Commit | Line | Data |
---|---|---|
9b62c83f GM |
1 | -*- coding: utf-8; mode: text; -*- |
2 | ||
3 | From README.multi-tty in the multi-tty branch. | |
4 | Some of this information may be out of date. | |
5 | ||
6 | ||
7 | THANKS | |
8 | ------ | |
9 | ||
10 | The following is a (sadly incomplete) list of people who have | |
11 | contributed to the project by testing, submitting patches, bug | |
12 | reports, and suggestions. Thanks! | |
13 | ||
14 | Bernard Adrian <bernadrian@free.fr> | |
15 | ARISAWA Akihiro <ari@mbf.ocn.ne.jp> | |
16 | Vincent Bernat <bernat@luffy.cx> | |
17 | Han Boetes <han@mijncomputer.nl> | |
18 | Francisco Borges <borges@let.rug.nl> | |
19 | Damien Cassou <damien.cassou@laposte.net> | |
20 | Robert J. Chassell <bob@rattlesnake.com> | |
21 | Romain Francoise <romain@orebokech.com> | |
22 | Ami Fischman <ami@fischman.org> | |
23 | Noah Friedman <friedman@splode.com> | |
24 | Friedrich Delgado Friedrichs <friedel@nomaden.org> | |
25 | Samium Gromoff <_deepfire@mail.ru> | |
26 | Mikhail Gusarov <dottedmag@dottedmag.net> | |
27 | Eric Hanchrow <offby1@blarg.net> | |
28 | IRIE Tetsuya <irie@t.email.ne.jp> | |
29 | Yoshiaki Kasahara <kasahara@nc.kyushu-u.ac.jp> | |
30 | Bas Kok <nekkobassu@yahoo.com> | |
31 | Jurej Kubelka <Juraj.Kubelka@email.cz> | |
32 | David Lichteblau <david@lichteblau.com> | |
33 | Richard Lewis <rtf@jabble.com> | |
34 | mace <mace@kirjakaapeli.lib.hel.fi> | |
35 | Suresh Madhu <madhu@cs.unm.edu> | |
36 | Xavier Mallard <zedek@gnu-rox.org> | |
37 | Istvan Marko <mi-mtty@kismala.com> | |
38 | Ted Morse <morse@ciholas.com> | |
39 | Gergely Nagy <algernon@debian.org> | |
40 | Dan Nicolaescu <dann@ics.uci.edu> | |
41 | Kalle Olavi Niemitalo <kon@iki.fi> | |
42 | Mark Plaksin <happy@mcplaksin.org> | |
43 | Frank Ruell <stoerte@dreamwarrior.net> | |
44 | Tom Schutzer-Weissmann <trmsw@yahoo.co.uk> | |
45 | Joakim Verona <joakim@verona.se> | |
46 | Dan Waber <dwaber@logolalia.com> | |
47 | and many others. | |
48 | ||
49 | Richard Stallman was kind enough to review an earlier version of my | |
50 | patches. | |
51 | ||
52 | ||
53 | STATUS | |
54 | ------ | |
55 | ||
56 | It still needs to be ported to Windows/Mac/DOS. Both multiple | |
57 | tty device support and simultaneous X and tty frame support works | |
58 | fine. Emacsclient has been extended to support opening new tty and X | |
59 | frames. It has been changed to open new Emacs frames by default. | |
60 | ||
61 | Tested on GNU/Linux, Solaris 8, FreeBSD and OpenBSD. | |
62 | ||
63 | Known problems: | |
64 | ||
65 | * GTK support. If you compile your Emacs with the GTK | |
66 | toolkit, some functionality of multi-tty may be lost. In | |
67 | particular, you may get crashes while working on multiple X | |
68 | displays at once. Previous releases of GTK had limitations | |
69 | and bugs that prevented full-blown multi-display support in | |
70 | Emacs. (GTK crashed when Emacs tries to disconnect from an | |
71 | X server.) Things are much improved in the current GTK | |
72 | version, but if you do experience crashes in libgtk, try | |
73 | compiling Emacs with the Lucid toolkit instead. | |
74 | ||
75 | * The single-kboard mode. | |
76 | ||
77 | If your multi-tty Emacs session seems to be frozen, you | |
78 | probably have a recursive editing session or a pending | |
79 | minibuffer prompt (which is a kind of recursive editing) on | |
80 | another display. To unfreeze your session, switch to that | |
81 | display and complete the recursive edit, for example by | |
82 | pressing C-] (`abort-recursive-edit'). | |
83 | ||
84 | I am sorry to say that currently there is no way to break | |
85 | out of this "single-kboard mode" from a frozen display. If | |
86 | you are unable to switch to the display that locks the | |
87 | others (for example because it is on a remote computer), | |
88 | then you can use emacsclient to break out of all recursive | |
89 | editing sessions: | |
90 | ||
91 | emacsclient -e '(top-level)' | |
92 | ||
93 | Note that this (perhaps) unintuitive behaviour is by design. | |
94 | Single-kboard mode is required because of an intrinsic Emacs | |
95 | limitation that is very hard to eliminate. (This limitation | |
96 | is related to the single-threaded nature of Emacs.) | |
97 | ||
98 | I plan to implement better user notification and support for | |
99 | breaking out of single-kboard mode from locked displays. | |
100 | ||
101 | * Mac and DOS support is broken, doesn't even | |
102 | compile. Multiple display support will probably not provide | |
103 | new Emacs features on these systems, but the multi-tty | |
104 | branch changed a few low-level interfaces, and the | |
105 | system-dependent source files need to be adapted | |
106 | accordingly. The changes are mostly trivial, so almost | |
107 | anyone can help, if only by compiling the branch and | |
108 | reporting the compiler errors. | |
109 | ||
110 | ||
111 | TESTING | |
112 | ------- | |
113 | ||
114 | To test the multi-tty feature, start up the Emacs server with the | |
115 | following commands: | |
116 | ||
117 | emacs | |
118 | M-x server-start | |
119 | ||
120 | and then (from a shell prompt on another terminal) start emacsclient | |
121 | with | |
122 | emacsclient -t /optional/file/names... (for a tty frame) | |
123 | emacsclient /optional/file/names... (for an X frame) | |
124 | ||
125 | (Make sure both emacs and emacsclient are multi-tty versions.) | |
126 | You'll hopefully have two fully working, independent frames on | |
127 | separate terminals. The new frame is closed automatically when you | |
128 | finish editing the specified files (C-x #), but delete-frame (C-x 5 0) | |
129 | also works. Of course, you can create frames on more than two tty | |
130 | devices. | |
131 | ||
132 | Creating new frames on the same tty with C-x 5 2 (make-frame-command) | |
133 | works, and behaves the same way as in previous Emacs versions. If you | |
134 | exit emacs, all terminals should be restored to their previous states. | |
135 | ||
136 | TIPS & TRICKS | |
137 | ------------- | |
138 | ||
139 | I think the best way to use the new Emacs is to have it running inside | |
140 | a disconnected GNU screen session, and always use emacsclient for | |
141 | normal work. One advantage of this is that not a single keystroke of | |
142 | your work will be lost if the display device that you are using | |
143 | crashes, or the network connection times out, or whatever. (I had an | |
144 | extremely unstable X server for some time while I was developing these | |
145 | patches, and running Emacs this way has saved me a number of M-x | |
146 | recover-session invocations.) | |
147 | ||
148 | I use the following two bash scripts to handle my Emacs sessions: | |
149 | ||
150 | -------------------------------------------------------connect-emacs-- | |
151 | #!/bin/bash | |
152 | # Usage: connect-emacs <name> <args>... | |
153 | # | |
154 | # Connects to the Emacs instance called NAME. Starts up the instance | |
155 | # if it is not already running. The rest of the arguments are passed | |
156 | # to emacsclient. | |
157 | ||
158 | name="$1" | |
159 | shift | |
160 | ||
161 | if [ -z "$name" ]; then | |
162 | echo "Usage: connect_emacs <name> <args>..." >&2 | |
163 | exit 1 | |
164 | fi | |
165 | preload-emacs "$name" wait | |
166 | /usr/bin/emacsclient.emacs-multi-tty -s "$name" "$@" | |
167 | ---------------------------------------------------------------------- | |
168 | ||
169 | -------------------------------------------------------preload-emacs-- | |
170 | #!/bin/bash | |
171 | # Usage: preload-emacs <name> [<waitp>] | |
172 | # | |
173 | # Preloads the Emacs instance called NAME in a detached screen | |
174 | # session. Does nothing if the instance is already running. If WAITP | |
175 | # is non-empty, the function waits until the server starts up and | |
176 | # creates its socket; otherwise it returns immediately. | |
177 | ||
178 | name="$1" | |
179 | waitp="$2" | |
180 | screendir="/var/run/screen/S-$USER" | |
181 | serverdir="/tmp/emacs$UID" | |
182 | emacs=/usr/bin/emacs-multi-tty # Or wherever you installed your multi-tty Emacs | |
183 | ||
184 | if [ -z "$name" ]; then | |
185 | echo "Usage: preload_emacs <name> [<waitp>]" >&2 | |
186 | exit 1 | |
187 | fi | |
188 | ||
189 | if [ ! -e "$screendir"/*."$name" ]; then | |
190 | if [ -e "$serverdir/$name" ]; then | |
191 | # Delete leftover socket (for the wait option) | |
192 | rm "$serverdir/$name" | |
193 | fi | |
194 | screen -dmS "$name" "$emacs" -nw --eval "(setq server-name \"$name\")" -f server-start | |
195 | fi | |
196 | if [ ! -z "$waitp" ]; then | |
197 | while [ ! -e "$serverdir/$name" ]; do sleep 0.1; done | |
198 | fi | |
199 | ---------------------------------------------------------------------- | |
200 | ||
201 | I have the following in my profile to have two instances automatically | |
202 | preloaded for editing and email: | |
203 | ||
204 | preload-emacs editor | |
205 | preload-emacs gnus | |
206 | ||
207 | It is useful to set up short aliases for connect-emacs. I use the | |
208 | following: | |
209 | ||
210 | alias edit="connect-emacs editor" | |
211 | alias e=edit | |
212 | alias et="connect-emacs editor -t" | |
213 | alias gnus="connect-emacs gnus" | |
214 | ||
215 | ||
216 | THINGS TO DO | |
217 | ------------ | |
218 | ||
219 | ** See if `tty-defined-color-alist' needs to be terminal-local. | |
220 | Update: Dan says it should be, so convert it. | |
221 | ||
222 | ** Mikhail Gusarov suggest to add a hook akin to | |
223 | `after-make-frame-functions' that is called whenever Emacs connects | |
224 | to a new terminal. Good idea! | |
225 | ||
226 | ** emacsclient -t on the console does not work after su. You have to | |
227 | use non-root accounts or start as root to see this. | |
228 | ||
229 | Login: root | |
230 | Password: | |
231 | # su lorentey | |
232 | $ emacsclient -t | |
233 | *ERROR*: Could not open file: /dev/tty1 | |
234 | ||
235 | The tty can be opened as /dev/tty by emacsclient, but not by Emacs. | |
236 | This seems to be a serious problem. Currently my only idea is to | |
237 | bring back the ugly pty proxy hack from the initial versions of | |
238 | multi-tty. Suggestions would be appreciated. | |
239 | ||
240 | Update: we could change emacsclient to pass its open file | |
241 | descriptor to the Emacs process. Unfortunately, this requires a | |
242 | new Lisp-level Emacs API, and as file descriptors are not otherwise | |
243 | exported to Lisp, this approach seems at least as ugly as the pty | |
244 | proxy idea. | |
245 | ||
246 | ** lisp/vc.el depends on the terminal type during load time. | |
247 | `vc-annotate-color-map' is one example that needs to be fixed. | |
248 | ||
249 | ** Understand how `quit_throw_to_read_char' works, and fix any bugs | |
250 | that come to light. | |
251 | ||
252 | ** See if getcjmp can be eliminated somehow. Why does Emacs allow | |
253 | asynchronous input processing while it's reading input anyway? | |
254 | ||
255 | ** `delete-frame' events are handled by `special-event-map' | |
256 | immediately when read by `read_char'. This is fine but it prevents | |
257 | higher-level keymaps from binding that event to get notified of the | |
258 | deleted frame. | |
259 | ||
260 | Sometimes it would be useful for Lisp code to be notified of frame | |
261 | deletions after they have happened, usually because they want to | |
262 | clean up after the deleted frame. Not all frame-local states can | |
263 | be stored as a frame parameter. For example, | |
264 | `display-splash-screen' uses `recursive-edit' with a special keymap | |
265 | override to create its buffer---and it leads to all kinds of | |
266 | nastiness if Emacs stays in this recursive edit mode after the | |
267 | frame containing the splash screen is deleted. Basically, the | |
268 | splash-screen implementation wants to throw out of the recursive | |
269 | edit when the frame is deleted; however, it is not legal to throw | |
270 | from `delete-frame-functions' because `delete-frame' must not fail. | |
271 | (Introducing `delete-frame-after-functions' would not help either | |
272 | because `delete-frame' may not fail at that time either.) | |
273 | ||
274 | Currently `fancy-splash-screens' installs a | |
275 | `delete-frame-functions' hook that sets up a timer to exit the | |
276 | recursive edit. This is an adequate solution, but it would perhaps | |
277 | be better to have something like a `frame-deleted' event that could | |
278 | be bound in the normal way. | |
279 | ||
280 | ** Trouble: `setenv' doesn't actually set environment variables in the | |
281 | Emacs process. This defeats the purpose of the elaborate | |
282 | `server-with-environment' magic around the `tgetent' call in | |
283 | `init_tty'. D'oh. | |
284 | ||
285 | ** (Possibly) create hooks in struct device for creating frames on a | |
286 | specific terminal, and eliminate the hackish terminal-related frame | |
287 | parameters (display, tty, tty-type). | |
288 | ||
289 | make_terminal_frame | |
290 | create_tty_output | |
291 | ||
292 | ** Decide whether to keep the C implementation of terminal parameters, | |
293 | or revert to the previous, purely Lisp code. It turned out that | |
294 | local environments do not need terminal parameters after all. | |
295 | ||
296 | ** Move Fsend_string_to_terminal to term.c, and declare get_named_tty | |
297 | as static, removing it from dispextern.h. | |
298 | Move fatal to emacs.c and declare it somewhere. | |
299 | ||
300 | ** Search for `suspend-emacs' references and replace them with | |
301 | `suspend-frame', if necessary. Ditto for `save-buffers-kill-emacs' | |
302 | vs. `save-buffers-kill-display'. | |
303 | ||
304 | ** Emacs crashes when a tty frame is resized so that there is no space | |
305 | for all its windows. (Tom Schutzer-Weissmann) | |
306 | ||
307 | ** Report GTK multi-display problems to GTK maintainers. For extra | |
308 | credit, fix them. | |
309 | ||
310 | Currently you can not connect to new X displays when you compile | |
311 | Emacs with GTK support. If you want to play around with GTK | |
312 | multidisplay (and don't mind core dumps), you can edit src/config.h | |
313 | and define HAVE_GTK_MULTIDISPLAY there by hand. | |
314 | ||
315 | http://bugzilla.gnome.org/show_bug.cgi?id=85715 | |
316 | ||
317 | Update: Han reports that GTK+ version 2.8.9 almost gets display | |
318 | disconnects right. GTK will probably be fully fixed by the time | |
319 | multi-tty gets into the trunk. | |
320 | ||
321 | Update: I am still having problems with GTK+ 2.8.10. I have the | |
322 | impression that the various multidisplay fixes will only get | |
323 | released in GTK+ 2.10. | |
324 | ||
325 | ** Audit `face-valid-attribute-values' usage in customize and | |
326 | elsewhere. Its return value depends on the current window system. | |
327 | Replace static initializers using it with runtime functions. For | |
328 | example, custom's buttons are broken on non-initial device types. | |
329 | ||
330 | ** Possibly turn off the double C-g feature when there is an X frame. | |
331 | C.f. (emacs)Emergency Escape. | |
332 | ||
333 | ** frames-on-display-list should also accept frames. | |
334 | ||
335 | ** Consider the `tty-type' frame parameter and the `display-tty-type' | |
336 | function. They serve the exact same purpose. I think it may be | |
337 | a good idea to eliminate one of them, preferably `tty-type'. | |
338 | ||
339 | ** The handling of lisp/term/*.el, and frame creation in general, is a | |
340 | big, big mess. How come the terminal-specific file is loaded by | |
341 | tty-create-frame-with-faces? I don't think it is necessary to load | |
342 | these files for each frame; once per terminal should be enough. | |
343 | Update: lisp/term/*.el is not loaded repeatedly anymore, but | |
344 | faces.el still needs to be cleaned up. | |
345 | ||
346 | ** Fix frame-set-background-mode in this branch. It was recently | |
347 | changed in CVS, and frame.el in multi-tty has not yet been adapted | |
348 | for the changes. (It needs to look at | |
349 | default-frame-background-mode.) (Update: maybe it is fixed now; | |
350 | needs testing.) (Note that the byte compiler has this to say about | |
351 | term/rxvt.el:) | |
352 | ||
353 | term/rxvt.el:309:17:Warning: assignment to free variable | |
354 | `default-frame-background-mode' | |
355 | ||
356 | ** I think `(set-)terminal-local-value' and the terminal parameter | |
357 | mechanism should be integrated into a single framework. | |
358 | ||
359 | (Update: `(set-)terminal-local-value' is now eliminated, but the | |
360 | terminal-local variables should still be accessible as terminal | |
361 | parameters. This also applies to `display-name' and similar | |
362 | functions.) | |
363 | ||
364 | ** Add the following hooks: after-delete-frame-hook (for server.el, | |
365 | instead of delete-frame-functions), | |
366 | after-delete-terminal-functions, after-create-terminal-functions. | |
367 | ||
368 | ** BULK RENAME: The `display-' prefix of new Lisp-level functions | |
369 | conflicts with stuff like `display-time-mode'. Use `device-' | |
370 | or `terminal-' instead. I think I prefer `terminal-'. | |
371 | ||
372 | It turns out that most of the offending Lisp functions were defined | |
373 | in the trunk. Therefore, compatibility aliases should be defined | |
374 | for the following names: | |
375 | ||
376 | display-color-cells terminal-color-cells | |
377 | display-color-p terminal-color-p | |
378 | display-graphic-p terminal-graphic-p | |
379 | display-grayscale-p terminal-grayscale-p | |
380 | display-images-p terminal-images-p | |
381 | display-mm-height terminal-mm-height | |
382 | display-mm-width terminal-mm-width | |
383 | display-mouse-p terminal-mouse-p | |
384 | display-multi-font-p terminal-multi-font-p | |
385 | display-multi-frame-p terminal-multi-frame-p | |
386 | display-pixel-height terminal-pixel-height | |
387 | display-pixel-width terminal-pixel-width | |
388 | display-pixels-per-inch terminal-pixels-per-inch | |
389 | display-planes terminal-planes | |
390 | display-popup-menus-p terminal-popup-menus-p | |
391 | display-save-under terminal-save-under | |
392 | display-screens terminal-screens | |
393 | display-supports-face-attributes-p terminal-supports-face-attributes-p | |
394 | display-visual-class terminal-visual-class | |
395 | framep-on-display framep-on-terminal | |
396 | frames-on-display-list frames-on-terminal-list | |
397 | ||
398 | The following functions were introduced in the multi-tty branch, and | |
399 | were renamed without aliases: | |
400 | ||
401 | delete-display delete-terminal | |
402 | display-controlling-tty-p controlling-tty-p | |
403 | display-list terminal-list | |
404 | display-live-p terminal-live-p | |
405 | display-name terminal-name | |
406 | display-tty-type tty-type | |
407 | frame-display frame-terminal | |
408 | selected-display selected-terminal | |
409 | ||
410 | ** The single-keyboard mode of MULTI_KBOARD is extremely confusing | |
411 | sometimes; Emacs does not respond to stimuli from other keyboards. | |
412 | At least a beep or a message would be important, if the single-mode | |
413 | is still required to prevent interference. (Reported by Dan | |
414 | Nicolaescu.) | |
415 | ||
416 | Update: selecting a region with the mouse enables single_kboard | |
417 | under X. This is very confusing. | |
418 | ||
419 | Update: After discussions with Richard Stallman, this will be | |
420 | resolved by having locked displays warn the user to wait, and | |
421 | introducing a complex protocol to remotely bail out of | |
422 | single-kboard mode by pressing C-g. | |
423 | ||
424 | Update: Warning the user is not trivial to implement, as Emacs has | |
425 | only one echo area, shared by all frames. Ideally the warning | |
426 | should not be displayed on the display that is locking the others. | |
427 | Perhaps the high probability of user confusion caused by | |
428 | single_kboard mode deserves a special case in the display code. | |
429 | Alternatively, it might be good enough to signal single_kboard mode | |
430 | by changing the modelines or some other frame-local display element | |
431 | on the locked out displays. | |
432 | ||
433 | Update: In fact struct kboard does have an echo_string slot. | |
434 | ||
435 | ** The session management module is prone to crashes when the X | |
436 | connection is closed and then later I try to connect to a new X | |
437 | session: | |
438 | ||
439 | #0 0xb7ebc806 in SmcGetIceConnection () from /usr/X11R6/lib/libSM.so.6 | |
440 | #1 0x080e6641 in x_session_check_input (bufp=0xbf86c9c0) at xsmfns.c:144 | |
441 | #2 0x080d3bbc in XTread_socket (device=0xa722ff8, expected=1, hold_quit=0xbf86ca90) at xterm.c:7037 | |
442 | #3 0x080fa404 in read_avail_input (expected=1) at keyboard.c:6696 | |
443 | #4 0x080fa4ca in handle_async_input () at keyboard.c:6900 | |
444 | #5 0x080d51fa in x_term_init (display_name=162628899, xrm_option=0x0, resource_name=0x857068c "emacs") at xterm.c:10622 | |
445 | #6 0x080d920e in x_display_info_for_name (name=162628899) at xfns.c:3975 | |
446 | #7 0x080d92f9 in check_x_display_info (object=1) at xfns.c:274 | |
447 | #8 0x080d97b8 in Fx_create_frame (parms=151221485) at xfns.c:3016 | |
448 | #9 0x0815bf72 in Ffuncall (nargs=2, args=0xbf86ceec) at eval.c:2851 | |
449 | ||
450 | I installed a workaround to prevent this. The X session manager is | |
451 | only contacted when the very first display in the Emacs session is | |
452 | an X display. Also, x_delete_display() on this display aborts | |
453 | session management, and XTread_socket only calls | |
454 | x_session_check_input when it is called for the display that the | |
455 | session was opened on. While this does not really fix the bug, it | |
456 | makes it much less frequent, because session manager support will | |
457 | not normally be enabled when Emacs can survive the shutdown of the | |
458 | X server. | |
459 | ||
460 | See if xsmfns.c should be updated. | |
461 | ||
462 | ** Hunt down display-related functions in frame.el and extend them all | |
463 | to accept display ids. | |
464 | ||
465 | ** rif->flush_display_optional (NULL) calls should be replaced by a | |
466 | new global function. | |
467 | ||
468 | ** The set-locale-environment hack (adding the DISPLAY option) should | |
469 | be replaced with a clean design. | |
470 | ||
471 | ** standard-display-table should be display-local. | |
472 | standard-display-european should be display-local. | |
473 | ||
474 | ** With iswitchb-default-method set to 'always-frame, only frames on | |
475 | the current display should be considered. This might involve | |
476 | extending `get-buffer-window'. | |
477 | ||
478 | ** Have a look at Vlocale_coding_system. Seems like it would be a | |
479 | tedious job to localize it, although most references use it for | |
480 | interfacing with libc and are therefore OK with the global | |
481 | definition. | |
482 | ||
483 | Exceptions found so far: x-select-text and | |
484 | x-cut-buffer-or-selection-value. | |
485 | ||
486 | ** Have a look at fatal_error_hook. | |
487 | ||
488 | ** Have a look at set_frame_matrix_frame. | |
489 | ||
490 | ** Check if we got term-setup-hook right. | |
491 | ||
492 | ** I think tip_frame should be display-local. | |
493 | ||
494 | ** Check display reference count handling in x_create_tip_frame. | |
495 | ||
496 | ** make-frame does not correctly handle extra parameters in its | |
497 | argument: | |
498 | ||
499 | (frame-parameter (make-frame (list (cons 'foobar 42))) 'foobar) | |
500 | => nil | |
501 | ||
502 | (This is likely an error in the CVS trunk.) | |
503 | ||
504 | ** Dan Nicolaescu suggests that -nw should be added as an alias for -t | |
505 | in emacsclient. Good idea. (Alas, implementing this is not | |
506 | trivial, getopt_long does not seem to support two-letter ``short'' | |
507 | options. Patches are welcome.) | |
508 | ||
509 | ** Mark Plaksin suggests that emacsclient should accept the same | |
510 | X-related command-line arguments as Emacs. Most of the X-related | |
511 | argument-handling is done in Lisp, so this should be quite easy to | |
512 | implement. (For example, Samium Gromoff wants emacsclient to | |
513 | support --geometry; implementing this would add that support.) | |
514 | ||
515 | ** Gergely Nagy suggests that C-x # should only kill the current | |
516 | frame, not any other emacsclient frame that may have the same file | |
517 | opened for editing. I think I agree with him. | |
518 | ||
519 | ** Very strange bug: visible-bell does not work on secondary | |
520 | terminals in xterm and konsole. The screen does flicker a bit, | |
521 | but it's so quick it isn't noticable. | |
522 | ||
523 | (Update: This is probably some problem with padding or whatnot on | |
524 | the secondary terminals.) | |
525 | ||
526 | ** Move baud_rate to struct display. | |
527 | ||
528 | ** Implement support for starting an interactive Emacs session without | |
529 | an initial frame. (The user would connect to it and open frames | |
530 | later, with emacsclient.) | |
531 | ||
532 | ** Fix Mac support (I can't do this entirely myself). Note that the | |
533 | current state of Mac-specific source files in the multi-tty tree | |
534 | are not useful; before starting work on Mac support, revert to | |
535 | pristine, pre-multi-tty versions. | |
536 | ||
537 | ** Fix DOS support (I can't do this entirely myself). Note that the | |
538 | current state of DOS-specific source files in the multi-tty tree | |
539 | are not useful; before starting work on DOS support, revert to | |
540 | pristine, pre-multi-tty versions. | |
541 | ||
542 | ** Fix Windows support. Currently bootstraping works on w32, but Emacs | |
543 | crashes on startup and none of the multi-tty features are | |
544 | implemented. Many XXX comments mark things that probably need | |
545 | updating, ChangeLogs will help in spotting changes to X specific | |
546 | files that may need porting. | |
547 | ||
548 | ** Do a grep on XXX and ?? for more issues. | |
549 | ||
550 | ** flow-ctrl.el must be updated. | |
551 | ||
552 | ** Fix stuff_char for multi-tty. Doesn't seem to be of high priority. | |
553 | ||
554 | DIARY OF CHANGES | |
555 | ---------------- | |
556 | ||
557 | (ex-TODO items with explanations.) | |
558 | ||
559 | -- Introduce a new struct for terminal devices. | |
560 | ||
561 | (Done, see struct tty_output. The list of members is not yet | |
562 | complete.) | |
563 | ||
564 | -- Change the bootstrap procedure to initialize tty_list. | |
565 | ||
566 | (Done, but needs review.) | |
567 | ||
568 | -- Change make-terminal-frame to support specifying another tty. | |
569 | ||
570 | (Done, new frame parameters: `tty' and `tty-type'.) | |
571 | ||
572 | -- Implement support for reading from multiple terminals. | |
573 | ||
574 | (Done, read_avail_input tries to read from each terminal, until one | |
575 | succeeds. MULTI_KBOARD is not used. Secondary terminals don't send | |
576 | SIGIO!) | |
577 | ||
578 | (Update: They do, now.) | |
579 | ||
580 | (Update2: After enabling X, they don't.) | |
581 | ||
582 | -- other-frame should cycle through the frames on the `current' | |
583 | terminal only. | |
584 | ||
585 | (Done, by trivially modifiying next_frame and prev_frame.) | |
586 | ||
587 | -- Support different terminal sizes. | |
588 | ||
589 | (Done, no problem.) | |
590 | ||
591 | -- Make sure terminal resizes are handled gracefully. (Could be | |
592 | problematic.) | |
593 | ||
594 | (Done. We don't get automatic SIGWINCH for additional ttys, | |
595 | though.) | |
596 | ||
597 | -- Extend emacsclient to automatically open a new tty when it connects | |
598 | to Emacs. | |
599 | ||
600 | (Done. It's an ugly hack, needs more work.) | |
601 | ||
602 | -- Redisplay must refresh the topmost frame on *all* terminals, not | |
603 | just the initial terminal. | |
604 | ||
605 | (Done, but introduced an ugly redisplay problems. Ugh.) | |
606 | ||
607 | -- Fix redisplay problems. | |
608 | ||
609 | (Done; it turned out that the entire Wcm structure must be moved | |
610 | inside tty_output. Why didn't I catch this earlier?) | |
611 | ||
612 | -- Provide a way for emacsclient to tell Emacs that the tty has been | |
613 | resized. | |
614 | ||
615 | (Done, simply forward the SIGWINCH signal.) | |
616 | ||
617 | -- Each keypress should automatically select the frame corresponding | |
618 | to the terminal that it was coming from. This means that Emacs | |
619 | must know from which terminal the last keyboard event came from. | |
620 | ||
621 | (Done, it was quite simple, the input event system already | |
622 | supported multiple frames.) | |
623 | ||
624 | -- Fix SIGIO issue with secondary terminals. | |
625 | ||
626 | (Done, emacsclient signals Emacs after writing to the proxy pseudo | |
627 | terminal. Note that this means that multi-tty does not work with | |
628 | raw ttys!) | |
629 | ||
630 | (Update: This is bullshit. There is a read_input_waiting function, | |
631 | extend that somehow.) | |
632 | ||
633 | (Update of update: The first update was not right either, extending | |
634 | read_input_waiting was not necessary. Secondary ttys do seem to | |
635 | send signals on input.) | |
636 | ||
637 | (Update^3: Not any more.) | |
638 | ||
639 | -- Make make-terminal-frame look up the `tty' and `tty-type' frame | |
640 | parameters from the currently selected terminal before the global | |
641 | default. | |
642 | ||
643 | (Done.) | |
644 | ||
645 | -- Put all cached terminal escape sequences into struct tty_output. | |
646 | Currently, they are still stored in global variables, so we don't | |
647 | really support multiple terminal types. | |
648 | ||
649 | (Done. It was not fun.) | |
650 | ||
651 | -- Implement sane error handling after initialization. (Currently | |
652 | emacs exits if you specify a bad terminal type.) The helpful error | |
653 | messages must still be provided when Emacs starts. | |
654 | ||
655 | (Done.) | |
656 | ||
657 | -- Implement terminal deletion, i.e., deleting local frames, closing | |
658 | the tty device and restoring its previous state without exiting | |
659 | Emacs. | |
660 | ||
661 | (Done, but at the moment only called when an error happens during | |
662 | initialization. There is a memory corruption error around this | |
663 | somewhere.) (Update: now it is fully enabled.) | |
664 | ||
665 | -- Implement automatic deletion of terminals when the last frame on | |
666 | that terminal is closed. | |
667 | ||
668 | (Done.) | |
669 | ||
670 | -- Restore tty screen after closing the terminal. | |
671 | ||
672 | (Done, we do the same as Emacs 21.2 for all terminals.) | |
673 | ||
674 | -- 'TERM=dumb src/emacs' does not restore the terminal state. | |
675 | ||
676 | (Done.) | |
677 | ||
678 | -- C-g should work on secondary terminals. | |
679 | ||
680 | (Done, but the binding is not configurable.) | |
681 | ||
682 | -- Deal with SIGHUP in Emacs and in emacsclient. (After this, the | |
683 | server-frames may be removed from server.el.) | |
684 | ||
685 | (Done, nothing to do. It seems that Emacs does not receive SIGHUP | |
686 | from secondary ttys, which is actually a good thing.) (Update: I | |
687 | think it would be a bad idea to remove server-frames.) | |
688 | ||
689 | -- Change emacsclient/server.el to support the -t argument better, | |
690 | i.e. automatically close the socket when the frame is closed. | |
691 | ||
692 | (Seems to be working OK.) | |
693 | ||
694 | -- Fix mysterious memory corruption error with tty deletion. To | |
695 | trigger it, try the following shell command: | |
696 | ||
697 | while true; do TERM=no-such-terminal-definition emacsclient -h; done | |
698 | ||
699 | Emacs usually dumps core after a few dozen iterations. (The bug | |
700 | seems to be related to the xfreeing or bzeroing of | |
701 | tty_output.Wcm. Maybe there are outside references to struct Wcm? | |
702 | Why were these vars collected into a struct before multi-tty | |
703 | support?) | |
704 | ||
705 | (Done. Whew. It turned out that the problem had nothing to do | |
706 | with hypothetical external references to Wcm, or any other | |
707 | tty_output component; it was simply that delete_tty closed the | |
708 | filehandles of secondary ttys twice, resulting in fclose doubly | |
709 | freeing memory. Utterly trivial matter. I love the C's memory | |
710 | management, it puts hair on your chest.) | |
711 | ||
712 | -- Support raw secondary terminals. (Note that SIGIO works only on | |
713 | the controlling terminal.) Hint: extend read_input_waiting for | |
714 | multiple ttys and hopefully this will be fixed. | |
715 | ||
716 | (Done, it seems to have been working already for some time. It | |
717 | seems F_SETOWN does work, after all. Not sure what made it fail | |
718 | earlier, but it seems to be fixed (there were several changes | |
719 | around request_sigio, maybe one of them did it). | |
720 | read_input_waiting is only used in sys_select, don't change | |
721 | it.) (Update: After adding X support, it's broken again.) | |
722 | (Update^2: No it isn't.) :-) | |
723 | ||
724 | -- Find out why does Emacs abort when it wants to close its | |
725 | controlling tty. Hint: chan_process[] array. Hey, maybe | |
726 | noninterrupt-IO would work, too? Update: no, there is no process | |
727 | for stdin/out. | |
728 | ||
729 | (Done. Added add/delete_keyboard_wait_descriptor to | |
730 | term_init/delete_tty. The hint was right, in a way.) | |
731 | ||
732 | -- Issue with SIGIO: it needs to be disabled during redisplay. See if | |
733 | fcntl kernel behaviour could be emulated by emacsclient. | |
734 | ||
735 | (Done. Simply disabled the SIGIO emulation hack in emacsclient.) | |
736 | (Update: it was added back.) (Update^2: and removed again.) | |
737 | ||
738 | -- server.el: There are issues with saving files in buffers of closed | |
739 | clients. Try editing a file with emacsclient -f, and (without | |
740 | saving it) do a delete-frame. The frame is closed without | |
741 | question, and a surprising confirmation prompt appears in another | |
742 | frame. | |
743 | ||
744 | (Done. delete-frame now asks for confirmation if it still has | |
745 | pending buffers, and modified buffers don't seem to be deleted.) | |
746 | ||
747 | -- emacsclient.el, server.el: Handle eval or file open errors when | |
748 | doing -t. | |
749 | ||
750 | (Done.) | |
751 | ||
752 | -- Make parts of struct tty_output accessible from Lisp. The device | |
753 | name and the type is sufficient. | |
754 | ||
755 | (Done, see frame-tty-name and frame-tty-type.) | |
756 | ||
757 | -- Export delete_tty to the Lisp environment, for emacsclient. | |
758 | ||
759 | (Done, see delete-tty.) | |
760 | ||
761 | -- Get rid of the accessor macros in termchar.h, or define macros for | |
762 | all members. | |
763 | ||
764 | (Done.) | |
765 | ||
766 | -- Move device-specific parameters (like costs) commonly used by | |
767 | device backends to a common, device-dependent structure. | |
768 | ||
769 | (Done. See struct display_method in termhooks.h.) | |
770 | ||
771 | -- Fix X support. | |
772 | ||
773 | (Done. Well, it seems to be working.) | |
774 | ||
775 | -- Allow simultaneous X and tty frames. (Handling input could be | |
776 | tricky. Or maybe not.) | |
777 | ||
778 | (Done. Allowed, that is. It is currently extremely unstable, to | |
779 | the point of being unusable. The rif variable causes constant | |
780 | core dumps. Handling input is indeed tricky.) | |
781 | ||
782 | -- Rewrite multi-tty input in terms of MULTI_KBOARD. | |
783 | ||
784 | (Done. In fact, there was no need to rewrite anything, I just | |
785 | added a kboard member to tty_display_info, and initialized the | |
786 | frame's kboard from there.) | |
787 | ||
788 | -- Fix rif issue with X-tty combo sessions. IMHO the best thing to do | |
789 | is to get rid of that global variable (and use the value value in | |
790 | display_method, which is guaranteed to be correct). | |
791 | ||
792 | (Done, did exactly that. Core dumps during combo sessions became | |
793 | much rarer. In fact, I have not yet met a single one.) | |
794 | ||
795 | -- Add multi-tty support to talk.el. | |
796 | ||
797 | (Done.) | |
798 | ||
799 | -- Clean up the source of emacsclient. It is a mess. | |
800 | ||
801 | (Done, eliminated stupid proxy-pty kludge.) | |
802 | ||
803 | -- Fix faces on tty frames during X-tty combo sessions. There is an | |
804 | init_frame_faces call in init_sys_modes, see if there is a problem | |
805 | with it. | |
806 | ||
807 | (Done, there was a stupid mistake in | |
808 | Ftty_supports_face_attributes_p. Colors are broken, though.) | |
809 | ||
810 | -- C-x 5 2, C-x 5 o, C-x 5 0 on an emacsclient frame unexpectedly | |
811 | exits emacsclient. This is a result of trying to be clever with | |
812 | delete-frame-functions. | |
813 | ||
814 | (Fixed, added delete-tty-after-functions, and changed server.el to | |
815 | use it.) | |
816 | ||
817 | -- Something with (maybe) multi-keyboard support broke function keys | |
818 | and arrows on ttys during X+tty combo sessions. Debug this. | |
819 | ||
820 | (I can't reproduce it, maybe the terminal type was wrong.) | |
821 | ||
822 | -- Fix input from raw ttys (again). | |
823 | ||
824 | (Now it seems to work all right.) | |
825 | ||
826 | -- During an X-tty combo session, a (message "Hello") from a tty frame | |
827 | goes to the X frame. Fix this. | |
828 | ||
829 | (Done. There was a safeguard against writing to the initial | |
830 | terminal frame during bootstrap which prevented echo_area_display | |
831 | from working correctly on a tty frame during a combo session.) | |
832 | ||
833 | -- If there are no frames on its controlling terminal, Emacs should | |
834 | exit if the user presses C-c there. | |
835 | ||
836 | (Done, as far as possible. See the SIGTERM comment in | |
837 | interrupt_signal on why this seems to be impossible to solve this | |
838 | in general.) | |
839 | ||
840 | -- During an X session, Emacs seems to read from stdin. Also, Emacs | |
841 | fails to start without a controlling tty. | |
842 | ||
843 | (Fixed by replacing the troublesome termcap display with a dummy | |
844 | bootstrap display during bootstrap. | |
845 | ||
846 | -- Do tty output through struct display, like graphical display | |
847 | backends. | |
848 | ||
849 | (Done.) | |
850 | ||
851 | -- Define an output_initial value for output_method for the initial | |
852 | frame that is dumped with Emacs. Checking for this frame (e.g. in | |
853 | cmd_error_internal) is ugly. | |
854 | ||
855 | (Done, breaking interactive temacs.) | |
856 | ||
857 | -- The command `emacsclient -t -e '(delete-frame)'' fails to exit. | |
858 | ||
859 | (Fixed.) | |
860 | ||
861 | -- frame-creation-function should always create a frame that is on the | |
862 | same display as the selected frame. Maybe frame-creation-function | |
863 | should simply be removed and make-frame changed to do the right | |
864 | thing. | |
865 | ||
866 | (Done, with a nice hack. frame-creation-function is now frame-local.) | |
867 | ||
868 | -- Fix C-g on raw ttys. | |
869 | ||
870 | (Done. I disabled the interrupt/quit keys on all secondary | |
871 | terminals, so Emacs sees C-g as normal input. This looks like an | |
872 | overkill, because emacsclient has extra code to pass SIGINT to | |
873 | Emacs, so C-g should remain the interrupt/quit key on emacsclient | |
874 | frames. See the next entry why implementing this distinction would | |
875 | be a bad idea.) | |
876 | ||
877 | -- Make sure C-g goes to the right frame with ttys. This is hard, as | |
878 | SIGINT doesn't have a tty parameter. :-( | |
879 | ||
880 | (Done, the previous change fixes this as a pleasant side effect.) | |
881 | ||
882 | -- I have seen a case when Emacs with multiple ttys fell into a loop | |
883 | eating 100% of CPU time. Strace showed this loop: | |
884 | ||
885 | getpid() = 30284 | |
886 | kill(30284, SIGIO) = 0 | |
887 | --- SIGIO (I/O possible) @ 0 (0) --- | |
888 | ioctl(6, FIONREAD, [0]) = -1 EIO (Input/output error) | |
889 | ioctl(5, FIONREAD, [0]) = -1 EIO (Input/output error) | |
890 | ioctl(0, FIONREAD, [0]) = 0 | |
891 | sigreturn() = ? (mask now []) | |
892 | gettimeofday({1072842297, 747760}, NULL) = 0 | |
893 | gettimeofday({1072842297, 747806}, NULL) = 0 | |
894 | select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0}) | |
895 | select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0}) | |
896 | gettimeofday({1072842297, 748245}, NULL) = 0 | |
897 | ||
898 | I have seen something similar with a single X frame, but have not | |
899 | been able to reproduce it for debugging. | |
900 | ||
901 | Update: This may have been caused by checking for nread != 0 | |
902 | instead of nread > 0 after calling read_socket_hook in | |
903 | read_avail_input. | |
904 | ||
905 | (Fixed. This was caused by unconditionally including stdin in | |
906 | input_wait_mask in init_process. The select call in | |
907 | wait_reading_process_input always returned immediately, indicating | |
908 | that there is pending input from stdin, which nobody read. | |
909 | ||
910 | Note that the above strace output seems to be an unrelated but | |
911 | similar bug. I think that is now fixed.) | |
912 | ||
913 | -- Exiting Emacs while there are emacsclient frames doesn't restore the | |
914 | ttys to their default states. | |
915 | ||
916 | (This seems to be fixed by some previous change.) | |
917 | ||
918 | -- Allow opening an X session after -nw. | |
919 | ||
920 | (Done.) | |
921 | ||
922 | -- Fix color handling during tty+X combo sessions. (It seems that tty | |
923 | sessions automatically convert the face colors to terminal colors | |
924 | when the face is loaded. This conversion must happen instead on | |
925 | the fly in write_glyphs, which might be problematic, as color | |
926 | approximation is currently done in lisp (term/tty-colors.el).) | |
927 | (Update: hm, colors seem to work fine if I start emacs with -nw and | |
928 | then create an X frame. Maybe it's just a small buglet somewhere.) | |
929 | ||
930 | (Seems to be fixed. The problem was in startup.el, it did not | |
931 | initialize tty colors when the initial window system was | |
932 | graphical.) | |
933 | ||
934 | -- emacs -nw --eval '(y-or-n-p "Foobar")' segfaults. (Reported by | |
935 | Romain Francoise) | |
936 | ||
937 | (Fixed, there was a keyboard initialization problem.) | |
938 | ||
939 | -- Fix interactive use of temacs. There are face-related SEGVs, most | |
940 | likely because of changes in realize_default_face, realize_face. | |
941 | ||
942 | (Fixed.) | |
943 | ||
944 | -- Don't exit Emacs when the last X connection fails during a | |
945 | multi-display session. | |
946 | ||
947 | (Fixed.) | |
948 | ||
949 | -- Dan Nicolaescu noticed that starting emacsclient on the same | |
950 | terminal device that is the controlling tty of the Emacs process | |
951 | gives unexpected results. | |
952 | ||
953 | (Fixed.) | |
954 | ||
955 | -- Istvan Marko reported that Emacs hang on ttys if it was started | |
956 | from a shell script. | |
957 | ||
958 | (Fixed. There was a bug in the multi-tty version of | |
959 | narrow_foreground_group. tcsetpgrp blocks if it is called from a | |
960 | process that is not in the same process group as the tty.) | |
961 | ||
962 | -- emacsclient -t from an Emacs term buffer does not work, complains | |
963 | about face problems. This can even lock up Emacs (if the recursive | |
964 | frame sets single_kboard). Update: the face problems are caused by | |
965 | bugs in term.el, not in multi-tty. The lockup is caused by | |
966 | single_kboard mode, and is not easily resolvable. The best thing to | |
967 | do is to simply refuse to create a tty frame of type `eterm'. | |
968 | ||
969 | (Fixed, changed emacsclient to check for TERM=eterm. The face | |
970 | complaints seem to be caused by bugs in term.el; they are not | |
971 | related to multi-tty.) | |
972 | ||
973 | -- Find out the best way to support suspending Emacs with multiple | |
974 | ttys. My guess: disable it on the controlling tty, but from other | |
975 | ttys pass it on to emacsclient somehow. (It is (I hope) trivial to | |
976 | extend emacsclient to handle suspend/resume. A `kill -STOP' almost | |
977 | works right now.) | |
978 | ||
979 | (Done. I needed to play with signal handling and the server | |
980 | protocol a bit to make emacsclient behave as a normal UNIX program | |
981 | wrt foreground/background process groups.) | |
982 | ||
983 | -- There is a flicker during the startup of `emacs -nw'; it's as if | |
984 | the terminal is initialized, reset and then initialialized again. | |
985 | Debug this. (Hint: narrow_foreground_group is called twice during | |
986 | startup.) | |
987 | ||
988 | (This is gone.) | |
989 | ||
990 | -- Robert Chassell has found serious copy-paste bugs with the | |
991 | multi-tty branch. There seem to be redisplay bugs while copying | |
992 | from X to a terminal frame. Copying accented characters do not | |
993 | work for me. | |
994 | ||
995 | (Patch-124 should fix this, by changing the interprogram-*-function | |
996 | variables to be frame-local, as suggested by Mark Plaksin | |
997 | (thanks!). I think that the redisplay bugs are in fact not bugs, | |
998 | but delays caused by single_kboard --> perhaps MULTI_KBOARD should | |
999 | be removed.) | |
1000 | ||
1001 | -- frame-creation-function was removed, which might be a bad idea. | |
1002 | Think up a compatible solution. | |
1003 | ||
1004 | (It was an internal interface that may be changed when necessary.) | |
1005 | ||
1006 | -- Change Lisp code not to (getenv "TERM"); use the `tty-type' frame | |
1007 | parameter or the frame-tty-type function instead. (M-x tags-search | |
1008 | "TERM" helps with this.) Update: Actually, all getenv invocations | |
1009 | should be checked for multi-tty compatibility, and an interface | |
1010 | must be implemented to get the remote client's environment. | |
1011 | ||
1012 | (Done. Only getenv calls in lisp/term/*.el were changed; other | |
1013 | calls should be mostly left as they are.) | |
1014 | ||
1015 | -- Add an elaborate mechanism for display-local variables. (There are | |
1016 | already a few of these; search for `terminal-local' in the Elisp | |
1017 | manual.) | |
1018 | ||
1019 | (Not needed. Display-local variables could be emulated by | |
1020 | frame-local variables.) | |
1021 | ||
1022 | -- Emacs assumes that all terminal frames have the same locale | |
1023 | settings as Emacs itself. This may lead to bogus results in a | |
1024 | multi-locale setup. (E.g., while logging in from a remote client | |
1025 | with a different locale.) | |
1026 | (Update after new bugreport by Friedrich Delgado Friedrichs: | |
1027 | (at least) the structs terminal_coding and keyboard_coding in | |
1028 | coding.c must be moved to struct display, and the Lisp interface | |
1029 | [set-]keyboard-coding-system must be adapted for the change.) | |
1030 | ||
1031 | (Fixed. Emacs now uses the locale settings as seen by the | |
1032 | emacsclient process for server tty frames.) | |
1033 | (Update: Not really; Vlocale_coding_system is still global.) | |
1034 | ||
1035 | -- Make `struct display' accessible to Lisp programs. Accessor functions: | |
1036 | ||
1037 | (displayp OBJECT): Returns t if OBJECT is a display. | |
1038 | => Implemented as display-live-p. | |
1039 | ||
1040 | (display-list): Returns list of currently active displays. | |
1041 | => Implemented. | |
1042 | ||
1043 | (selected-display): Returns the display object of the selected frame. | |
1044 | => Not strictly necessary, but implemented anyway. | |
1045 | ||
1046 | (frame-display FRAME): Returns the display object of FRAME. | |
1047 | => Implemented. | |
1048 | ||
1049 | (display-frames DISPLAY): Returns a list of frames on DISPLAY. | |
1050 | => Already implemented, see frames-on-display-list. | |
1051 | ||
1052 | (display-type DISPLAY): Returns the type of DISPLAY, as a | |
1053 | symbol. (See `framep'.) | |
1054 | => Implemented as display-live-p. | |
1055 | ||
1056 | (display-device DISPLAY): Returns the name of the device that | |
1057 | DISPLAY uses, as a string. (E.g: "/dev/pts/16", or | |
1058 | ":0.0") | |
1059 | => Implemented as display-name. | |
1060 | ||
1061 | etc. | |
1062 | ||
1063 | See next issue why this is necessary. | |
1064 | ||
1065 | (Update: The consensus on emacs-devel seems to be to do this via | |
1066 | integer identifiers. That's fine by me.) | |
1067 | ||
1068 | (Done.) | |
1069 | ||
1070 | -- The following needs to be supported: | |
1071 | ||
1072 | $ emacsclient -t | |
1073 | C-z | |
1074 | $ emacsclient -t | |
1075 | (This fails now.) | |
1076 | ||
1077 | The cleanest way to solve this is to allow multiple displays on the | |
1078 | same terminal device; each new emacsclient process should create | |
1079 | its own display. As displays are currently identified by their | |
1080 | device names, this is not possible until struct display becomes | |
1081 | accessible as a Lisp-level object. | |
1082 | ||
1083 | (Done.) | |
1084 | ||
1085 | -- Miles Bader suggests that C-x C-c on an emacsclient frame should | |
1086 | only close the frame, not exit the entire Emacs session. Update: | |
1087 | see above for a function that does this. Maybe this should be the | |
1088 | new default? | |
1089 | ||
1090 | (Done. This is the new default. No complaints so far.) | |
1091 | ||
1092 | -- Clean up the frame-local variable system. I think it's ugly and | |
1093 | error-prone. But maybe I just haven't yet fully understood it. | |
1094 | ||
1095 | (Nothing to do. It doesn't seem ugly any more. It's rather clever.) | |
1096 | ||
1097 | -- Support multiple character locales. A version of | |
1098 | `set-locale-environment' needs to be written for setting up | |
1099 | display-local settings on ttys. I think calling | |
1100 | set-display-table-and-terminal-coding-system and | |
1101 | set-keyboard-coding-system would be enough. The language | |
1102 | environment itself should remain a global setting. | |
1103 | ||
1104 | (Done, by an ugly hack.) | |
1105 | ||
1106 | -- The terminal customization files in term/*.el tend to change global | |
1107 | parameters, which may confuse Emacs with multiple displays. Change | |
1108 | them to tweak only frame-local settings, if possible. (They tend | |
1109 | to call define-key to set function key sequences a lot.) | |
1110 | ||
1111 | (Done, by making `function-key-map' terminal-local (i.e., part of | |
1112 | struct kboard). This has probably covered all the remaining problems.) | |
1113 | ||
1114 | -- Make `function-key-map' and `key-translation-map' terminal-local. | |
1115 | ||
1116 | (Done.) | |
1117 | ||
1118 | -- Implement `terminal-local-value' and `set-terminal-local-value' to | |
1119 | allow deterministic access to terminal local bindings. The | |
1120 | encode-kb package can not set up `key-translation-map' without | |
1121 | these. The terminal-local bindings seem to be independent of what | |
1122 | frame is selected. | |
1123 | ||
1124 | (Done.) | |
1125 | ||
1126 | -- xt-mouse.el needs to be adapted for multi-tty. It currently | |
1127 | signals an error on kill-emacs under X, which prevents the user | |
1128 | from exiting Emacs. (Reported by Mnemonikk on freenode.) | |
1129 | ||
1130 | (Done, I hope.) | |
1131 | ||
1132 | ||
1133 | -- Having {reset,init}_all_sys_modes in set-input-mode breaks arrow | |
1134 | keys on non-selected terminals under screen, and sometimes on other | |
1135 | terminal types as well. The other function keys continue to work | |
1136 | fine. Sometimes faces on these screens become garbled. | |
1137 | ||
1138 | This only seems to affect displays that are of the same terminfo | |
1139 | type as the selected one. Interestingly, in screen Emacs normally | |
1140 | reports the up arrow key as `M-o A', but after the above SNAFU, it | |
1141 | complains about `M-[ a'. UNIX ttys are a complete mystery to me, | |
1142 | but it seems the reset-reinitialize cycle somehow leaves the | |
1143 | non-selected terminals in a different state than usual. I have no | |
1144 | idea how this could happen. | |
1145 | ||
1146 | Currently set-input-mode resets only the currently selected | |
1147 | terminal, which seems to somehow work around the problem. | |
1148 | ||
1149 | Update: | |
1150 | ||
1151 | Dan Nicolaescu <dann@ics.uci.edu> writes: | |
1152 | > Some terminals have 2 modes for cursor keys: Application Mode where | |
1153 | > the cursor keys transmit the codes defined in the terminfo entry, and | |
1154 | > Cursor mode. Applications have to send the smkx and rmkx terminfo | |
1155 | > strings to switch between the 2 modes. So Emacs (and emacsclient) have | |
1156 | > to send smkx when initializing and rmkx when quitting (or on | |
1157 | > suspend). | |
1158 | ||
1159 | (I think patch-370 fixed this.) | |
1160 | ||
1161 | -- This long-standing bug (first reported by Han Boetes) seems to come | |
1162 | and go all the time. It is time to track it down and fix it. | |
1163 | ||
1164 | emacs | |
1165 | M-x server-start | |
1166 | ||
1167 | # From another xterm: | |
1168 | emacsclient -e '(y-or-n-p "Do you want me to crash? ")' | |
1169 | # Notice how the answer ends up in the *scratch* buffer | |
1170 | M-x garbage-collect | |
1171 | SIGSEGV | |
1172 | ||
1173 | (Fixed in patch-414 after detailed analysis by Kalle Olavi Niemitalo.) | |
1174 | ||
1175 | -- normal-erase-is-backspace-mode in simple.el needs to be updated for | |
1176 | multi-tty (rep. by Dan Waber). (The Delete key is broken on X | |
1177 | because of this.) | |
1178 | ||
1179 | (Fixed in patch-427.) | |
1180 | ||
1181 | -- I think keyboard-translate-table should be made terminal-local. | |
1182 | ||
1183 | (Done in patch-431.) | |
1184 | ||
1185 | -- The semantics of terminal-local variables are confusing; it is not | |
1186 | clear what binding is in effect in any given time. See if | |
1187 | current_kboard (or at least the terminal-local bindings exported to | |
1188 | Lisp) might be changed to be tied to the selected frame instead. | |
1189 | Currently, `function-key-map' and `key-translation-map' may be | |
1190 | accessed reliably only using the hackish | |
1191 | `(set-)terminal-local-value' functions. | |
1192 | ||
1193 | Perhaps there should be a difference between `last-command' &co. | |
1194 | and these more conventional configuration variables. | |
1195 | (E.g. `symbol-value' would use current_kboard to access | |
1196 | `last-command', but SELECTED_FRAME()->display->kboard to get the | |
1197 | value of `function-key-map'. | |
1198 | ||
1199 | (Fixed in patch-434.) | |
1200 | ||
1201 | -- If the first key pressed on a new tty terminal is a function key, | |
1202 | it is not recognized correctly. May be related to the bug below. | |
1203 | ||
1204 | (Seems to have been fixed as a side effect of patch-434. "The bug | |
1205 | below" was the set-input-mode madness.) | |
1206 | ||
1207 | (Update: this bug was fixed for good in patch-449. It was tracked | |
1208 | down to a bug in `read_key_sequence': it failed to reinitialize its | |
1209 | local function-key-map/key-translation-map references when it | |
1210 | switched keyboards. I don't understand why did this bug only | |
1211 | appear on brand new frames, though!) | |
1212 | ||
1213 | -- Disable connecting to a new X display when we use the GTK toolkit. | |
1214 | ||
1215 | (Disabled in patch-450.) | |
1216 | ||
1217 | -- Implement automatic forwarding of client environment variables to | |
1218 | forked processes, as discussed on the multi-tty list. Terminal | |
1219 | parameters are now accessible in C code, so the biggest obstacle is | |
1220 | gone. The `getenv_internal' and `child_setup' functions in | |
1221 | callproc.c must be changed to support the following variable: | |
1222 | ||
1223 | terminal-local-environment-variables is a variable defined in ... | |
1224 | ||
1225 | Enable or disable terminal-local environment variables. | |
1226 | ||
1227 | If set to t, `getenv', `setenv' and subprocess creation | |
1228 | functions use the environment variables of the emacsclient | |
1229 | process that created the selected frame, ignoring | |
1230 | `process-environment'. | |
1231 | ||
1232 | If set to nil, Emacs uses `process-environment' and ignores | |
1233 | the client environment. | |
1234 | ||
1235 | Otherwise, `terminal-local-environment-variables' should be a | |
1236 | list of variable names (represented by Lisp strings) to look | |
1237 | up in the client environment. The rest will come from | |
1238 | `process-environment'. | |
1239 | ||
1240 | (Implemented in patch-461; `terminal-getenv', `terminal-setenv' and | |
1241 | `with-terminal-environment' are now replaced by extensions to | |
1242 | `getenv' and `setenv', and the new `local-environment-variables' | |
1243 | facility. Yay!) | |
1244 | ||
1245 | (Updated in patch-465 to fix the semantics of let-binding | |
1246 | `process-environment'. `process-environment' was changed to | |
1247 | override all local/global environment variables, and a new variable | |
1248 | `global-environment' was introduced to have `process-environment's | |
1249 | old meaning.) | |
1250 | ||
1251 | (Updated in patch-466 to fix the case when two emacsclient sessions | |
1252 | share the same terminal, but have different environment. The local | |
1253 | environment lists are now stored as frame parameters, so the | |
1254 | C-level terminal parameters are not strictly necessary any more.) | |
1255 | ||
1256 | -- `Fdelete_frame' is called from various critical places where it is | |
1257 | not acceptable for the frame deletion to fail, e.g. from | |
1258 | x_connection_closed after an X error. `Fdelete_frame' now protects | |
1259 | against `delete-frame-functions' throwing an error and preventing a | |
1260 | frame delete. (patch-475) | |
1261 | ||
1262 | -- Fix set-input-mode for multi-tty. It's a truly horrible interface; | |
1263 | what if we'd blow it up into several separate functions (with a | |
1264 | compatibility definition)? | |
1265 | ||
1266 | (Done. See `set-input-interrupt-mode', `set-output-flow-control', | |
1267 | `set-input-meta-mode' and `set-quit-char'.) (patch-457) | |
1268 | ||
1269 | -- Let-binding `overriding-terminal-local-map' on a brand new frame | |
1270 | does not seem to work correctly. (See `fancy-splash-screens'.) | |
1271 | The keymap seems to be set up right, but events go to another | |
1272 | terminal. Or is it `unread-command-events' that gets Emacs | |
1273 | confused? Investigate. | |
1274 | ||
1275 | (Emacs was confused because a process filter entered | |
1276 | `recursive-edit' while Emacs was reading input. I added support | |
1277 | for this in the input system.) (patch-489) | |
1278 | ||
1279 | -- I smell something funny around pop_kboard's "deleted kboard" case. | |
1280 | Determine what are the circumstances of this case, and fix any | |
1281 | bug that comes to light. | |
1282 | ||
1283 | (It happens simply because single_kboard's terminal is sometimes | |
1284 | deleted while executing a command on it, for example the one that | |
1285 | kills the terminal. There was no bug here, but I rewrote the whole | |
1286 | single_kboard mess anyway.) (patch-489) | |
1287 | ||
1288 | -- Understand Emacs's low-level input system (it's black magic) :-) | |
1289 | What exactly does interrupt_input do? I tried to disable it for | |
1290 | raw secondary tty support, but it does not seem to do anything | |
1291 | useful. (Update: Look again. X unconditionally enables this, maybe | |
1292 | that's why raw terminal support is broken again. I really do need | |
1293 | to understand input.) | |
1294 | (Update: I am starting to understand the read_key_sequence->read-char | |
1295 | ->kbd_buffer_get_event->read_avail_input->read_socket_hook path. Yay!) | |
1296 | ||
1297 | (Update: OK, it all seems so easy now (NOT). Input could be done | |
1298 | synchronously (with wait_reading_process_input), or asynchronously | |
1299 | by SIGIO or polling (SIGALRM). C-g either sets the Vquit_flag, | |
1300 | signals a 'quit condition (when immediate_quit), or throws to | |
1301 | `getcjmp' when Emacs was waiting for input when the C-g event | |
1302 | arrived.) | |
1303 | ||
1304 | -- Replace wrong_kboard_jmpbuf with a special return value of | |
1305 | read_char. It is absurd that we use setjmp/longjmp just to return | |
1306 | to the immediate caller. | |
1307 | ||
1308 | (Done in patch-500.) | |
1309 | ||
1310 | -- `tool-bar-mode', `scroll-bar-mode', `menu-bar-mode' and | |
1311 | 'fringe-mode' are modes global to the entire Emacs session, not | |
1312 | just a single frame or a single terminal. This means that their | |
1313 | status sometimes differs from what's actually displayed on the | |
1314 | screen. As a consequence, the Options | Show/Hide menu sometimes | |
1315 | shows incorrect status, and you have to select an option twice for | |
1316 | it to have any visible effect on the current frame. | |
1317 | ||
1318 | Change Emacs so that the status of the items in the Options | | |
1319 | Show/Hide menu correspond to the current frame. | |
1320 | ||
1321 | (Done in patch-537.) | |
1322 | ||
1323 | -- The `default-directory' variable should somehow be set to the | |
1324 | cwd of the emacsclient process when the user runs emacsclient | |
1325 | without file arguments. Perhaps it is OK to just override the | |
1326 | directory of the *scratch* buffer. | |
1327 | ||
1328 | (Done in patch-539.) | |
1329 | ||
1330 | -- The borders on tooltip frames on X are messed up. More | |
1331 | specifically, the frame's internal border (internal-border-width | |
1332 | frame parameter) is not filled with the correct background color. | |
1333 | ||
1334 | It seems the border contents are drawn onto by the | |
1335 | update_single_window call in `x-show-tip'. After some debugging, I | |
1336 | think the window's background color is not set up | |
1337 | correctly---calling `x_clear_area' fills the specified area with | |
1338 | black, not light yellow. | |
1339 | ||
1340 | (Done in patch-544. A background_pixel field was defined both in | |
1341 | struct frame and struct x_output, and Emacs got confused between | |
1342 | them.) | |
1343 | ||
1344 | \f | |
1345 | This file is part of GNU Emacs. | |
1346 | ||
1347 | GNU Emacs is free software; you can redistribute it and/or modify | |
1348 | it under the terms of the GNU General Public License as published by | |
1349 | the Free Software Foundation; either version 3, or (at your option) | |
1350 | any later version. | |
1351 | ||
1352 | GNU Emacs is distributed in the hope that it will be useful, | |
1353 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
1354 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
1355 | GNU General Public License for more details. | |
1356 | ||
1357 | You should have received a copy of the GNU General Public License | |
1358 | along with GNU Emacs; see the file COPYING. If not, write to the | |
1359 | Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, | |
1360 | Boston, MA 02110-1301, USA. | |
1361 | ||
1362 | ;;; arch-tag: 8da1619e-2e79-41a8-9ac9-a0485daad17d |