occurs, all in-use objects must be reachable somehow or
|
|
|
4585 other from one of the roots of accessibility. The roots
|
|
|
4586 of accessibility are:
|
|
|
4587
|
|
|
4588 @enumerate
|
|
|
4589 @item
|
|
|
4590 All objects that have been @code{staticpro()}d. This is used for
|
|
|
4591 any global C variables that hold Lisp objects. A call to
|
|
|
4592 @code{staticpro()} happens implicitly as a result of any symbols
|
|
|
4593 declared with @code{defsymbol()} and any variables declared with
|
|
|
4594 @code{DEFVAR_FOO()}. You need to explicitly call @code{staticpro()}
|
|
|
4595 (in the @code{vars_of_foo()} method of a module) for other global
|
|
|
4596 C variables holding Lisp objects. (This typically includes
|
|
|
4597 internal lists and such things.)
|
|
|
4598
|
|
|
4599 Note that @code{obarray} is one of the @code{staticpro()}d things.
|
|
|
4600 Therefore, all functions and variables get marked through this.
|
|
|
4601 @item
|
|
|
4602 Any shadowed bindings that are sitting on the @code{specpdl} stack.
|
|
|
4603 @item
|
|
|
4604 Any objects sitting in currently active (Lisp) stack frames,
|
|
|
4605 catches, and condition cases.
|
|
|
4606 @item
|
|
|
4607 A couple of special-case places where active objects are
|
|
|
4608 located.
|
|
|
4609 @item
|
|
|
4610 Anything currently marked with @code{GCPRO}.
|
|
|
4611 @end enumerate
|
|
|
4612
|
|
|
4613 Marking with @code{GCPRO} is necessary because some C functions (quite
|
|
|
4614 a lot, in fact), allocate objects during their operation. Quite
|
|
|
4615 frequently, there will be no other pointer to the object while the
|
|
|
4616 function is running, and if a garbage collection occurs and the object
|
|
|
4617 needs to be referenced again, bad things will happen. The solution is
|
|
|
4618 to mark those objects with @code{GCPRO}. Unfortunately this is easy to
|
|
|
4619 forget, and there is basically no way around this problem. Here are
|
|
|
4620 some rules, though:
|
|
|
4621
|
|
|
4622 @enumerate
|
|
|
4623 @item
|
|
|
4624 For every @code{GCPRO@var{n}}, there have to be declarations of
|
|
|
4625 @code{struct gcpro gcpro1, gcpro2}, etc.
|
|
|
4626
|
|
|
4627 @item
|
|
|
4628 You @emph{must} @code{UNGCPRO} anything that's @code{GCPRO}ed, and you
|
|
|
4629 @emph{must not} @code{UNGCPRO} if you haven't @code{GCPRO}ed. Getting
|
|
|
4630 either of these wrong will lead to crashes, often in completely random
|
|
|
4631 places unrelated to where the problem lies.
|
|
|
4632
|
|
|
4633 @item
|
|
|
4634 The way this actually works is that all currently active @code{GCPRO}s
|
|
|
4635 are chained through the @code{struct gcpro} local variables, with the
|
|
|
4636 variable @samp{gcprolist} pointing to the head of the list and the nth
|
|
|
4637 local @code{gcpro} variable pointing to the first @code{gcpro} variable
|
|
|
4638 in the next enclosing stack frame. Each @code{GCPRO}ed thing is an
|
|
|
4639 lvalue, and the @code{struct gcpro} local variable contains a pointer to
|
|
|
4640 this lvalue. This is why things will mess up badly if you don't pair up
|
|
440
|
4641 the @code{GCPRO}s and @code{UNGCPRO}s---you will end up with
|
|
428
|
4642 @code{gcprolist}s containing pointers to @code{struct gcpro}s or local
|
|
|
4643 @code{Lisp_Object} variables in no-longer-active stack frames.
|
|
|
4644
|
|
|
4645 @item
|
|
|
4646 It is actually possible for a single @code{struct gcpro} to
|
|
|
4647 protect a contiguous array of any number of values, rather than
|
|
|
4648 just a single lvalue. To effect this, call @code{GCPRO@var{n}} as usual on
|
|
|
4649 the first object in the array and then set @code{gcpro@var{n}.nvars}.
|
|
|
4650
|
|
|
4651 @item
|
|
|
4652 @strong{Strings are relocated.} What this means in practice is that the
|
|
|
4653 pointer obtained using @code{XSTRING_DATA()} is liable to change at any
|
|
|
4654 time, and you should never keep it around past any function call, or
|
|
|
4655 pass it as an argument to any function that might cause a garbage
|
|
|
4656 collection. This is why a number of functions accept either a
|
|
|
4657 ``non-relocatable'' @code{char *} pointer or a relocatable Lisp string,
|
|
|
4658 and only access the Lisp string's data at the very last minute. In some
|
|
|
4659 cases, you may end up having to @code{alloca()} some space and copy the
|
|
|
4660 string's data into it.
|
|
|
4661
|
|
|
4662 @item
|
|
|
4663 By convention, if you have to nest @code{GCPRO}'s, use @code{NGCPRO@var{n}}
|
|
|
4664 (along with @code{struct gcpro ngcpro1, ngcpro2}, etc.), @code{NNGCPRO@var{n}},
|
|
|
4665 etc. This avoids compiler warnings about shadowed locals.
|
|
|
4666
|
|
|
4667 @item
|
|
|
4668 It is @emph{always} better to err on the side of extra @code{GCPRO}s
|
|
|
4669 rather than too few. The extra cycles spent on this are
|
|
|
4670 almost never going to make a whit of difference in the
|
|
|
4671 speed of anything.
|
|
|
4672
|
|
|
4673 @item
|
|
|
4674 The general rule to follow is that caller, not callee, @code{GCPRO}s.
|
|
|
4675 That is, you should not have to explicitly @code{GCPRO} any Lisp objects
|
|
|
4676 that are passed in as parameters.
|
|
|
4677
|
|
|
4678 One exception from this rule is if you ever plan to change the parameter
|
|
|
4679 value, and store a new object in it. In that case, you @emph{must}
|
|
|
4680 @code{GCPRO} the parameter, because otherwise the new object will not be
|
|
|
4681 protected.
|
|
|
4682
|
|
|
4683 So, if you create any Lisp objects (remember, this happens in all sorts
|
|
|
4684 of circumstances, e.g. with @code{Fcons()}, etc.), you are responsible
|
|
|
4685 for @code{GCPRO}ing them, unless you are @emph{absolutely sure} that
|
|
|
4686 there's no possibility that a garbage-collection can occur while you
|
|
|
4687 need to use the object. Even then, consider @code{GCPRO}ing.
|
|
|
4688
|
|
|
4689 @item
|
|
|
4690 A garbage collection can occur whenever anything calls @code{Feval}, or
|
|
|
4691 whenever a QUIT can occur where execution can continue past
|
|
|
4692 this. (Remember, this is almost anywhere.)
|
|
|
4693
|
|
|
4694 @item
|
|
|
4695 If you have the @emph{least smidgeon of doubt} about whether
|
|
|
4696 you need to @code{GCPRO}, you should @code{GCPRO}.
|
|
|
4697
|
|
|
4698 @item
|
|
|
4699 Beware of @code{GCPRO}ing something that is uninitialized. If you have
|
|
|
4700 any shade of doubt about this, initialize all your variables to @code{Qnil}.
|
|
|
4701
|
|
|
4702 @item
|
|
|
4703 Be careful of traps, like calling @code{Fcons()} in the argument to
|
|
|
4704 another function. By the ``caller protects'' law, you should be
|
|
|
4705 @code{GCPRO}ing the newly-created cons, but you aren't. A certain
|
|
|
4706 number of functions that are commonly called on freshly created stuff
|
|
|
4707 (e.g. @code{nconc2()}, @code{Fsignal()}), break the ``caller protects''
|
|
|
4708 law and go ahead and @code{GCPRO} their arguments so as to simplify
|
|
|
4709 things, but make sure and check if it's OK whenever doing something like
|
|
|
4710 this.
|
|
|
4711
|
|
|
4712 @item
|
|
|
4713 Once again, remember to @code{GCPRO}! Bugs resulting from insufficient
|
|
|
4714 @code{GCPRO}ing are intermittent and extremely difficult to track down,
|
|
|
4715 often showing up in crashes inside of @code{garbage-collect} or in
|
|
|
4716 weirdly corrupted objects or even in incorrect values in a totally
|
|
|
4717 different section of code.
|
|
|
4718 @end enumerate
|
|
|
4719
|
|
|
4720 @cindex garbage collection, conservative
|
|
|
4721 @cindex conservative garbage collection
|
|
|
4722 Given the extremely error-prone nature of the @code{GCPRO} scheme, and
|
|
|
4723 the difficulties in tracking down, it should be considered a deficiency
|
|
|
4724 in the XEmacs code. A solution to this problem would involve
|
|
|
4725 implementing so-called @dfn{conservative} garbage collection for the C
|
|
|
4726 stack. That involves looking through all of stack memory and treating
|
|
|
4727 anything that looks like a reference to an object as a reference. This
|
|
|
4728 will result in a few objects not getting collected when they should, but
|
|
|
4729 it obviates the need for @code{GCPRO}ing, and allows garbage collection
|
|
|
4730 to happen at any point at all, such as during object allocation.
|
|
|
4731
|
|
|
4732 @node Garbage Collection - Step by Step
|
|
|
4733 @section Garbage Collection - Step by Step
|
|
|
4734 @cindex garbage collection step by step
|
|
|
4735
|
|
|
4736 @menu
|
|
|
4737 * Invocation::
|
|
|
4738 * garbage_collect_1::
|
|
|
4739 * mark_object::
|
|
|
4740 * gc_sweep::
|
|
|
4741 * sweep_lcrecords_1::
|
|
|
4742 * compact_string_chars::
|
|
|
4743 * sweep_strings::
|
|
|
4744 * sweep_bit_vectors_1::
|
|
|
4745 @end menu
|
|
|
4746
|
|
|
4747 @node Invocation
|
|
|
4748 @subsection Invocation
|
|
|
4749 @cindex garbage collection, invocation
|
|
|
4750
|
|
|
4751 The first thing that anyone should know about garbage collection is:
|
|
|
4752 when and how the garbage collector is invoked. One might think that this
|
|
|
4753 could happen every time new memory is allocated, e.g. new objects are
|
|
|
4754 created, but this is @emph{not} the case. Instead, we have the following
|
|
|
4755 situation:
|
|
|
4756
|
|
|
4757 The entry point of any process of garbage collection is an invocation
|
|
|
4758 of the function @code{garbage_collect_1} in file @code{alloc.c}. The
|
|
|
4759 invocation can occur @emph{explicitly} by calling the function
|
|
|
4760 @code{Fgarbage_collect} (in addition this function provides information
|
|
|
4761 about the freed memory), or can occur @emph{implicitly} in four different
|
|
|
4762 situations:
|
|
|
4763 @enumerate
|
|
|
4764 @item
|
|
|
4765 In function @code{main_1} in file @code{emacs.c}. This function is called
|
|
|
4766 at each startup of xemacs. The garbage collection is invoked after all
|
|
|
4767 initial creations are completed, but only if a special internal error
|
|
|
4768 checking-constant @code{ERROR_CHECK_GC} is defined.
|
|
|
4769 @item
|
|
|
4770 In function @code{disksave_object_finalization} in file
|
|
|
4771 @code{alloc.c}. The only purpose of this function is to clear the
|
|
|
4772 objects from memory which need not be stored with xemacs when we dump out
|
|
|
4773 an executable. This is only done by @code{Fdump_emacs} or by
|
|
|
4774 @code{Fdump_emacs_data} respectively (both in @code{emacs.c}). The
|
|
|
4775 actual clearing is accomplished by making these objects unreachable and
|
|
|
4776 starting a garbage collection. The function is only used while building
|
|
|
4777 xemacs.
|
|
|
4778 @item
|
|
|
4779 In function @code{Feval / eval} in file @code{eval.c}. Each time the
|
|
|
4780 well known and often used function eval is called to evaluate a form,
|
|
|
4781 one of the first things that could happen, is a potential call of
|
|
|
4782 @code{garbage_collect_1}. There exist three global variables,
|
|
|
4783 @code{consing_since_gc} (counts the created cons-cells since the last
|
|
|
4784 garbage collection), @code{gc_cons_threshold} (a specified threshold
|
|
|
4785 after which a garbage collection occurs) and @code{always_gc}. If
|
|
|
4786 @code{always_gc} is set or if the threshold is exceeded, the garbage
|
|
|
4787 collection will start.
|
|
|
4788 @item
|
|
|
4789 In function @code{Ffuncall / funcall} in file @code{eval.c}. This
|
|
|
4790 function evaluates calls of elisp functions and works according to
|
|
|
4791 @code{Feval}.
|
|
|
4792 @end enumerate
|
|
|
4793
|
|
|
4794 The upshot is that garbage collection can basically occur everywhere
|
|
|
4795 @code{Feval}, respectively @code{Ffuncall}, is used - either directly or
|
|
|
4796 through another function. Since calls to these two functions are
|
|
|
4797 hidden in various other functions, many calls to
|
|
|
4798 @code{garabge_collect_1} are not obviously foreseeable, and therefore
|
|
|
4799 unexpected. Instances where they are used that are worth remembering are
|
|
|
4800 various elisp commands, as for example @code{or},
|
|
|
4801 @code{and}, @code{if}, @code{cond}, @code{while}, @code{setq}, etc.,
|
|
|
4802 miscellaneous @code{gui_item_...} functions, everything related to
|
|
|
4803 @code{eval} (@code{Feval_buffer}, @code{call0}, ...) and inside
|
|
|
4804 @code{Fsignal}. The latter is used to handle signals, as for example the
|
|
|
4805 ones raised by every @code{QUITE}-macro triggered after pressing Ctrl-g.
|
|
|
4806
|
|
|
4807 @node garbage_collect_1
|
|
|
4808 @subsection @code{garbage_collect_1}
|
|
|
4809 @cindex @code{garbage_collect_1}
|
|
|
4810
|
|
|
4811 We can now describe exactly what happens after the invocation takes
|
|
|
4812 place.
|
|
|
4813 @enumerate
|
|
|
4814 @item
|
|
|
4815 There are several cases in which the garbage collector is left immediately:
|
|
|
4816 when we are already garbage collecting (@code{gc_in_progress}), when
|
|
|
4817 the garbage collection is somehow forbidden
|
|
|
4818 (@code{gc_currently_forbidden}), when we are currently displaying something
|
|
|
4819 (@code{in_display}) or when we are preparing for the armageddon of the
|
|
|
4820 whole system (@code{preparing_for_armageddon}).
|
|
|
4821 @item
|
|
|
4822 Next the correct frame in which to put
|
|
|
4823 all the output occurring during garbage collecting is determined. In
|
|
|
4824 order to be able to restore the old display's state after displaying the
|
|
|
4825 message, some data about the current cursor position has to be
|
|
|
4826 saved. The variables @code{pre_gc_curser} and @code{cursor_changed} take
|
|
|
4827 care of that.
|
|
|
4828 @item
|
|
|
4829 The state of @code{gc_currently_forbidden} must be restored after
|
|
|
4830 the garbage collection, no matter what happens during the process. We
|
|
|
4831 accomplish this by @code{record_unwind_protect}ing the suitable function
|
|
|
4832 @code{restore_gc_inhibit} together with the current value of
|
|
|
4833 @code{gc_currently_forbidden}.
|
|
|
4834 @item
|
|
|
4835 If we are concurrently running an interactive xemacs session, the next step
|
|
|
4836 is simply to show the garbage collector's cursor/message.
|
|
|
4837 @item
|
|
|
4838 The following steps are the intrinsic steps of the garbage collector,
|
|
|
4839 therefore @code{gc_in_progress} is set.
|
|
|
4840 @item
|
|
|
4841 For debugging purposes, it is possible to copy the current C stack
|
|
|
4842 frame. However, this seems to be a currently unused feature.
|
|
|
4843 @item
|
|
|
4844 Before actually starting to go over all live objects, references to
|
|
|
4845 objects that are no longer used are pruned. We only have to do this for events
|
|
|
4846 (@code{clear_event_resource}) and for specifiers
|
|
|
4847 (@code{cleanup_specifiers}).
|
|
|
4848 @item
|
|
|
4849 Now the mark phase begins and marks all accessible elements. In order to
|
|
|
4850 start from
|
|
|
4851 all slots that serve as roots of accessibility, the function
|
|
|
4852 @code{mark_object} is called for each root individually to go out from
|
|
|
4853 there to mark all reachable objects. All roots that are traversed are
|
|
|
4854 shown in their processed order:
|
|
|
4855 @itemize @bullet
|
|
|
4856 @item
|
|
|
4857 all constant symbols and static variables that are registered via
|
|
|
4858 @code{staticpro}@ in the array @code{staticvec}.
|
|
|
4859 @xref{Adding Global Lisp Variables}.
|
|
|
4860 @item
|
|
|
4861 all Lisp objects that are created in C functions and that must be
|
|
|
4862 protected from freeing them. They are registered in the global
|
|
|
4863 list @code{gcprolist}.
|
|
|
4864 @xref{GCPROing}.
|
|
|
4865 @item
|
|
|
4866 all local variables (i.e. their name fields @code{symbol} and old
|
|
|
4867 values @code{old_values}) that are bound during the evaluation by the Lisp
|
|
|
4868 engine. They are stored in @code{specbinding} structs pushed on a stack
|
|
|
4869 called @code{specpdl}.
|
|
|
4870 @xref{Dynamic Binding; The specbinding Stack; Unwind-Protects}.
|
|
|
4871 @item
|
|
|
4872 all catch blocks that the Lisp engine encounters during the evaluation
|
|
|
4873 cause the creation of structs @code{catchtag} inserted in the list
|
|
|
4874 @code{catchlist}. Their tag (@code{tag}) and value (@code{val} fields
|
|
|
4875 are freshly created objects and therefore have to be marked.
|
|
|
4876 @xref{Catch and Throw}.
|
|
|
4877 @item
|
|
|
4878 every function application pushes new structs @code{backtrace}
|
|
|
4879 on the call stack of the Lisp engine (@code{backtrace_list}). The unique
|
|
|
4880 parts that have to be marked are the fields for each function
|
|
|
4881 (@code{function}) and all their arguments (@code{args}).
|
|
|
4882 @xref{Evaluation}.
|
|
|
4883 @item
|
|
|
4884 all objects that are used by the redisplay engine that must not be freed
|
|
|
4885 are marked by a special function called @code{mark_redisplay} (in
|
|
|
4886 @code{redisplay.c}).
|
|
|
4887 @item
|
|
|
4888 all objects created for profiling purposes are allocated by C functions
|
|
|
4889 instead of using the lisp allocation mechanisms. In order to receive the
|
|
|
4890 right ones during the sweep phase, they also have to be marked
|
|
|
4891 manually. That is done by the function @code{mark_profiling_info}
|
|
|
4892 @end itemize
|
|
|
4893 @item
|
|
436
|
4894 Hash tables in XEmacs belong to a kind of special objects that
|
|
428
|
4895 make use of a concept often called 'weak pointers'.
|
|
|
4896 To make a long story short, these kind of pointers are not followed
|
|
|
4897 during the estimation of the live objects during garbage collection.
|
|
|
4898 Any object referenced only by weak pointers is collected
|
|
|
4899 anyway, and the reference to it is cleared. In hash tables there are
|
|
|
4900 different usage patterns of them, manifesting in different types of hash
|
|
|
4901 tables, namely 'non-weak', 'weak', 'key-weak' and 'value-weak'
|
|
|
4902 (internally also 'key-car-weak' and 'value-car-weak') hash tables, each
|
|
|
4903 clearing entries depending on different conditions. More information can
|
|
|
4904 be found in the documentation to the function @code{make-hash-table}.
|
|
|
4905
|
|
|
4906 Because there are complicated dependency rules about when and what to
|
|
|
4907 mark while processing weak hash tables, the standard @code{marker}
|
|
|
4908 method is only active if it is marking non-weak hash tables. As soon as
|
|
|
4909 a weak component is in the table, the hash table entries are ignored
|
|
|
4910 while marking. Instead their marking is done each separately by the
|
|
|
4911 function @code{finish_marking_weak_hash_tables}. This function iterates
|
|
|
4912 over each hash table entry @code{hentries} for each weak hash table in
|
|
|
4913 @code{Vall_weak_hash_tables}. Depending on the type of a table, the
|
|
|
4914 appropriate action is performed.
|
|
|
4915 If a table is acting as @code{HASH_TABLE_KEY_WEAK}, and a key already marked,
|
|
|
4916 everything reachable from the @code{value} component is marked. If it is
|
|
|
4917 acting as a @code{HASH_TABLE_VALUE_WEAK} and the value component is
|
|
|
4918 already marked, the marking starts beginning only from the
|
|
|
4919 @code{key} component.
|
|
|
4920 If it is a @code{HASH_TABLE_KEY_CAR_WEAK} and the car
|
|
|
4921 of the key entry is already marked, we mark both the @code{key} and
|
|
|
4922 @code{value} components.
|
|
|
4923 Finally, if the table is of the type @code{HASH_TABLE_VALUE_CAR_WEAK}
|
|
|
4924 and the car of the value components is already marked, again both the
|
|
|
4925 @code{key} and the @code{value} components get marked.
|
|
|
4926
|
|
|
4927 Again, there are lists with comparable properties called weak
|
|
|
4928 lists. There exist different peculiarities of their types called
|
|
|
4929 @code{simple}, @code{assoc}, @code{key-assoc} and
|
|
|
4930 @code{value-assoc}. You can find further details about them in the
|
|
|
4931 description to the function @code{make-weak-list}. The scheme of their
|
|
|
4932 marking is similar: all weak lists are listed in @code{Qall_weak_lists},
|
|
|
4933 therefore we iterate over them. The marking is advanced until we hit an
|
|
|
4934 already marked pair. Then we know that during a former run all
|
|
|
4935 the rest has been marked completely. Again, depending on the special
|
|
|
4936 type of the weak list, our jobs differ. If it is a @code{WEAK_LIST_SIMPLE}
|
|
|
4937 and the elem is marked, we mark the @code{cons} part. If it is a
|
|
|
4938 @code{WEAK_LIST_ASSOC} and not a pair or a pair with both marked car and
|
|
|
4939 cdr, we mark the @code{cons} and the @code{elem}. If it is a
|
|
|
4940 @code{WEAK_LIST_KEY_ASSOC} and not a pair or a pair with a marked car of
|
|
|
4941 the elem, we mark the @code{cons} and the @code{elem}. Finally, if it is
|
|
|
4942 a @code{WEAK_LIST_VALUE_ASSOC} and not a pair or a pair with a marked
|
|
|
4943 cdr of the elem, we mark both the @code{cons} and the @code{elem}.
|
|
|
4944
|
|
|
4945 Since, by marking objects in reach from weak hash tables and weak lists,
|
|
|
4946 other objects could get marked, this perhaps implies further marking of
|
|
|
4947 other weak objects, both finishing functions are redone as long as
|
|
|
4948 yet unmarked objects get freshly marked.
|
|
|
4949
|
|
|
4950 @item
|
|
|
4951 After completing the special marking for the weak hash tables and for the weak
|
|
|
4952 lists, all entries that point to objects that are going to be swept in
|
|
|
4953 the further process are useless, and therefore have to be removed from
|
|
|
4954 the table or the list.
|
|
|
4955
|
|
|
4956 The function @code{prune_weak_hash_tables} does the job for weak hash
|
|
|
4957 tables. Totally unmarked hash tables are removed from the list
|
|
|
4958 @code{Vall_weak_hash_tables}. The other ones are treated more carefully
|
|
|
4959 by scanning over all entries and removing one as soon as one of
|
|
|
4960 the components @code{key} and @code{value} is unmarked.
|
|
|
4961
|
|
|
4962 The same idea applies to the weak lists. It is accomplished by
|
|
|
4963 @code{prune_weak_lists}: An unmarked list is pruned from
|
|
|
4964 @code{Vall_weak_lists} immediately. A marked list is treated more
|
|
|
4965 carefully by going over it and removing just the unmarked pairs.
|
|
|
4966
|
|
|
4967 @item
|
|
|
4968 The function @code{prune_specifiers} checks all listed specifiers held
|
|
|
4969 in @code{Vall_speficiers} and removes the ones from the lists that are
|
|
|
4970 unmarked.
|
|
|
4971
|
|
|
4972 @item
|
|
|
4973 All syntax tables are stored in a list called
|
|
|
4974 @code{Vall_syntax_tables}. The function @code{prune_syntax_tables} walks
|
|
|
4975 through it and unlinks the tables that are unmarked.
|
|
|
4976
|
|
|
4977 @item
|
|
|
4978 Next, we will attack the complete sweeping - the function
|
|
|
4979 @code{gc_sweep} which holds the predominance.
|
|
|
4980 @item
|
|
|
4981 First, all the variables with respect to garbage collection are
|
|
|
4982 reset. @code{consing_since_gc} - the counter of the created cells since
|
|
|
4983 the last garbage collection - is set back to 0, and
|
|
|
4984 @code{gc_in_progress} is not @code{true} anymore.
|
|
|
4985 @item
|
|
|
4986 In case the session is interactive, the displayed cursor and message are
|
|
|
4987 removed again.
|
|
|
4988 @item
|
|
|
4989 The state of @code{gc_inhibit} is restored to the former value by
|
|
|
4990 unwinding the stack.
|
|
|
4991 @item
|
|
|
4992 A small memory reserve is always held back that can be reached by
|
|
|
4993 @code{breathing_space}. If nothing more is left, we create a new reserve
|
|
|
4994 and exit.
|
|
|
4995 @end enumerate
|
|
|
4996
|
|
|
4997 @node mark_object
|
|
|
4998 @subsection @code{mark_object}
|
|
|
4999 @cindex @code{mark_object}
|
|
|
5000
|
|
|
5001 The first thing that is checked while marking an object is whether the
|
|
|
5002 object is a real Lisp object @code{Lisp_Type_Record} or just an integer
|
|
|
5003 or a character. Integers and characters are the only two types that are
|
|
|
5004 stored directly - without another level of indirection, and therefore they
|
|
438
|
5005 don't have to be marked and collected.
|
|
428
|
5006 @xref{How Lisp Objects Are Represented in C}.
|
|
|
5007
|
|
|
5008 The second case is the one we have to handle. It is the one when we are
|
|
|
5009 dealing with a pointer to a Lisp object. But, there exist also three
|
|
|
5010 possibilities, that prevent us from doing anything while marking: The
|
|
|
5011 object is read only which prevents it from being garbage collected,
|
|
|
5012 i.e. marked (@code{C_READONLY_RECORD_HEADER}). The object in question is
|
|
|
5013 already marked, and need not be marked for the second time (checked by
|
|
|
5014 @code{MARKED_RECORD_HEADER_P}). If it is a special, unmarkable object
|
|
|
5015 (@code{UNMARKABLE_RECORD_HEADER_P}, apparently, these are objects that
|
|
|
5016 sit in some CONST space, and can therefore not be marked, see
|
|
|
5017 @code{this_one_is_unmarkable} in @code{alloc.c}).
|
|
|
5018
|
|
|
5019 Now, the actual marking is feasible. We do so by once using the macro
|
|
|
5020 @code{MARK_RECORD_HEADER} to mark the object itself (actually the
|
|
|
5021 special flag in the lrecord header), and calling its special marker
|
|
|
5022 "method" @code{marker} if available. The marker method marks every
|
|
|
5023 other object that is in reach from our current object. Note, that these
|
|
|
5024 marker methods should not call @code{mark_object} recursively, but
|
|
|
5025 instead should return the next object from where further marking has to
|
|
|
5026 be performed.
|
|
|
5027
|
|
|
5028 In case another object was returned, as mentioned before, we reiterate
|
|
|
5029 the whole @code{mark_object} process beginning with this next object.
|
|
|
5030
|
|
|
5031 @node gc_sweep
|
|
|
5032 @subsection @code{gc_sweep}
|
|
|
5033 @cindex @code{gc_sweep}
|
|
|
5034
|
|
|
5035 The job of this function is to free all unmarked records from memory. As
|
|
|
5036 we know, there are different types of objects implemented and managed, and
|
|
|
5037 consequently different ways to free them from memory.
|
|
|
5038 @xref{Introduction to Allocation}.
|
|
|
5039
|
|
|
5040 We start with all objects stored through @code{lcrecords}. All
|
|
|
5041 bulkier objects are allocated and handled using that scheme of
|
|
|
5042 @code{lcrecords}. Each object is @code{malloc}ed separately
|
|
|
5043 instead of placing it in one of the contiguous frob blocks. All types
|
|
|
5044 that are currently stored
|
|
438
|
5045 using @code{lcrecords}'s @code{alloc_lcrecord} and
|
|
428
|
5046 @code{make_lcrecord_list} are the types: vectors, buffers,
|
|
|
5047 char-table, char-table-entry, console, weak-list, database, device,
|
|
|
5048 ldap, hash-table, command-builder, extent-auxiliary, extent-info, face,
|
|
|
5049 coding-system, frame, image-instance, glyph, popup-data, gui-item,
|
|
|
5050 keymap, charset, color_instance, font_instance, opaque, opaque-list,
|
|
|
5051 process, range-table, specifier, symbol-value-buffer-local,
|
|
|
5052 symbol-value-lisp-magic, symbol-value-varalias, toolbar-button,
|
|
|
5053 tooltalk-message, tooltalk-pattern, window, and window-configuration. We
|
|
|
5054 take care of them in the fist place
|
|
|
5055 in order to be able to handle and to finalize items stored in them more
|
|
|
5056 easily. The function @code{sweep_lcrecords_1} as described below is
|
|
|
5057 doing the whole job for us.
|
|
|
5058 For a description about the internals: @xref{lrecords}.
|
|
|
5059
|
|
|
5060 Our next candidates are the other objects that behave quite differently
|
|
|
5061 than everything else: the strings. They consists of two parts, a
|
|
|
5062 fixed-size portion (@code{struct Lisp_string}) holding the string's
|
|
|
5063 length, its property list and a pointer to the second part, and the
|
|
|
5064 actual string data, which is stored in string-chars blocks comparable to
|
|
|
5065 frob blocks. In this block, the data is not only freed, but also a
|
|
|
5066 compression of holes is made, i.e. all strings are relocated together.
|
|
|
5067 @xref{String}. This compacting phase is performed by the function
|
|
|
5068 @code{compact_string_chars}, the actual sweeping by the function
|
|
|
5069 @code{sweep_strings} is described below.
|
|
|
5070
|
|
|
5071 After that, the other types are swept step by step using functions
|
|
|
5072 @code{sweep_conses}, @code{sweep_bit_vectors_1},
|
|
|
5073 @code{sweep_compiled_functions}, @code{sweep_floats},
|
|
|
5074 @code{sweep_symbols}, @code{sweep_extents}, @code{sweep_markers} and
|
|
|
5075 @code{sweep_extents}. They are the fixed-size types cons, floats,
|
|
|
5076 compiled-functions, symbol, marker, extent, and event stored in
|
|
|
5077 so-called "frob blocks", and therefore we can basically do the same on
|
|
|
5078 every type objects, using the same macros, especially defined only to
|
|
|
5079 handle everything with respect to fixed-size blocks. The only fixed-size
|
|
|
5080 type that is not handled here are the fixed-size portion of strings,
|
|
|
5081 because we took special care of them earlier.
|
|
|
5082
|
|
|
5083 The only big exceptions are bit vectors stored differently and
|
|
|
5084 therefore treated differently by the function @code{sweep_bit_vectors_1}
|
|
|
5085 described later.
|
|
|
5086
|
|
|
5087 At first, we need some brief information about how
|
|
|
5088 these fixed-size types are managed in general, in order to understand
|
|
|
5089 how the sweeping is done. They have all a fixed size, and are therefore
|
|
|
5090 stored in big blocks of memory - allocated at once - that can hold a
|
|
|
5091 certain amount of objects of one type. The macro
|
|
|
5092 @code{DECLARE_FIXED_TYPE_ALLOC} creates the suitable structures for
|
|
|
5093 every type. More precisely, we have the block struct
|
|
|
5094 (holding a pointer to the previous block @code{prev} and the
|
|
|
5095 objects in @code{block[]}), a pointer to current block
|
|
|
5096 (@code{current_..._block)}) and its last index
|
|
|
5097 (@code{current_..._block_index}), and a pointer to the free list that
|
|
|
5098 will be created. Also a macro @code{FIXED_TYPE_FROM_BLOCK} plus some
|
|
|
5099 related macros exists that are used to obtain a new object, either from
|
|
|
5100 the free list @code{ALLOCATE_FIXED_TYPE_1} if there is an unused object
|
|
|
5101 of that type stored or by allocating a completely new block using
|
|
|
5102 @code{ALLOCATE_FIXED_TYPE_FROM_BLOCK}.
|
|
|
5103
|
|
|
5104 The rest works as follows: all of them define a
|
|
|
5105 macro @code{UNMARK_...} that is used to unmark the object. They define a
|
|
|
5106 macro @code{ADDITIONAL_FREE_...} that defines additional work that has
|
|
|
5107 to be done when converting an object from in use to not in use (so far,
|
|
|
5108 only markers use it in order to unchain them). Then, they all call
|
|
|
5109 the macro @code{SWEEP_FIXED_TYPE_BLOCK} instantiated with their type name
|
|
|
5110 and their struct name.
|
|
|
5111
|
|
|
5112 This call in particular does the following: we go over all blocks
|
|
|
5113 starting with the current moving towards the oldest.
|
|
|
5114 For each block, we look at every object in it. If the object already
|
|
|
5115 freed (checked with @code{FREE_STRUCT_P} using the first pointer of the
|
|
|
5116 object), or if it is
|
|
|
5117 set to read only (@code{C_READONLY_RECORD_HEADER_P}, nothing must be
|
|
|
5118 done. If it is unmarked (checked with @code{MARKED_RECORD_HEADER_P}), it
|
|
|
5119 is put in the free list and set free (using the macro
|
|
|
5120 @code{FREE_FIXED_TYPE}, otherwise it stays in the block, but is unmarked
|
|
|
5121 (by @code{UNMARK_...}). While going through one block, we note if the
|
|
|
5122 whole block is empty. If so, the whole block is freed (using
|
|
|
5123 @code{xfree}) and the free list state is set to the state it had before
|
|
|
5124 handling this block.
|
|
|
5125
|
|
|
5126 @node sweep_lcrecords_1
|
|
|
5127 @subsection @code{sweep_lcrecords_1}
|
|
|
5128 @cindex @code{sweep_lcrecords_1}
|
|
|
5129
|
|
|
5130 After nullifying the complete lcrecord statistics, we go over all
|
|
|
5131 lcrecords two separate times. They are all chained together in a list with
|
|
|
5132 a head called @code{all_lcrecords}.
|
|
|
5133
|
|
|
5134 The first loop calls for each object its @code{finalizer} method, but only
|
|
|
5135 in the case that it is not read only
|
|
|
5136 (@code{C_READONLY_RECORD_HEADER_P)}, it is not already marked
|
|
|
5137 (@code{MARKED_RECORD_HEADER_P}), it is not already in a free list (list of
|
|
|
5138 freed objects, field @code{free}) and finally it owns a finalizer
|
|
|
5139 method.
|
|
|
5140
|
|
|
5141 The second loop actually frees the appropriate objects again by iterating
|
|
|
5142 through the whole list. In case an object is read only or marked, it
|
|
|
5143 has to persist, otherwise it is manually freed by calling
|
|
|
5144 @code{xfree}. During this loop, the lcrecord statistics are kept up to
|
|
|
5145 date by calling @code{tick_lcrecord_stats} with the right arguments,
|
|
|
5146
|
|
|
5147 @node compact_string_chars
|
|
|
5148 @subsection @code{compact_string_chars}
|
|
|
5149 @cindex @code{compact_string_chars}
|
|
|
5150
|
|
|
5151 The purpose of this function is to compact all the data parts of the
|
|
|
5152 strings that are held in so-called @code{string_chars_block}, i.e. the
|
|
|
5153 strings that do not exceed a certain maximal length.
|
|
|
5154
|
|
|
5155 The procedure with which this is done is as follows. We are keeping two
|
|
|
5156 positions in the @code{string_chars_block}s using two pointer/integer
|
|
|
5157 pairs, namely @code{from_sb}/@code{from_pos} and
|
|
|
5158 @code{to_sb}/@code{to_pos}. They stand for the actual positions, from
|
|
|
5159 where to where, to copy the actually handled string.
|
|
|
5160
|
|
|
5161 While going over all chained @code{string_char_block}s and their held
|
|
|
5162 strings, staring at @code{first_string_chars_block}, both pointers
|
|
|
5163 are advanced and eventually a string is copied from @code{from_sb} to
|
|
|
5164 @code{to_sb}, depending on the status of the pointed at strings.
|
|
|
5165
|
|
|
5166 More precisely, we can distinguish between the following actions.
|
|
|
5167 @itemize @bullet
|
|
|
5168 @item
|
|
|
5169 The string at @code{from_sb}'s position could be marked as free, which
|
|
|
5170 is indicated by an invalid pointer to the pointer that should point back
|
|
|
5171 to the fixed size string object, and which is checked by
|
|
|
5172 @code{FREE_STRUCT_P}. In this case, the @code{from_sb}/@code{from_pos}
|
|
|
5173 is advanced to the next string, and nothing has to be copied.
|
|
|
5174 @item
|
|
|
5175 Also, if a string object itself is unmarked, nothing has to be
|
|
|
5176 copied. We likewise advance the @code{from_sb}/@code{from_pos}
|
|
|
5177 pair as described above.
|
|
|
5178 @item
|
|
|
5179 In all other cases, we have a marked string at hand. The string data
|
|
|
5180 must be moved from the from-position to the to-position. In case
|
|
|
5181 there is not enough space in the actual @code{to_sb}-block, we advance
|
|
|
5182 this pointer to the beginning of the next block before copying. In case the
|
|
|
5183 from and to positions are different, we perform the
|
|
|
5184 actual copying using the library function @code{memmove}.
|
|
|
5185 @end itemize
|
|
|
5186
|
|
|
5187 After compacting, the pointer to the current
|
|
|
5188 @code{string_chars_block}, sitting in @code{current_string_chars_block},
|
|
|
5189 is reset on the last block to which we moved a string,
|
|
|
5190 i.e. @code{to_block}, and all remaining blocks (we know that they just
|
|
|
5191 carry garbage) are explicitly @code{xfree}d.
|
|
|
5192
|
|
|
5193 @node sweep_strings
|
|
|
5194 @subsection @code{sweep_strings}
|
|
|
5195 @cindex @code{sweep_strings}
|
|
|
5196
|
|
|
5197 The sweeping for the fixed sized string objects is essentially exactly
|
|
|
5198 the same as it is for all other fixed size types. As before, the freeing
|
|
|
5199 into the suitable free list is done by using the macro
|
|
|
5200 @code{SWEEP_FIXED_SIZE_BLOCK} after defining the right macros
|
|
|
5201 @code{UNMARK_string} and @code{ADDITIONAL_FREE_string}. These two
|
|
|
5202 definitions are a little bit special compared to the ones used
|
|
|
5203 for the other fixed size types.
|
|
|
5204
|
|
|
5205 @code{UNMARK_string} is defined the same way except some additional code
|
|
|
5206 used for updating the bookkeeping information.
|
|
|
5207
|
|
|
5208 For strings, @code{ADDITIONAL_FREE_string} has to do something in
|
|
|
5209 addition: in case, the string was not allocated in a
|
|
|
5210 @code{string_chars_block} because it exceeded the maximal length, and
|
|
|
5211 therefore it was @code{malloc}ed separately, we know also @code{xfree}
|
|
|
5212 it explicitly.
|
|
|
5213
|
|
|
5214 @node sweep_bit_vectors_1
|
|
|
5215 @subsection @code{sweep_bit_vectors_1}
|
|
|
5216 @cindex @code{sweep_bit_vectors_1}
|
|
|
5217
|
|
|
5218 Bit vectors are also one of the rare types that are @code{malloc}ed
|
|
|
5219 individually. Consequently, while sweeping, all further needless
|
|
|
5220 bit vectors must be freed by hand. This is done, as one might imagine,
|
|
|
5221 the expected way: since they are all registered in a list called
|
|
|
5222 @code{all_bit_vectors}, all elements of that list are traversed,
|
|
|
5223 all unmarked bit vectors are unlinked by calling @code{xfree} and all of
|
|
|
5224 them become unmarked.
|
|
|
5225 In addition, the bookkeeping information used for garbage
|
|
|
5226 collector's output purposes is updated.
|
|
|
5227
|
|
|
5228 @node Integers and Characters
|
|
|
5229 @section Integers and Characters
|
|
|
5230
|
|
|
5231 Integer and character Lisp objects are created from integers using the
|
|
|
5232 macros @code{XSETINT()} and @code{XSETCHAR()} or the equivalent
|
|
|
5233 functions @code{make_int()} and @code{make_char()}. (These are actually
|
|
|
5234 macros on most systems.) These functions basically just do some moving
|
|
|
5235 of bits around, since the integral value of the object is stored
|
|
|
5236 directly in the @code{Lisp_Object}.
|
|
|
5237
|
|
|
5238 @code{XSETINT()} and the like will truncate values given to them that
|
|
|
5239 are too big; i.e. you won't get the value you expected but the tag bits
|
|
|
5240 will at least be correct.
|
|
|
5241
|
|
|
5242 @node Allocation from Frob Blocks
|
|
|
5243 @section Allocation from Frob Blocks
|
|
|
5244
|
|
|
5245 The uninitialized memory required by a @code{Lisp_Object} of a particular type
|
|
|
5246 is allocated using
|
|
|
5247 @code{ALLOCATE_FIXED_TYPE()}. This only occurs inside of the
|
|
|
5248 lowest-level object-creating functions in @file{alloc.c}:
|
|
|
5249 @code{Fcons()}, @code{make_float()}, @code{Fmake_byte_code()},
|
|
|
5250 @code{Fmake_symbol()}, @code{allocate_extent()},
|
|
|
5251 @code{allocate_event()}, @code{Fmake_marker()}, and
|
|
|
5252 @code{make_uninit_string()}. The idea is that, for each type, there are
|
|
|
5253 a number of frob blocks (each 2K in size); each frob block is divided up
|
|
|
5254 into object-sized chunks. Each frob block will have some of these
|
|
|
5255 chunks that are currently assigned to objects, and perhaps some that are
|
|
|
5256 free. (If a frob block has nothing but free chunks, it is freed at the
|
|
|
5257 end of the garbage collection cycle.) The free chunks are stored in a
|
|
|
5258 free list, which is chained by storing a pointer in the first four bytes
|
|
|
5259 of the chunk. (Except for the free chunks at the end of the last frob
|
|
|
5260 block, which are handled using an index which points past the end of the
|
|
|
5261 last-allocated chunk in the last frob block.)
|
|
|
5262 @code{ALLOCATE_FIXED_TYPE()} first tries to retrieve a chunk from the
|
|
|
5263 free list; if that fails, it calls
|
|
|
5264 @code{ALLOCATE_FIXED_TYPE_FROM_BLOCK()}, which looks at the end of the
|
|
|
5265 last frob block for space, and creates a new frob block if there is
|
|
|
5266 none. (There are actually two versions of these macros, one of which is
|
|
|
5267 more defensive but less efficient and is used for error-checking.)
|
|
|
5268
|
|
|
5269 @node lrecords
|
|
|
5270 @section lrecords
|
|
|
5271
|
|
|
5272 [see @file{lrecord.h}]
|
|
|
5273
|
|
|
5274 All lrecords have at the beginning of their structure a @code{struct
|
|
|
5275 lrecord_header}. This just contains a pointer to a @code{struct
|
|
|
5276 lrecord_implementation}, which is a structure containing method pointers
|
|
|
5277 and such. There is one of these for each type, and it is a global,
|
|
|
5278 constant, statically-declared structure that is declared in the
|
|
|
5279 @code{DEFINE_LRECORD_IMPLEMENTATION()} macro. (This macro actually
|
|
|
5280 declares an array of two @code{struct lrecord_implementation}
|
|
|
5281 structures. The first one contains all the standard method pointers,
|
|
|
5282 and is used in all normal circumstances. During garbage collection,
|
|
|
5283 however, the lrecord is @dfn{marked} by bumping its implementation
|
|
|
5284 pointer by one, so that it points to the second structure in the array.
|
|
|
5285 This structure contains a special indication in it that it's a
|
|
|
5286 @dfn{marked-object} structure: the finalize method is the special
|
|
|
5287 function @code{this_marks_a_marked_record()}, and all other methods are
|
|
|
5288 null pointers. At the end of garbage collection, all lrecords will
|
|
|
5289 either be reclaimed or unmarked by decrementing their implementation
|
|
|
5290 pointers, so this second structure pointer will never remain past
|
|
|
5291 garbage collection.
|
|
|
5292
|
|
|
5293 Simple lrecords (of type (c) above) just have a @code{struct
|
|
|
5294 lrecord_header} at their beginning. lcrecords, however, actually have a
|
|
|
5295 @code{struct lcrecord_header}. This, in turn, has a @code{struct
|
|
|
5296 lrecord_header} at its beginning, so sanity is preserved; but it also
|
|
|
5297 has a pointer used to chain all lcrecords together, and a special ID
|
|
|
5298 field used to distinguish one lcrecord from another. (This field is used
|
|
|
5299 only for debugging and could be removed, but the space gain is not
|
|
|
5300 significant.)
|
|
|
5301
|
|
|
5302 Simple lrecords are created using @code{ALLOCATE_FIXED_TYPE()}, just
|
|
|
5303 like for other frob blocks. The only change is that the implementation
|
|
|
5304 pointer must be initialized correctly. (The implementation structure for
|
|
|
5305 an lrecord, or rather the pointer to it, is named @code{lrecord_float},
|
|
|
5306 @code{lrecord_extent}, @code{lrecord_buffer}, etc.)
|
|
|
5307
|
|
|
5308 lcrecords are created using @code{alloc_lcrecord()}. This takes a
|
|
|
5309 size to allocate and an implementation pointer. (The size needs to be
|
|
|
5310 passed because some lcrecords, such as window configurations, are of
|
|
|
5311 variable size.) This basically just @code{malloc()}s the storage,
|
|
|
5312 initializes the @code{struct lcrecord_header}, and chains the lcrecord
|
|
|
5313 onto the head of the list of all lcrecords, which is stored in the
|
|
|
5314 variable @code{all_lcrecords}. The calls to @code{alloc_lcrecord()}
|
|
|
5315 generally occur in the lowest-level allocation function for each lrecord
|
|
|
5316 type.
|
|
|
5317
|
|
|
5318 Whenever you create an lrecord, you need to call either
|
|
|
5319 @code{DEFINE_LRECORD_IMPLEMENTATION()} or
|
|
|
5320 @code{DEFINE_LRECORD_SEQUENCE_IMPLEMENTATION()}. This needs to be
|
|
|
5321 specified in a C file, at the top level. What this actually does is
|
|
|
5322 define and initialize the implementation structure for the lrecord. (And
|
|
|
5323 possibly declares a function @code{error_check_foo()} that implements
|
|
|
5324 the @code{XFOO()} macro when error-checking is enabled.) The arguments
|
|
|
5325 to the macros are the actual type name (this is used to construct the C
|
|
|
5326 variable name of the lrecord implementation structure and related
|
|
|
5327 structures using the @samp{##} macro concatenation operator), a string
|
|
|
5328 that names the type on the Lisp level (this may not be the same as the C
|
|
|
5329 type name; typically, the C type name has underscores, while the Lisp
|
|
|
5330 string has dashes), various method pointers, and the name of the C
|
|
|
5331 structure that contains the object. The methods are used to encapsulate
|
|
|
5332 type-specific information about the object, such as how to print it or
|
|
|
5333 mark it for garbage collection, so that it's easy to add new object
|
|
|
5334 types without having to add a specific case for each new type in a bunch
|
|
|
5335 of different places.
|
|
|
5336
|
|
|
5337 The difference between @code{DEFINE_LRECORD_IMPLEMENTATION()} and
|
|
|
5338 @code{DEFINE_LRECORD_SEQUENCE_IMPLEMENTATION()} is that the former is
|
|
|
5339 used for fixed-size object types and the latter is for variable-size
|
|
|
5340 object types. Most object types are fixed-size; some complex
|
|
|
5341 types, however (e.g. window configurations), are variable-size.
|
|
|
5342 Variable-size object types have an extra method, which is called
|
|
|
5343 to determine the actual size of a particular object of that type.
|
|
|
5344 (Currently this is only used for keeping allocation statistics.)
|
|
|
5345
|
|
|
5346 For the purpose of keeping allocation statistics, the allocation
|
|
|
5347 engine keeps a list of all the different types that exist. Note that,
|
|
|
5348 since @code{DEFINE_LRECORD_IMPLEMENTATION()} is a macro that is
|
|
|
5349 specified at top-level, there is no way for it to add to the list of all
|
|
|
5350 existing types. What happens instead is that each implementation
|
|
|
5351 structure contains in it a dynamically assigned number that is
|
|
|
5352 particular to that type. (Or rather, it contains a pointer to another
|
|
|
5353 structure that contains this number. This evasiveness is done so that
|
|
|
5354 the implementation structure can be declared const.) In the sweep stage
|
|
|
5355 of garbage collection, each lrecord is examined to see if its
|
|
|
5356 implementation structure has its dynamically-assigned number set. If
|
|
|
5357 not, it must be a new type, and it is added to the list of known types
|
|
|
5358 and a new number assigned. The number is used to index into an array
|
|
|
5359 holding the number of objects of each type and the total memory
|
|
|
5360 allocated for objects of that type. The statistics in this array are
|
|
|
5361 also computed during the sweep stage. These statistics are returned by
|
|
|
5362 the call to @code{garbage-collect} and are printed out at the end of the
|
|
|
5363 loadup phase.
|
|
|
5364
|
|
|
5365 Note that for every type defined with a @code{DEFINE_LRECORD_*()}
|
|
|
5366 macro, there needs to be a @code{DECLARE_LRECORD_IMPLEMENTATION()}
|
|
|
5367 somewhere in a @file{.h} file, and this @file{.h} file needs to be
|
|
|
5368 included by @file{inline.c}.
|
|
|
5369
|
|
|
5370 Furthermore, there should generally be a set of @code{XFOOBAR()},
|
|
|
5371 @code{FOOBARP()}, etc. macros in a @file{.h} (or occasionally @file{.c})
|
|
|
5372 file. To create one of these, copy an existing model and modify as
|
|
|
5373 necessary.
|
|
|
5374
|
|
|
5375 The various methods in the lrecord implementation structure are:
|
|
|
5376
|
|
|
5377 @enumerate
|
|
|
5378 @item
|
|
|
5379 @cindex mark method
|
|
|
5380 A @dfn{mark} method. This is called during the marking stage and passed
|
|
|
5381 a function pointer (usually the @code{mark_object()} function), which is
|
|
|
5382 used to mark an object. All Lisp objects that are contained within the
|
|
|
5383 object need to be marked by applying this function to them. The mark
|
|
|
5384 method should also return a Lisp object, which should be either nil or
|
|
|
5385 an object to mark. (This can be used in lieu of calling
|
|
|
5386 @code{mark_object()} on the object, to reduce the recursion depth, and
|
|
|
5387 consequently should be the most heavily nested sub-object, such as a
|
|
|
5388 long list.)
|
|
|
5389
|
|
|
5390 @strong{Please note:} When the mark method is called, garbage collection
|
|
|
5391 is in progress, and special precautions need to be taken when accessing
|
|
|
5392 objects; see section (B) above.
|
|
|
5393
|
|
|
5394 If your mark method does not need to do anything, it can be
|
|
|
5395 @code{NULL}.
|
|
|
5396
|
|
|
5397 @item
|
|
|
5398 A @dfn{print} method. This is called to create a printed representation
|
|
|
5399 of the object, whenever @code{princ}, @code{prin1}, or the like is
|
|
|
5400 called. It is passed the object, a stream to which the output is to be
|
|
|
5401 directed, and an @code{escapeflag} which indicates whether the object's
|
|
|
5402 printed representation should be @dfn{escaped} so that it is
|
|
|
5403 readable. (This corresponds to the difference between @code{princ} and
|
|
|
5404 @code{prin1}.) Basically, @dfn{escaped} means that strings will have
|
|
|
5405 quotes around them and confusing characters in the strings such as
|
|
|
5406 quotes, backslashes, and newlines will be backslashed; and that special
|
|
|
5407 care will be taken to make symbols print in a readable fashion
|
|
|
5408 (e.g. symbols that look like numbers will be backslashed). Other
|
|
|
5409 readable objects should perhaps pass @code{escapeflag} on when
|
|
|
5410 sub-objects are printed, so that readability is preserved when necessary
|
|
|
5411 (or if not, always pass in a 1 for @code{escapeflag}). Non-readable
|
|
|
5412 objects should in general ignore @code{escapeflag}, except that some use
|
|
|
5413 it as an indication that more verbose output should be given.
|
|
|
5414
|
|
|
5415 Sub-objects are printed using @code{print_internal()}, which takes
|
|
|
5416 exactly the same arguments as are passed to the print method.
|
|
|
5417
|
|
|
5418 Literal C strings should be printed using @code{write_c_string()},
|
|
|
5419 or @code{write_string_1()} for non-null-terminated strings.
|
|
|
5420
|
|
|
5421 Functions that do not have a readable representation should check the
|
|
|
5422 @code{print_readably} flag and signal an error if it is set.
|
|
|
5423
|
|
|
5424 If you specify NULL for the print method, the
|
|
|
5425 @code{default_object_printer()} will be used.
|
|
|
5426
|
|
|
5427 @item
|
|
|
5428 A @dfn{finalize} method. This is called at the beginning of the sweep
|
|
|
5429 stage on lcrecords that are about to be freed, and should be used to
|
|
|
5430 perform any extra object cleanup. This typically involves freeing any
|
|
|
5431 extra @code{malloc()}ed memory associated with the object, releasing any
|
|
|
5432 operating-system and window-system resources associated with the object
|
|
|
5433 (e.g. pixmaps, fonts), etc.
|
|
|
5434
|
|
|
5435 The finalize method can be NULL if nothing needs to be done.
|
|
|
5436
|
|
|
5437 WARNING #1: The finalize method is also called at the end of the dump
|
|
|
5438 phase; this time with the for_disksave parameter set to non-zero. The
|
|
|
5439 object is @emph{not} about to disappear, so you have to make sure to
|
|
|
5440 @emph{not} free any extra @code{malloc()}ed memory if you're going to
|
|
|
5441 need it later. (Also, signal an error if there are any operating-system
|
|
|
5442 and window-system resources here, because they can't be dumped.)
|
|
|
5443
|
|
|
5444 Finalize methods should, as a rule, set to zero any pointers after
|
|
|
5445 they've been freed, and check to make sure pointers are not zero before
|
|
|
5446 freeing. Although I'm pretty sure that finalize methods are not called
|
|
|
5447 twice on the same object (except for the @code{for_disksave} proviso),
|
|
|
5448 we've gotten nastily burned in some cases by not doing this.
|
|
|
5449
|
|
|
5450 WARNING #2: The finalize method is @emph{only} called for
|
|
|
5451 lcrecords, @emph{not} for simply lrecords. If you need a
|
|
|
5452 finalize method for simple lrecords, you have to stick
|
|
|
5453 it in the @code{ADDITIONAL_FREE_foo()} macro in @file{alloc.c}.
|
|
|
5454
|
|
|
5455 WARNING #3: Things are in an @emph{extremely} bizarre state
|
|
|
5456 when @code{ADDITIONAL_FREE_foo()} is called, so you have to
|
|
|
5457 be incredibly careful when writing one of these functions.
|
|
|
5458 See the comment in @code{gc_sweep()}. If you ever have to add
|
|
|
5459 one of these, consider using an lcrecord or dealing with
|
|
|
5460 the problem in a different fashion.
|
|
|
5461
|
|
|
5462 @item
|
|
|
5463 An @dfn{equal} method. This compares the two objects for similarity,
|
|
|
5464 when @code{equal} is called. It should compare the contents of the
|
|
|
5465 objects in some reasonable fashion. It is passed the two objects and a
|
|
|
5466 @dfn{depth} value, which is used to catch circular objects. To compare
|
|
|
5467 sub-Lisp-objects, call @code{internal_equal()} and bump the depth value
|
|
|
5468 by one. If this value gets too high, a @code{circular-object} error
|
|
|
5469 will be signaled.
|
|
|
5470
|
|
|
5471 If this is NULL, objects are @code{equal} only when they are @code{eq},
|
|
|
5472 i.e. identical.
|
|
|
5473
|
|
|
5474 @item
|
|
|
5475 A @dfn{hash} method. This is used to hash objects when they are to be
|
|
|
5476 compared with @code{equal}. The rule here is that if two objects are
|
|
|
5477 @code{equal}, they @emph{must} hash to the same value; i.e. your hash
|
|
|
5478 function should use some subset of the sub-fields of the object that are
|
|
|
5479 compared in the ``equal'' method. If you specify this method as
|
|
|
5480 @code{NULL}, the object's pointer will be used as the hash, which will
|
|
|
5481 @emph{fail} if the object has an @code{equal} method, so don't do this.
|
|
|
5482
|
|
|
5483 To hash a sub-Lisp-object, call @code{internal_hash()}. Bump the
|
|
|
5484 depth by one, just like in the ``equal'' method.
|
|
|
5485
|
|
|
5486 To convert a Lisp object directly into a hash value (using
|
|
|
5487 its pointer), use @code{LISP_HASH()}. This is what happens when
|
|
|
5488 the hash method is NULL.
|
|
|
5489
|
|
|
5490 To hash two or more values together into a single value, use
|
|
|
5491 @code{HASH2()}, @code{HASH3()}, @code{HASH4()}, etc.
|
|
|
5492
|
|
|
5493 @item
|
|
|
5494 @dfn{getprop}, @dfn{putprop}, @dfn{remprop}, and @dfn{plist} methods.
|
|
|
5495 These are used for object types that have properties. I don't feel like
|
|
|
5496 documenting them here. If you create one of these objects, you have to
|
|
|
5497 use different macros to define them,
|
|
|
5498 i.e. @code{DEFINE_LRECORD_IMPLEMENTATION_WITH_PROPS()} or
|
|
|
5499 @code{DEFINE_LRECORD_SEQUENCE_IMPLEMENTATION_WITH_PROPS()}.
|
|
|
5500
|
|
|
5501 @item
|
|
|
5502 A @dfn{size_in_bytes} method, when the object is of variable-size.
|
|
|
5503 (i.e. declared with a @code{_SEQUENCE_IMPLEMENTATION} macro.) This should
|
|
|
5504 simply return the object's size in bytes, exactly as you might expect.
|
|
|
5505 For an example, see the methods for window configurations and opaques.
|
|
|
5506 @end enumerate
|
|
|
5507
|
|
|
5508 @node Low-level allocation
|
|
|
5509 @section Low-level allocation
|
|
|
5510
|
|
|
5511 Memory that you want to allocate directly should be allocated using
|
|
|
5512 @code{xmalloc()} rather than @code{malloc()}. This implements
|
|
|
5513 error-checking on the return value, and once upon a time did some more
|
|
|
5514 vital stuff (i.e. @code{BLOCK_INPUT}, which is no longer necessary).
|
|
|
5515 Free using @code{xfree()}, and realloc using @code{xrealloc()}. Note
|
|
|
5516 that @code{xmalloc()} will do a non-local exit if the memory can't be
|
|
|
5517 allocated. (Many functions, however, do not expect this, and thus XEmacs
|
|
|
5518 will likely crash if this happens. @strong{This is a bug.} If you can,
|
|
|
5519 you should strive to make your function handle this OK. However, it's
|
|
|
5520 difficult in the general circumstance, perhaps requiring extra
|
|
|
5521 unwind-protects and such.)
|
|
|
5522
|
|
|
5523 Note that XEmacs provides two separate replacements for the standard
|
|
|
5524 @code{malloc()} library function. These are called @dfn{old GNU malloc}
|
|
|
5525 (@file{malloc.c}) and @dfn{new GNU malloc} (@file{gmalloc.c}),
|
|
|
5526 respectively. New GNU malloc is better in pretty much every way than
|
|
|
5527 old GNU malloc, and should be used if possible. (It used to be that on
|
|
|
5528 some systems, the old one worked but the new one didn't. I think this
|
|
|
5529 was due specifically to a bug in SunOS, which the new one now works
|
|
|
5530 around; so I don't think the old one ever has to be used any more.) The
|
|
|
5531 primary difference between both of these mallocs and the standard system
|
|
|
5532 malloc is that they are much faster, at the expense of increased space.
|
|
|
5533 The basic idea is that memory is allocated in fixed chunks of powers of
|
|
|
5534 two. This allows for basically constant malloc time, since the various
|
|
|
5535 chunks can just be kept on a number of free lists. (The standard system
|
|
|
5536 malloc typically allocates arbitrary-sized chunks and has to spend some
|
|
|
5537 time, sometimes a significant amount of time, walking the heap looking
|
|
|
5538 for a free block to use and cleaning things up.) The new GNU malloc
|
|
|
5539 improves on things by allocating large objects in chunks of 4096 bytes
|
|
|
5540 rather than in ever larger powers of two, which results in ever larger
|
|
|
5541 wastage. There is a slight speed loss here, but it's of doubtful
|
|
|
5542 significance.
|
|
|
5543
|
|
|
5544 NOTE: Apparently there is a third-generation GNU malloc that is
|
|
|
5545 significantly better than the new GNU malloc, and should probably
|
|
|
5546 be included in XEmacs.
|
|
|
5547
|
|
|
5548 There is also the relocating allocator, @file{ralloc.c}. This actually
|
|
|
5549 moves blocks of memory around so that the @code{sbrk()} pointer shrunk
|
|
|
5550 and virtual memory released back to the system. On some systems,
|
|
|
5551 this is a big win. On all systems, it causes a noticeable (and
|
|
|
5552 sometimes huge) speed penalty, so I turn it off by default.
|
|
|
5553 @file{ralloc.c} only works with the new GNU malloc in @file{gmalloc.c}.
|
|
|
5554 There are also two versions of @file{ralloc.c}, one that uses @code{mmap()}
|
|
|
5555 rather than block copies to move data around. This purports to
|
|
|
5556 be faster, although that depends on the amount of data that would
|
|
|
5557 have had to be block copied and the system-call overhead for
|
|
|
5558 @code{mmap()}. I don't know exactly how this works, except that the
|
|
|
5559 relocating-allocation routines are pretty much used only for
|
|
|
5560 the memory allocated for a buffer, which is the biggest consumer
|
|
|
5561 of space, esp. of space that may get freed later.
|
|
|
5562
|
|
|
5563 Note that the GNU mallocs have some ``memory warning'' facilities.
|
|
|
5564 XEmacs taps into them and issues a warning through the standard
|
|
|
5565 warning system, when memory gets to 75%, 85%, and 95% full.
|
|
|
5566 (On some systems, the memory warnings are not functional.)
|
|
|
5567
|
|
|
5568 Allocated memory that is going to be used to make a Lisp object
|
|
|
5569 is created using @code{allocate_lisp_storage()}. This calls @code{xmalloc()}
|
|
|
5570 but also verifies that the pointer to the memory can fit into
|
|
|
5571 a Lisp word (remember that some bits are taken away for a type
|
|
|
5572 tag and a mark bit). If not, an error is issued through @code{memory_full()}.
|
|
|
5573 @code{allocate_lisp_storage()} is called by @code{alloc_lcrecord()},
|
|
|
5574 @code{ALLOCATE_FIXED_TYPE()}, and the vector and bit-vector creation
|
|
|
5575 routines. These routines also call @code{INCREMENT_CONS_COUNTER()} at the
|
|
|
5576 appropriate times; this keeps statistics on how much memory is
|
|
|
5577 allocated, so that garbage-collection can be invoked when the
|
|
|
5578 threshold is reached.
|
|
|
5579
|
|
|
5580 @node Pure Space
|
|
|
5581 @section Pure Space
|
|
|
5582
|
|
|
5583 Not yet documented.
|
|
|
5584
|
|
|
5585 @node Cons
|
|
|
5586 @section Cons
|
|
|
5587
|
|
|
5588 Conses are allocated in standard frob blocks. The only thing to
|
|
|
5589 note is that conses can be explicitly freed using @code{free_cons()}
|
|
|
5590 and associated functions @code{free_list()} and @code{free_alist()}. This
|
|
|
5591 immediately puts the conses onto the cons free list, and decrements
|
|
|
5592 the statistics on memory allocation appropriately. This is used
|
|
|
5593 to good effect by some extremely commonly-used code, to avoid
|
|
|
5594 generating extra objects and thereby triggering GC sooner.
|
|
|
5595 However, you have to be @emph{extremely} careful when doing this.
|
|
|
5596 If you mess this up, you will get BADLY BURNED, and it has happened
|
|
|
5597 before.
|
|
|
5598
|
|
|
5599 @node Vector
|
|
|
5600 @section Vector
|
|
|
5601
|
|
|
5602 As mentioned above, each vector is @code{malloc()}ed individually, and
|
|
|
5603 all are threaded through the variable @code{all_vectors}. Vectors are
|
|
|
5604 marked strangely during garbage collection, by kludging the size field.
|
|
|
5605 Note that the @code{struct Lisp_Vector} is declared with its
|
|
|
5606 @code{contents} field being a @emph{stretchy} array of one element. It
|
|
|
5607 is actually @code{malloc()}ed with the right size, however, and access
|
|
|
5608 to any element through the @code{contents} array works fine.
|
|
|
5609
|
|
|
5610 @node Bit Vector
|
|
|
5611 @section Bit Vector
|
|
|
5612
|
|
|
5613 Bit vectors work exactly like vectors, except for more complicated
|
|
|
5614 code to access an individual bit, and except for the fact that bit
|
|
|
5615 vectors are lrecords while vectors are not. (The only difference here is
|
|
|
5616 that there's an lrecord implementation pointer at the beginning and the
|
|
|
5617 tag field in bit vector Lisp words is ``lrecord'' rather than
|
|
|
5618 ``vector''.)
|
|
|
5619
|
|
|
5620 @node Symbol
|
|
|
5621 @section Symbol
|
|
|
5622
|
|
|
5623 Symbols are also allocated in frob blocks. Note that the code
|
|
|
5624 exists for symbols to be either lrecords (category (c) above)
|
|
|
5625 or simple types (category (b) above), and are lrecords by
|
|
|
5626 default (I think), although there is no good reason for this.
|
|
|
5627
|
|
|
5628 Note that symbols in the awful horrible obarray structure are
|
|
|
5629 chained through their @code{next} field.
|
|
|
5630
|
|
|
5631 Remember that @code{intern} looks up a symbol in an obarray, creating
|
|
|
5632 one if necessary.
|
|
|
5633
|
|
|
5634 @node Marker
|
|
|
5635 @section Marker
|
|
|
5636
|
|
|
5637 Markers are allocated in frob blocks, as usual. They are kept
|
|
|
5638 in a buffer unordered, but in a doubly-linked list so that they
|
|
|
5639 can easily be removed. (Formerly this was a singly-linked list,
|
|
|
5640 but in some cases garbage collection took an extraordinarily
|
|
|
5641 long time due to the O(N^2) time required to remove lots of
|
|
|
5642 markers from a buffer.) Markers are removed from a buffer in
|
|
|
5643 the finalize stage, in @code{ADDITIONAL_FREE_marker()}.
|
|
|
5644
|
|
|
5645 @node String
|
|
|
5646 @section String
|
|
|
5647
|
|
|
5648 As mentioned above, strings are a special case. A string is logically
|
|
|
5649 two parts, a fixed-size object (containing the length, property list,
|
|
|
5650 and a pointer to the actual data), and the actual data in the string.
|
|
|
5651 The fixed-size object is a @code{struct Lisp_String} and is allocated in
|
|
|
5652 frob blocks, as usual. The actual data is stored in special
|
|
|
5653 @dfn{string-chars blocks}, which are 8K blocks of memory.
|
|
|
5654 Currently-allocated strings are simply laid end to end in these
|
|
|
5655 string-chars blocks, with a pointer back to the @code{struct Lisp_String}
|
|
|
5656 stored before each string in the string-chars block. When a new string
|
|
|
5657 needs to be allocated, the remaining space at the end of the last
|
|
|
5658 string-chars block is used if there's enough, and a new string-chars
|
|
|
5659 block is created otherwise.
|
|
|
5660
|
|
|
5661 There are never any holes in the string-chars blocks due to the string
|
|
|
5662 compaction and relocation that happens at the end of garbage collection.
|
|
|
5663 During the sweep stage of garbage collection, when objects are
|
|
|
5664 reclaimed, the garbage collector goes through all string-chars blocks,
|
|
|
5665 looking for unused strings. Each chunk of string data is preceded by a
|
|
|
5666 pointer to the corresponding @code{struct Lisp_String}, which indicates
|
|
|
5667 both whether the string is used and how big the string is, i.e. how to
|
|
|
5668 get to the next chunk of string data. Holes are compressed by
|
|
|
5669 block-copying the next string into the empty space and relocating the
|
|
|
5670 pointer stored in the corresponding @code{struct Lisp_String}.
|
|
|
5671 @strong{This means you have to be careful with strings in your code.}
|
|
|
5672 See the section above on @code{GCPRO}ing.
|
|
|
5673
|
|
|
5674 Note that there is one situation not handled: a string that is too big
|
|
|
5675 to fit into a string-chars block. Such strings, called @dfn{big
|
|
|
5676 strings}, are all @code{malloc()}ed as their own block. (#### Although it
|
|
|
5677 would make more sense for the threshold for big strings to be somewhat
|
|
|
5678 lower, e.g. 1/2 or 1/4 the size of a string-chars block. It seems that
|
|
440
|
5679 this was indeed the case formerly---indeed, the threshold was set at
|
|
|
5680 1/8---but Mly forgot about this when rewriting things for 19.8.)
|
|
428
|
5681
|
|
|
5682 Note also that the string data in string-chars blocks is padded as
|
|
|
5683 necessary so that proper alignment constraints on the @code{struct
|
|
|
5684 Lisp_String} back pointers are maintained.
|
|
|
5685
|
|
|
5686 Finally, strings can be resized. This happens in Mule when a
|
|
|
5687 character is substituted with a different-length character, or during
|
|
|
5688 modeline frobbing. (You could also export this to Lisp, but it's not
|
|
|
5689 done so currently.) Resizing a string is a potentially tricky process.
|
|
|
5690 If the change is small enough that the padding can absorb it, nothing
|
|
|
5691 other than a simple memory move needs to be done. Keep in mind,
|
|
|
5692 however, that the string can't shrink too much because the offset to the
|
|
|
5693 next string in the string-chars block is computed by looking at the
|
|
|
5694 length and rounding to the nearest multiple of four or eight. If the
|
|
|
5695 string would shrink or expand beyond the correct padding, new string
|
|
|
5696 data needs to be allocated at the end of the last string-chars block and
|
|
|
5697 the data moved appropriately. This leaves some dead string data, which
|
|
|
5698 is marked by putting a special marker of 0xFFFFFFFF in the @code{struct
|
|
|
5699 Lisp_String} pointer before the data (there's no real @code{struct
|
|
|
5700 Lisp_String} to point to and relocate), and storing the size of the dead
|
|
|
5701 string data (which would normally be obtained from the now-non-existent
|
|
|
5702 @code{struct Lisp_String}) at the beginning of the dead string data gap.
|
|
|
5703 The string compactor recognizes this special 0xFFFFFFFF marker and
|
|
|
5704 handles it correctly.
|
|
|
5705
|
|
|
5706 @node Compiled Function
|
|
|
5707 @section Compiled Function
|
|
|
5708
|
|
|
5709 Not yet documented.
|
|
|
5710
|
|
|
5711 @node Events and the Event Loop, Evaluation; Stack Frames; Bindings, Allocation of Objects in XEmacs Lisp, Top
|
|
|
5712 @chapter Events and the Event Loop
|
|
|
5713
|
|
|
5714 @menu
|
|
|
5715 * Introduction to Events::
|
|
|
5716 * Main Loop::
|
|
|
5717 * Specifics of the Event Gathering Mechanism::
|
|
|
5718 * Specifics About the Emacs Event::
|
|
|
5719 * The Event Stream Callback Routines::
|
|
|
5720 * Other Event Loop Functions::
|
|
|
5721 * Converting Events::
|
|
|
5722 * Dispatching Events; The Command Builder::
|
|
|
5723 @end menu
|
|
|
5724
|
|
|
5725 @node Introduction to Events
|
|
|
5726 @section Introduction to Events
|
|
|
5727
|
|
|
5728 An event is an object that encapsulates information about an
|
|
|
5729 interesting occurrence in the operating system. Events are
|
|
|
5730 generated either by user action, direct (e.g. typing on the
|
|
|
5731 keyboard or moving the mouse) or indirect (moving another
|
|
|
5732 window, thereby generating an expose event on an Emacs frame),
|
|
|
5733 or as a result of some other typically asynchronous action happening,
|
|
|
5734 such as output from a subprocess being ready or a timer expiring.
|
|
|
5735 Events come into the system in an asynchronous fashion (typically
|
|
|
5736 through a callback being called) and are converted into a
|
|
|
5737 synchronous event queue (first-in, first-out) in a process that
|
|
|
5738 we will call @dfn{collection}.
|
|
|
5739
|
|
|
5740 Note that each application has its own event queue. (It is
|
|
|
5741 immaterial whether the collection process directly puts the
|
|
|
5742 events in the proper application's queue, or puts them into
|
|
|
5743 a single system queue, which is later split up.)
|
|
|
5744
|
|
|
5745 The most basic level of event collection is done by the
|
|
|
5746 operating system or window system. Typically, XEmacs does
|
|
|
5747 its own event collection as well. Often there are multiple
|
|
|
5748 layers of collection in XEmacs, with events from various
|
|
|
5749 sources being collected into a queue, which is then combined
|
|
|
5750 with other sources to go into another queue (i.e. a second
|
|
|
5751 level of collection), with perhaps another level on top of
|
|
|
5752 this, etc.
|
|
|
5753
|
|
|
5754 XEmacs has its own types of events (called @dfn{Emacs events}),
|
|
|
5755 which provides an abstract layer on top of the system-dependent
|
|
|
5756 nature of the most basic events that are received. Part of the
|
|
|
5757 complex nature of the XEmacs event collection process involves
|
|
|
5758 converting from the operating-system events into the proper
|
|
440
|
5759 Emacs events---there may not be a one-to-one correspondence.
|
|
428
|
5760
|
|
|
5761 Emacs events are documented in @file{events.h}; I'll discuss them
|
|
|
5762 later.
|
|
|
5763
|
|
|
5764 @node Main Loop
|
|
|
5765 @section Main Loop
|
|
|
5766
|
|
|
5767 The @dfn{command loop} is the top-level loop that the editor is always
|
|
|
5768 running. It loops endlessly, calling @code{next-event} to retrieve an
|
|
|
5769 event and @code{dispatch-event} to execute it. @code{dispatch-event} does
|
|
|
5770 the appropriate thing with non-user events (process, timeout,
|
|
|
5771 magic, eval, mouse motion); this involves calling a Lisp handler
|
|
|
5772 function, redrawing a newly-exposed part of a frame, reading
|
|
|
5773 subprocess output, etc. For user events, @code{dispatch-event}
|
|
|
5774 looks up the event in relevant keymaps or menubars; when a
|
|
|
5775 full key sequence or menubar selection is reached, the appropriate
|
|
|
5776 function is executed. @code{dispatch-event} may have to keep state
|
|
|
5777 across calls; this is done in the ``command-builder'' structure
|
|
|
5778 associated with each console (remember, there's usually only
|
|
|
5779 one console), and the engine that looks up keystrokes and
|
|
|
5780 constructs full key sequences is called the @dfn{command builder}.
|
|
|
5781 This is documented elsewhere.
|
|
|
5782
|
|
|
5783 The guts of the command loop are in @code{command_loop_1()}. This
|
|
440
|
5784 function doesn't catch errors, though---that's the job of
|
|
428
|
5785 @code{command_loop_2()}, which is a condition-case (i.e. error-trapping)
|
|
|
5786 wrapper around @code{command_loop_1()}. @code{command_loop_1()} never
|
|
|
5787 returns, but may get thrown out of.
|
|
|
5788
|
|
|
5789 When an error occurs, @code{cmd_error()} is called, which usually
|
|
|
5790 invokes the Lisp error handler in @code{command-error}; however, a
|
|
|
5791 default error handler is provided if @code{command-error} is @code{nil}
|
|
|
5792 (e.g. during startup). The purpose of the error handler is simply to
|
|
|
5793 display the error message and do associated cleanup; it does not need to
|
|
|
5794 throw anywhere. When the error handler finishes, the condition-case in
|
|
|
5795 @code{command_loop_2()} will finish and @code{command_loop_2()} will
|
|
|
5796 reinvoke @code{command_loop_1()}.
|
|
|
5797
|
|
|
5798 @code{command_loop_2()} is invoked from three places: from
|
|
|
5799 @code{initial_command_loop()} (called from @code{main()} at the end of
|
|
|
5800 internal initialization), from the Lisp function @code{recursive-edit},
|
|
|
5801 and from @code{call_command_loop()}.
|
|
|
5802
|
|
|
5803 @code{call_command_loop()} is called when a macro is started and when
|
|
|
5804 the minibuffer is entered; normal termination of the macro or minibuffer
|
|
|
5805 causes a throw out of the recursive command loop. (To
|
|
|
5806 @code{execute-kbd-macro} for macros and @code{exit} for minibuffers.
|
|
|
5807 Note also that the low-level minibuffer-entering function,
|
|
|
5808 @code{read-minibuffer-internal}, provides its own error handling and
|
|
|
5809 does not need @code{command_loop_2()}'s error encapsulation; so it tells
|
|
|
5810 @code{call_command_loop()} to invoke @code{command_loop_1()} directly.)
|
|
|
5811
|
|
|
5812 Note that both read-minibuffer-internal and recursive-edit set up a
|
|
|
5813 catch for @code{exit}; this is why @code{abort-recursive-edit}, which
|
|
|
5814 throws to this catch, exits out of either one.
|
|
|
5815
|
|
|
5816 @code{initial_command_loop()}, called from @code{main()}, sets up a
|
|
|
5817 catch for @code{top-level} when invoking @code{command_loop_2()},
|
|
|
5818 allowing functions to throw all the way to the top level if they really
|
|
|
5819 need to. Before invoking @code{command_loop_2()},
|
|
|
5820 @code{initial_command_loop()} calls @code{top_level_1()}, which handles
|
|
|
5821 all of the startup stuff (creating the initial frame, handling the
|
|
|
5822 command-line options, loading the user's @file{.emacs} file, etc.). The
|
|
|
5823 function that actually does this is in Lisp and is pointed to by the
|
|
|
5824 variable @code{top-level}; normally this function is
|
|
|
5825 @code{normal-top-level}. @code{top_level_1()} is just an error-handling
|
|
|
5826 wrapper similar to @code{command_loop_2()}. Note also that
|
|
|
5827 @code{initial_command_loop()} sets up a catch for @code{top-level} when
|
|
|
5828 invoking @code{top_level_1()}, just like when it invokes
|
|
|
5829 @code{command_loop_2()}.
|
|
|
5830
|
|
|
5831 @node Specifics of the Event Gathering Mechanism
|
|
|
5832 @section Specifics of the Event Gathering Mechanism
|
|
|
5833
|
|
|
5834 Here is an approximate diagram of the collection processes
|
|
|
5835 at work in XEmacs, under TTY's (TTY's are simpler than X
|
|
|
5836 so we'll look at this first):
|
|
|
5837
|
|
|
5838 @noindent
|
|
|
5839 @example
|
|
|
5840 asynch. asynch. asynch. asynch. [Collectors in
|
|
|
5841 kbd events kbd events process process the OS]
|
|
|
5842 | | output output
|
|
|
5843 | | | |
|
|
|
5844 | | | | SIGINT, [signal handlers
|
|
|
5845 | | | | SIGQUIT, in XEmacs]
|
|
|
5846 V V V V SIGWINCH,
|
|
|
5847 file file file file SIGALRM
|
|
|
5848 desc. desc. desc. desc. |
|
|
|
5849 (TTY) (TTY) (pipe) (pipe) |
|
|
|
5850 | | | | fake timeouts
|
|
|
5851 | | | | file |
|
|
|
5852 | | | | desc. |
|
|
|
5853 | | | | (pipe) |
|
|
|
5854 | | | | | |
|
|
|
5855 | | | | | |
|
|
|
5856 | | | | | |
|
|
|
5857 V V V V V V
|
|
|
5858 ------>-----------<----------------<----------------
|
|
|
5859 |
|
|
|
5860 |
|
|
|
5861 | [collected using select() in emacs_tty_next_event()
|
|
|
5862 | and converted to the appropriate Emacs event]
|
|
|
5863 |
|
|
|
5864 |
|
|
|
5865 V (above this line is TTY-specific)
|
|
|
5866 Emacs -----------------------------------------------
|
|
|
5867 event (below this line is the generic event mechanism)
|
|
|
5868 |
|
|
|
5869 |
|
|
|
5870 was there if not, call
|
|
|
5871 a SIGINT? emacs_tty_next_event()
|
|
|
5872 | |
|
|
|
5873 | |
|
|
|
5874 | |
|
|
|
5875 V V
|
|
|
5876 --->------<----
|
|
|
5877 |
|
|
|
5878 | [collected in event_stream_next_event();
|
|
|
5879 | SIGINT is converted using maybe_read_quit_event()]
|
|
|
5880 V
|
|
|
5881 Emacs
|
|
|
5882 event
|
|
|
5883 |
|
|
|
5884 \---->------>----- maybe_kbd_translate() ---->---\
|
|
|
5885 |
|
|
|
5886 |
|
|
|
5887 |
|
|
|
5888 command event queue |
|
|
|
5889 if not from command
|
|
|
5890 (contains events that were event queue, call
|
|
|
5891 read earlier but not processed, event_stream_next_event()
|
|
|
5892 typically when waiting in a |
|
|
|
5893 sit-for, sleep-for, etc. for |
|
|
|
5894 a particular event to be received) |
|
|
|
5895 | |
|
|
|
5896 | |
|
|
|
5897 V V
|
|
|
5898 ---->------------------------------------<----
|
|
|
5899 |
|
|
|
5900 | [collected in
|
|
|
5901 | next_event_internal()]
|
|
|
5902 |
|
|
|
5903 unread- unread- event from |
|
|
|
5904 command- command- keyboard else, call
|
|
|
5905 events event macro next_event_internal()
|
|
|
5906 | | | |
|
|
|
5907 | | | |
|
|
|
5908 | | | |
|
|
|
5909 V V V V
|
|
|
5910 --------->----------------------<------------
|
|
|
5911 |
|
|
|
5912 | [collected in `next-event', which may loop
|
|
|
5913 | more than once if the event it gets is on
|
|
|
5914 | a dead frame, device, etc.]
|
|
|
5915 |
|
|
|
5916 |
|
|
|
5917 V
|
|
|
5918 feed into top-level event loop,
|
|
|
5919 which repeatedly calls `next-event'
|
|
|
5920 and then dispatches the event
|
|
|
5921 using `dispatch-event'
|
|
|
5922 @end example
|
|
|
5923
|
|
|
5924 Notice the separation between TTY-specific and generic event mechanism.
|
|
|
5925 When using the Xt-based event loop, the TTY-specific stuff is replaced
|
|
|
5926 but the rest stays the same.
|
|
|
5927
|
|
|
5928 It's also important to realize that only one different kind of
|
|
|
5929 system-specific event loop can be operating at a time, and must be able
|
|
|
5930 to receive all kinds of events simultaneously. For the two existing
|
|
|
5931 event loops (implemented in @file{event-tty.c} and @file{event-Xt.c},
|
|
|
5932 respectively), the TTY event loop @emph{only} handles TTY consoles,
|
|
|
5933 while the Xt event loop handles @emph{both} TTY and X consoles. This
|
|
|
5934 situation is different from all of the output handlers, where you simply
|
|
|
5935 have one per console type.
|
|
|
5936
|
|
|
5937 Here's the Xt Event Loop Diagram (notice that below a certain point,
|
|
|
5938 it's the same as the above diagram):
|
|
|
5939
|
|
|
5940 @example
|
|
|
5941 asynch. asynch. asynch. asynch. [Collectors in
|
|
|
5942 kbd kbd process process the OS]
|
|
|
5943 events events output output
|
|
|
5944 | | | |
|
|
|
5945 | | | | asynch. asynch. [Collectors in the
|
|
|
5946 | | | | X X OS and X Window System]
|
|
|
5947 | | | | events events
|
|
|
5948 | | | | | |
|
|
|
5949 | | | | | |
|
|
|
5950 | | | | | | SIGINT, [signal handlers
|
|
|
5951 | | | | | | SIGQUIT, in XEmacs]
|
|
|
5952 | | | | | | SIGWINCH,
|
|
|
5953 | | | | | | SIGALRM
|
|
|
5954 | | | | | | |
|
|
|
5955 | | | | | | |
|
|
|
5956 | | | | | | | timeouts
|
|
|
5957 | | | | | | | |
|
|
|
5958 | | | | | | | |
|
|
|
5959 | | | | | | V |
|
|
|
5960 V V V V V V fake |
|
|
|
5961 file file file file file file file |
|
|
|
5962 desc. desc. desc. desc. desc. desc. desc. |
|
|
|
5963 (TTY) (TTY) (pipe) (pipe) (socket) (socket) (pipe) |
|
|
|
5964 | | | | | | | |
|
|
|
5965 | | | | | | | |
|
|
|
5966 | | | | | | | |
|
|
|
5967 V V V V V V V V
|
|
|
5968 --->----------------------------------------<---------<------
|
|
|
5969 | | |
|
|
|
5970 | | |[collected using select() in
|
|
|
5971 | | | _XtWaitForSomething(), called
|
|
|
5972 | | | from XtAppProcessEvent(), called
|
|
|
5973 | | | in emacs_Xt_next_event();
|
|
|
5974 | | | dispatched to various callbacks]
|
|
|
5975 | | |
|
|
|
5976 | | |
|
|
|
5977 emacs_Xt_ p_s_callback(), | [popup_selection_callback]
|
|
|
5978 event_handler() x_u_v_s_callback(),| [x_update_vertical_scrollbar_
|
|
|
5979 | x_u_h_s_callback(),| callback]
|
|
|
5980 | search_callback() | [x_update_horizontal_scrollbar_
|
|
|
5981 | | | callback]
|
|
|
5982 | | |
|
|
|
5983 | | |
|
|
|
5984 enqueue_Xt_ signal_special_ |
|
|
|
5985 dispatch_event() Xt_user_event() |
|
|
|
5986 [maybe multiple | |
|
|
|
5987 times, maybe 0 | |
|
|
|
5988 times] | |
|
|
|
5989 | enqueue_Xt_ |
|
|
|
5990 | dispatch_event() |
|
|
|
5991 | | |
|
|
|
5992 | | |
|
|
|
5993 V V |
|
|
|
5994 -->----------<-- |
|
|
|
5995 | |
|
|
|
5996 | |
|
|
|
5997 dispatch Xt_what_callback()
|
|
|
5998 event sets flags
|
|
|
5999 queue |
|
|
|
6000 | |
|
|
|
6001 | |
|
|
|
6002 | |
|
|
|
6003 | |
|
|
|
6004 ---->-----------<--------
|
|
|
6005 |
|
|
|
6006 |
|
|
|
6007 | [collected and converted as appropriate in
|
|
|
6008 | emacs_Xt_next_event()]
|
|
|
6009 |
|
|
|
6010 |
|
|
|
6011 V (above this line is Xt-specific)
|
|
|
6012 Emacs ------------------------------------------------
|
|
|
6013 event (below this line is the generic event mechanism)
|
|
|
6014 |
|
|
|
6015 |
|
|
|
6016 was there if not, call
|
|
|
6017 a SIGINT? emacs_Xt_next_event()
|
|
|
6018 | |
|
|
|
6019 | |
|
|
|
6020 | |
|
|
|
6021 V V
|
|
|
6022 --->-------<----
|
|
|
6023 |
|
|
|
6024 | [collected in event_stream_next_event();
|
|
|
6025 | SIGINT is converted using maybe_read_quit_event()]
|
|
|
6026 V
|
|
|
6027 Emacs
|
|
|
6028 event
|
|
|
6029 |
|
|
|
6030 \---->------>----- maybe_kbd_translate() -->-----\
|
|
|
6031 |
|
|
|
6032 |
|
|
|
6033 |
|
|
|
6034 command event queue |
|
|
|
6035 if not from command
|
|
|
6036 (contains events that were event queue, call
|
|
|
6037 read earlier but not processed, event_stream_next_event()
|
|
|
6038 typically when waiting in a |
|
|
|
6039 sit-for, sleep-for, etc. for |
|
|
|
6040 a particular event to be received) |
|
|
|
6041 | |
|
|
|
6042 | |
|
|
|
6043 V V
|
|
|
6044 ---->----------------------------------<------
|
|
|
6045 |
|
|
|
6046 | [collected in
|
|
|
6047 | next_event_internal()]
|
|
|
6048 |
|
|
|
6049 unread- unread- event from |
|
|
|
6050 command- command- keyboard else, call
|
|
|
6051 events event macro next_event_internal()
|
|
|
6052 | | | |
|
|
|
6053 | | | |
|
|
|
6054 | | | |
|
|
|
6055 V V V V
|
|
|
6056 --------->----------------------<------------
|
|
|
6057 |
|
|
|
6058 | [collected in `next-event', which may loop
|
|
|
6059 | more than once if the event it gets is on
|
|
|
6060 | a dead frame, device, etc.]
|
|
|
6061 |
|
|
|
6062 |
|
|
|
6063 V
|
|
|
6064 feed into top-level event loop,
|
|
|
6065 which repeatedly calls `next-event'
|
|
|
6066 and then dispatches the event
|
|
|
6067 using `dispatch-event'
|
|
|
6068 @end example
|
|
|
6069
|
|
|
6070 @node Specifics About the Emacs Event
|
|
|
6071 @section Specifics About the Emacs Event
|
|
|
6072
|
|
|
6073 @node The Event Stream Callback Routines
|
|
|
6074 @section The Event Stream Callback Routines
|
|
|
6075
|
|
|
6076 @node Other Event Loop Functions
|
|
|
6077 @section Other Event Loop Functions
|
|
|
6078
|
|
|
6079 @code{detect_input_pending()} and @code{input-pending-p} look for
|
|
|
6080 input by calling @code{event_stream->event_pending_p} and looking in
|
|
|
6081 @code{[V]unread-command-event} and the @code{command_event_queue} (they
|
|
|
6082 do not check for an executing keyboard macro, though).
|
|
|
6083
|
|
|
6084 @code{discard-input} cancels any command events pending (and any
|
|
|
6085 keyboard macros currently executing), and puts the others onto the
|
|
|
6086 @code{command_event_queue}. There is a comment about a ``race
|
|
|
6087 condition'', which is not a good sign.
|
|
|
6088
|
|
|
6089 @code{next-command-event} and @code{read-char} are higher-level
|
|
|
6090 interfaces to @code{next-event}. @code{next-command-event} gets the
|
|
|
6091 next @dfn{command} event (i.e. keypress, mouse event, menu selection,
|
|
|
6092 or scrollbar action), calling @code{dispatch-event} on any others.
|
|
|
6093 @code{read-char} calls @code{next-command-event} and uses
|
|
|
6094 @code{event_to_character()} to return the character equivalent. With
|
|
|
6095 the right kind of input method support, it is possible for (read-char)
|
|
|
6096 to return a Kanji character.
|
|
|
6097
|
|
|
6098 @node Converting Events
|
|
|
6099 @section Converting Events
|
|
|
6100
|
|
|
6101 @code{character_to_event()}, @code{event_to_character()},
|
|
|
6102 @code{event-to-character}, and @code{character-to-event} convert between
|
|
|
6103 characters and keypress events corresponding to the characters. If the
|
|
|
6104 event was not a keypress, @code{event_to_character()} returns -1 and
|
|
|
6105 @code{event-to-character} returns @code{nil}. These functions convert
|
|
|
6106 between character representation and the split-up event representation
|
|
|
6107 (keysym plus mod keys).
|
|
|
6108
|
|
|
6109 @node Dispatching Events; The Command Builder
|
|
|
6110 @section Dispatching Events; The Command Builder
|
|
|
6111
|
|
|
6112 Not yet documented.
|
|
|
6113
|
|
|
6114 @node Evaluation; Stack Frames; Bindings, Symbols and Variables, Events and the Event Loop, Top
|
|
|
6115 @chapter Evaluation; Stack Frames; Bindings
|
|
|
6116
|
|
|
6117 @menu
|
|
|
6118 * Evaluation::
|
|
|
6119 * Dynamic Binding; The specbinding Stack; Unwind-Protects::
|
|
|
6120 * Simple Special Forms::
|
|
|
6121 * Catch and Throw::
|
|
|
6122 @end menu
|
|
|
6123
|
|
|
6124 @node Evaluation
|
|
|
6125 @section Evaluation
|
|
|
6126
|
|
|
6127 @code{Feval()} evaluates the form (a Lisp object) that is passed to
|
|
|
6128 it. Note that evaluation is only non-trivial for two types of objects:
|
|
|
6129 symbols and conses. A symbol is evaluated simply by calling
|
|
|
6130 @code{symbol-value} on it and returning the value.
|
|
|
6131
|
|
|
6132 Evaluating a cons means calling a function. First, @code{eval} checks
|
|
|
6133 to see if garbage-collection is necessary, and calls
|
|
|
6134 @code{garbage_collect_1()} if so. It then increases the evaluation
|
|
|
6135 depth by 1 (@code{lisp_eval_depth}, which is always less than
|
|
|
6136 @code{max_lisp_eval_depth}) and adds an element to the linked list of
|
|
|
6137 @code{struct backtrace}'s (@code{backtrace_list}). Each such structure
|
|
|
6138 contains a pointer to the function being called plus a list of the
|
|
|
6139 function's arguments. Originally these values are stored unevalled, and
|
|
|
6140 as they are evaluated, the backtrace structure is updated. Garbage
|
|
|
6141 collection pays attention to the objects pointed to in the backtrace
|
|
|
6142 structures (garbage collection might happen while a function is being
|
|
|
6143 called or while an argument is being evaluated, and there could easily
|
|
|
6144 be no other references to the arguments in the argument list; once an
|
|
|
6145 argument is evaluated, however, the unevalled version is not needed by
|
|
|
6146 eval, and so the backtrace structure is changed).
|
|
|
6147
|
|
|
6148 At this point, the function to be called is determined by looking at
|
|
|
6149 the car of the cons (if this is a symbol, its function definition is
|
|
|
6150 retrieved and the process repeated). The function should then consist
|
|
|
6151 of either a @code{Lisp_Subr} (built-in function written in C), a
|
|
|
6152 @code{Lisp_Compiled_Function} object, or a cons whose car is one of the
|
|
|
6153 symbols @code{autoload}, @code{macro} or @code{lambda}.
|
|
|
6154
|
|
|
6155 If the function is a @code{Lisp_Subr}, the lisp object points to a
|
|
|
6156 @code{struct Lisp_Subr} (created by @code{DEFUN()}), which contains a
|
|
|
6157 pointer to the C function, a minimum and maximum number of arguments
|
|
|
6158 (or possibly the special constants @code{MANY} or @code{UNEVALLED}), a
|
|
|
6159 pointer to the symbol referring to that subr, and a couple of other
|
|
|
6160 things. If the subr wants its arguments @code{UNEVALLED}, they are
|
|
|
6161 passed raw as a list. Otherwise, an array of evaluated arguments is
|
|
|
6162 created and put into the backtrace structure, and either passed whole
|
|
|
6163 (@code{MANY}) or each argument is passed as a C argument.
|
|
|
6164
|
|
|
6165 If the function is a @code{Lisp_Compiled_Function},
|
|
|
6166 @code{funcall_compiled_function()} is called. If the function is a
|
|
|
6167 lambda list, @code{funcall_lambda()} is called. If the function is a
|
|
|
6168 macro, [..... fill in] is done. If the function is an autoload,
|
|
|
6169 @code{do_autoload()} is called to load the definition and then eval
|
|
|
6170 starts over [explain this more].
|
|
|
6171
|
|
|
6172 When @code{Feval()} exits, the evaluation depth is reduced by one, the
|
|
|
6173 debugger is called if appropriate, and the current backtrace structure
|
|
|
6174 is removed from the list.
|
|
|
6175
|
|
|
6176 Both @code{funcall_compiled_function()} and @code{funcall_lambda()} need
|
|
|
6177 to go through the list of formal parameters to the function and bind
|
|
|
6178 them to the actual arguments, checking for @code{&rest} and
|
|
|
6179 @code{&optional} symbols in the formal parameters and making sure the
|
|
|
6180 number of actual arguments is correct.
|
|
|
6181 @code{funcall_compiled_function()} can do this a little more
|
|
|
6182 efficiently, since the formal parameter list can be checked for sanity
|
|
|
6183 when the compiled function object is created.
|
|
|
6184
|
|
|
6185 @code{funcall_lambda()} simply calls @code{Fprogn} to execute the code
|
|
|
6186 in the lambda list.
|
|
|
6187
|
|
|
6188 @code{funcall_compiled_function()} calls the real byte-code interpreter
|
|
|
6189 @code{execute_optimized_program()} on the byte-code instructions, which
|
|
|
6190 are converted into an internal form for faster execution.
|
|
|
6191
|
|
|
6192 When a compiled function is executed for the first time by
|
|
|
6193 @code{funcall_compiled_function()}, or when it is @code{Fpurecopy()}ed
|
|
|
6194 during the dump phase of building XEmacs, the byte-code instructions are
|
|
|
6195 converted from a @code{Lisp_String} (which is inefficient to access,
|
|
|
6196 especially in the presence of MULE) into a @code{Lisp_Opaque} object
|
|
|
6197 containing an array of unsigned char, which can be directly executed by
|
|
|
6198 the byte-code interpreter. At this time the byte code is also analyzed
|
|
|
6199 for validity and transformed into a more optimized form, so that
|
|
|
6200 @code{execute_optimized_program()} can really fly.
|
|
|
6201
|
|
|
6202 Here are some of the optimizations performed by the internal byte-code
|
|
|
6203 transformer:
|
|
|
6204 @enumerate
|
|
|
6205 @item
|
|
|
6206 References to the @code{constants} array are checked for out-of-range
|
|
|
6207 indices, so that the byte interpreter doesn't have to.
|
|
|
6208 @item
|
|
|
6209 References to the @code{constants} array that will be used as a Lisp
|
|
|
6210 variable are checked for being correct non-constant (i.e. not @code{t},
|
|
|
6211 @code{nil}, or @code{keywordp}) symbols, so that the byte interpreter
|
|
|
6212 doesn't have to.
|
|
|
6213 @item
|
|
|
6214 The maxiumum number of variable bindings in the byte-code is
|
|
|
6215 pre-computed, so that space on the @code{specpdl} stack can be
|
|
|
6216 pre-reserved once for the whole function execution.
|
|
|
6217 @item
|
|
|
6218 All byte-code jumps are relative to the current program counter instead
|
|
|
6219 of the start of the program, thereby saving a register.
|
|
|
6220 @item
|
|
|
6221 One-byte relative jumps are converted from the byte-code form of unsigned
|
|
|
6222 chars offset by 127 to machine-friendly signed chars.
|
|
|
6223 @end enumerate
|
|
|
6224
|
|
|
6225 Of course, this transformation of the @code{instructions} should not be
|
|
|
6226 visible to the user, so @code{Fcompiled_function_instructions()} needs
|
|
|
6227 to know how to convert the optimized opaque object back into a Lisp
|
|
|
6228 string that is identical to the original string from the @file{.elc}
|
|
|
6229 file. (Actually, the resulting string may (rarely) contain slightly
|
|
|
6230 different, yet equivalent, byte code.)
|
|
|
6231
|
|
|
6232 @code{Ffuncall()} implements Lisp @code{funcall}. @code{(funcall fun
|
|
|
6233 x1 x2 x3 ...)} is equivalent to @code{(eval (list fun (quote x1) (quote
|
|
|
6234 x2) (quote x3) ...))}. @code{Ffuncall()} contains its own code to do
|
|
|
6235 the evaluation, however, and is very similar to @code{Feval()}.
|
|
|
6236
|
|
|
6237 From the performance point of view, it is worth knowing that most of the
|
|
|
6238 time in Lisp evaluation is spent executing @code{Lisp_Subr} and
|
|
|
6239 @code{Lisp_Compiled_Function} objects via @code{Ffuncall()} (not
|
|
|
6240 @code{Feval()}).
|
|
|
6241
|
|
|
6242 @code{Fapply()} implements Lisp @code{apply}, which is very similar to
|
|
|
6243 @code{funcall} except that if the last argument is a list, the result is the
|
|
|
6244 same as if each of the arguments in the list had been passed separately.
|
|
|
6245 @code{Fapply()} does some business to expand the last argument if it's a
|
|
|
6246 list, then calls @code{Ffuncall()} to do the work.
|
|
|
6247
|
|
|
6248 @code{apply1()}, @code{call0()}, @code{call1()}, @code{call2()}, and
|
|
|
6249 @code{call3()} call a function, passing it the argument(s) given (the
|
|
|
6250 arguments are given as separate C arguments rather than being passed as
|
|
|
6251 an array). @code{apply1()} uses @code{Fapply()} while the others use
|
|
|
6252 @code{Ffuncall()} to do the real work.
|
|
|
6253
|
|
|
6254 @node Dynamic Binding; The specbinding Stack; Unwind-Protects
|
|
|
6255 @section Dynamic Binding; The specbinding Stack; Unwind-Protects
|
|
|
6256
|
|
|
6257 @example
|
|
|
6258 struct specbinding
|
|
|
6259 @{
|
|
|
6260 Lisp_Object symbol;
|
|
|
6261 Lisp_Object old_value;
|
|
|
6262 Lisp_Object (*func) (Lisp_Object); /* for unwind-protect */
|
|
|
6263 @};
|
|
|
6264 @end example
|
|
|
6265
|
|
|
6266 @code{struct specbinding} is used for local-variable bindings and
|
|
|
6267 unwind-protects. @code{specpdl} holds an array of @code{struct specbinding}'s,
|
|
|
6268 @code{specpdl_ptr} points to the beginning of the free bindings in the
|
|
|
6269 array, @code{specpdl_size} specifies the total number of binding slots
|
|
|
6270 in the array, and @code{max_specpdl_size} specifies the maximum number
|
|
|
6271 of bindings the array can be expanded to hold. @code{grow_specpdl()}
|
|
|
6272 increases the size of the @code{specpdl} array, multiplying its size by
|
|
|
6273 2 but never exceeding @code{max_specpdl_size} (except that if this
|
|
|
6274 number is less than 400, it is first set to 400).
|
|
|
6275
|
|
|
6276 @code{specbind()} binds a symbol to a value and is used for local
|
|
|
6277 variables and @code{let} forms. The symbol and its old value (which
|
|
|
6278 might be @code{Qunbound}, indicating no prior value) are recorded in the
|
|
|
6279 specpdl array, and @code{specpdl_size} is increased by 1.
|
|
|
6280
|
|
|
6281 @code{record_unwind_protect()} implements an @dfn{unwind-protect},
|
|
|
6282 which, when placed around a section of code, ensures that some specified
|
|
|
6283 cleanup routine will be executed even if the code exits abnormally
|
|
|
6284 (e.g. through a @code{throw} or quit). @code{record_unwind_protect()}
|
|
|
6285 simply adds a new specbinding to the @code{specpdl} array and stores the
|
|
|
6286 appropriate information in it. The cleanup routine can either be a C
|
|
|
6287 function, which is stored in the @code{func} field, or a @code{progn}
|
|
|
6288 form, which is stored in the @code{old_value} field.
|
|
|
6289
|
|
|
6290 @code{unbind_to()} removes specbindings from the @code{specpdl} array
|
|
|
6291 until the specified position is reached. Each specbinding can be one of
|
|
|
6292 three types:
|
|
|
6293
|
|
|
6294 @enumerate
|
|
|
6295 @item
|
|
|
6296 an unwind-protect with a C cleanup function (@code{func} is not 0, and
|
|
|
6297 @code{old_value} holds an argument to be passed to the function);
|
|
|
6298 @item
|
|
|
6299 an unwind-protect with a Lisp form (@code{func} is 0, @code{symbol}
|
|
|
6300 is @code{nil}, and @code{old_value} holds the form to be executed with
|
|
|
6301 @code{Fprogn()}); or
|
|
|
6302 @item
|
|
|
6303 a local-variable binding (@code{func} is 0, @code{symbol} is not
|
|
|
6304 @code{nil}, and @code{old_value} holds the old value, which is stored as
|
|
|
6305 the symbol's value).
|
|
|
6306 @end enumerate
|
|
|
6307
|
|
|
6308 @node Simple Special Forms
|
|
|
6309 @section Simple Special Forms
|
|
|
6310
|
|
|
6311 @code{or}, @code{and}, @code{if}, @code{cond}, @code{progn},
|
|
|
6312 @code{prog1}, @code{prog2}, @code{setq}, @code{quote}, @code{function},
|
|
|
6313 @code{let*}, @code{let}, @code{while}
|
|
|
6314
|
|
|
6315 All of these are very simple and work as expected, calling
|
|
|
6316 @code{Feval()} or @code{Fprogn()} as necessary and (in the case of
|
|
|
6317 @code{let} and @code{let*}) using @code{specbind()} to create bindings
|
|
|
6318 and @code{unbind_to()} to undo the bindings when finished.
|
|
|
6319
|
|
|
6320 Note that, with the exeption of @code{Fprogn}, these functions are
|
|
|
6321 typically called in real life only in interpreted code, since the byte
|
|
|
6322 compiler knows how to convert calls to these functions directly into
|
|
|
6323 byte code.
|
|
|
6324
|
|
|
6325 @node Catch and Throw
|
|
|
6326 @section Catch and Throw
|
|
|
6327
|
|
|
6328 @example
|
|
|
6329 struct catchtag
|
|
|
6330 @{
|
|
|
6331 Lisp_Object tag;
|
|
|
6332 Lisp_Object val;
|
|
|
6333 struct catchtag *next;
|
|
|
6334 struct gcpro *gcpro;
|
|
|
6335 jmp_buf jmp;
|
|
|
6336 struct backtrace *backlist;
|
|
|
6337 int lisp_eval_depth;
|
|
|
6338 int pdlcount;
|
|
|
6339 @};
|
|
|
6340 @end example
|
|
|
6341
|
|
|
6342 @code{catch} is a Lisp function that places a catch around a body of
|
|
|
6343 code. A catch is a means of non-local exit from the code. When a catch
|
|
|
6344 is created, a tag is specified, and executing a @code{throw} to this tag
|
|
|
6345 will exit from the body of code caught with this tag, and its value will
|
|
|
6346 be the value given in the call to @code{throw}. If there is no such
|
|
|
6347 call, the code will be executed normally.
|
|
|
6348
|
|
|
6349 Information pertaining to a catch is held in a @code{struct catchtag},
|
|
|
6350 which is placed at the head of a linked list pointed to by
|
|
|
6351 @code{catchlist}. @code{internal_catch()} is passed a C function to
|
|
|
6352 call (@code{Fprogn()} when Lisp @code{catch} is called) and arguments to
|
|
|
6353 give it, and places a catch around the function. Each @code{struct
|
|
|
6354 catchtag} is held in the stack frame of the @code{internal_catch()}
|
|
|
6355 instance that created the catch.
|
|
|
6356
|
|
|
6357 @code{internal_catch()} is fairly straightforward. It stores into the
|
|
|
6358 @code{struct catchtag} the tag name and the current values of
|
|
|
6359 @code{backtrace_list}, @code{lisp_eval_depth}, @code{gcprolist}, and the
|
|
|
6360 offset into the @code{specpdl} array, sets a jump point with @code{_setjmp()}
|
|
|
6361 (storing the jump point into the @code{struct catchtag}), and calls the
|
|
|
6362 function. Control will return to @code{internal_catch()} either when
|
|
|
6363 the function exits normally or through a @code{_longjmp()} to this jump
|
|
|
6364 point. In the latter case, @code{throw} will store the value to be
|
|
|
6365 returned into the @code{struct catchtag} before jumping. When it's
|
|
|
6366 done, @code{internal_catch()} removes the @code{struct catchtag} from
|
|
|
6367 the catchlist and returns the proper value.
|
|
|
6368
|
|
|
6369 @code{Fthrow()} goes up through the catchlist until it finds one with
|
|
|
6370 a matching tag. It then calls @code{unbind_catch()} to restore
|
|
|
6371 everything to what it was when the appropriate catch was set, stores the
|
|
|
6372 return value in the @code{struct catchtag}, and jumps (with
|
|
|
6373 @code{_longjmp()}) to its jump point.
|
|
|
6374
|
|
|
6375 @code{unbind_catch()} removes all catches from the catchlist until it
|
|
|
6376 finds the correct one. Some of the catches might have been placed for
|
|
|
6377 error-trapping, and if so, the appropriate entries on the handlerlist
|
|
|
6378 must be removed (see ``errors''). @code{unbind_catch()} also restores
|
|
|
6379 the values of @code{gcprolist}, @code{backtrace_list}, and
|
|
|
6380 @code{lisp_eval}, and calls @code{unbind_to()} to undo any specbindings
|
|
|
6381 created since the catch.
|
|
|
6382
|
|
|
6383
|
|
|
6384 @node Symbols and Variables, Buffers and Textual Representation, Evaluation; Stack Frames; Bindings, Top
|
|
|
6385 @chapter Symbols and Variables
|
|
|
6386
|
|
|
6387 @menu
|
|
|
6388 * Introduction to Symbols::
|
|
|
6389 * Obarrays::
|
|
|
6390 * Symbol Values::
|
|
|
6391 @end menu
|
|
|
6392
|
|
|
6393 @node Introduction to Symbols
|
|
|
6394 @section Introduction to Symbols
|
|
|
6395
|
|
|
6396 A symbol is basically just an object with four fields: a name (a
|
|
|
6397 string), a value (some Lisp object), a function (some Lisp object), and
|
|
|
6398 a property list (usually a list of alternating keyword/value pairs).
|
|
|
6399 What makes symbols special is that there is usually only one symbol with
|
|
|
6400 a given name, and the symbol is referred to by name. This makes a
|
|
|
6401 symbol a convenient way of calling up data by name, i.e. of implementing
|
|
|
6402 variables. (The variable's value is stored in the @dfn{value slot}.)
|
|
|
6403 Similarly, functions are referenced by name, and the definition of the
|
|
|
6404 function is stored in a symbol's @dfn{function slot}. This means that
|
|
|
6405 there can be a distinct function and variable with the same name. The
|
|
|
6406 property list is used as a more general mechanism of associating
|
|
|
6407 additional values with particular names, and once again the namespace is
|
|
|
6408 independent of the function and variable namespaces.
|
|
|
6409
|
|
|
6410 @node Obarrays
|
|
|
6411 @section Obarrays
|
|
|
6412
|
|
|
6413 The identity of symbols with their names is accomplished through a
|
|
|
6414 structure called an obarray, which is just a poorly-implemented hash
|
|
|
6415 table mapping from strings to symbols whose name is that string. (I say
|
|
|
6416 ``poorly implemented'' because an obarray appears in Lisp as a vector
|
|
|
6417 with some hidden fields rather than as its own opaque type. This is an
|
|
|
6418 Emacs Lisp artifact that should be fixed.)
|
|
|
6419
|
|
|
6420 Obarrays are implemented as a vector of some fixed size (which should
|
|
|
6421 be a prime for best results), where each ``bucket'' of the vector
|
|
|
6422 contains one or more symbols, threaded through a hidden @code{next}
|
|
|
6423 field in the symbol. Lookup of a symbol in an obarray, and adding a
|
|
|
6424 symbol to an obarray, is accomplished through standard hash-table
|
|
|
6425 techniques.
|
|
|
6426
|
|
|
6427 The standard Lisp function for working with symbols and obarrays is
|
|
|
6428 @code{intern}. This looks up a symbol in an obarray given its name; if
|
|
|
6429 it's not found, a new symbol is automatically created with the specified
|
|
|
6430 name, added to the obarray, and returned. This is what happens when the
|
|
|
6431 Lisp reader encounters a symbol (or more precisely, encounters the name
|
|
|
6432 of a symbol) in some text that it is reading. There is a standard
|
|
|
6433 obarray called @code{obarray} that is used for this purpose, although
|
|
|
6434 the Lisp programmer is free to create his own obarrays and @code{intern}
|
|
|
6435 symbols in them.
|
|
|
6436
|
|
|
6437 Note that, once a symbol is in an obarray, it stays there until
|
|
|
6438 something is done about it, and the standard obarray @code{obarray}
|
|
|
6439 always stays around, so once you use any particular variable name, a
|
|
|
6440 corresponding symbol will stay around in @code{obarray} until you exit
|
|
|
6441 XEmacs.
|
|
|
6442
|
|
|
6443 Note that @code{obarray} itself is a variable, and as such there is a
|
|
|
6444 symbol in @code{obarray} whose name is @code{"obarray"} and which
|
|
|
6445 contains @code{obarray} as its value.
|
|
|
6446
|
|
|
6447 Note also that this call to @code{intern} occurs only when in the Lisp
|
|
|
6448 reader, not when the code is executed (at which point the symbol is
|
|
|
6449 already around, stored as such in the definition of the function).
|
|
|
6450
|
|
|
6451 You can create your own obarray using @code{make-vector} (this is
|
|
|
6452 horrible but is an artifact) and intern symbols into that obarray.
|
|
|
6453 Doing that will result in two or more symbols with the same name.
|
|
|
6454 However, at most one of these symbols is in the standard @code{obarray}:
|
|
|
6455 You cannot have two symbols of the same name in any particular obarray.
|
|
|
6456 Note that you cannot add a symbol to an obarray in any fashion other
|
|
|
6457 than using @code{intern}: i.e. you can't take an existing symbol and put
|
|
|
6458 it in an existing obarray. Nor can you change the name of an existing
|
|
|
6459 symbol. (Since obarrays are vectors, you can violate the consistency of
|
|
|
6460 things by storing directly into the vector, but let's ignore that
|
|
|
6461 possibility.)
|
|
|
6462
|
|
|
6463 Usually symbols are created by @code{intern}, but if you really want,
|
|
|
6464 you can explicitly create a symbol using @code{make-symbol}, giving it
|
|
|
6465 some name. The resulting symbol is not in any obarray (i.e. it is
|
|
|
6466 @dfn{uninterned}), and you can't add it to any obarray. Therefore its
|
|
|
6467 primary purpose is as a symbol to use in macros to avoid namespace
|
|
|
6468 pollution. It can also be used as a carrier of information, but cons
|
|
|
6469 cells could probably be used just as well.
|
|
|
6470
|
|
|
6471 You can also use @code{intern-soft} to look up a symbol but not create
|
|
|
6472 a new one, and @code{unintern} to remove a symbol from an obarray. This
|
|
|
6473 returns the removed symbol. (Remember: You can't put the symbol back
|
|
|
6474 into any obarray.) Finally, @code{mapatoms} maps over all of the symbols
|
|
|
6475 in an obarray.
|
|
|
6476
|
|
|
6477 @node Symbol Values
|
|
|
6478 @section Symbol Values
|
|
|
6479
|
|
|
6480 The value field of a symbol normally contains a Lisp object. However,
|
|
|
6481 a symbol can be @dfn{unbound}, meaning that it logically has no value.
|
|
|
6482 This is internally indicated by storing a special Lisp object, called
|
|
|
6483 @dfn{the unbound marker} and stored in the global variable
|
|
|
6484 @code{Qunbound}. The unbound marker is of a special Lisp object type
|
|
|
6485 called @dfn{symbol-value-magic}. It is impossible for the Lisp
|
|
|
6486 programmer to directly create or access any object of this type.
|
|
|
6487
|
|
|
6488 @strong{You must not let any ``symbol-value-magic'' object escape to
|
|
|
6489 the Lisp level.} Printing any of these objects will cause the message
|
|
|
6490 @samp{INTERNAL EMACS BUG} to appear as part of the print representation.
|
|
|
6491 (You may see this normally when you call @code{debug_print()} from the
|
|
|
6492 debugger on a Lisp object.) If you let one of these objects escape to
|
|
|
6493 the Lisp level, you will violate a number of assumptions contained in
|
|
|
6494 the C code and make the unbound marker not function right.
|
|
|
6495
|
|
|
6496 When a symbol is created, its value field (and function field) are set
|
|
|
6497 to @code{Qunbound}. The Lisp programmer can restore these conditions
|
|
|
6498 later using @code{makunbound} or @code{fmakunbound}, and can query to
|
|
|
6499 see whether the value of function fields are @dfn{bound} (i.e. have a
|
|
|
6500 value other than @code{Qunbound}) using @code{boundp} and
|
|
|
6501 @code{fboundp}. The fields are set to a normal Lisp object using
|
|
|
6502 @code{set} (or @code{setq}) and @code{fset}.
|
|
|
6503
|
|
|
6504 Other symbol-value-magic objects are used as special markers to
|
|
|
6505 indicate variables that have non-normal properties. This includes any
|
|
|
6506 variables that are tied into C variables (setting the variable magically
|
|
|
6507 sets some global variable in the C code, and likewise for retrieving the
|
|
|
6508 variable's value), variables that magically tie into slots in the
|
|
|
6509 current buffer, variables that are buffer-local, etc. The
|
|
|
6510 symbol-value-magic object is stored in the value cell in place of
|
|
|
6511 a normal object, and the code to retrieve a symbol's value
|
|
|
6512 (i.e. @code{symbol-value}) knows how to do special things with them.
|
|
|
6513 This means that you should not just fetch the value cell directly if you
|
|
|
6514 want a symbol's value.
|
|
|
6515
|
|
|
6516 The exact workings of this are rather complex and involved and are
|
|
|
6517 well-documented in comments in @file{buffer.c}, @file{symbols.c}, and
|
|
|
6518 @file{lisp.h}.
|
|
|
6519
|
|
|
6520 @node Buffers and Textual Representation, MULE Character Sets and Encodings, Symbols and Variables, Top
|
|
|
6521 @chapter Buffers and Textual Representation
|
|
|
6522
|
|
|
6523 @menu
|
|
|
6524 * Introduction to Buffers:: A buffer holds a block of text such as a file.
|
|
|
6525 * The Text in a Buffer:: Representation of the text in a buffer.
|
|
|
6526 * Buffer Lists:: Keeping track of all buffers.
|
|
|
6527 * Markers and Extents:: Tagging locations within a buffer.
|
|
|
6528 * Bufbytes and Emchars:: Representation of individual characters.
|
|
|
6529 * The Buffer Object:: The Lisp object corresponding to a buffer.
|
|
|
6530 @end menu
|
|
|
6531
|
|
|
6532 @node Introduction to Buffers
|
|
|
6533 @section Introduction to Buffers
|
|
|
6534
|
|
|
6535 A buffer is logically just a Lisp object that holds some text.
|
|
|
6536 In this, it is like a string, but a buffer is optimized for
|
|
|
6537 frequent insertion and deletion, while a string is not. Furthermore:
|
|
|
6538
|
|
|
6539 @enumerate
|
|
|
6540 @item
|
|
|
6541 Buffers are @dfn{permanent} objects, i.e. once you create them, they
|
|
|
6542 remain around, and need to be explicitly deleted before they go away.
|
|
|
6543 @item
|
|
|
6544 Each buffer has a unique name, which is a string. Buffers are
|
|
|
6545 normally referred to by name. In this respect, they are like
|
|
|
6546 symbols.
|
|
|
6547 @item
|
|
|
6548 Buffers have a default insertion position, called @dfn{point}.
|
|
|
6549 Inserting text (unless you explicitly give a position) goes at point,
|
|
|
6550 and moves point forward past the text. This is what is going on when
|
|
|
6551 you type text into Emacs.
|
|
|
6552 @item
|
|
|
6553 Buffers have lots of extra properties associated with them.
|
|
|
6554 @item
|
|
|
6555 Buffers can be @dfn{displayed}. What this means is that there
|
|
|
6556 exist a number of @dfn{windows}, which are objects that correspond
|
|
|
6557 to some visible section of your display, and each window has
|
|
|
6558 an associated buffer, and the current contents of the buffer
|
|
|
6559 are shown in that section of the display. The redisplay mechanism
|
|
|
6560 (which takes care of doing this) knows how to look at the
|
|
|
6561 text of a buffer and come up with some reasonable way of displaying
|
|
|
6562 this. Many of the properties of a buffer control how the
|
|
|
6563 buffer's text is displayed.
|
|
|
6564 @item
|
|
|
6565 One buffer is distinguished and called the @dfn{current buffer}. It is
|
|
|
6566 stored in the variable @code{current_buffer}. Buffer operations operate
|
|
|
6567 on this buffer by default. When you are typing text into a buffer, the
|
|
|
6568 buffer you are typing into is always @code{current_buffer}. Switching
|
|
|
6569 to a different window changes the current buffer. Note that Lisp code
|
|
|
6570 can temporarily change the current buffer using @code{set-buffer} (often
|
|
|
6571 enclosed in a @code{save-excursion} so that the former current buffer
|
|
|
6572 gets restored when the code is finished). However, calling
|
|
|
6573 @code{set-buffer} will NOT cause a permanent change in the current
|
|
|
6574 buffer. The reason for this is that the top-level event loop sets
|
|
|
6575 @code{current_buffer} to the buffer of the selected window, each time
|
|
|
6576 it finishes executing a user command.
|
|
|
6577 @end enumerate
|
|
|
6578
|
|
|
6579 Make sure you understand the distinction between @dfn{current buffer}
|
|
|
6580 and @dfn{buffer of the selected window}, and the distinction between
|
|
|
6581 @dfn{point} of the current buffer and @dfn{window-point} of the selected
|
|
|
6582 window. (This latter distinction is explained in detail in the section
|
|
|
6583 on windows.)
|
|
|
6584
|
|
|
6585 @node The Text in a Buffer
|
|
|
6586 @section The Text in a Buffer
|
|
|
6587
|
|
|
6588 The text in a buffer consists of a sequence of zero or more
|
|
|
6589 characters. A @dfn{character} is an integer that logically represents
|
|
|
6590 a letter, number, space, or other unit of text. Most of the characters
|
|
|
6591 that you will typically encounter belong to the ASCII set of characters,
|
|
|
6592 but there are also characters for various sorts of accented letters,
|
|
|
6593 special symbols, Chinese and Japanese ideograms (i.e. Kanji, Katakana,
|
|
|
6594 etc.), Cyrillic and Greek letters, etc. The actual number of possible
|
|
|
6595 characters is quite large.
|
|
|
6596
|
|
|
6597 For now, we can view a character as some non-negative integer that
|
|
|
6598 has some shape that defines how it typically appears (e.g. as an
|
|
|
6599 uppercase A). (The exact way in which a character appears depends on the
|
|
|
6600 font used to display the character.) The internal type of characters in
|
|
|
6601 the C code is an @code{Emchar}; this is just an @code{int}, but using a
|
|
|
6602 symbolic type makes the code clearer.
|
|
|
6603
|
|
|
6604 Between every character in a buffer is a @dfn{buffer position} or
|
|
|
6605 @dfn{character position}. We can speak of the character before or after
|
|
|
6606 a particular buffer position, and when you insert a character at a
|
|
|
6607 particular position, all characters after that position end up at new
|
|
|
6608 positions. When we speak of the character @dfn{at} a position, we
|
|
|
6609 really mean the character after the position. (This schizophrenia
|
|
|
6610 between a buffer position being ``between'' a character and ``on'' a
|
|
|
6611 character is rampant in Emacs.)
|
|
|
6612
|
|
|
6613 Buffer positions are numbered starting at 1. This means that
|
|
|
6614 position 1 is before the first character, and position 0 is not
|
|
|
6615 valid. If there are N characters in a buffer, then buffer
|
|
|
6616 position N+1 is after the last one, and position N+2 is not valid.
|
|
|
6617
|
|
|
6618 The internal makeup of the Emchar integer varies depending on whether
|
|
|
6619 we have compiled with MULE support. If not, the Emchar integer is an
|
|
|
6620 8-bit integer with possible values from 0 - 255. 0 - 127 are the
|
|
|
6621 standard ASCII characters, while 128 - 255 are the characters from the
|
|
|
6622 ISO-8859-1 character set. If we have compiled with MULE support, an
|
|
|
6623 Emchar is a 19-bit integer, with the various bits having meanings
|
|
|
6624 according to a complex scheme that will be detailed later. The
|
|
|
6625 characters numbered 0 - 255 still have the same meanings as for the
|
|
|
6626 non-MULE case, though.
|
|
|
6627
|
|
|
6628 Internally, the text in a buffer is represented in a fairly simple
|
|
|
6629 fashion: as a contiguous array of bytes, with a @dfn{gap} of some size
|
|
|
6630 in the middle. Although the gap is of some substantial size in bytes,
|
|
|
6631 there is no text contained within it: From the perspective of the text
|
|
|
6632 in the buffer, it does not exist. The gap logically sits at some buffer
|
|
|
6633 position, between two characters (or possibly at the beginning or end of
|
|
|
6634 the buffer). Insertion of text in a buffer at a particular position is
|
|
|
6635 always accomplished by first moving the gap to that position
|
|
|
6636 (i.e. through some block moving of text), then writing the text into the
|
|
|
6637 beginning of the gap, thereby shrinking the gap. If the gap shrinks
|
|
|
6638 down to nothing, a new gap is created. (What actually happens is that a
|
|
|
6639 new gap is ``created'' at the end of the buffer's text, which requires
|
|
|
6640 nothing more than changing a couple of indices; then the gap is
|
|
|
6641 ``moved'' to the position where the insertion needs to take place by
|
|
|
6642 moving up in memory all the text after that position.) Similarly,
|
|
|
6643 deletion occurs by moving the gap to the place where the text is to be
|
|
|
6644 deleted, and then simply expanding the gap to include the deleted text.
|
|
|
6645 (@dfn{Expanding} and @dfn{shrinking} the gap as just described means
|
|
|
6646 just that the internal indices that keep track of where the gap is
|
|
|
6647 located are changed.)
|
|
|
6648
|
|
|
6649 Note that the total amount of memory allocated for a buffer text never
|
|
|
6650 decreases while the buffer is live. Therefore, if you load up a
|
|
|
6651 20-megabyte file and then delete all but one character, there will be a
|
|
|
6652 20-megabyte gap, which won't get any smaller (except by inserting
|
|
|
6653 characters back again). Once the buffer is killed, the memory allocated
|
|
|
6654 for the buffer text will be freed, but it will still be sitting on the
|
|
|
6655 heap, taking up virtual memory, and will not be released back to the
|
|
|
6656 operating system. (However, if you have compiled XEmacs with rel-alloc,
|
|
|
6657 the situation is different. In this case, the space @emph{will} be
|
|
|
6658 released back to the operating system. However, this tends to result in a
|
|
|
6659 noticeable speed penalty.)
|
|
|
6660
|
|
|
6661 Astute readers may notice that the text in a buffer is represented as
|
|
|
6662 an array of @emph{bytes}, while (at least in the MULE case) an Emchar is
|
|
|
6663 a 19-bit integer, which clearly cannot fit in a byte. This means (of
|
|
|
6664 course) that the text in a buffer uses a different representation from
|
|
|
6665 an Emchar: specifically, the 19-bit Emchar becomes a series of one to
|
|
|
6666 four bytes. The conversion between these two representations is complex
|
|
|
6667 and will be described later.
|
|
|
6668
|
|
|
6669 In the non-MULE case, everything is very simple: An Emchar
|
|
|
6670 is an 8-bit value, which fits neatly into one byte.
|
|
|
6671
|
|
|
6672 If we are given a buffer position and want to retrieve the
|
|
|
6673 character at that position, we need to follow these steps:
|
|
|
6674
|
|
|
6675 @enumerate
|
|
|
6676 @item
|
|
|
6677 Pretend there's no gap, and convert the buffer position into a @dfn{byte
|
|
|
6678 index} that indexes to the appropriate byte in the buffer's stream of
|
|
|
6679 textual bytes. By convention, byte indices begin at 1, just like buffer
|
|
|
6680 positions. In the non-MULE case, byte indices and buffer positions are
|
|
|
6681 identical, since one character equals one byte.
|
|
|
6682 @item
|
|
|
6683 Convert the byte index into a @dfn{memory index}, which takes the gap
|
|
|
6684 into account. The memory index is a direct index into the block of
|
|
|
6685 memory that stores the text of a buffer. This basically just involves
|
|
|
6686 checking to see if the byte index is past the gap, and if so, adding the
|
|
|
6687 size of the gap to it. By convention, memory indices begin at 1, just
|
|
|
6688 like buffer positions and byte indices, and when referring to the
|
|
|
6689 position that is @dfn{at} the gap, we always use the memory position at
|
|
|
6690 the @emph{beginning}, not at the end, of the gap.
|
|
|
6691 @item
|
|
|
6692 Fetch the appropriate bytes at the determined memory position.
|
|
|
6693 @item
|
|
|
6694 Convert these bytes into an Emchar.
|
|
|
6695 @end enumerate
|
|
|
6696
|
|
|
6697 In the non-Mule case, (3) and (4) boil down to a simple one-byte
|
|
|
6698 memory access.
|
|
|
6699
|
|
|
6700 Note that we have defined three types of positions in a buffer:
|
|
|
6701
|
|
|
6702 @enumerate
|
|
|
6703 @item
|
|
|
6704 @dfn{buffer positions} or @dfn{character positions}, typedef @code{Bufpos}
|
|
|
6705 @item
|
|
|
6706 @dfn{byte indices}, typedef @code{Bytind}
|
|
|
6707 @item
|
|
|
6708 @dfn{memory indices}, typedef @code{Memind}
|
|
|
6709 @end enumerate
|
|
|
6710
|
|
|
6711 All three typedefs are just @code{int}s, but defining them this way makes
|
|
|
6712 things a lot clearer.
|
|
|
6713
|
|
|
6714 Most code works with buffer positions. In particular, all Lisp code
|
|
|
6715 that refers to text in a buffer uses buffer positions. Lisp code does
|
|
|
6716 not know that byte indices or memory indices exist.
|
|
|
6717
|
|
|
6718 Finally, we have a typedef for the bytes in a buffer. This is a
|
|
|
6719 @code{Bufbyte}, which is an unsigned char. Referring to them as
|
|
|
6720 Bufbytes underscores the fact that we are working with a string of bytes
|
|
|
6721 in the internal Emacs buffer representation rather than in one of a
|
|
|
6722 number of possible alternative representations (e.g. EUC-encoded text,
|
|
|
6723 etc.).
|
|
|
6724
|
|
|
6725 @node Buffer Lists
|
|
|
6726 @section Buffer Lists
|
|
|
6727
|
|
|
6728 Recall earlier that buffers are @dfn{permanent} objects, i.e. that
|
|
|
6729 they remain around until explicitly deleted. This entails that there is
|
|
|
6730 a list of all the buffers in existence. This list is actually an
|
|
|
6731 assoc-list (mapping from the buffer's name to the buffer) and is stored
|
|
|
6732 in the global variable @code{Vbuffer_alist}.
|
|
|
6733
|
|
|
6734 The order of the buffers in the list is important: the buffers are
|
|
|
6735 ordered approximately from most-recently-used to least-recently-used.
|
|
|
6736 Switching to a buffer using @code{switch-to-buffer},
|
|
|
6737 @code{pop-to-buffer}, etc. and switching windows using
|
|
|
6738 @code{other-window}, etc. usually brings the new current buffer to the
|
|
|
6739 front of the list. @code{switch-to-buffer}, @code{other-buffer},
|
|
|
6740 etc. look at the beginning of the list to find an alternative buffer to
|
|
|
6741 suggest. You can also explicitly move a buffer to the end of the list
|
|
|
6742 using @code{bury-buffer}.
|
|
|
6743
|
|
|
6744 In addition to the global ordering in @code{Vbuffer_alist}, each frame
|
|
|
6745 has its own ordering of the list. These lists always contain the same
|
|
|
6746 elements as in @code{Vbuffer_alist} although possibly in a different
|
|
|
6747 order. @code{buffer-list} normally returns the list for the selected
|
|
|
6748 frame. This allows you to work in separate frames without things
|
|
|
6749 interfering with each other.
|
|
|
6750
|
|
|
6751 The standard way to look up a buffer given a name is
|
|
|
6752 @code{get-buffer}, and the standard way to create a new buffer is
|
|
|
6753 @code{get-buffer-create}, which looks up a buffer with a given name,
|
|
|
6754 creating a new one if necessary. These operations correspond exactly
|
|
|
6755 with the symbol operations @code{intern-soft} and @code{intern},
|
|
|
6756 respectively. You can also force a new buffer to be created using
|
|
|
6757 @code{generate-new-buffer}, which takes a name and (if necessary) makes
|
|
|
6758 a unique name from this by appending a number, and then creates the
|
|
|
6759 buffer. This is basically like the symbol operation @code{gensym}.
|
|
|
6760
|
|
|
6761 @node Markers and Extents
|
|
|
6762 @section Markers and Extents
|
|
|
6763
|
|
|
6764 Among the things associated with a buffer are things that are
|
|
|
6765 logically attached to certain buffer positions. This can be used to
|
|
|
6766 keep track of a buffer position when text is inserted and deleted, so
|
|
|
6767 that it remains at the same spot relative to the text around it; to
|
|
|
6768 assign properties to particular sections of text; etc. There are two
|
|
|
6769 such objects that are useful in this regard: they are @dfn{markers} and
|
|
|
6770 @dfn{extents}.
|
|
|
6771
|
|
|
6772 A @dfn{marker} is simply a flag placed at a particular buffer
|
|
|
6773 position, which is moved around as text is inserted and deleted.
|
|
|
6774 Markers are used for all sorts of purposes, such as the @code{mark} that
|
|
|
6775 is the other end of textual regions to be cut, copied, etc.
|
|
|
6776
|
|
|
6777 An @dfn{extent} is similar to two markers plus some associated
|
|
|
6778 properties, and is used to keep track of regions in a buffer as text is
|
|
|
6779 inserted and deleted, and to add properties (e.g. fonts) to particular
|
|
|
6780 regions of text. The external interface of extents is explained
|
|
|
6781 elsewhere.
|
|
|
6782
|
|
|
6783 The important thing here is that markers and extents simply contain
|
|
|
6784 buffer positions in them as integers, and every time text is inserted or
|
|
|
6785 deleted, these positions must be updated. In order to minimize the
|
|
|
6786 amount of shuffling that needs to be done, the positions in markers and
|
|
|
6787 extents (there's one per marker, two per extent) and stored in Meminds.
|
|
|
6788 This means that they only need to be moved when the text is physically
|
|
|
6789 moved in memory; since the gap structure tries to minimize this, it also
|
|
|
6790 minimizes the number of marker and extent indices that need to be
|
|
|
6791 adjusted. Look in @file{insdel.c} for the details of how this works.
|
|
|
6792
|
|
|
6793 One other important distinction is that markers are @dfn{temporary}
|
|
|
6794 while extents are @dfn{permanent}. This means that markers disappear as
|
|
|
6795 soon as there are no more pointers to them, and correspondingly, there
|
|
|
6796 is no way to determine what markers are in a buffer if you are just
|
|
|
6797 given the buffer. Extents remain in a buffer until they are detached
|
|
|
6798 (which could happen as a result of text being deleted) or the buffer is
|
|
|
6799 deleted, and primitives do exist to enumerate the extents in a buffer.
|
|
|
6800
|
|
|
6801 @node Bufbytes and Emchars
|
|
|
6802 @section Bufbytes and Emchars
|
|
|
6803
|
|
|
6804 Not yet documented.
|
|
|
6805
|
|
|
6806 @node The Buffer Object
|
|
|
6807 @section The Buffer Object
|
|
|
6808
|
|
|
6809 Buffers contain fields not directly accessible by the Lisp programmer.
|
|
|
6810 We describe them here, naming them by the names used in the C code.
|
|
|
6811 Many are accessible indirectly in Lisp programs via Lisp primitives.
|
|
|
6812
|
|
|
6813 @table @code
|
|
|
6814 @item name
|
|
|
6815 The buffer name is a string that names the buffer. It is guaranteed to
|
|
|
6816 be unique. @xref{Buffer Names,,, lispref, XEmacs Lisp Programmer's
|
|
|
6817 Manual}.
|
|
|
6818
|
|
|
6819 @item save_modified
|
|
|
6820 This field contains the time when the buffer was last saved, as an
|
|
|
6821 integer. @xref{Buffer Modification,,, lispref, XEmacs Lisp Programmer's
|
|
|
6822 Manual}.
|
|
|
6823
|
|
|
6824 @item modtime
|
|
|
6825 This field contains the modification time of the visited file. It is
|
|
|
6826 set when the file is written or read. Every time the buffer is written
|
|
|
6827 to the file, this field is compared to the modification time of the
|
|
|
6828 file. @xref{Buffer Modification,,, lispref, XEmacs Lisp Programmer's
|
|
|
6829 Manual}.
|
|
|
6830
|
|
|
6831 @item auto_save_modified
|
|
|
6832 This field contains the time when the buffer was last auto-saved.
|
|
|
6833
|
|
|
6834 @item last_window_start
|
|
|
6835 This field contains the @code{window-start} position in the buffer as of
|
|
|
6836 the last time the buffer was displayed in a window.
|
|
|
6837
|
|
|
6838 @item undo_list
|
|
|
6839 This field points to the buffer's undo list. @xref{Undo,,, lispref,
|
|
|
6840 XEmacs Lisp Programmer's Manual}.
|
|
|
6841
|
|
|
6842 @item syntax_table_v
|
|
|
6843 This field contains the syntax table for the buffer. @xref{Syntax
|
|
|
6844 Tables,,, lispref, XEmacs Lisp Programmer's Manual}.
|
|
|
6845
|
|
|
6846 @item downcase_table
|
|
|
6847 This field contains the conversion table for converting text to lower
|
|
|
6848 case. @xref{Case Tables,,, lispref, XEmacs Lisp Programmer's Manual}.
|
|
|
6849
|
|
|
6850 @item upcase_table
|
|
|
6851 This field contains the conversion table for converting text to upper
|
|
|
6852 case. @xref{Case Tables,,, lispref, XEmacs Lisp Programmer's Manual}.
|
|
|
6853
|
|
|
6854 @item case_canon_table
|
|
|
6855 This field contains the conversion table for canonicalizing text for
|
|
|
6856 case-folding search. @xref{Case Tables,,, lispref, XEmacs Lisp
|
|
|
6857 Programmer's Manual}.
|
|
|
6858
|
|
|
6859 @item case_eqv_table
|
|
|
6860 This field contains the equivalence table for case-folding search.
|
|
|
6861 @xref{Case Tables,,, lispref, XEmacs Lisp Programmer's Manual}.
|
|
|
6862
|
|
|
6863 @item display_table
|
|
|
6864 This field contains the buffer's display table, or @code{nil} if it
|
|
|
6865 doesn't have one. @xref{Display Tables,,, lispref, XEmacs Lisp
|
|
|
6866 Programmer's Manual}.
|
|
|
6867
|
|
|
6868 @item markers
|
|
|
6869 This field contains the chain of all markers that currently point into
|
|
|
6870 the buffer. Deletion of text in the buffer, and motion of the buffer's
|
|
|
6871 gap, must check each of these markers and perhaps update it.
|
|
|
6872 @xref{Markers,,, lispref, XEmacs Lisp Programmer's Manual}.
|
|
|
6873
|
|
|
6874 @item backed_up
|
|
|
6875 This field is a flag that tells whether a backup file has been made for
|
|
|
6876 the visited file of this buffer.
|
|
|
6877
|
|
|
6878 @item mark
|
|
|
6879 This field contains the mark for the buffer. The mark is a marker,
|
|
|
6880 hence it is also included on the list @code{markers}. @xref{The Mark,,,
|
|
|
6881 lispref, XEmacs Lisp Programmer's Manual}.
|
|
|
6882
|
|
|
6883 @item mark_active
|
|
|
6884 This field is non-@code{nil} if the buffer's mark is active.
|
|
|
6885
|
|
|
6886 @item local_var_alist
|
|
|
6887 This field contains the association list describing the variables local
|
|
|
6888 in this buffer, and their values, with the exception of local variables
|
|
|
6889 that have special slots in the buffer object. (Those slots are omitted
|
|
|
6890 from this table.) @xref{Buffer-Local Variables,,, lispref, XEmacs Lisp
|
|
|
6891 Programmer's Manual}.
|
|
|
6892
|
|
|
6893 @item modeline_format
|
|
|
6894 This field contains a Lisp object which controls how to display the mode
|
|
|
6895 line for this buffer. @xref{Modeline Format,,, lispref, XEmacs Lisp
|
|
|
6896 Programmer's Manual}.
|
|
|
6897
|
|
|
6898 @item base_buffer
|
|
|
6899 This field holds the buffer's base buffer (if it is an indirect buffer),
|
|
|
6900 or @code{nil}.
|
|
|
6901 @end table
|
|
|
6902
|
|
|
6903 @node MULE Character Sets and Encodings, The Lisp Reader and Compiler, Buffers and Textual Representation, Top
|
|
|
6904 @chapter MULE Character Sets and Encodings
|
|
|
6905
|
|
|
6906 Recall that there are two primary ways that text is represented in
|
|
|
6907 XEmacs. The @dfn{buffer} representation sees the text as a series of
|
|
|
6908 bytes (Bufbytes), with a variable number of bytes used per character.
|
|
|
6909 The @dfn{character} representation sees the text as a series of integers
|
|
|
6910 (Emchars), one per character. The character representation is a cleaner
|
|
|
6911 representation from a theoretical standpoint, and is thus used in many
|
|
|
6912 cases when lots of manipulations on a string need to be done. However,
|
|
|
6913 the buffer representation is the standard representation used in both
|
|
|
6914 Lisp strings and buffers, and because of this, it is the ``default''
|
|
|
6915 representation that text comes in. The reason for using this
|
|
|
6916 representation is that it's compact and is compatible with ASCII.
|
|
|
6917
|
|
|
6918 @menu
|
|
|
6919 * Character Sets::
|
|
|
6920 * Encodings::
|
|
|
6921 * Internal Mule Encodings::
|
|
|
6922 * CCL::
|
|
|
6923 @end menu
|
|
|
6924
|
|
|
6925 @node Character Sets
|
|
|
6926 @section Character Sets
|
|
|
6927
|
|
|
6928 A character set (or @dfn{charset}) is an ordered set of characters. A
|
|
|
6929 particular character in a charset is indexed using one or more
|
|
|
6930 @dfn{position codes}, which are non-negative integers. The number of
|
|
|
6931 position codes needed to identify a particular character in a charset is
|
|
|
6932 called the @dfn{dimension} of the charset. In XEmacs/Mule, all charsets
|
|
|
6933 have dimension 1 or 2, and the size of all charsets (except for a few
|
|
|
6934 special cases) is either 94, 96, 94 by 94, or 96 by 96. The range of
|
|
|
6935 position codes used to index characters from any of these types of
|
|
|
6936 character sets is as follows:
|
|
|
6937
|
|
|
6938 @example
|
|
|
6939 Charset type Position code 1 Position code 2
|
|
|
6940 ------------------------------------------------------------
|
|
|
6941 94 33 - 126 N/A
|
|
|
6942 96 32 - 127 N/A
|
|
|
6943 94x94 33 - 126 33 - 126
|
|
|
6944 96x96 32 - 127 32 - 127
|
|
|
6945 @end example
|
|
|
6946
|
|
|
6947 Note that in the above cases position codes do not start at an
|
|
|
6948 expected value such as 0 or 1. The reason for this will become clear
|
|
|
6949 later.
|
|
|
6950
|
|
|
6951 For example, Latin-1 is a 96-character charset, and JISX0208 (the
|
|
|
6952 Japanese national character set) is a 94x94-character charset.
|
|
|
6953
|
|
|
6954 [Note that, although the ranges above define the @emph{valid} position
|
|
|
6955 codes for a charset, some of the slots in a particular charset may in
|
|
|
6956 fact be empty. This is the case for JISX0208, for example, where (e.g.)
|
|
|
6957 all the slots whose first position code is in the range 118 - 127 are
|
|
|
6958 empty.]
|
|
|
6959
|
|
|
6960 There are three charsets that do not follow the above rules. All of
|
|
|
6961 them have one dimension, and have ranges of position codes as follows:
|
|
|
6962
|
|
|
6963 @example
|
|
|
6964 Charset name Position code 1
|
|
|
6965 ------------------------------------
|
|
|
6966 ASCII 0 - 127
|
|
|
6967 Control-1 0 - 31
|
|
|
6968 Composite 0 - some large number
|
|
|
6969 @end example
|
|
|
6970
|
|
|
6971 (The upper bound of the position code for composite characters has not
|
|
|
6972 yet been determined, but it will probably be at least 16,383).
|
|
|
6973
|
|
|
6974 ASCII is the union of two subsidiary character sets: Printing-ASCII
|
|
|
6975 (the printing ASCII character set, consisting of position codes 33 -
|
|
|
6976 126, like for a standard 94-character charset) and Control-ASCII (the
|
|
|
6977 non-printing characters that would appear in a binary file with codes 0
|
|
|
6978 - 32 and 127).
|
|
|
6979
|
|
|
6980 Control-1 contains the non-printing characters that would appear in a
|
|
|
6981 binary file with codes 128 - 159.
|
|
|
6982
|
|
|
6983 Composite contains characters that are generated by overstriking one
|
|
|
6984 or more characters from other charsets.
|
|
|
6985
|
|
|
6986 Note that some characters in ASCII, and all characters in Control-1,
|
|
|
6987 are @dfn{control} (non-printing) characters. These have no printed
|
|
|
6988 representation but instead control some other function of the printing
|
|
|
6989 (e.g. TAB or 8 moves the current character position to the next tab
|
|
|
6990 stop). All other characters in all charsets are @dfn{graphic}
|
|
|
6991 (printing) characters.
|
|
|
6992
|
|
|
6993 When a binary file is read in, the bytes in the file are assigned to
|
|
|
6994 character sets as follows:
|
|
|
6995
|
|
|
6996 @example
|
|
|
6997 Bytes Character set Range
|
|
|
6998 --------------------------------------------------
|
|
|
6999 0 - 127 ASCII 0 - 127
|
|
|
7000 128 - 159 Control-1 0 - 31
|
|
|
7001 160 - 255 Latin-1 32 - 127
|
|
|
7002 @end example
|
|
|
7003
|
|
|
7004 This is a bit ad-hoc but gets the job done.
|
|
|
7005
|
|
|
7006 @node Encodings
|
|
|
7007 @section Encodings
|
|
|
7008
|
|
|
7009 An @dfn{encoding} is a way of numerically representing characters from
|
|
|
7010 one or more character sets. If an encoding only encompasses one
|
|
|
7011 character set, then the position codes for the characters in that
|
|
|
7012 character set could be used directly. This is not possible, however, if
|
|
|
7013 more than one character set is to be used in the encoding.
|
|
|
7014
|
|
|
7015 For example, the conversion detailed above between bytes in a binary
|
|
|
7016 file and characters is effectively an encoding that encompasses the
|
|
|
7017 three character sets ASCII, Control-1, and Latin-1 in a stream of 8-bit
|
|
|
7018 bytes.
|
|
|
7019
|
|
|
7020 Thus, an encoding can be viewed as a way of encoding characters from a
|
|
|
7021 specified group of character sets using a stream of bytes, each of which
|
|
|
7022 contains a fixed number of bits (but not necessarily 8, as in the common
|
|
|
7023 usage of ``byte'').
|
|
|
7024
|
|
|
7025 Here are descriptions of a couple of common
|
|
|
7026 encodings:
|
|
|
7027
|
|
|
7028 @menu
|
|
|
7029 * Japanese EUC (Extended Unix Code)::
|
|
|
7030 * JIS7::
|
|
|
7031 @end menu
|
|
|
7032
|
|
|
7033 @node Japanese EUC (Extended Unix Code)
|
|
|
7034 @subsection Japanese EUC (Extended Unix Code)
|
|
|
7035
|
|
|
7036 This encompasses the character sets Printing-ASCII, Japanese-JISX0201,
|
|
|
7037 and Japanese-JISX0208-Kana (half-width katakana, the right half of
|
|
|
7038 JISX0201). It uses 8-bit bytes.
|
|
|
7039
|
|
|
7040 Note that Printing-ASCII and Japanese-JISX0201-Kana are 94-character
|
|
|
7041 charsets, while Japanese-JISX0208 is a 94x94-character charset.
|
|
|
7042
|
|
|
7043 The encoding is as follows:
|
|
|
7044
|
|
|
7045 @example
|
|
|
7046 Character set Representation (PC=position-code)
|
|
|
7047 ------------- --------------
|
|
|
7048 Printing-ASCII PC1
|
|
|
7049 Japanese-JISX0201-Kana 0x8E | PC1 + 0x80
|
|
|
7050 Japanese-JISX0208 PC1 + 0x80 | PC2 + 0x80
|
|
|
7051 Japanese-JISX0212 PC1 + 0x80 | PC2 + 0x80
|
|
|
7052 @end example
|
|
|
7053
|
|
|
7054
|
|
|
7055 @node JIS7
|
|
|
7056 @subsection JIS7
|
|
|
7057
|
|
|
7058 This encompasses the character sets Printing-ASCII,
|
|
|
7059 Japanese-JISX0201-Roman (the left half of JISX0201; this character set
|
|
|
7060 is very similar to Printing-ASCII and is a 94-character charset),
|
|
|
7061 Japanese-JISX0208, and Japanese-JISX0201-Kana. It uses 7-bit bytes.
|
|
|
7062
|
|
|
7063 Unlike Japanese EUC, this is a @dfn{modal} encoding, which
|
|
|
7064 means that there are multiple states that the encoding can
|
|
|
7065 be in, which affect how the bytes are to be interpreted.
|
|
|
7066 Special sequences of bytes (called @dfn{escape sequences})
|
|
|
7067 are used to change states.
|
|
|
7068
|
|
|
7069 The encoding is as follows:
|
|
|
7070
|
|
|
7071 @example
|
|
|
7072 Character set Representation (PC=position-code)
|
|
|
7073 ------------- --------------
|
|
|
7074 Printing-ASCII PC1
|
|
|
7075 Japanese-JISX0201-Roman PC1
|
|
|
7076 Japanese-JISX0201-Kana PC1
|
|
|
7077 Japanese-JISX0208 PC1 PC2
|
|
|
7078
|
|
|
7079
|
|
|
7080 Escape sequence ASCII equivalent Meaning
|
|
|
7081 --------------- ---------------- -------
|
|
|
7082 0x1B 0x28 0x4A ESC ( J invoke Japanese-JISX0201-Roman
|
|
|
7083 0x1B 0x28 0x49 ESC ( I invoke Japanese-JISX0201-Kana
|
|
|
7084 0x1B 0x24 0x42 ESC $ B invoke Japanese-JISX0208
|
|
|
7085 0x1B 0x28 0x42 ESC ( B invoke Printing-ASCII
|
|
|
7086 @end example
|
|
|
7087
|
|
|
7088 Initially, Printing-ASCII is invoked.
|
|
|
7089
|
|
|
7090 @node Internal Mule Encodings
|
|
|
7091 @section Internal Mule Encodings
|
|
|
7092
|
|
|
7093 In XEmacs/Mule, each character set is assigned a unique number, called a
|
|
|
7094 @dfn{leading byte}. This is used in the encodings of a character.
|
|
|
7095 Leading bytes are in the range 0x80 - 0xFF (except for ASCII, which has
|
|
|
7096 a leading byte of 0), although some leading bytes are reserved.
|
|
|
7097
|
|
|
7098 Charsets whose leading byte is in the range 0x80 - 0x9F are called
|
|
|
7099 @dfn{official} and are used for built-in charsets. Other charsets are
|
|
|
7100 called @dfn{private} and have leading bytes in the range 0xA0 - 0xFF;
|
|
|
7101 these are user-defined charsets.
|
|
|
7102
|
|
|
7103 More specifically:
|
|
|
7104
|
|
|
7105 @example
|
|
|
7106 Character set Leading byte
|
|
|
7107 ------------- ------------
|
|
|
7108 ASCII 0
|
|
|
7109 Composite 0x80
|
|
|
7110 Dimension-1 Official 0x81 - 0x8D
|
|
|
7111 (0x8E is free)
|
|
|
7112 Control-1 0x8F
|
|
|
7113 Dimension-2 Official 0x90 - 0x99
|
|
|
7114 (0x9A - 0x9D are free;
|
|
|
7115 0x9E and 0x9F are reserved)
|
|
|
7116 Dimension-1 Private 0xA0 - 0xEF
|
|
|
7117 Dimension-2 Private 0xF0 - 0xFF
|
|
|
7118 @end example
|
|
|
7119
|
|
|
7120 There are two internal encodings for characters in XEmacs/Mule. One is
|
|
|
7121 called @dfn{string encoding} and is an 8-bit encoding that is used for
|
|
|
7122 representing characters in a buffer or string. It uses 1 to 4 bytes per
|
|
|
7123 character. The other is called @dfn{character encoding} and is a 19-bit
|
|
|
7124 encoding that is used for representing characters individually in a
|
|
|
7125 variable.
|
|
|
7126
|
|
|
7127 (In the following descriptions, we'll ignore composite characters for
|
|
|
7128 the moment. We also give a general (structural) overview first,
|
|
|
7129 followed later by the exact details.)
|
|
|
7130
|
|
|
7131 @menu
|
|
|
7132 * Internal String Encoding::
|
|
|
7133 * Internal Character Encoding::
|
|
|
7134 @end menu
|
|
|
7135
|
|
|
7136 @node Internal String Encoding
|
|
|
7137 @subsection Internal String Encoding
|
|
|
7138
|
|
|
7139 ASCII characters are encoded using their position code directly. Other
|
|
|
7140 characters are encoded using their leading byte followed by their
|
|
|
7141 position code(s) with the high bit set. Characters in private character
|
|
|
7142 sets have their leading byte prefixed with a @dfn{leading byte prefix},
|
|
|
7143 which is either 0x9E or 0x9F. (No character sets are ever assigned these
|
|
|
7144 leading bytes.) Specifically:
|
|
|
7145
|
|
|
7146 @example
|
|
|
7147 Character set Encoding (PC=position-code, LB=leading-byte)
|
|
|
7148 ------------- --------
|
|
|
7149 ASCII PC-1 |
|
|
|
7150 Control-1 LB | PC1 + 0xA0 |
|
|
|
7151 Dimension-1 official LB | PC1 + 0x80 |
|
|
|
7152 Dimension-1 private 0x9E | LB | PC1 + 0x80 |
|
|
|
7153 Dimension-2 official LB | PC1 + 0x80 | PC2 + 0x80 |
|
|
|
7154 Dimension-2 private 0x9F | LB | PC1 + 0x80 | PC2 + 0x80
|
|
|
7155 @end example
|
|
|
7156
|
|
|
7157 The basic characteristic of this encoding is that the first byte
|
|
|
7158 of all characters is in the range 0x00 - 0x9F, and the second and
|
|
|
7159 following bytes of all characters is in the range 0xA0 - 0xFF.
|
|
|
7160 This means that it is impossible to get out of sync, or more
|
|
|
7161 specifically:
|
|
|
7162
|
|
|
7163 @enumerate
|
|
|
7164 @item
|
|
|
7165 Given any byte position, the beginning of the character it is
|
|
|
7166 within can be determined in constant time.
|
|
|
7167 @item
|
|
|
7168 Given any byte position at the beginning of a character, the
|
|
|
7169 beginning of the next character can be determined in constant
|
|
|
7170 time.
|
|
|
7171 @item
|
|
|
7172 Given any byte position at the beginning of a character, the
|
|
|
7173 beginning of the previous character can be determined in constant
|
|
|
7174 time.
|
|
|
7175 @item
|
|
|
7176 Textual searches can simply treat encoded strings as if they
|
|
|
7177 were encoded in a one-byte-per-character fashion rather than
|
|
|
7178 the actual multi-byte encoding.
|
|
|
7179 @end enumerate
|
|
|
7180
|
|
|
7181 None of the standard non-modal encodings meet all of these
|
|
|
7182 conditions. For example, EUC satisfies only (2) and (3), while
|
|
|
7183 Shift-JIS and Big5 (not yet described) satisfy only (2). (All
|
|
|
7184 non-modal encodings must satisfy (2), in order to be unambiguous.)
|
|
|
7185
|
|
|
7186 @node Internal Character Encoding
|
|
|
7187 @subsection Internal Character Encoding
|
|
|
7188
|
|
|
7189 One 19-bit word represents a single character. The word is
|
|
|
7190 separated into three fields:
|
|
|
7191
|
|
|
7192 @example
|
|
|
7193 Bit number: 18 17 16 15 14 13 12 11 10 09 08 07 06 05 04 03 02 01 00
|
|
|
7194 <------------> <------------------> <------------------>
|
|
|
7195 Field: 1 2 3
|
|
|
7196 @end example
|
|
|
7197
|
|
|
7198 Note that fields 2 and 3 hold 7 bits each, while field 1 holds 5 bits.
|
|
|
7199
|
|
|
7200 @example
|
|
|
7201 Character set Field 1 Field 2 Field 3
|
|
|
7202 ------------- ------- ------- -------
|
|
|
7203 ASCII 0 0 PC1
|
|
|
7204 range: (00 - 7F)
|
|
|
7205 Control-1 0 1 PC1
|
|
|
7206 range: (00 - 1F)
|
|
|
7207 Dimension-1 official 0 LB - 0x80 PC1
|
|
|
7208 range: (01 - 0D) (20 - 7F)
|
|
|
7209 Dimension-1 private 0 LB - 0x80 PC1
|
|
|
7210 range: (20 - 6F) (20 - 7F)
|
|
|
7211 Dimension-2 official LB - 0x8F PC1 PC2
|
|
|
7212 range: (01 - 0A) (20 - 7F) (20 - 7F)
|
|
|
7213 Dimension-2 private LB - 0xE1 PC1 PC2
|
|
|
7214 range: (0F - 1E) (20 - 7F) (20 - 7F)
|
|
|
7215 Composite 0x1F ? ?
|
|
|
7216 @end example
|
|
|
7217
|
|
|
7218 Note that character codes 0 - 255 are the same as the ``binary encoding''
|
|
|
7219 described above.
|
|
|
7220
|
|
|
7221 @node CCL
|
|
|
7222 @section CCL
|
|
|
7223
|
|
|
7224 @example
|
|
|
7225 CCL PROGRAM SYNTAX:
|
|
|
7226 CCL_PROGRAM := (CCL_MAIN_BLOCK
|
|
|
7227 [ CCL_EOF_BLOCK ])
|
|
|
7228
|
|
|
7229 CCL_MAIN_BLOCK := CCL_BLOCK
|
|
|
7230 CCL_EOF_BLOCK := CCL_BLOCK
|
|
|
7231
|
|
|
7232 CCL_BLOCK := STATEMENT | (STATEMENT [STATEMENT ...])
|
|
|
7233 STATEMENT :=
|
|
|
7234 SET | IF | BRANCH | LOOP | REPEAT | BREAK
|
|
|
7235 | READ | WRITE
|
|
|
7236
|
|
|
7237 SET := (REG = EXPRESSION) | (REG SELF_OP EXPRESSION)
|
|
|
7238 | INT-OR-CHAR
|
|
|
7239
|
|
|
7240 EXPRESSION := ARG | (EXPRESSION OP ARG)
|
|
|
7241
|
|
|
7242 IF := (if EXPRESSION CCL_BLOCK CCL_BLOCK)
|
|
|
7243 BRANCH := (branch EXPRESSION CCL_BLOCK [CCL_BLOCK ...])
|
|
|
7244 LOOP := (loop STATEMENT [STATEMENT ...])
|
|
|
7245 BREAK := (break)
|
|
|
7246 REPEAT := (repeat)
|
|
|
7247 | (write-repeat [REG | INT-OR-CHAR | string])
|
|
|
7248 | (write-read-repeat REG [INT-OR-CHAR | string | ARRAY]?)
|
|
|
7249 READ := (read REG) | (read REG REG)
|
|
|
7250 | (read-if REG ARITH_OP ARG CCL_BLOCK CCL_BLOCK)
|
|
|
7251 | (read-branch REG CCL_BLOCK [CCL_BLOCK ...])
|
|
|
7252 WRITE := (write REG) | (write REG REG)
|
|
|
7253 | (write INT-OR-CHAR) | (write STRING) | STRING
|
|
|
7254 | (write REG ARRAY)
|
|
|
7255 END := (end)
|
|
|
7256
|
|
|
7257 REG := r0 | r1 | r2 | r3 | r4 | r5 | r6 | r7
|
|
|
7258 ARG := REG | INT-OR-CHAR
|
|
|
7259 OP := + | - | * | / | % | & | '|' | ^ | << | >> | <8 | >8 | //
|
|
|
7260 | < | > | == | <= | >= | !=
|
|
|
7261 SELF_OP :=
|
|
|
7262 += | -= | *= | /= | %= | &= | '|=' | ^= | <<= | >>=
|
|
|
7263 ARRAY := '[' INT-OR-CHAR ... ']'
|
|
|
7264 INT-OR-CHAR := INT | CHAR
|
|
|
7265
|
|
|
7266 MACHINE CODE:
|
|
|
7267
|
|
|
7268 The machine code consists of a vector of 32-bit words.
|
|
|
7269 The first such word specifies the start of the EOF section of the code;
|
|
|
7270 this is the code executed to handle any stuff that needs to be done
|
|
|
7271 (e.g. designating back to ASCII and left-to-right mode) after all
|
|
|
7272 other encoded/decoded data has been written out. This is not used for
|
|
|
7273 charset CCL programs.
|
|
|
7274
|
|
|
7275 REGISTER: 0..7 -- refered by RRR or rrr
|
|
|
7276
|
|
|
7277 OPERATOR BIT FIELD (27-bit): XXXXXXXXXXXXXXX RRR TTTTT
|
|
|
7278 TTTTT (5-bit): operator type
|
|
|
7279 RRR (3-bit): register number
|
|
|
7280 XXXXXXXXXXXXXXXX (15-bit):
|
|
|
7281 CCCCCCCCCCCCCCC: constant or address
|
|
|
7282 000000000000rrr: register number
|
|
|
7283
|
|
|
7284 AAAA: 00000 +
|
|
|
7285 00001 -
|
|
|
7286 00010 *
|
|
|
7287 00011 /
|
|
|
7288 00100 %
|
|
|
7289 00101 &
|
|
|
7290 00110 |
|
|
|
7291 00111 ~
|
|
|
7292
|
|
|
7293 01000 <<
|
|
|
7294 01001 >>
|
|
|
7295 01010 <8
|
|
|
7296 01011 >8
|
|
|
7297 01100 //
|
|
|
7298 01101 not used
|
|
|
7299 01110 not used
|
|
|
7300 01111 not used
|
|
|
7301
|
|
|
7302 10000 <
|
|
|
7303 10001 >
|
|
|
7304 10010 ==
|
|
|
7305 10011 <=
|
|
|
7306 10100 >=
|
|
|
7307 10101 !=
|
|
|
7308
|
|
|
7309 OPERATORS: TTTTT RRR XX..
|
|
|
7310
|
|
|
7311 SetCS: 00000 RRR C...C RRR = C...C
|
|
|
7312 SetCL: 00001 RRR ..... RRR = c...c
|
|
|
7313 c.............c
|
|
|
7314 SetR: 00010 RRR ..rrr RRR = rrr
|
|
|
7315 SetA: 00011 RRR ..rrr RRR = array[rrr]
|
|
|
7316 C.............C size of array = C...C
|
|
|
7317 c.............c contents = c...c
|
|
|
7318
|
|
|
7319 Jump: 00100 000 c...c jump to c...c
|
|
|
7320 JumpCond: 00101 RRR c...c if (!RRR) jump to c...c
|
|
|
7321 WriteJump: 00110 RRR c...c Write1 RRR, jump to c...c
|
|
|
7322 WriteReadJump: 00111 RRR c...c Write1, Read1 RRR, jump to c...c
|
|
|
7323 WriteCJump: 01000 000 c...c Write1 C...C, jump to c...c
|
|
|
7324 C...C
|
|
|
7325 WriteCReadJump: 01001 RRR c...c Write1 C...C, Read1 RRR,
|
|
|
7326 C.............C and jump to c...c
|
|
|
7327 WriteSJump: 01010 000 c...c WriteS, jump to c...c
|
|
|
7328 C.............C
|
|
|
7329 S.............S
|
|
|
7330 ...
|
|
|
7331 WriteSReadJump: 01011 RRR c...c WriteS, Read1 RRR, jump to c...c
|
|
|
7332 C.............C
|
|
|
7333 S.............S
|
|
|
7334 ...
|
|
|
7335 WriteAReadJump: 01100 RRR c...c WriteA, Read1 RRR, jump to c...c
|
|
|
7336 C.............C size of array = C...C
|
|
|
7337 c.............c contents = c...c
|
|
|
7338 ...
|
|
|
7339 Branch: 01101 RRR C...C if (RRR >= 0 && RRR < C..)
|
|
|
7340 c.............c branch to (RRR+1)th address
|
|
|
7341 Read1: 01110 RRR ... read 1-byte to RRR
|
|
|
7342 Read2: 01111 RRR ..rrr read 2-byte to RRR and rrr
|
|
|
7343 ReadBranch: 10000 RRR C...C Read1 and Branch
|
|
|
7344 c.............c
|
|
|
7345 ...
|
|
|
7346 Write1: 10001 RRR ..... write 1-byte RRR
|
|
|
7347 Write2: 10010 RRR ..rrr write 2-byte RRR and rrr
|
|
|
7348 WriteC: 10011 000 ..... write 1-char C...CC
|
|
|
7349 C.............C
|
|
|
7350 WriteS: 10100 000 ..... write C..-byte of string
|
|
|
7351 C.............C
|
|
|
7352 S.............S
|
|
|
7353 ...
|
|
|
7354 WriteA: 10101 RRR ..... write array[RRR]
|
|
|
7355 C.............C size of array = C...C
|
|
|
7356 c.............c contents = c...c
|
|
|
7357 ...
|
|
|
7358 End: 10110 000 ..... terminate the execution
|
|
|
7359
|
|
|
7360 SetSelfCS: 10111 RRR C...C RRR AAAAA= C...C
|
|
|
7361 ..........AAAAA
|
|
|
7362 SetSelfCL: 11000 RRR ..... RRR AAAAA= c...c
|
|
|
7363 c.............c
|
|
|
7364 ..........AAAAA
|
|
|
7365 SetSelfR: 11001 RRR ..Rrr RRR AAAAA= rrr
|
|
|
7366 ..........AAAAA
|
|
|
7367 SetExprCL: 11010 RRR ..Rrr RRR = rrr AAAAA c...c
|
|
|
7368 c.............c
|
|
|
7369 ..........AAAAA
|
|
|
7370 SetExprR: 11011 RRR ..rrr RRR = rrr AAAAA Rrr
|
|
|
7371 ............Rrr
|
|
|
7372 ..........AAAAA
|
|
|
7373 JumpCondC: 11100 RRR c...c if !(RRR AAAAA C..) jump to c...c
|
|
|
7374 C.............C
|
|
|
7375 ..........AAAAA
|
|
|
7376 JumpCondR: 11101 RRR c...c if !(RRR AAAAA rrr) jump to c...c
|
|
|
7377 ............rrr
|
|
|
7378 ..........AAAAA
|
|
|
7379 ReadJumpCondC: 11110 RRR c...c Read1 and JumpCondC
|
|
|
7380 C.............C
|
|
|
7381 ..........AAAAA
|
|
|
7382 ReadJumpCondR: 11111 RRR c...c Read1 and JumpCondR
|
|
|
7383 ............rrr
|
|
|
7384 ..........AAAAA
|
|
|
7385 @end example
|
|
|
7386
|
|
|
7387 @node The Lisp Reader and Compiler, Lstreams, MULE Character Sets and Encodings, Top
|
|
|
7388 @chapter The Lisp Reader and Compiler
|
|
|
7389
|
|
|
7390 Not yet documented.
|
|
|
7391
|
|
|
7392 @node Lstreams, Consoles; Devices; Frames; Windows, The Lisp Reader and Compiler, Top
|
|
|
7393 @chapter Lstreams
|
|
|
7394
|
|
|
7395 An @dfn{lstream} is an internal Lisp object that provides a generic
|
|
|
7396 buffering stream implementation. Conceptually, you send data to the
|
|
|
7397 stream or read data from the stream, not caring what's on the other end
|
|
|
7398 of the stream. The other end could be another stream, a file
|
|
|
7399 descriptor, a stdio stream, a fixed block of memory, a reallocating
|
|
|
7400 block of memory, etc. The main purpose of the stream is to provide a
|
|
|
7401 standard interface and to do buffering. Macros are defined to read or
|
|
|
7402 write characters, so the calling functions do not have to worry about
|
|
|
7403 blocking data together in order to achieve efficiency.
|
|
|
7404
|
|
|
7405 @menu
|
|
|
7406 * Creating an Lstream:: Creating an lstream object.
|
|
|
7407 * Lstream Types:: Different sorts of things that are streamed.
|
|
|
7408 * Lstream Functions:: Functions for working with lstreams.
|
|
|
7409 * Lstream Methods:: Creating new lstream types.
|
|
|
7410 @end menu
|
|
|
7411
|
|
|
7412 @node Creating an Lstream
|
|
|
7413 @section Creating an Lstream
|
|
|
7414
|
|
|
7415 Lstreams come in different types, depending on what is being interfaced
|
|
|
7416 to. Although the primitive for creating new lstreams is
|
|
|
7417 @code{Lstream_new()}, generally you do not call this directly. Instead,
|
|
|
7418 you call some type-specific creation function, which creates the lstream
|
|
|
7419 and initializes it as appropriate for the particular type.
|
|
|
7420
|
|
|
7421 All lstream creation functions take a @var{mode} argument, specifying
|
|
|
7422 what mode the lstream should be opened as. This controls whether the
|
|
|
7423 lstream is for input and output, and optionally whether data should be
|
|
|
7424 blocked up in units of MULE characters. Note that some types of
|
|
|
7425 lstreams can only be opened for input; others only for output; and
|
|
|
7426 others can be opened either way. #### Richard Mlynarik thinks that
|
|
|
7427 there should be a strict separation between input and output streams,
|
|
|
7428 and he's probably right.
|
|
|
7429
|
|
|
7430 @var{mode} is a string, one of
|
|
|
7431
|
|
|
7432 @table @code
|
|
|
7433 @item "r"
|
|
|
7434 Open for reading.
|
|
|
7435 @item "w"
|
|
|
7436 Open for writing.
|
|
|
7437 @item "rc"
|
|
|
7438 Open for reading, but ``read'' never returns partial MULE characters.
|
|
|
7439 @item "wc"
|
|
|
7440 Open for writing, but never writes partial MULE characters.
|
|
|
7441 @end table
|
|
|
7442
|
|
|
7443 @node Lstream Types
|
|
|
7444 @section Lstream Types
|
|
|
7445
|
|
|
7446 @table @asis
|
|
|
7447 @item stdio
|
|
|
7448
|
|
|
7449 @item filedesc
|
|
|
7450
|
|
|
7451 @item lisp-string
|
|
|
7452
|
|
|
7453 @item fixed-buffer
|
|
|
7454
|
|
|
7455 @item resizing-buffer
|
|
|
7456
|
|
|
7457 @item dynarr
|
|
|
7458
|
|
|
7459 @item lisp-buffer
|
|
|
7460
|
|
|
7461 @item print
|
|
|
7462
|
|
|
7463 @item decoding
|
|
|
7464
|
|
|
7465 @item encoding
|
|
|
7466 @end table
|
|
|
7467
|
|
|
7468 @node Lstream Functions
|
|
|
7469 @section Lstream Functions
|
|
|
7470
|
|
|
7471 @deftypefun {Lstream *} Lstream_new (Lstream_implementation *@var{imp}, CONST char *@var{mode})
|
|
|
7472 Allocate and return a new Lstream. This function is not really meant to
|
|
|
7473 be called directly; rather, each stream type should provide its own
|
|
|
7474 stream creation function, which creates the stream and does any other
|
|
|
7475 necessary creation stuff (e.g. opening a file).
|
|
|
7476 @end deftypefun
|
|
|
7477
|
|
|
7478 @deftypefun void Lstream_set_buffering (Lstream *@var{lstr}, Lstream_buffering @var{buffering}, int @var{buffering_size})
|
|
|
7479 Change the buffering of a stream. See @file{lstream.h}. By default the
|
|
|
7480 buffering is @code{STREAM_BLOCK_BUFFERED}.
|
|
|
7481 @end deftypefun
|
|
|
7482
|
|
|
7483 @deftypefun int Lstream_flush (Lstream *@var{lstr})
|
|
|
7484 Flush out any pending unwritten data in the stream. Clear any buffered
|
|
|
7485 input data. Returns 0 on success, -1 on error.
|
|
|
7486 @end deftypefun
|
|
|
7487
|
|
|
7488 @deftypefn Macro int Lstream_putc (Lstream *@var{stream}, int @var{c})
|
|
|
7489 Write out one byte to the stream. This is a macro and so it is very
|
|
|
7490 efficient. The @var{c} argument is only evaluated once but the @var{stream}
|
|
|
7491 argument is evaluated more than once. Returns 0 on success, -1 on
|
|
|
7492 error.
|
|
|
7493 @end deftypefn
|
|
|
7494
|
|
|
7495 @deftypefn Macro int Lstream_getc (Lstream *@var{stream})
|
|
|
7496 Read one byte from the stream. This is a macro and so it is very
|
|
|
7497 efficient. The @var{stream} argument is evaluated more than once. Return
|
|
|
7498 value is -1 for EOF or error.
|
|
|
7499 @end deftypefn
|
|
|
7500
|
|
|
7501 @deftypefn Macro void Lstream_ungetc (Lstream *@var{stream}, int @var{c})
|
|
|
7502 Push one byte back onto the input queue. This will be the next byte
|
|
|
7503 read from the stream. Any number of bytes can be pushed back and will
|
|
440
|
7504 be read in the reverse order they were pushed back---most recent
|
|
|
7505 first. (This is necessary for consistency---if there are a number of
|
|
428
|
7506 bytes that have been unread and I read and unread a byte, it needs to be
|
|
|
7507 the first to be read again.) This is a macro and so it is very
|
|
|
7508 efficient. The @var{c} argument is only evaluated once but the @var{stream}
|
|
|
7509 argument is evaluated more than once.
|
|
|
7510 @end deftypefn
|
|
|
7511
|
|
|
7512 @deftypefun int Lstream_fputc (Lstream *@var{stream}, int @var{c})
|
|
|
7513 @deftypefunx int Lstream_fgetc (Lstream *@var{stream})
|
|
|
7514 @deftypefunx void Lstream_fungetc (Lstream *@var{stream}, int @var{c})
|
|
|
7515 Function equivalents of the above macros.
|
|
|
7516 @end deftypefun
|
|
|
7517
|
|
|
7518 @deftypefun ssize_t Lstream_read (Lstream *@var{stream}, void *@var{data}, size_t @var{size})
|
|
|
7519 Read @var{size} bytes of @var{data} from the stream. Return the number
|
|
|
7520 of bytes read. 0 means EOF. -1 means an error occurred and no bytes
|
|
|
7521 were read.
|
|
|
7522 @end deftypefun
|
|
|
7523
|
|
|
7524 @deftypefun ssize_t Lstream_write (Lstream *@var{stream}, void *@var{data}, size_t @var{size})
|
|
|
7525 Write @var{size} bytes of @var{data} to the stream. Return the number
|
|
|
7526 of bytes written. -1 means an error occurred and no bytes were written.
|
|
|
7527 @end deftypefun
|
|
|
7528
|
|
|
7529 @deftypefun void Lstream_unread (Lstream *@var{stream}, void *@var{data}, size_t @var{size})
|
|
|
7530 Push back @var{size} bytes of @var{data} onto the input queue. The next
|
|
|
7531 call to @code{Lstream_read()} with the same size will read the same
|
|
|
7532 bytes back. Note that this will be the case even if there is other
|
|
|
7533 pending unread data.
|
|
|
7534 @end deftypefun
|
|
|
7535
|
|
|
7536 @deftypefun int Lstream_close (Lstream *@var{stream})
|
|
|
7537 Close the stream. All data will be flushed out.
|
|
|
7538 @end deftypefun
|
|
|
7539
|
|
|
7540 @deftypefun void Lstream_reopen (Lstream *@var{stream})
|
|
|
7541 Reopen a closed stream. This enables I/O on it again. This is not
|
|
|
7542 meant to be called except from a wrapper routine that reinitializes
|
|
440
|
7543 variables and such---the close routine may well have freed some
|
|
428
|
7544 necessary storage structures, for example.
|
|
|
7545 @end deftypefun
|
|
|
7546
|
|
|
7547 @deftypefun void Lstream_rewind (Lstream *@var{stream})
|
|
|
7548 Rewind the stream to the beginning.
|
|
|
7549 @end deftypefun
|
|
|
7550
|
|
|
7551 @node Lstream Methods
|
|
|
7552 @section Lstream Methods
|
|
|
7553
|
|
|
7554 @deftypefn {Lstream Method} ssize_t reader (Lstream *@var{stream}, unsigned char *@var{data}, size_t @var{size})
|
|
|
7555 Read some data from the stream's end and store it into @var{data}, which
|
|
|
7556 can hold @var{size} bytes. Return the number of bytes read. A return
|
|
|
7557 value of 0 means no bytes can be read at this time. This may be because
|
|
|
7558 of an EOF, or because there is a granularity greater than one byte that
|
|
|
7559 the stream imposes on the returned data, and @var{size} is less than
|
|
|
7560 this granularity. (This will happen frequently for streams that need to
|
|
|
7561 return whole characters, because @code{Lstream_read()} calls the reader
|
|
|
7562 function repeatedly until it has the number of bytes it wants or until 0
|
|
|
7563 is returned.) The lstream functions do not treat a 0 return as EOF or
|
|
|
7564 do anything special; however, the calling function will interpret any 0
|
|
|
7565 it gets back as EOF. This will normally not happen unless the caller
|
|
|
7566 calls @code{Lstream_read()} with a very small size.
|
|
|
7567
|
|
|
7568 This function can be @code{NULL} if the stream is output-only.
|
|
|
7569 @end deftypefn
|
|
|
7570
|
|
|
7571 @deftypefn {Lstream Method} ssize_t writer (Lstream *@var{stream}, CONST unsigned char *@var{data}, size_t @var{size})
|
|
|
7572 Send some data to the stream's end. Data to be sent is in @var{data}
|
|
|
7573 and is @var{size} bytes. Return the number of bytes sent. This
|
|
|
7574 function can send and return fewer bytes than is passed in; in that
|
|
|
7575 case, the function will just be called again until there is no data left
|
|
|
7576 or 0 is returned. A return value of 0 means that no more data can be
|
|
|
7577 currently stored, but there is no error; the data will be squirreled
|
|
|
7578 away until the writer can accept data. (This is useful, e.g., if you're
|
|
|
7579 dealing with a non-blocking file descriptor and are getting
|
|
|
7580 @code{EWOULDBLOCK} errors.) This function can be @code{NULL} if the
|
|
|
7581 stream is input-only.
|
|
|
7582 @end deftypefn
|
|
|
7583
|
|
|
7584 @deftypefn {Lstream Method} int rewinder (Lstream *@var{stream})
|
|
|
7585 Rewind the stream. If this is @code{NULL}, the stream is not seekable.
|
|
|
7586 @end deftypefn
|
|
|
7587
|
|
|
7588 @deftypefn {Lstream Method} int seekable_p (Lstream *@var{stream})
|
|
440
|
7589 Indicate whether this stream is seekable---i.e. it can be rewound.
|
|
428
|
7590 This method is ignored if the stream does not have a rewind method. If
|
|
|
7591 this method is not present, the result is determined by whether a rewind
|
|
|
7592 method is present.
|
|
|
7593 @end deftypefn
|
|
|
7594
|
|
|
7595 @deftypefn {Lstream Method} int flusher (Lstream *@var{stream})
|
|
|
7596 Perform any additional operations necessary to flush the data in this
|
|
|
7597 stream.
|
|
|
7598 @end deftypefn
|
|
|
7599
|
|
|
7600 @deftypefn {Lstream Method} int pseudo_closer (Lstream *@var{stream})
|
|
|
7601 @end deftypefn
|
|
|
7602
|
|
|
7603 @deftypefn {Lstream Method} int closer (Lstream *@var{stream})
|
|
|
7604 Perform any additional operations necessary to close this stream down.
|
|
|
7605 May be @code{NULL}. This function is called when @code{Lstream_close()}
|
|
|
7606 is called or when the stream is garbage-collected. When this function
|
|
|
7607 is called, all pending data in the stream will already have been written
|
|
|
7608 out.
|
|
|
7609 @end deftypefn
|
|
|
7610
|
|
|
7611 @deftypefn {Lstream Method} Lisp_Object marker (Lisp_Object @var{lstream}, void (*@var{markfun}) (Lisp_Object))
|
|
|
7612 Mark this object for garbage collection. Same semantics as a standard
|
|
|
7613 @code{Lisp_Object} marker. This function can be @code{NULL}.
|
|
|
7614 @end deftypefn
|
|
|
7615
|
|
|
7616 @node Consoles; Devices; Frames; Windows, The Redisplay Mechanism, Lstreams, Top
|
|
|
7617 @chapter Consoles; Devices; Frames; Windows
|
|
|
7618
|
|
|
7619 @menu
|
|
|
7620 * Introduction to Consoles; Devices; Frames; Windows::
|
|
|
7621 * Point::
|
|
|
7622 * Window Hierarchy::
|
|
|
7623 * The Window Object::
|
|
|
7624 @end menu
|
|
|
7625
|
|
|
7626 @node Introduction to Consoles; Devices; Frames; Windows
|
|
|
7627 @section Introduction to Consoles; Devices; Frames; Windows
|
|
|
7628
|
|
|
7629 A window-system window that you see on the screen is called a
|
|
|
7630 @dfn{frame} in Emacs terminology. Each frame is subdivided into one or
|
|
|
7631 more non-overlapping panes, called (confusingly) @dfn{windows}. Each
|
|
|
7632 window displays the text of a buffer in it. (See above on Buffers.) Note
|
|
|
7633 that buffers and windows are independent entities: Two or more windows
|
|
|
7634 can be displaying the same buffer (potentially in different locations),
|
|
|
7635 and a buffer can be displayed in no windows.
|
|
|
7636
|
|
|
7637 A single display screen that contains one or more frames is called
|
|
|
7638 a @dfn{display}. Under most circumstances, there is only one display.
|
|
|
7639 However, more than one display can exist, for example if you have
|
|
|
7640 a @dfn{multi-headed} console, i.e. one with a single keyboard but
|
|
|
7641 multiple displays. (Typically in such a situation, the various
|
|
|
7642 displays act like one large display, in that the mouse is only
|
|
|
7643 in one of them at a time, and moving the mouse off of one moves
|
|
|
7644 it into another.) In some cases, the different displays will
|
|
|
7645 have different characteristics, e.g. one color and one mono.
|
|
|
7646
|
|
|
7647 XEmacs can display frames on multiple displays. It can even deal
|
|
|
7648 simultaneously with frames on multiple keyboards (called @dfn{consoles} in
|
|
|
7649 XEmacs terminology). Here is one case where this might be useful: You
|
|
|
7650 are using XEmacs on your workstation at work, and leave it running.
|
|
|
7651 Then you go home and dial in on a TTY line, and you can use the
|
|
|
7652 already-running XEmacs process to display another frame on your local
|
|
|
7653 TTY.
|
|
|
7654
|
|
|
7655 Thus, there is a hierarchy console -> display -> frame -> window.
|
|
|
7656 There is a separate Lisp object type for each of these four concepts.
|
|
|
7657 Furthermore, there is logically a @dfn{selected console},
|
|
|
7658 @dfn{selected display}, @dfn{selected frame}, and @dfn{selected window}.
|
|
|
7659 Each of these objects is distinguished in various ways, such as being the
|
|
|
7660 default object for various functions that act on objects of that type.
|
|
|
7661 Note that every containing object rememembers the ``selected'' object
|
|
|
7662 among the objects that it contains: e.g. not only is there a selected
|
|
|
7663 window, but every frame remembers the last window in it that was
|
|
|
7664 selected, and changing the selected frame causes the remembered window
|
|
|
7665 within it to become the selected window. Similar relationships apply
|
|
|
7666 for consoles to devices and devices to frames.
|
|
|
7667
|
|
|
7668 @node Point
|
|
|
7669 @section Point
|
|
|
7670
|
|
|
7671 Recall that every buffer has a current insertion position, called
|
|
|
7672 @dfn{point}. Now, two or more windows may be displaying the same buffer,
|
|
|
7673 and the text cursor in the two windows (i.e. @code{point}) can be in
|
|
|
7674 two different places. You may ask, how can that be, since each
|
|
|
7675 buffer has only one value of @code{point}? The answer is that each window
|
|
|
7676 also has a value of @code{point} that is squirreled away in it. There
|
|
|
7677 is only one selected window, and the value of ``point'' in that buffer
|
|
|
7678 corresponds to that window. When the selected window is changed
|
|
|
7679 from one window to another displaying the same buffer, the old
|
|
|
7680 value of @code{point} is stored into the old window's ``point'' and the
|
|
|
7681 value of @code{point} from the new window is retrieved and made the
|
|
|
7682 value of @code{point} in the buffer. This means that @code{window-point}
|
|
|
7683 for the selected window is potentially inaccurate, and if you
|
|
|
7684 want to retrieve the correct value of @code{point} for a window,
|
|
|
7685 you must special-case on the selected window and retrieve the
|
|
|
7686 buffer's point instead. This is related to why @code{save-window-excursion}
|
|
|
7687 does not save the selected window's value of @code{point}.
|
|
|
7688
|
|
|
7689 @node Window Hierarchy
|
|
|
7690 @section Window Hierarchy
|
|
|
7691 @cindex window hierarchy
|
|
|
7692 @cindex hierarchy of windows
|
|
|
7693
|
|
|
7694 If a frame contains multiple windows (panes), they are always created
|
|
|
7695 by splitting an existing window along the horizontal or vertical axis.
|
|
|
7696 Terminology is a bit confusing here: to @dfn{split a window
|
|
|
7697 horizontally} means to create two side-by-side windows, i.e. to make a
|
|
|
7698 @emph{vertical} cut in a window. Likewise, to @dfn{split a window
|
|
|
7699 vertically} means to create two windows, one above the other, by making
|
|
|
7700 a @emph{horizontal} cut.
|
|
|
7701
|
|
|
7702 If you split a window and then split again along the same axis, you
|
|
|
7703 will end up with a number of panes all arranged along the same axis.
|
|
|
7704 The precise way in which the splits were made should not be important,
|
|
|
7705 and this is reflected internally. Internally, all windows are arranged
|
|
|
7706 in a tree, consisting of two types of windows, @dfn{combination} windows
|
|
|
7707 (which have children, and are covered completely by those children) and
|
|
|
7708 @dfn{leaf} windows, which have no children and are visible. Every
|
|
|
7709 combination window has two or more children, all arranged along the same
|
|
|
7710 axis. There are (logically) two subtypes of windows, depending on
|
|
|
7711 whether their children are horizontally or vertically arrayed. There is
|
|
|
7712 always one root window, which is either a leaf window (if the frame
|
|
|
7713 contains only one window) or a combination window (if the frame contains
|
|
|
7714 more than one window). In the latter case, the root window will have
|
|
|
7715 two or more children, either horizontally or vertically arrayed, and
|
|
|
7716 each of those children will be either a leaf window or another
|
|
|
7717 combination window.
|
|
|
7718
|
|
|
7719 Here are some rules:
|
|
|
7720
|
|
|
7721 @enumerate
|
|
|
7722 @item
|
|
|
7723 Horizontal combination windows can never have children that are
|
|
|
7724 horizontal combination windows; same for vertical.
|
|
|
7725
|
|
|
7726 @item
|
|
|
7727 Only leaf windows can be split (obviously) and this splitting does one
|
|
|
7728 of two things: (a) turns the leaf window into a combination window and
|
|
|
7729 creates two new leaf children, or (b) turns the leaf window into one of
|
|
|
7730 the two new leaves and creates the other leaf. Rule (1) dictates which
|
|
|
7731 of these two outcomes happens.
|
|
|
7732
|
|
|
7733 @item
|
|
|
7734 Every combination window must have at least two children.
|
|
|
7735
|
|
|
7736 @item
|
|
|
7737 Leaf windows can never become combination windows. They can be deleted,
|
|
|
7738 however. If this results in a violation of (3), the parent combination
|
|
|
7739 window also gets deleted.
|
|
|
7740
|
|
|
7741 @item
|
|
|
7742 All functions that accept windows must be prepared to accept combination
|
|
|
7743 windows, and do something sane (e.g. signal an error if so).
|
|
|
7744 Combination windows @emph{do} escape to the Lisp level.
|
|
|
7745
|
|
|
7746 @item
|
|
|
7747 All windows have three fields governing their contents:
|
|
|
7748 these are @dfn{hchild} (a list of horizontally-arrayed children),
|
|
|
7749 @dfn{vchild} (a list of vertically-arrayed children), and @dfn{buffer}
|
|
|
7750 (the buffer contained in a leaf window). Exactly one of
|
|
|
7751 these will be non-nil. Remember that @dfn{horizontally-arrayed}
|
|
|
7752 means ``side-by-side'' and @dfn{vertically-arrayed} means
|
|
|
7753 @dfn{one above the other}.
|
|
|
7754
|
|
|
7755 @item
|
|
|
7756 Leaf windows also have markers in their @code{start} (the
|
|
|
7757 first buffer position displayed in the window) and @code{pointm}
|
|
440
|
7758 (the window's stashed value of @code{point}---see above) fields,
|
|
428
|
7759 while combination windows have nil in these fields.
|
|
|
7760
|
|
|
7761 @item
|
|
|
7762 The list of children for a window is threaded through the
|
|
|
7763 @code{next} and @code{prev} fields of each child window.
|
|
|
7764
|
|
|
7765 @item
|
|
|
7766 @strong{Deleted windows can be undeleted}. This happens as a result of
|
|
|
7767 restoring a window configuration, and is unlike frames, displays, and
|
|
|
7768 consoles, which, once deleted, can never be restored. Deleting a window
|
|
|
7769 does nothing except set a special @code{dead} bit to 1 and clear out the
|
|
|
7770 @code{next}, @code{prev}, @code{hchild}, and @code{vchild} fields, for
|
|
|
7771 GC purposes.
|
|
|
7772
|
|
|
7773 @item
|
|
440
|
7774 Most frames actually have two top-level windows---one for the
|
|
428
|
7775 minibuffer and one (the @dfn{root}) for everything else. The modeline
|
|
|
7776 (if present) separates these two. The @code{next} field of the root
|
|
|
7777 points to the minibuffer, and the @code{prev} field of the minibuffer
|
|
|
7778 points to the root. The other @code{next} and @code{prev} fields are
|
|
|
7779 @code{nil}, and the frame points to both of these windows.
|
|
|
7780 Minibuffer-less frames have no minibuffer window, and the @code{next}
|
|
|
7781 and @code{prev} of the root window are @code{nil}. Minibuffer-only
|
|
|
7782 frames have no root window, and the @code{next} of the minibuffer window
|
|
|
7783 is @code{nil} but the @code{prev} points to itself. (#### This is an
|
|
|
7784 artifact that should be fixed.)
|
|
|
7785 @end enumerate
|
|
|
7786
|
|
|
7787 @node The Window Object
|
|
|
7788 @section The Window Object
|
|
|
7789
|
|
|
7790 Windows have the following accessible fields:
|
|
|
7791
|
|
|
7792 @table @code
|
|
|
7793 @item frame
|
|
|
7794 The frame that this window is on.
|
|
|
7795
|
|
|
7796 @item mini_p
|
|
|
7797 Non-@code{nil} if this window is a minibuffer window.
|
|
|
7798
|
|
|
7799 @item buffer
|
|
|
7800 The buffer that the window is displaying. This may change often during
|
|
|
7801 the life of the window.
|
|
|
7802
|
|
|
7803 @item dedicated
|
|
|
7804 Non-@code{nil} if this window is dedicated to its buffer.
|
|
|
7805
|
|
|
7806 @item pointm
|
|
|
7807 @cindex window point internals
|
|
|
7808 This is the value of point in the current buffer when this window is
|
|
|
7809 selected; when it is not selected, it retains its previous value.
|
|
|
7810
|
|
|
7811 @item start
|
|
|
7812 The position in the buffer that is the first character to be displayed
|
|
|
7813 in the window.
|
|
|
7814
|
|
|
7815 @item force_start
|
|
|
7816 If this flag is non-@code{nil}, it says that the window has been
|
|
|
7817 scrolled explicitly by the Lisp program. This affects what the next
|
|
|
7818 redisplay does if point is off the screen: instead of scrolling the
|
|
|
7819 window to show the text around point, it moves point to a location that
|
|
|
7820 is on the screen.
|
|
|
7821
|
|
|
7822 @item last_modified
|
|
|
7823 The @code{modified} field of the window's buffer, as of the last time
|
|
|
7824 a redisplay completed in this window.
|
|
|
7825
|
|
|
7826 @item last_point
|
|
|
7827 The buffer's value of point, as of the last time
|
|
|
7828 a redisplay completed in this window.
|
|
|
7829
|
|
|
7830 @item left
|
|
|
7831 This is the left-hand edge of the window, measured in columns. (The
|
|
|
7832 leftmost column on the screen is @w{column 0}.)
|
|
|
7833
|
|
|
7834 @item top
|
|
|
7835 This is the top edge of the window, measured in lines. (The top line on
|
|
|
7836 the screen is @w{line 0}.)
|
|
|
7837
|
|
|
7838 @item height
|
|
|
7839 The height of the window, measured in lines.
|
|
|
7840
|
|
|
7841 @item width
|
|
|
7842 The width of the window, measured in columns.
|
|
|
7843
|
|
|
7844 @item next
|
|
|
7845 This is the window that is the next in the chain of siblings. It is
|
|
|
7846 @code{nil} in a window that is the rightmost or bottommost of a group of
|
|
|
7847 siblings.
|
|
|
7848
|
|
|
7849 @item prev
|
|
|
7850 This is the window that is the previous in the chain of siblings. It is
|
|
|
7851 @code{nil} in a window that is the leftmost or topmost of a group of
|
|
|
7852 siblings.
|
|
|
7853
|
|
|
7854 @item parent
|
|
|
7855 Internally, XEmacs arranges windows in a tree; each group of siblings has
|
|
|
7856 a parent window whose area includes all the siblings. This field points
|
|
|
7857 to a window's parent.
|
|
|
7858
|
|
|
7859 Parent windows do not display buffers, and play little role in display
|
|
|
7860 except to shape their child windows. Emacs Lisp programs usually have
|
|
|
7861 no access to the parent windows; they operate on the windows at the
|
|
|
7862 leaves of the tree, which actually display buffers.
|
|
|
7863
|
|
|
7864 @item hscroll
|
|
|
7865 This is the number of columns that the display in the window is scrolled
|
|
|
7866 horizontally to the left. Normally, this is 0.
|
|
|
7867
|
|
|
7868 @item use_time
|
|
|
7869 This is the last time that the window was selected. The function
|
|
|
7870 @code{get-lru-window} uses this field.
|
|
|
7871
|
|
|
7872 @item display_table
|
|
|
7873 The window's display table, or @code{nil} if none is specified for it.
|
|
|
7874
|
|
|
7875 @item update_mode_line
|
|
|
7876 Non-@code{nil} means this window's mode line needs to be updated.
|
|
|
7877
|
|
|
7878 @item base_line_number
|
|
|
7879 The line number of a certain position in the buffer, or @code{nil}.
|
|
|
7880 This is used for displaying the line number of point in the mode line.
|
|
|
7881
|
|
|
7882 @item base_line_pos
|
|
|
7883 The position in the buffer for which the line number is known, or
|
|
|
7884 @code{nil} meaning none is known.
|
|
|
7885
|
|
|
7886 @item region_showing
|
|
|
7887 If the region (or part of it) is highlighted in this window, this field
|
|
|
7888 holds the mark position that made one end of that region. Otherwise,
|
|
|
7889 this field is @code{nil}.
|
|
|
7890 @end table
|
|
|
7891
|
|
|
7892 @node The Redisplay Mechanism, Extents, Consoles; Devices; Frames; Windows, Top
|
|
|
7893 @chapter The Redisplay Mechanism
|
|
|
7894
|
|
|
7895 The redisplay mechanism is one of the most complicated sections of
|
|
|
7896 XEmacs, especially from a conceptual standpoint. This is doubly so
|
|
|
7897 because, unlike for the basic aspects of the Lisp interpreter, the
|
|
|
7898 computer science theories of how to efficiently handle redisplay are not
|
|
|
7899 well-developed.
|
|
|
7900
|
|
|
7901 When working with the redisplay mechanism, remember the Golden Rules
|
|
|
7902 of Redisplay:
|
|
|
7903
|
|
|
7904 @enumerate
|
|
|
7905 @item
|
|
|
7906 It Is Better To Be Correct Than Fast.
|
|
|
7907 @item
|
|
|
7908 Thou Shalt Not Run Elisp From Within Redisplay.
|
|
|
7909 @item
|
|
|
7910 It Is Better To Be Fast Than Not To Be.
|
|
|
7911 @end enumerate
|
|
|
7912
|
|
|
7913 @menu
|
|
|
7914 * Critical Redisplay Sections::
|
|
|
7915 * Line Start Cache::
|
|
|
7916 * Redisplay Piece by Piece::
|
|
|
7917 @end menu
|
|
|
7918
|
|
|
7919 @node Critical Redisplay Sections
|
|
|
7920 @section Critical Redisplay Sections
|
|
|
7921 @cindex critical redisplay sections
|
|
|
7922
|
|
|
7923 Within this section, we are defenseless and assume that the
|
|
|
7924 following cannot happen:
|
|
|
7925
|
|
|
7926 @enumerate
|
|
|
7927 @item
|
|
|
7928 garbage collection
|
|
|
7929 @item
|
|
|
7930 Lisp code evaluation
|
|
|
7931 @item
|
|
|
7932 frame size changes
|
|
|
7933 @end enumerate
|
|
|
7934
|
|
|
7935 We ensure (3) by calling @code{hold_frame_size_changes()}, which
|
|
|
7936 will cause any pending frame size changes to get put on hold
|
|
|
7937 till after the end of the critical section. (1) follows
|
|
|
7938 automatically if (2) is met. #### Unfortunately, there are
|
|
|
7939 some places where Lisp code can be called within this section.
|
|
|
7940 We need to remove them.
|
|
|
7941
|
|
|
7942 If @code{Fsignal()} is called during this critical section, we
|
|
|
7943 will @code{abort()}.
|
|
|
7944
|
|
|
7945 If garbage collection is called during this critical section,
|
|
|
7946 we simply return. #### We should abort instead.
|
|
|
7947
|
|
|
7948 #### If a frame-size change does occur we should probably
|
|
|
7949 actually be preempting redisplay.
|
|
|
7950
|
|
|
7951 @node Line Start Cache
|
|
|
7952 @section Line Start Cache
|
|
|
7953 @cindex line start cache
|
|
|
7954
|
|
|
7955 The traditional scrolling code in Emacs breaks in a variable height
|
|
|
7956 world. It depends on the key assumption that the number of lines that
|
|
|
7957 can be displayed at any given time is fixed. This led to a complete
|
|
|
7958 separation of the scrolling code from the redisplay code. In order to
|
|
|
7959 fully support variable height lines, the scrolling code must actually be
|
|
|
7960 tightly integrated with redisplay. Only redisplay can determine how
|
|
|
7961 many lines will be displayed on a screen for any given starting point.
|
|
|
7962
|
|
|
7963 What is ideally wanted is a complete list of the starting buffer
|
|
|
7964 position for every possible display line of a buffer along with the
|
|
|
7965 height of that display line. Maintaining such a full list would be very
|
|
|
7966 expensive. We settle for having it include information for all areas
|
|
|
7967 which we happen to generate anyhow (i.e. the region currently being
|
|
|
7968 displayed) and for those areas we need to work with.
|
|
|
7969
|
|
|
7970 In order to ensure that the cache accurately represents what redisplay
|
|
|
7971 would actually show, it is necessary to invalidate it in many
|
|
|
7972 situations. If the buffer changes, the starting positions may no longer
|
|
|
7973 be correct. If a face or an extent has changed then the line heights
|
|
|
7974 may have altered. These events happen frequently enough that the cache
|
|
|
7975 can end up being constantly disabled. With this potentially constant
|
|
|
7976 invalidation when is the cache ever useful?
|
|
|
7977
|
|
|
7978 Even if the cache is invalidated before every single usage, it is
|
|
|
7979 necessary. Scrolling often requires knowledge about display lines which
|
|
|
7980 are actually above or below the visible region. The cache provides a
|
|
|
7981 convenient light-weight method of storing this information for multiple
|
|
|
7982 display regions. This knowledge is necessary for the scrolling code to
|
|
|
7983 always obey the First Golden Rule of Redisplay.
|
|
|
7984
|
|
|
7985 If the cache already contains all of the information that the scrolling
|
|
|
7986 routines happen to need so that it doesn't have to go generate it, then
|
|
|
7987 we are able to obey the Third Golden Rule of Redisplay. The first thing
|
|
|
7988 we do to help out the cache is to always add the displayed region. This
|
|
|
7989 region had to be generated anyway, so the cache ends up getting the
|
|
|
7990 information basically for free. In those cases where a user is simply
|
|
|
7991 scrolling around viewing a buffer there is a high probability that this
|
|
|
7992 is sufficient to always provide the needed information. The second
|
|
|
7993 thing we can do is be smart about invalidating the cache.
|
|
|
7994
|
|
440
|
7995 TODO---Be smart about invalidating the cache. Potential places:
|
|
428
|
7996
|
|
|
7997 @itemize @bullet
|
|
|
7998 @item
|
|
|
7999 Insertions at end-of-line which don't cause line-wraps do not alter the
|
|
|
8000 starting positions of any display lines. These types of buffer
|
|
|
8001 modifications should not invalidate the cache. This is actually a large
|
|
|
8002 optimization for redisplay speed as well.
|
|
|
8003 @item
|
|
|
8004 Buffer modifications frequently only affect the display of lines at and
|
|
|
8005 below where they occur. In these situations we should only invalidate
|
|
|
8006 the part of the cache starting at where the modification occurs.
|
|
|
8007 @end itemize
|
|
|
8008
|
|
|
8009 In case you're wondering, the Second Golden Rule of Redisplay is not
|
|
|
8010 applicable.
|
|
|
8011
|
|
|
8012 @node Redisplay Piece by Piece
|
|
|
8013 @section Redisplay Piece by Piece
|
|
|
8014 @cindex Redisplay Piece by Piece
|
|
|
8015
|
|
|
8016 As you can begin to see redisplay is complex and also not well
|
|
|
8017 documented. Chuck no longer works on XEmacs so this section is my take
|
|
|
8018 on the workings of redisplay.
|
|
|
8019
|
|
|
8020 Redisplay happens in three phases:
|
|
|
8021
|
|
|
8022 @enumerate
|
|
|
8023 @item
|
|
|
8024 Determine desired display in area that needs redisplay.
|
|
|
8025 Implemented by @code{redisplay.c}
|
|
|
8026 @item
|
|
|
8027 Compare desired display with current display
|
|
|
8028 Implemented by @code{redisplay-output.c}
|
|
|
8029 @item
|
|
|
8030 Output changes Implemented by @code{redisplay-output.c},
|
|
|
8031 @code{redisplay-x.c}, @code{redisplay-msw.c} and @code{redisplay-tty.c}
|
|
|
8032 @end enumerate
|
|
|
8033
|
|
|
8034 Steps 1 and 2 are device-independant and relatively complex. Step 3 is
|
|
|
8035 mostly device-dependent.
|
|
|
8036
|
|
|
8037 Determining the desired display
|
|
|
8038
|
|
|
8039 Display attributes are stored in @code{display_line} structures. Each
|
|
|
8040 @code{display_line} consists of a set of @code{display_block}'s and each
|
|
|
8041 @code{display_block} contains a number of @code{rune}'s. Generally
|
|
|
8042 dynarr's of @code{display_line}'s are held by each window representing
|
|
|
8043 the current display and the desired display.
|
|
|
8044
|
|
|
8045 The @code{display_line} structures are tighly tied to buffers which
|
|
|
8046 presents a problem for redisplay as this connection is bogus for the
|
|
|
8047 modeline. Hence the @code{display_line} generation routines are
|
|
|
8048 duplicated for generating the modeline. This means that the modeline
|
|
|
8049 display code has many bugs that the standard redisplay code does not.
|
|
|
8050
|
|
|
8051 The guts of @code{display_line} generation are in
|
|
|
8052 @code{create_text_block}, which creates a single display line for the
|
|
|
8053 desired locale. This incrementally parses the characters on the current
|
|
|
8054 line and generates redisplay structures for each.
|
|
|
8055
|
|
|
8056 Gutter redisplay is different. Because the data to display is stored in
|
|
|
8057 a string we cannot use @code{create_text_block}. Instead we use
|
|
|
8058 @code{create_text_string_block} which performs the same function as
|
|
|
8059 @code{create_text_block} but for strings. Many of the complexities of
|
|
|
8060 @code{create_text_block} to do with cursor handling and selective
|
|
|
8061 display have been removed.
|
|
|
8062
|
|
|
8063 @node Extents, Faces, The Redisplay Mechanism, Top
|
|
|
8064 @chapter Extents
|
|
|
8065
|
|
|
8066 @menu
|
|
|
8067 * Introduction to Extents:: Extents are ranges over text, with properties.
|
|
|
8068 * Extent Ordering:: How extents are ordered internally.
|
|
|
8069 * Format of the Extent Info:: The extent information in a buffer or string.
|
|
|
8070 * Zero-Length Extents:: A weird special case.
|
|
|
8071 * Mathematics of Extent Ordering:: A rigorous foundation.
|
|
|
8072 * Extent Fragments:: Cached information useful for redisplay.
|
|
|
8073 @end menu
|
|
|
8074
|
|
|
8075 @node Introduction to Extents
|
|
|
8076 @section Introduction to Extents
|
|
|
8077
|
|
|
8078 Extents are regions over a buffer, with a start and an end position
|
|
|
8079 denoting the region of the buffer included in the extent. In
|
|
|
8080 addition, either end can be closed or open, meaning that the endpoint
|
|
|
8081 is or is not logically included in the extent. Insertion of a character
|
|
|
8082 at a closed endpoint causes the character to go inside the extent;
|
|
|
8083 insertion at an open endpoint causes the character to go outside.
|
|
|
8084
|
|
|
8085 Extent endpoints are stored using memory indices (see @file{insdel.c}),
|
|
|
8086 to minimize the amount of adjusting that needs to be done when
|
|
|
8087 characters are inserted or deleted.
|
|
|
8088
|
|
|
8089 (Formerly, extent endpoints at the gap could be either before or
|
|
|
8090 after the gap, depending on the open/closedness of the endpoint.
|
|
|
8091 The intent of this was to make it so that insertions would
|
|
|
8092 automatically go inside or out of extents as necessary with no
|
|
|
8093 further work needing to be done. It didn't work out that way,
|
|
|
8094 however, and just ended up complexifying and buggifying all the
|
|
|
8095 rest of the code.)
|
|
|
8096
|
|
|
8097 @node Extent Ordering
|
|
|
8098 @section Extent Ordering
|
|
|
8099
|
|
|
8100 Extents are compared using memory indices. There are two orderings
|
|
|
8101 for extents and both orders are kept current at all times. The normal
|
|
|
8102 or @dfn{display} order is as follows:
|
|
|
8103
|
|
|
8104 @example
|
|
|
8105 Extent A is ``less than'' extent B,
|
|
|
8106 that is, earlier in the display order,
|
|
|
8107 if: A-start < B-start,
|
|
|
8108 or if: A-start = B-start, and A-end > B-end
|
|
|
8109 @end example
|
|
|
8110
|
|
|
8111 So if two extents begin at the same position, the larger of them is the
|
|
|
8112 earlier one in the display order (@code{EXTENT_LESS} is true).
|
|
|
8113
|
|
|
8114 For the e-order, the same thing holds:
|
|
|
8115
|
|
|
8116 @example
|
|
|
8117 Extent A is ``less than'' extent B in e-order,
|
|
|
8118 that is, later in the buffer,
|
|
|
8119 if: A-end < B-end,
|
|
|
8120 or if: A-end = B-end, and A-start > B-start
|
|
|
8121 @end example
|
|
|
8122
|
|
|
8123 So if two extents end at the same position, the smaller of them is the
|
|
|
8124 earlier one in the e-order (@code{EXTENT_E_LESS} is true).
|
|
|
8125
|
|
|
8126 The display order and the e-order are complementary orders: any
|
|
|
8127 theorem about the display order also applies to the e-order if you swap
|
|
|
8128 all occurrences of ``display order'' and ``e-order'', ``less than'' and
|
|
|
8129 ``greater than'', and ``extent start'' and ``extent end''.
|
|
|
8130
|
|
|
8131 @node Format of the Extent Info
|
|
|
8132 @section Format of the Extent Info
|
|
|
8133
|
|
|
8134 An extent-info structure consists of a list of the buffer or string's
|
|
|
8135 extents and a @dfn{stack of extents} that lists all of the extents over
|
|
|
8136 a particular position. The stack-of-extents info is used for
|
|
440
|
8137 optimization purposes---it basically caches some info that might
|
|
428
|
8138 be expensive to compute. Certain otherwise hard computations are easy
|
|
|
8139 given the stack of extents over a particular position, and if the
|
|
|
8140 stack of extents over a nearby position is known (because it was
|
|
|
8141 calculated at some prior point in time), it's easy to move the stack
|
|
|
8142 of extents to the proper position.
|
|
|
8143
|
|
|
8144 Given that the stack of extents is an optimization, and given that
|
|
|
8145 it requires memory, a string's stack of extents is wiped out each
|
|
|
8146 time a garbage collection occurs. Therefore, any time you retrieve
|
|
|
8147 the stack of extents, it might not be there. If you need it to
|
|
|
8148 be there, use the @code{_force} version.
|
|
|
8149
|
|
|
8150 Similarly, a string may or may not have an extent_info structure.
|
|
|
8151 (Generally it won't if there haven't been any extents added to the
|
|
|
8152 string.) So use the @code{_force} version if you need the extent_info
|
|
|
8153 structure to be there.
|
|
|
8154
|
|
|
8155 A list of extents is maintained as a double gap array: one gap array
|
|
|
8156 is ordered by start index (the @dfn{display order}) and the other is
|
|
|
8157 ordered by end index (the @dfn{e-order}). Note that positions in an
|
|
|
8158 extent list should logically be conceived of as referring @emph{to} a
|
|
|
8159 particular extent (as is the norm in programs) rather than sitting
|
|
|
8160 between two extents. Note also that callers of these functions should
|
|
|
8161 not be aware of the fact that the extent list is implemented as an
|
|
|
8162 array, except for the fact that positions are integers (this should be
|
|
|
8163 generalized to handle integers and linked list equally well).
|
|
|
8164
|
|
|
8165 @node Zero-Length Extents
|
|
|
8166 @section Zero-Length Extents
|
|
|
8167
|
|
|
8168 Extents can be zero-length, and will end up that way if their endpoints
|
|
|
8169 are explicitly set that way or if their detachable property is nil
|
|
|
8170 and all the text in the extent is deleted. (The exception is open-open
|
|
|
8171 zero-length extents, which are barred from existing because there is
|
|
|
8172 no sensible way to define their properties. Deletion of the text in
|
|
|
8173 an open-open extent causes it to be converted into a closed-open
|
|
|
8174 extent.) Zero-length extents are primarily used to represent
|
|
|
8175 annotations, and behave as follows:
|
|
|
8176
|
|
|
8177 @enumerate
|
|
|
8178 @item
|
|
|
8179 Insertion at the position of a zero-length extent expands the extent
|
|
|
8180 if both endpoints are closed; goes after the extent if it is closed-open;
|
|
|
8181 and goes before the extent if it is open-closed.
|
|
|
8182
|
|
|
8183 @item
|
|
|
8184 Deletion of a character on a side of a zero-length extent whose
|
|
|
8185 corresponding endpoint is closed causes the extent to be detached if
|
|
|
8186 it is detachable; if the extent is not detachable or the corresponding
|
|
|
8187 endpoint is open, the extent remains in the buffer, moving as necessary.
|
|
|
8188 @end enumerate
|
|
|
8189
|
|
|
8190 Note that closed-open, non-detachable zero-length extents behave
|
|
|
8191 exactly like markers and that open-closed, non-detachable zero-length
|
|
|
8192 extents behave like the ``point-type'' marker in Mule.
|
|
|
8193
|
|
|
8194 @node Mathematics of Extent Ordering
|
|
|
8195 @section Mathematics of Extent Ordering
|
|
|
8196 @cindex extent mathematics
|
|
|
8197 @cindex mathematics of extents
|
|
|
8198 @cindex extent ordering
|
|
|
8199
|
|
|
8200 @cindex display order of extents
|
|
|
8201 @cindex extents, display order
|
|
|
8202 The extents in a buffer are ordered by ``display order'' because that
|
|
|
8203 is that order that the redisplay mechanism needs to process them in.
|
|
|
8204 The e-order is an auxiliary ordering used to facilitate operations
|
|
|
8205 over extents. The operations that can be performed on the ordered
|
|
|
8206 list of extents in a buffer are
|
|
|
8207
|
|
|
8208 @enumerate
|
|
|
8209 @item
|
|
|
8210 Locate where an extent would go if inserted into the list.
|
|
|
8211 @item
|
|
|
8212 Insert an extent into the list.
|
|
|
8213 @item
|
|
|
8214 Remove an extent from the list.
|
|
|
8215 @item
|
|
|
8216 Map over all the extents that overlap a range.
|
|
|
8217 @end enumerate
|
|
|
8218
|
|
|
8219 (4) requires being able to determine the first and last extents
|
|
|
8220 that overlap a range.
|
|
|
8221
|
|
|
8222 NOTE: @dfn{overlap} is used as follows:
|
|
|
8223
|
|
|
8224 @itemize @bullet
|
|
|
8225 @item
|
|
|
8226 two ranges overlap if they have at least one point in common.
|
|
|
8227 Whether the endpoints are open or closed makes a difference here.
|
|
|
8228 @item
|
|
|
8229 a point overlaps a range if the point is contained within the
|
|
|
8230 range; this is equivalent to treating a point @math{P} as the range
|
|
|
8231 @math{[P, P]}.
|
|
|
8232 @item
|
|
|
8233 In the case of an @emph{extent} overlapping a point or range, the extent
|
|
|
8234 is normally treated as having closed endpoints. This applies
|
|
|
8235 consistently in the discussion of stacks of extents and such below.
|
|
|
8236 Note that this definition of overlap is not necessarily consistent with
|
|
|
8237 the extents that @code{map-extents} maps over, since @code{map-extents}
|
|
|
8238 sometimes pays attention to whether the endpoints of an extents are open
|
|
|
8239 or closed. But for our purposes, it greatly simplifies things to treat
|
|
|
8240 all extents as having closed endpoints.
|
|
|
8241 @end itemize
|
|
|
8242
|
|
|
8243 First, define @math{>}, @math{<}, @math{<=}, etc. as applied to extents
|
|
|
8244 to mean comparison according to the display order. Comparison between
|
|
|
8245 an extent @math{E} and an index @math{I} means comparison between
|
|
|
8246 @math{E} and the range @math{[I, I]}.
|
|
|
8247
|
|
|
8248 Also define @math{e>}, @math{e<}, @math{e<=}, etc. to mean comparison
|
|
|
8249 according to the e-order.
|
|
|
8250
|
|
|
8251 For any range @math{R}, define @math{R(0)} to be the starting index of
|
|
|
8252 the range and @math{R(1)} to be the ending index of the range.
|
|
|
8253
|
|
|
8254 For any extent @math{E}, define @math{E(next)} to be the extent directly
|
|
|
8255 following @math{E}, and @math{E(prev)} to be the extent directly
|
|
|
8256 preceding @math{E}. Assume @math{E(next)} and @math{E(prev)} can be
|
|
|
8257 determined from @math{E} in constant time. (This is because we store
|
|
|
8258 the extent list as a doubly linked list.)
|
|
|
8259
|
|
|
8260 Similarly, define @math{E(e-next)} and @math{E(e-prev)} to be the
|
|
|
8261 extents directly following and preceding @math{E} in the e-order.
|
|
|
8262
|
|
|
8263 Now:
|
|
|
8264
|
|
|
8265 Let @math{R} be a range.
|
|
|
8266 Let @math{F} be the first extent overlapping @math{R}.
|
|
|
8267 Let @math{L} be the last extent overlapping @math{R}.
|
|
|
8268
|
|
|
8269 Theorem 1: @math{R(1)} lies between @math{L} and @math{L(next)},
|
|
|
8270 i.e. @math{L <= R(1) < L(next)}.
|
|
|
8271
|
|
|
8272 This follows easily from the definition of display order. The
|
|
|
8273 basic reason that this theorem applies is that the display order
|
|
|
8274 sorts by increasing starting index.
|
|
|
8275
|
|
|
8276 Therefore, we can determine @math{L} just by looking at where we would
|
|
|
8277 insert @math{R(1)} into the list, and if we know @math{F} and are moving
|
|
|
8278 forward over extents, we can easily determine when we've hit @math{L} by
|
|
|
8279 comparing the extent we're at to @math{R(1)}.
|
|
|
8280
|
|
|
8281 @example
|
|
|
8282 Theorem 2: @math{F(e-prev) e< [1, R(0)] e<= F}.
|
|
|
8283 @end example
|
|
|
8284
|
|
|
8285 This is the analog of Theorem 1, and applies because the e-order
|
|
|
8286 sorts by increasing ending index.
|
|
|
8287
|
|
|
8288 Therefore, @math{F} can be found in the same amount of time as
|
|
|
8289 operation (1), i.e. the time that it takes to locate where an extent
|
|
|
8290 would go if inserted into the e-order list.
|
|
|
8291
|
|
|
8292 If the lists were stored as balanced binary trees, then operation (1)
|
|
|
8293 would take logarithmic time, which is usually quite fast. However,
|
|
|
8294 currently they're stored as simple doubly-linked lists, and instead we
|
|
|
8295 do some caching to try to speed things up.
|
|
|
8296
|
|
|
8297 Define a @dfn{stack of extents} (or @dfn{SOE}) as the set of extents
|
|
|
8298 (ordered in the display order) that overlap an index @math{I}, together
|
|
|
8299 with the SOE's @dfn{previous} extent, which is an extent that precedes
|
|
|
8300 @math{I} in the e-order. (Hopefully there will not be very many extents
|
|
|
8301 between @math{I} and the previous extent.)
|
|
|
8302
|
|
|
8303 Now:
|
|
|
8304
|
|
|
8305 Let @math{I} be an index, let @math{S} be the stack of extents on
|
|
|
8306 @math{I}, let @math{F} be the first extent in @math{S}, and let @math{P}
|
|
|
8307 be @math{S}'s previous extent.
|
|
|
8308
|
|
|
8309 Theorem 3: The first extent in @math{S} is the first extent that overlaps
|
|
|
8310 any range @math{[I, J]}.
|
|
|
8311
|
|
|
8312 Proof: Any extent that overlaps @math{[I, J]} but does not include
|
|
|
8313 @math{I} must have a start index @math{> I}, and thus be greater than
|
|
|
8314 any extent in @math{S}.
|
|
|
8315
|
|
|
8316 Therefore, finding the first extent that overlaps a range @math{R} is
|
|
|
8317 the same as finding the first extent that overlaps @math{R(0)}.
|
|
|
8318
|
|
|
8319 Theorem 4: Let @math{I2} be an index such that @math{I2 > I}, and let
|
|
|
8320 @math{F2} be the first extent that overlaps @math{I2}. Then, either
|
|
|
8321 @math{F2} is in @math{S} or @math{F2} is greater than any extent in
|
|
|
8322 @math{S}.
|
|
|
8323
|
|
|
8324 Proof: If @math{F2} does not include @math{I} then its start index is
|
|
|
8325 greater than @math{I} and thus it is greater than any extent in
|
|
|
8326 @math{S}, including @math{F}. Otherwise, @math{F2} includes @math{I}
|
|
|
8327 and thus is in @math{S}, and thus @math{F2 >= F}.
|
|
|
8328
|
|
|
8329 @node Extent Fragments
|
|
|
8330 @section Extent Fragments
|
|
|
8331 @cindex extent fragment
|
|
|
8332
|
|
|
8333 Imagine that the buffer is divided up into contiguous, non-overlapping
|
|
|
8334 @dfn{runs} of text such that no extent starts or ends within a run
|
|
|
8335 (extents that abut the run don't count).
|
|
|
8336
|
|
|
8337 An extent fragment is a structure that holds data about the run that
|
|
|
8338 contains a particular buffer position (if the buffer position is at the
|
|
440
|
8339 junction of two runs, the run after the position is used)---the
|
|
428
|
8340 beginning and end of the run, a list of all of the extents in that run,
|
|
|
8341 the @dfn{merged face} that results from merging all of the faces
|
|
|
8342 corresponding to those extents, the begin and end glyphs at the
|
|
|
8343 beginning of the run, etc. This is the information that redisplay needs
|
|
|
8344 in order to display this run.
|
|
|
8345
|
|
|
8346 Extent fragments have to be very quick to update to a new buffer
|
|
|
8347 position when moving linearly through the buffer. They rely on the
|
|
|
8348 stack-of-extents code, which does the heavy-duty algorithmic work of
|
|
|
8349 determining which extents overly a particular position.
|
|
|
8350
|
|
|
8351 @node Faces, Glyphs, Extents, Top
|
|
|
8352 @chapter Faces
|
|
|
8353
|
|
|
8354 Not yet documented.
|
|
|
8355
|
|
|
8356 @node Glyphs, Specifiers, Faces, Top
|
|
|
8357 @chapter Glyphs
|
|
|
8358
|
|
|
8359 Glyphs are graphical elements that can be displayed in XEmacs buffers or
|
|
|
8360 gutters. We use the term graphical element here in the broadest possible
|
|
|
8361 sense since glyphs can be as mundane as text to as arcane as a native
|
|
|
8362 tab widget.
|
|
|
8363
|
|
|
8364 In XEmacs, glyphs represent the uninstantiated state of graphical
|
|
|
8365 elements, i.e. they hold all the information necessary to produce an
|
|
|
8366 image on-screen but the image does not exist at this stage.
|
|
|
8367
|
|
|
8368 Glyphs are lazily instantiated by calling one of the glyph
|
|
|
8369 functions. This usually occurs within redisplay when
|
|
|
8370 @code{Fglyph_height} is called. Instantiation causes an image-instance
|
|
|
8371 to be created and cached. This cache is on a device basis for all glyphs
|
|
|
8372 except glyph-widgets, and on a window basis for glyph widgets. The
|
|
|
8373 caching is done by @code{image_instantiate} and is necessary because it
|
|
|
8374 is generally possible to display an image-instance in multiple
|
|
|
8375 domains. For instance if we create a Pixmap, we can actually display
|
|
|
8376 this on multiple windows - even though we only need a single Pixmap
|
|
|
8377 instance to do this. If caching wasn't done then it would be necessary
|
|
|
8378 to create image-instances for every displayable occurrance of a glyph -
|
|
|
8379 and every usage - and this would be extremely memory and cpu intensive.
|
|
|
8380
|
|
|
8381 Widget-glyphs (a.k.a native widgets) are not cached in this way. This is
|
|
|
8382 because widget-glyph image-instances on screen are toolkit windows, and
|
|
|
8383 thus cannot be reused in multiple XEmacs domains. Thus widget-glyphs are
|
|
|
8384 cached on a window basis.
|
|
|
8385
|
|
|
8386 Any action on a glyph first consults the cache before actually
|
|
|
8387 instantiating a widget.
|
|
|
8388
|
|
440
|
8389 @section Widget-Glyphs in the MS-Windows Environment
|
|
428
|
8390
|
|
|
8391 To Do
|
|
|
8392
|
|
|
8393 @section Widget-Glyphs in the X Environment
|
|
|
8394
|
|
|
8395 Widget-glyphs under X make heavy use of lwlib for manipulating the
|
|
|
8396 native toolkit objects. This is primarily so that different toolkits can
|
|
|
8397 be supported for widget-glyphs, just as they are supported for features
|
|
|
8398 such as menubars etc.
|
|
|
8399
|
|
|
8400 Lwlib is extremely poorly documented and quite hairy so here is my
|
|
|
8401 understanding of what goes on.
|
|
|
8402
|
|
|
8403 Lwlib maintains a set of widget_instances which mirror the hierarchical
|
|
|
8404 state of Xt widgets. I think this is so that widgets can be updated and
|
|
|
8405 manipulated generically by the lwlib library. For instance
|
|
|
8406 update_one_widget_instance can cope with multiple types of widget and
|
|
|
8407 multiple types of toolkit. Each element in the widget hierarchy is updated
|
|
|
8408 from its corresponding widget_instance by walking the widget_instance
|
|
|
8409 tree recursively.
|
|
|
8410
|
|
|
8411 This has desirable properties such as lw_modify_all_widgets which is
|
|
|
8412 called from glyphs-x.c and updates all the properties of a widget
|
|
|
8413 without having to know what the widget is or what toolkit it is from.
|
|
|
8414 Unfortunately this also has hairy properrties such as making the lwlib
|
|
|
8415 code quite complex. And of course lwlib has to know at some level what
|
|
|
8416 the widget is and how to set its properties.
|
|
|
8417
|
|
|
8418 @node Specifiers, Menus, Glyphs, Top
|
|
|
8419 @chapter Specifiers
|
|
|
8420
|
|
|
8421 Not yet documented.
|
|
|
8422
|
|
|
8423 @node Menus, Subprocesses, Specifiers, Top
|
|
|
8424 @chapter Menus
|
|
|
8425
|
|
|
8426 A menu is set by setting the value of the variable
|
|
|
8427 @code{current-menubar} (which may be buffer-local) and then calling
|
|
|
8428 @code{set-menubar-dirty-flag} to signal a change. This will cause the
|
|
|
8429 menu to be redrawn at the next redisplay. The format of the data in
|
|
|
8430 @code{current-menubar} is described in @file{menubar.c}.
|
|
|
8431
|
|
|
8432 Internally the data in current-menubar is parsed into a tree of
|
|
|
8433 @code{widget_value's} (defined in @file{lwlib.h}); this is accomplished
|
|
|
8434 by the recursive function @code{menu_item_descriptor_to_widget_value()},
|
|
|
8435 called by @code{compute_menubar_data()}. Such a tree is deallocated
|
|
|
8436 using @code{free_widget_value()}.
|
|
|
8437
|
|
|
8438 @code{update_screen_menubars()} is one of the external entry points.
|
|
|
8439 This checks to see, for each screen, if that screen's menubar needs to
|
|
|
8440 be updated. This is the case if
|
|
|
8441
|
|
|
8442 @enumerate
|
|
|
8443 @item
|
|
|
8444 @code{set-menubar-dirty-flag} was called since the last redisplay. (This
|
|
|
8445 function sets the C variable menubar_has_changed.)
|
|
|
8446 @item
|
|
|
8447 The buffer displayed in the screen has changed.
|
|
|
8448 @item
|
|
|
8449 The screen has no menubar currently displayed.
|
|
|
8450 @end enumerate
|
|
|
8451
|
|
|
8452 @code{set_screen_menubar()} is called for each such screen. This
|
|
|
8453 function calls @code{compute_menubar_data()} to create the tree of
|
|
|
8454 widget_value's, then calls @code{lw_create_widget()},
|
|
|
8455 @code{lw_modify_all_widgets()}, and/or @code{lw_destroy_all_widgets()}
|
|
|
8456 to create the X-Toolkit widget associated with the menu.
|
|
|
8457
|
|
|
8458 @code{update_psheets()}, the other external entry point, actually
|
|
|
8459 changes the menus being displayed. It uses the widgets fixed by
|
|
|
8460 @code{update_screen_menubars()} and calls various X functions to ensure
|
|
|
8461 that the menus are displayed properly.
|
|
|
8462
|
|
|
8463 The menubar widget is set up so that @code{pre_activate_callback()} is
|
|
|
8464 called when the menu is first selected (i.e. mouse button goes down),
|
|
|
8465 and @code{menubar_selection_callback()} is called when an item is
|
|
|
8466 selected. @code{pre_activate_callback()} calls the function in
|
|
|
8467 activate-menubar-hook, which can change the menubar (this is described
|
|
|
8468 in @file{menubar.c}). If the menubar is changed,
|
|
|
8469 @code{set_screen_menubars()} is called.
|
|
|
8470 @code{menubar_selection_callback()} enqueues a menu event, putting in it
|
|
|
8471 a function to call (either @code{eval} or @code{call-interactively}) and
|
|
|
8472 its argument, which is the callback function or form given in the menu's
|
|
|
8473 description.
|
|
|
8474
|
|
|
8475 @node Subprocesses, Interface to X Windows, Menus, Top
|
|
|
8476 @chapter Subprocesses
|
|
|
8477
|
|
|
8478 The fields of a process are:
|
|
|
8479
|
|
|
8480 @table @code
|
|
|
8481 @item name
|
|
|
8482 A string, the name of the process.
|
|
|
8483
|
|
|
8484 @item command
|
|
|
8485 A list containing the command arguments that were used to start this
|
|
|
8486 process.
|
|
|
8487
|
|
|
8488 @item filter
|
|
|
8489 A function used to accept output from the process instead of a buffer,
|
|
|
8490 or @code{nil}.
|
|
|
8491
|
|
|
8492 @item sentinel
|
|
|
8493 A function called whenever the process receives a signal, or @code{nil}.
|
|
|
8494
|
|
|
8495 @item buffer
|
|
|
8496 The associated buffer of the process.
|
|
|
8497
|
|
|
8498 @item pid
|
|
|
8499 An integer, the Unix process @sc{id}.
|
|
|
8500
|
|
|
8501 @item childp
|
|
|
8502 A flag, non-@code{nil} if this is really a child process.
|
|
|
8503 It is @code{nil} for a network connection.
|
|
|
8504
|
|
|
8505 @item mark
|
|
|
8506 A marker indicating the position of the end of the last output from this
|
|
|
8507 process inserted into the buffer. This is often but not always the end
|
|
|
8508 of the buffer.
|
|
|
8509
|
|
|
8510 @item kill_without_query
|
|
|
8511 If this is non-@code{nil}, killing XEmacs while this process is still
|
|
|
8512 running does not ask for confirmation about killing the process.
|
|
|
8513
|
|
|
8514 @item raw_status_low
|
|
|
8515 @itemx raw_status_high
|
|
|
8516 These two fields record 16 bits each of the process status returned by
|
|
|
8517 the @code{wait} system call.
|
|
|
8518
|
|
|
8519 @item status
|
|
|
8520 The process status, as @code{process-status} should return it.
|
|
|
8521
|
|
|
8522 @item tick
|
|
|
8523 @itemx update_tick
|
|
|
8524 If these two fields are not equal, a change in the status of the process
|
|
|
8525 needs to be reported, either by running the sentinel or by inserting a
|
|
|
8526 message in the process buffer.
|
|
|
8527
|
|
|
8528 @item pty_flag
|
|
|
8529 Non-@code{nil} if communication with the subprocess uses a @sc{pty};
|
|
|
8530 @code{nil} if it uses a pipe.
|
|
|
8531
|
|
|
8532 @item infd
|
|
|
8533 The file descriptor for input from the process.
|
|
|
8534
|
|
|
8535 @item outfd
|
|
|
8536 The file descriptor for output to the process.
|
|
|
8537
|
|
|
8538 @item subtty
|
|
|
8539 The file descriptor for the terminal that the subprocess is using. (On
|
|
|
8540 some systems, there is no need to record this, so the value is
|
|
|
8541 @code{-1}.)
|
|
|
8542
|
|
|
8543 @item tty_name
|
|
|
8544 The name of the terminal that the subprocess is using,
|
|
|
8545 or @code{nil} if it is using pipes.
|
|
|
8546 @end table
|
|
|
8547
|
|
|
8548 @node Interface to X Windows, Index, Subprocesses, Top
|
|
|
8549 @chapter Interface to X Windows
|
|
|
8550
|
|
|
8551 Not yet documented.
|
|
|
8552
|
|
|
8553 @include index.texi
|
|
|
8554
|
|
|
8555 @c Print the tables of contents
|
|
|
8556 @summarycontents
|
|
|
8557 @contents
|
|
|
8558 @c That's all
|
|
|
8559
|
|
|
8560 @bye
|
|
|
8561
|
