1 -*- coding: utf-8; mode: text; -*-
5 The ultimate goal of this branch is to implement support for opening
6 multiple, different tty devices and simultaneous X and tty frames from
7 a single Emacs session.
11 Emacs is notoriously slow at startup, so most people use another
12 editor or emacsclient for quick editing jobs from the console.
13 Unfortunately, emacsclient was very awkward to use, because it did not
14 support opening a new Emacs frame on the current virtual console.
15 Now, with multi-tty support, it can do that. (Emacsclient starts up
18 Some Gnus users (including me) run Gnus in an X frame in its own Emacs
19 instance, which they typically leave running for weeks. It would be
20 nice if they could connect to this instance from a remote ssh session
21 and check their messages without opening a remote X frame or resorting
28 I'm Károly Lőrentey. My address: lorentey@elte.hu.
30 Patches or suggestions are welcome!
32 Retrieving the latest version of the branch:
34 tla register-archive lorentey@elte.hu--2004 http://lorentey.web.elte.hu/arch/2004/
35 tla get lorentey@elte.hu--2004/emacs--multi-tty <directory>
37 (I use a recent arch development snapshot, but any of the released
38 versions of arch will do fine, I think.)
40 If you don't have arch, the branch has a homepage from which you can
41 download conventional patches against Emacs CVS HEAD:
43 http://lorentey.web.elte.hu/project/emacs.html
48 Multi-tty support is stable, I think most of the problems were fixed.
49 (It still needs testing on other architectures, though.) Please let
50 me know if you find any bugs in it. Emacsclient has been extended to
51 support opening a new terminal frame.
53 To try it out, compile and run the multi-tty branch with the following
63 and then (from a shell prompt on another terminal) start emacsclient
66 lib-src/emacsclient -t /optional/file/names...
68 You'll hopefully have two fully working, independent frames on
69 separate terminals. The new frame is closed automatically when you
70 have finished editing the specified files (C-x #), but delete-frame
71 (C-x 5 0) also works. Of course, you can create frames on more than
74 Creating new frames on the same tty with C-x 5 2 works, and they
75 behave the same way as in previous Emacs versions. If you exit emacs,
76 all terminals should be restored to their previous states.
78 This is work in progress, and probably full of bugs. You should
79 always run emacs from gdb, so that you'll have a live instance to
80 debug if something goes wrong. Please send me your reports.
84 * Suspending Emacs is disabled if there are multiple tty
85 devices. Also, there is no way to suspend emacsclient. This
88 * X support is (I hope) working, but at the moment there are
89 problems with simultaneous X and tty devices, so don't do
90 that - start a separate Emacs with -nw and run the server
93 * Mac, Windows and DOS support is broken, probably doesn't
94 even compile -- this will be solved later.
96 * Only tested on my GNU/Linux box.
103 ** Support for multiple terminal devices has been added. You can
104 specify a terminal device (`tty' parameter) and a terminal type
105 (`tty-type' parameter) to `make-terminal-frame'. `tty' must be a
106 terminal device created by the updated emacsclient, or there will
107 be problems with terminal input and window resizes.
109 You can test for the presence of multiple terminal support by
110 testing for the `multi-tty' feature.
112 ** A make-frame-on-tty function has been added to make it easier to
113 create frames on new terminals.
115 ** New functions: frame-tty-name, frame-tty-type for accessing
116 terminal parameters, and delete-tty for closing the terminal
119 ** Emacsclient has been extended to support opening a new terminal
130 ** Fix color handling during X+tty combo sessions. (It seems that tty
131 sessions automatically convert the face colors to terminal colors
132 when the face is loaded. This conversion must happen instead on
133 the fly in write_glyphs, which might be problematic, as color
134 approximation is currently done in lisp (term/tty-colors.el).)
136 ** Fix interactive use of temacs. There are face-related SEGVs, most
137 likely because of changes in realize_default_face, realize_face.
139 ** Very strange bug: visible-bell does not work on secondary
140 terminals. This might be something xterm (konsole) specific.
142 ** Allow opening an X session after -nw.
144 ** Find out the best way to support suspending Emacs with multiple
145 ttys. My guess: disable it on the controlling tty, but from other
146 ttys pass it on to emacsclient somehow. (It is (I hope) trivial to
147 extend emacsclient to handle suspend/resume. A `kill -STOP' almost
150 ** Exiting Emacs while there are emacsclient frames doesn't restore the
151 ttys to their default states.
153 ** Move baud_rate to struct display.
155 ** Implement support for starting an interactive Emacs session without
156 an initial frame. (The user would connect to it and open frames
157 later, with emacsclient.) Not necessarily a good idea.
159 ** Fix Mac support (I can't do this myself). Note that the current
160 state of Mac-specific source files in the multi-tty tree are not
161 useful; before starting work on Mac support, revert to pristine,
162 pre-multi-tty versions.
164 ** Fix W32 support (I can't do this myself). Note that the current
165 state of W32-specific source files in the multi-tty tree are not
166 useful; before starting work on W32 support, revert to pristine,
167 pre-multi-tty versions.
169 ** Fix DOS support (I can't do this myself). Note that the current
170 state of DOS-specific source files in the multi-tty tree are not
171 useful; before starting work on DOS support, revert to pristine,
172 pre-multi-tty versions.
174 ** Do a grep on XXX and ?? for more issues.
176 ** Understand Emacs's low-level input system (it seems complicated) :-)
177 What does interrupt_input do? I tried to disable it for raw
178 secondary tty support, but it does not seem to do anything useful.
179 (Update: Look again. X unconditionally enables this, maybe that's
180 why raw terminal support is broken again. I really do need to
183 ** I have seen a case when Emacs with multiple ttys fell into a loop
184 eating 100% of CPU time. Strace showed this loop:
187 kill(30284, SIGIO) = 0
188 --- SIGIO (I/O possible) @ 0 (0) ---
189 ioctl(6, FIONREAD, [0]) = -1 EIO (Input/output error)
190 ioctl(5, FIONREAD, [0]) = -1 EIO (Input/output error)
191 ioctl(0, FIONREAD, [0]) = 0
192 sigreturn() = ? (mask now [])
193 gettimeofday({1072842297, 747760}, NULL) = 0
194 gettimeofday({1072842297, 747806}, NULL) = 0
195 select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0})
196 select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0})
197 gettimeofday({1072842297, 748245}, NULL) = 0
199 I have seen something similar with a single X frame, but have not
200 been able to reproduce it for debugging.
202 Update: This may have been caused by checking for nread != 0
203 instead of nread > 0 after calling read_socket_hook in
206 ** emacsclient -t from an Emacs term buffer does not work, complains
207 about face problems. This can even lock up Emacs (if the recursive
208 frame sets single_kboard). Update: the face problems are caused by
209 bugs in term.el, not in multi-tty. The lockup is caused by
210 single_kboard mode, and is not easily solvable. The best thing to
211 do is to simply refuse to create a tty frame of type `eterm'.
216 (ex-TODO items with explanations.)
218 -- Introduce a new struct for terminal devices.
220 (Done, see struct tty_output. The list of members is not yet
223 -- Change the bootstrap procedure to initialize tty_list.
225 (Done, but needs review.)
227 -- Change make-terminal-frame to support specifying another tty.
229 (Done, new frame parameters: `tty' and `tty-type'.)
231 -- Implement support for reading from multiple terminals.
233 (Done, read_avail_input tries to read from each terminal, until one
234 succeeds. MULTI_KBOARD is not used. Secondary terminals don't send
237 (Update: They do, now.)
239 (Update2: After enabling X, they don't.)
241 -- other-frame should cycle through the frames on the `current'
244 (Done, by trivially modifiying next_frame and prev_frame.)
246 -- Support different terminal sizes.
250 -- Make sure terminal resizes are handled gracefully. (Could be
253 (Done. We don't get automatic SIGWINCH for additional ttys,
256 -- Extend emacsclient to automatically open a new tty when it connects
259 (Done. It's an ugly hack, needs more work.)
261 -- Redisplay must refresh the topmost frame on *all* terminals, not
262 just the initial terminal.
264 (Done, but introduced an ugly redisplay problems. Ugh.)
266 -- Fix redisplay problems.
268 (Done; it turned out that the entire Wcm structure must be moved
269 inside tty_output. Why didn't I catch this earlier?)
271 -- Provide a way for emacsclient to tell Emacs that the tty has been
274 (Done, simply forward the SIGWINCH signal.)
276 -- Each keypress should automatically select the frame corresponding
277 to the terminal that it was coming from. This means that Emacs
278 must know from which terminal the last keyboard event came from.
280 (Done, it was quite simple, the input event system already
281 supported multiple frames.)
283 -- Fix SIGIO issue with secondary terminals.
285 (Done, emacsclient signals Emacs after writing to the proxy pseudo
286 terminal. Note that this means that multi-tty does not work with
289 (Update: This is bullshit. There is a read_input_waiting function,
290 extend that somehow.)
292 (Update of update: The first update was not right either, extending
293 read_input_waiting was not necessary. Secondary ttys do seem to
294 send signals on input.)
296 (Update^3: Not any more.)
298 -- Make make-terminal-frame look up the `tty' and `tty-type' frame
299 parameters from the currently selected terminal before the global
304 -- Put all cached terminal escape sequences into struct tty_output.
305 Currently, they are still stored in global variables, so we don't
306 really support multiple terminal types.
308 (Done. It was not fun.)
310 -- Implement sane error handling after initialization. (Currently
311 emacs exits if you specify a bad terminal type.) The helpful error
312 messages must still be provided when Emacs starts.
316 -- Implement terminal deletion, i.e., deleting local frames, closing
317 the tty device and restoring its previous state without exiting
320 (Done, but at the moment only called when an error happens during
321 initialization. There is a memory corruption error around this
322 somewhere.) (Update: now it is fully enabled.)
324 -- Implement automatic deletion of terminals when the last frame on
325 that terminal is closed.
329 -- Restore tty screen after closing the terminal.
331 (Done, we do the same as Emacs 21.2 for all terminals.)
333 -- 'TERM=dumb src/emacs' does not restore the terminal state.
337 -- C-g should work on secondary terminals.
339 (Done, but the binding is not configurable.)
341 -- Deal with SIGHUP in Emacs and in emacsclient. (After this, the
342 server-frames may be removed from server.el.)
344 (Done, nothing to do. It seems that Emacs does not receive SIGHUP
345 from secondary ttys, which is actually a good thing.) (Update: I
346 think it would be a bad idea to remove server-frames.)
348 -- Change emacsclient/server.el to support the -t argument better,
349 i.e. automatically close the socket when the frame is closed.
351 (Seems to be working OK.)
353 -- Fix mysterious memory corruption error with tty deletion. To
354 trigger it, try the following shell command:
356 while true; do TERM=no-such-terminal-definition emacsclient -h; done
358 Emacs usually dumps core after a few dozen iterations. (The bug
359 seems to be related to the xfreeing or bzeroing of
360 tty_output.Wcm. Maybe there are outside references to struct Wcm?
361 Why were these vars collected into a struct before multi-tty
364 (Done. Whew. It turned out that the problem had nothing to do
365 with hypothetical external references to Wcm, or any other
366 tty_output component; it was simply that delete_tty closed the
367 filehandles of secondary ttys twice, resulting in fclose doubly
368 freeing memory. Utterly trivial matter. I love the C's memory
369 management, it puts hair on your chest.)
371 -- Support raw secondary terminals. (Note that SIGIO works only on
372 the controlling terminal.) Hint: extend read_input_waiting for
373 multiple ttys and hopefully this will be fixed.
375 (Done, it seems to have been working already for some time. It
376 seems F_SETOWN does work, after all. Not sure what made it fail
377 earlier, but it seems to be fixed (there were several changes
378 around request_sigio, maybe one of them did it).
379 read_input_waiting is only used in sys_select, don't change
380 it.) (Update: After adding X support, it's broken again.)
382 -- Find out why does Emacs abort when it wants to close its
383 controlling tty. Hint: chan_process[] array. Hey, maybe
384 noninterrupt-IO would work, too? Update: no, there is no process
387 (Done. Added add/delete_keyboard_wait_descriptor to
388 term_init/delete_tty. The hint was right, in a way.)
390 -- Issue with SIGIO: it needs to be disabled during redisplay. See if
391 fcntl kernel behaviour could be emulated by emacsclient.
393 (Done. Simply disabled the SIGIO emulation hack in emacsclient.)
394 (Update: it was added back.)
396 -- server.el: There are issues with saving files in buffers of closed
397 clients. Try editing a file with emacsclient -f, and (without
398 saving it) do a delete-frame. The frame is closed without
399 question, and a surprising confirmation prompt appears in another
402 (Done. delete-frame now asks for confirmation if it still has
403 pending buffers, and modified buffers don't seem to be deleted.)
405 -- emacsclient.el, server.el: Handle eval or file open errors when
410 -- Make parts of struct tty_output accessible from Lisp. The device
411 name and the type is sufficient.
413 (Done, see frame-tty-name and frame-tty-type.)
415 -- Export delete_tty to the Lisp environment, for emacsclient.
417 (Done, see delete-tty.)
419 -- Get rid of the accessor macros in termchar.h, or define macros for
424 -- Move device-specific parameters (like costs) commonly used by
425 device backends to a common, device-dependent structure.
427 (Done. See struct display_method in termhooks.h.)
431 (Done. Well, it seems to be working.)
433 -- Allow simultaneous X and tty frames. (Handling input could be
434 tricky. Or maybe not.)
436 (Done. Allowed, that is. It is currently extremely unstable, to
437 the point of being unusable. The rif variable causes constant
438 core dumps. Handling input is indeed tricky.)
440 -- Rewrite multi-tty input in terms of MULTI_KBOARD.
442 (Done. In fact, there was no need to rewrite anything, I just
443 added a kboard member to tty_display_info, and initialized the
444 frame's kboard from there.)
446 -- Fix rif issue with X-tty combo sessions. IMHO the best thing to do
447 is to get rid of that global variable (and use the value value in
448 display_method, which is guaranteed to be correct).
450 (Done, did exactly that. Core dumps during combo sessions became
451 much rarer. In fact, I have not yet met a single one.)
453 -- Add multi-tty support to talk.el.
457 -- Clean up the source of emacsclient. It is a mess.
459 (Done, eliminated stupid proxy-pty kludge.)
461 -- Fix faces on tty frames during X-tty combo sessions. There is an
462 init_frame_faces call in init_sys_modes, see if there is a problem
465 (Done, there was a stupid mistake in
466 Ftty_supports_face_attributes_p. Colors are broken, though.)
468 -- C-x 5 2, C-x 5 o, C-x 5 0 on an emacsclient frame unexpectedly
469 exits emacsclient. This is a result of trying to be clever with
470 delete-frame-functions.
472 (Fixed, added delete-tty-after-functions, and changed server.el to
475 -- Something with (maybe) multi-keyboard support broke function keys
476 and arrows on ttys during X+tty combo sessions. Debug this.
478 (I can't reproduce it, maybe the terminal type was wrong.)
480 -- Fix input from raw ttys (again).
482 (Now it seems to work all right.)
484 -- During an X-tty combo session, a (message "Hello") from a tty frame
485 goes to the X frame. Fix this.
487 (Done. There was a safeguard against writing to the initial
488 terminal frame during bootstrap which prevented echo_area_display
489 from working correctly on a tty frame during a combo session.)
491 -- If there are no frames on its controlling terminal, Emacs should
492 exit if the uses presses C-c there.
494 (Done, as far as possible. See the SIGTERM comment in
495 interrupt_signal on why this seems to be impossible to solve this
498 -- During an X session, Emacs seems to read from stdin. Also, Emacs
499 fails to start without a controlling tty.
501 (Fixed by replacing the troublesome termcap display with a dummy
502 bootstrap display during bootstrap.
504 -- Do tty output through struct display, like graphical display
509 -- Define an output_initial value for output_method for the initial
510 frame that is dumped with Emacs. Checking for this frame (e.g. in
511 cmd_error_internal) is ugly.
513 (Done, broking interactive temacs.)
515 -- The command `emacsclient -t -e '(delete-frame)'' fails to exit.
519 -- frame-creation-function should always create a frame that is on the
520 same display as the selected frame. Maybe frame-creation-function
521 should simply be removed and make-frame changed to do the right
524 (Done, with a nice hack. frame-creation-function is now frame-local.)
526 -- Fix C-g on raw ttys.
528 (Done. I disabled the interrupt/quit keys on all secondary
529 terminals, so Emacs sees C-g as normal input. This looks like an
530 overkill, because emacsclient has extra code to pass SIGINT to
531 Emacs, so C-g should remain the interrupt/quit key on emacsclient
532 frames. See the next entry why implementing this distinction would
535 -- Make sure C-g goes to the right frame with ttys. This is hard, as
536 SIGINT doesn't have a tty parameter. :-(
538 (Done, the previous change fixes this as a pleasant side effect.)
541 ;;; arch-tag: 8da1619e-2e79-41a8-9ac9-a0485daad17d