1 -*- coding: utf-8; mode: text; -*-
5 This branch implements support for opening multiple, different tty
6 devices and simultaneous X and tty frames from a single Emacs session.
9 Emacs is notoriously slow at startup, so most people use another
10 editor or emacsclient for quick editing jobs from the console.
11 Unfortunately, emacsclient was very awkward to use, because it did not
12 support opening a new Emacs frame on the current virtual console.
13 Now, with multi-tty support, it can do that. (Emacsclient starts up
16 Some Gnus users (including me) run Gnus in an X frame in its own Emacs
17 instance, which they typically leave running for weeks. It would be
18 nice if they could connect to this instance from a remote ssh session
19 and check their messages without opening a remote X frame or resorting
25 I'm Károly Lőrentey. My address: lorentey@elte.hu.
27 Comments, bug reports, suggestions and patches are welcome; send them
28 to multi-tty@lists.fnord.hu.
30 The following is a (sadly incomplete) list of people who have
31 contributed to the project by testing, submitting patches, bug
32 reports, and suggestions. Thanks!
34 Bernard Adrian <bernadrian@free.fr>
35 ARISAWA Akihiro <ari@mbf.ocn.ne.jp>
36 Vincent Bernat <bernat@luffy.cx>
37 Han Boetes <han@mijncomputer.nl>
38 Francisco Borges <borges@let.rug.nl>
39 Damien Cassou <damien.cassou@laposte.net>
40 Robert J. Chassell <bob@rattlesnake.com>
41 Romain Francoise <romain@orebokech.com>
42 Ami Fischman <ami@fischman.org>
43 Friedrich Delgado Friedrichs <friedel@nomaden.org>
44 IRIE Tetsuya <irie@t.email.ne.jp>
45 Yoshiaki Kasahara <kasahara@nc.kyushu-u.ac.jp>
46 Bas Kok <nekkobassu@yahoo.com>
47 Jurej Kubelka <Juraj.Kubelka@email.cz>
48 David Lichteblau <david@lichteblau.com>
49 Xavier Mallard <zedek@gnu-rox.org>
50 Istvan Marko <mi-mtty@kismala.com>
51 Ted Morse <morse@ciholas.com>
52 Gergely Nagy <algernon@debian.org>
53 Dan Nicolaescu <dann@ics.uci.edu>
54 Kalle Olavi Niemitalo <kon@iki.fi>
55 Mark Plaksin <happy@mcplaksin.org>
56 Frank Ruell <stoerte@dreamwarrior.net>
57 Tom Schutzer-Weissmann <trmsw@yahoo.co.uk>
58 Joakim Verona <joakim@verona.se>
59 Dan Waber <dwaber@logolalia.com>
62 Richard Stallman was kind enough to review an earlier version of my
69 The multi-tty mailing list (discussion & bug reports):
71 Address: multi-tty@lists.fnord.hu
72 Signup: http://lists.fnord.hu/mailman/listinfo/multi-tty
73 Archive: http://news.gmane.org/gmane.emacs.multi-tty/
75 Commit notifications (read-only):
77 Address: multi-tty-commits@lists.fnord.hu
78 Signup: http://lists.fnord.hu/mailman/listinfo/multi-tty-commits
84 The branch is now very stable and almost full-featured. All of the
85 major problems have been fixed, only a few minor issues remain. (It
86 still needs to be ported to Windows/Mac/DOS, though.) Both multiple
87 tty device support and simultaneous X and tty frame support works
88 fine. Emacsclient has been extended to support opening new tty and X
89 frames. It has been changed open new Emacs frames by default.
91 The multi-tty branch has been scheduled for inclusion in the next
92 major release of Emacs (version 23). I expect the merge into the
93 development trunk to occur sometime during next year (2006), after the
94 merge of the Unicode branch.
96 Tested on GNU/Linux, Solaris 8, FreeBSD and OpenBSD. Please let me
97 know if you succeed or fail to use it on other platforms---I'll have a
98 few tricky test cases for you.
102 * GTK support. If you compile your Emacs with the GTK
103 toolkit, some functionality of multi-tty will be lost. In
104 particular, you will not be able to work on multiple X
105 displays at once. Current releases of GTK have limitations
106 and bugs that prevent full-blown multi-display support in
107 Emacs. (GTK crashes when Emacs tries to disconnect from an
108 X server.) Use the Lucid toolkit if you want to see a
109 complete feature set.
111 * The single-kboard mode.
113 If your multi-tty Emacs session seems to be frozen, you
114 probably have a recursive editing session or a pending
115 minibuffer prompt (which is a kind of recursive editing) on
116 another display. To unfreeze your session, switch to that
117 display and complete the recursive edit, for example by
118 pressing C-] (`abort-recursive-edit').
120 I am sorry to say that currently there is no way to break
121 out of this "single-kboard mode" from a frozen display. If
122 you are unable to switch to the display that locks the
123 others (for example because it is on a remote computer),
124 then you can use emacsclient to break out of all recursive
127 emacsclient -e '(top-level)'
129 Note that this (perhaps) unintuitive behaviour is by design.
130 Single-kboard mode is required because of an intrinsic Emacs
131 limitation that is very hard to eliminate. (This limitation
132 is related to the single-threaded nature of Emacs.)
134 I plan to implement better user notification and support for
135 breaking out of single-kboard mode from locked displays.
137 * Mac, Windows and DOS support is broken, doesn't even
138 compile. Multiple display support will probably not provide
139 new Emacs features on these systems, but the multi-tty
140 branch changed a few low-level interfaces, and the
141 system-dependent source files need to be adapted
142 accordingly. The changes are mostly trivial, so almost
143 anyone can help, if only by compiling the branch and
144 reporting the compiler errors.
147 HOW TO GET THE BRANCH
148 ---------------------
150 The branch uses Bazaar 1 (http://bazaar.canonical.com) for version control.
152 Retrieving the latest version of the branch:
154 baz register-archive -f http://aszt.inf.elte.hu/~lorentey/mirror/arch/2004
155 baz get lorentey@elte.hu--2004/emacs--multi-tty <directory>
157 This incantation uses an archive mirror that is hosted on a
158 high-bandwidth site. Please note that on average there is a two-hour
159 delay for commits to arrive on this mirror. My primary mirror is on the
160 low-bandwidth http://lorentey.hu/ site:
162 baz register-archive -f http://lorentey.hu/arch/2004/
163 baz get lorentey@elte.hu--2004/emacs--multi-tty <directory>
165 This is "instantly" updated, but very slow from outside Hungary.
166 (By "instantly" I mean as soon as I connect the notebook I work on to
167 a network. It could take days.)
169 The Arch supermirror provides mirroring services for all public Arch
170 repositories. We have a mirror there, too, if you prefer.
172 baz register-archive -f http://mirrors.sourcecontrol.net/lorentey%40elte.hu--2004
173 baz get lorentey@elte.hu--2004/emacs--multi-tty <directory>
175 My GPG key id is 0FB27A3F; it is available from
176 hkp://wwwkeys.eu.pgp.net/, or from my homepage at
177 http://lorentey.hu/rolam/gpg.html)
179 Don't worry if the above checkout takes a few minutes to complete;
180 once you have a source tree, updating it to the latest revision will
181 be _much_ faster. Use the following command for the update:
185 You can find more information about Bazaar on
186 http://bazaar.canonical.com/. It's a distributed source control
187 system that is somewhat less broken than competing projects.
189 If you don't have Bazaar, the branch has a homepage from which you can
190 download conventional patches against Emacs CVS HEAD:
192 http://lorentey.hu/project/emacs.html
194 I suggest you use Bazaar whenever feasible.
200 If you run Debian, or a distribution based on Debian, you are welcome
201 to use our binary packages; put these lines in your /etc/apt/sources.list:
204 deb http://aszt.inf.elte.hu/~lorentey/mirror/apt unstable multi-tty
205 deb-src http://aszt.inf.elte.hu/~lorentey/mirror/apt unstable multi-tty
207 Note that these packages are intended solely to provide an easy way to
208 test the new multi-tty features. They are not to be taken as Emacs
209 releases, and it's a mistake to expect robust operation or any kind of
210 timely support for them. Do not install them, unless you'd like to
211 have your editor crash on you.
217 The multi-tty branch is compiled the same way as Emacs itself:
219 make maintainer-clean # (If you have compiled Emacs before)
221 ./configure --without-gtk <your favourite options>
225 If you have strange compilation errors, they may be caused by old
226 *.elc files that are left over from an earlier bootstrap. The `make
227 maintainer-clean' target deletes them, so it is a good idea to run
228 that before reporting a bug. (Emacs requires a clean recompilation
229 after certain kinds of source code changes.)
234 To test the multi-tty branch, start up the Emacs server with the
240 and then (from a shell prompt on another terminal) start emacsclient
242 emacsclient -t /optional/file/names... (for a tty frame)
243 emacsclient /optional/file/names... (for an X frame)
245 (Make sure both emacs and emacsclient are multi-tty versions.)
246 You'll hopefully have two fully working, independent frames on
247 separate terminals. The new frame is closed automatically when you
248 finish editing the specified files (C-x #), but delete-frame (C-x 5 0)
249 also works. Of course, you can create frames on more than two tty
252 Creating new frames on the same tty with C-x 5 2 (make-frame-command)
253 works, and they behave the same way as in previous Emacs versions. If
254 you exit emacs, all terminals should be restored to their previous
257 This is work in progress, and probably full of bugs. It is a good
258 idea to run emacs from gdb, so that you'll have a live instance to
259 debug if something goes wrong. Please send me your bug reports on our
260 mailing list: multi-tty@lists.fnord.hu
265 I think the best way to use the new Emacs is to have it running inside
266 a disconnected GNU screen session, and always use emacsclient for
267 normal work. One advantage of this is that not a single keystroke of
268 your work will be lost if the display device that you are using
269 crashes, or the network connection times out, or whatever. (I had an
270 extremely unstable X server for some time while I was developing these
271 patches, and running Emacs this way has saved me a number of M-x
272 recover-session invocations.)
274 I use the following two bash scripts to handle my Emacs sessions:
276 -------------------------------------------------------connect-emacs--
278 # Usage: connect-emacs <name> <args>...
280 # Connects to the Emacs instance called NAME. Starts up the instance
281 # if it is not already running. The rest of the arguments are passed
287 if [ -z "$name" ]; then
288 echo "Usage: connect_emacs <name> <args>..." >&2
291 preload-emacs "$name" wait
292 /usr/bin/emacsclient.emacs-multi-tty -s "$name" "$@"
293 ----------------------------------------------------------------------
295 -------------------------------------------------------preload-emacs--
297 # Usage: preload-emacs <name> [<waitp>]
299 # Preloads the Emacs instance called NAME in a detached screen
300 # session. Does nothing if the instance is already running. If WAITP
301 # is non-empty, the function waits until the server starts up and
302 # creates its socket; otherwise it returns immediately.
306 screendir="/var/run/screen/S-$USER"
307 serverdir="/tmp/emacs$UID"
308 emacs=/usr/bin/emacs-multi-tty # Or wherever you installed your multi-tty Emacs
310 if [ -z "$name" ]; then
311 echo "Usage: preload_emacs <name> [<waitp>]" >&2
315 if [ ! -e "$screendir"/*."$name" ]; then
316 if [ -e "$serverdir/$name" ]; then
317 # Delete leftover socket (for the wait option)
318 rm "$serverdir/$name"
320 screen -dmS "$name" "$emacs" -nw --eval "(setq server-name \"$name\")" -f server-start
322 if [ ! -z "$waitp" ]; then
323 while [ ! -e "$serverdir/$name" ]; do sleep 0.1; done
325 ----------------------------------------------------------------------
327 I have the following in my profile to have two instances automatically
328 preloaded for editing and email:
333 It is useful to set up short aliases for connect-emacs. I use the
336 alias edit="connect-emacs editor"
338 alias et="connect-emacs editor -t"
339 alias gnus="connect-emacs gnus"
351 For the NEWS file: (Needs much, much work)
353 ** Support for multiple terminal devices has been added.
355 *** You can specify a terminal device (`tty' parameter) and a terminal
356 type (`tty-type' parameter) to `make-terminal-frame'.
358 *** You can test for the presence of multiple terminal support by
359 testing for the `multi-tty' feature.
361 *** Emacsclient has been extended to support opening a new terminal
362 frame. Its behaviour has been changed to open a new Emacs frame by
363 default. Use the -c option to get the old behavior of opening
364 files in the currently selected Emacs frame.
366 *** A make-frame-on-tty function has been added to make it easier to
367 create frames on new terminals.
369 *** New functions: frame-tty-name, frame-tty-type, delete-tty,
370 terminal-local-value, set-terminal-local-value
372 terminal-id, terminal-parameters, terminal-parameter,
373 set-terminal-parameter
375 *** New variables: global-key-translation-map
377 *** The keymaps key-translation-map and function-key-map are now
380 ** Support for simultaneous graphical and terminal frames has been
383 *** The function `make-frame-on-display' now works during a terminal
384 session, and `make-frame-on-tty' works during a graphical session.
386 *** The `window-system' variable has been made frame-local.
388 *** The new `initial-window-system' variable contains the
389 `window-system' value for the first frame.
391 *** talk.el has been extended for multiple tty support.
393 *** C-z now invokes `suspend-frame', C-x C-c now invokes
394 `save-buffers-kill-frame'.
398 (The rest of this file consists of my development notes and as such it
399 is probably not very interesting for anyone else.)
404 ** Trouble: `setenv' doesn't actually set environment variables in the
405 Emacs process. This defeats the purpose of the elaborate
406 `server-with-environment' magic around the `tgetent' call in
409 ** (Possibly) create hooks in struct device for creating frames on a
410 specific terminal, and eliminate the hackish terminal-related frame
411 parameters (display, tty, tty-type).
417 ** Move Fsend_string_to_terminal to term.c, and declare get_named_tty
418 as static, removing it from dispextern.h.
419 Move fatal to emacs.c and declare it somewhere.
421 ** Search for `suspend-emacs' references and replace them with
422 `suspend-frame', if necessary. Ditto for `save-buffers-kill-emacs'
423 vs. `save-buffers-kill-display'.
425 ** Emacs crashes when a tty frame is resized so that there is no space
426 for all its windows. (Tom Schutzer-Weissmann)
428 ** Report GTK multi-display problems to GTK maintainers. For extra
431 Currently you can not connect to new X displays when you compile
432 Emacs with GTK support. If you want to play around with GTK
433 multidisplay (and don't mind core dumps), you can edit src/config.h
434 and define HAVE_GTK_MULTIDISPLAY there by hand.
436 ** Audit `face-valid-attribute-values' usage in customize and
437 elsewhere. Its return value depends on the current window system.
438 Replace static initializers using it with runtime functions. For
439 example, custom's buttons are broken on non-initial device types.
441 ** Possibly turn off the double C-g feature when there is an X frame.
442 C.f. (emacs)Emergency Escape.
444 ** frames-on-display-list should also accept frames.
446 ** I smell something funny around pop_kboard's "deleted kboard" case.
447 Determine what are the circumstances of this case, and fix any
448 bug that comes to light.
450 ** Consider the `tty-type' frame parameter and the `display-tty-type'
451 function. They serve the exact same purpose. I think it may be
452 a good idea to eliminate one of them, preferably `tty-type'.
454 ** The handling of lisp/term/*.el, and frame creation in general, is a
455 big, big mess. How come the terminal-specific file is loaded by
456 tty-create-frame-with-faces? I don't think it is necessary to load
457 these files for each frame; once per terminal should be enough.
458 Update: lisp/term/*.el is not loaded repeatedly anymore, but
459 faces.el still needs to be cleaned up.
461 ** Fix frame-set-background-mode in this branch. It was recently
462 changed in CVS, and frame.el in multi-tty has not yet been adapted
463 for the changes. (It needs to look at
464 default-frame-background-mode.) (Update: maybe it is fixed now;
467 ** I think `(set-)terminal-local-value' and the terminal parameter
468 mechanism should be integrated into a single framework.
470 (Update: `(set-)terminal-local-value' is now eliminated, but the
471 terminal-local variables should still be accessible as terminal
472 parameters. This also applies to `display-name' and similar
475 ** Add the following hooks: after-delete-frame-hook (for server.el,
476 instead of delete-frame-functions),
477 after-delete-terminal-functions, after-create-terminal-functions.
479 ** Fix set-input-mode for multi-tty. It's a truly horrible interface;
480 what if we'd blow it up into several separate functions (with a
481 compatibility definition)?
483 ** BULK RENAME: The `display-' prefix of new Lisp-level functions
484 conflicts with stuff like `display-time-mode'. Use `device-'
485 or `terminal-' instead. I think I prefer `terminal-'.
487 It turns out that most of the offending Lisp functions were defined
488 in the trunk. Therefore, compatibility aliases should be defined
489 for the following names:
491 display-color-cells terminal-color-cells
492 display-color-p terminal-color-p
493 display-graphic-p terminal-graphic-p
494 display-grayscale-p terminal-grayscale-p
495 display-images-p terminal-images-p
496 display-mm-height terminal-mm-height
497 display-mm-width terminal-mm-width
498 display-mouse-p terminal-mouse-p
499 display-multi-font-p terminal-multi-font-p
500 display-multi-frame-p terminal-multi-frame-p
501 display-pixel-height terminal-pixel-height
502 display-pixel-width terminal-pixel-width
503 display-pixels-per-inch terminal-pixels-per-inch
504 display-planes terminal-planes
505 display-popup-menus-p terminal-popup-menus-p
506 display-save-under terminal-save-under
507 display-screens terminal-screens
508 display-supports-face-attributes-p terminal-supports-face-attributes-p
509 display-visual-class terminal-visual-class
510 framep-on-display framep-on-terminal
511 frames-on-display-list frames-on-terminal-list
513 The following functions were introduced in the multi-tty branch, and
514 can be renamed without aliases:
516 display-controlling-tty-p terminal-controlling-tty-p
517 display-list terminal-list
518 display-live-p terminal-live-p
519 display-name terminal-name
520 display-tty-type terminal-tty-type
521 frame-display terminal-of-frame
522 delete-display delete-terminal
524 ** The single-keyboard mode of MULTI_KBOARD is extremely confusing
525 sometimes; Emacs does not respond to stimuli from other keyboards.
526 At least a beep or a message would be important, if the single-mode
527 is still required to prevent interference. (Reported by Dan
530 Update: selecting a region with the mouse enables single_kboard
531 under X. This is very confusing.
533 Update: After discussions with Richard Stallman, this will be
534 resolved by having locked displays warn the user to wait, and
535 introducing a complex protocol to remotely bail out of
536 single-kboard mode by pressing C-g.
538 Update: Warning the user is not trivial to implement, as Emacs has
539 only one echo area, shared by all frames. Ideally the warning
540 should not be displayed on the display that is locking the others.
541 Perhaps the high probability of user confusion caused by
542 single_kboard mode deserves a special case in the display code.
543 Alternatively, it might be good enough to signal single_kboard mode
544 by changing the modelines or some other frame-local display element
545 on the locked out displays.
547 ** The session management module is prone to crashes when the X
548 connection is closed and then later I try to connect to a new X
551 #0 0xb7ebc806 in SmcGetIceConnection () from /usr/X11R6/lib/libSM.so.6
552 #1 0x080e6641 in x_session_check_input (bufp=0xbf86c9c0) at xsmfns.c:144
553 #2 0x080d3bbc in XTread_socket (device=0xa722ff8, expected=1, hold_quit=0xbf86ca90) at xterm.c:7037
554 #3 0x080fa404 in read_avail_input (expected=1) at keyboard.c:6696
555 #4 0x080fa4ca in handle_async_input () at keyboard.c:6900
556 #5 0x080d51fa in x_term_init (display_name=162628899, xrm_option=0x0, resource_name=0x857068c "emacs") at xterm.c:10622
557 #6 0x080d920e in x_display_info_for_name (name=162628899) at xfns.c:3975
558 #7 0x080d92f9 in check_x_display_info (object=1) at xfns.c:274
559 #8 0x080d97b8 in Fx_create_frame (parms=151221485) at xfns.c:3016
560 #9 0x0815bf72 in Ffuncall (nargs=2, args=0xbf86ceec) at eval.c:2851
562 I installed a workaround to prevent this. The X session manager is
563 only contacted when the very first display in the Emacs session is
564 an X display. Also, x_delete_display() on this display aborts
565 session management, and XTread_socket only calls
566 x_session_check_input when it is called for the display that the
567 session was opened on. While this does not really fix the bug, it
568 makes it much less frequent, because session manager support will
569 not normally be enabled when Emacs can survive the shutdown of the
572 See if xsmfns.c should be updated.
574 ** Hunt down display-related functions in frame.el and extend them all
575 to accept display ids.
577 ** rif->flush_display_optional (NULL) calls should be replaced by a
580 ** The set-locale-environment hack (adding the DISPLAY option) should
581 be replaced with a clean design.
583 ** standard-display-table should be display-local.
584 standard-display-european should be display-local.
586 ** With iswitchb-default-method set to 'always-frame, only frames on
587 the current display should be considered. This might involve
588 extending `get-buffer-window'.
590 ** Have a look at Vlocale_coding_system. Seems like it would be a
591 tedious job to localize it, although most references use it for
592 interfacing with libc and are therefore OK with the global
595 Exceptions found so far: x-select-text and
596 x-cut-buffer-or-selection-value.
598 ** Have a look at fatal_error_hook.
600 ** Have a look at set_frame_matrix_frame.
602 ** Check if we got term-setup-hook right.
604 ** I think tip_frame should be display-local.
606 ** Check display reference count handling in x_create_tip_frame.
608 ** make-frame does not correctly handle extra parameters in its
611 (frame-parameter (make-frame (list (cons 'foobar 42))) 'foobar)
614 (This is likely an error in the CVS trunk.)
616 ** Dan Nicolaescu suggests that -nw should be added as an alias for -t
617 in emacsclient. Good idea. (Alas, implementing this is not
618 trivial, getopt_long does not seem to support two-letter ``short''
619 options. Patches are welcome.)
621 ** Mark Plaksin suggests that emacsclient should accept the same
622 X-related command-line arguments as Emacs. Most of the X-related
623 argument-handling is done in Lisp, so this should be quite easy to
626 ** Gergely Nagy suggests that C-x # should only kill the current
627 frame, not any other emacsclient frame that may have the same file
628 opened for editing. I think I agree with him.
630 ** Very strange bug: visible-bell does not work on secondary
631 terminals in xterm and konsole. The screen does flicker a bit,
632 but it's so quick it isn't noticable.
634 ** Move baud_rate to struct display.
636 ** Implement support for starting an interactive Emacs session without
637 an initial frame. (The user would connect to it and open frames
638 later, with emacsclient.)
640 ** Fix Mac support (I can't do this entirely myself). Note that the
641 current state of Mac-specific source files in the multi-tty tree
642 are not useful; before starting work on Mac support, revert to
643 pristine, pre-multi-tty versions.
645 ** Fix W32 support (I can't do this entirely myself). Note that the
646 current state of W32-specific source files in the multi-tty tree
647 are not useful; before starting work on W32 support, revert to
648 pristine, pre-multi-tty versions.
650 ** Fix DOS support (I can't do this entirely myself). Note that the
651 current state of DOS-specific source files in the multi-tty tree
652 are not useful; before starting work on DOS support, revert to
653 pristine, pre-multi-tty versions.
655 ** Do a grep on XXX and ?? for more issues.
657 ** Understand Emacs's low-level input system (it's black magic) :-)
658 What exactly does interrupt_input do? I tried to disable it for
659 raw secondary tty support, but it does not seem to do anything
660 useful. (Update: Look again. X unconditionally enables this, maybe
661 that's why raw terminal support is broken again. I really do need
662 to understand input.)
663 (Update: I am starting to understand the read_key_sequence->read-char
664 ->kbd_buffer_get_event->read_avail_input->read_socket_hook path. Yay!)
666 ** flow-ctrl.el must be updated.
668 ** Fix stuff_char for multi-tty. Doesn't seem to be of high priority.
673 (ex-TODO items with explanations.)
675 -- Introduce a new struct for terminal devices.
677 (Done, see struct tty_output. The list of members is not yet
680 -- Change the bootstrap procedure to initialize tty_list.
682 (Done, but needs review.)
684 -- Change make-terminal-frame to support specifying another tty.
686 (Done, new frame parameters: `tty' and `tty-type'.)
688 -- Implement support for reading from multiple terminals.
690 (Done, read_avail_input tries to read from each terminal, until one
691 succeeds. MULTI_KBOARD is not used. Secondary terminals don't send
694 (Update: They do, now.)
696 (Update2: After enabling X, they don't.)
698 -- other-frame should cycle through the frames on the `current'
701 (Done, by trivially modifiying next_frame and prev_frame.)
703 -- Support different terminal sizes.
707 -- Make sure terminal resizes are handled gracefully. (Could be
710 (Done. We don't get automatic SIGWINCH for additional ttys,
713 -- Extend emacsclient to automatically open a new tty when it connects
716 (Done. It's an ugly hack, needs more work.)
718 -- Redisplay must refresh the topmost frame on *all* terminals, not
719 just the initial terminal.
721 (Done, but introduced an ugly redisplay problems. Ugh.)
723 -- Fix redisplay problems.
725 (Done; it turned out that the entire Wcm structure must be moved
726 inside tty_output. Why didn't I catch this earlier?)
728 -- Provide a way for emacsclient to tell Emacs that the tty has been
731 (Done, simply forward the SIGWINCH signal.)
733 -- Each keypress should automatically select the frame corresponding
734 to the terminal that it was coming from. This means that Emacs
735 must know from which terminal the last keyboard event came from.
737 (Done, it was quite simple, the input event system already
738 supported multiple frames.)
740 -- Fix SIGIO issue with secondary terminals.
742 (Done, emacsclient signals Emacs after writing to the proxy pseudo
743 terminal. Note that this means that multi-tty does not work with
746 (Update: This is bullshit. There is a read_input_waiting function,
747 extend that somehow.)
749 (Update of update: The first update was not right either, extending
750 read_input_waiting was not necessary. Secondary ttys do seem to
751 send signals on input.)
753 (Update^3: Not any more.)
755 -- Make make-terminal-frame look up the `tty' and `tty-type' frame
756 parameters from the currently selected terminal before the global
761 -- Put all cached terminal escape sequences into struct tty_output.
762 Currently, they are still stored in global variables, so we don't
763 really support multiple terminal types.
765 (Done. It was not fun.)
767 -- Implement sane error handling after initialization. (Currently
768 emacs exits if you specify a bad terminal type.) The helpful error
769 messages must still be provided when Emacs starts.
773 -- Implement terminal deletion, i.e., deleting local frames, closing
774 the tty device and restoring its previous state without exiting
777 (Done, but at the moment only called when an error happens during
778 initialization. There is a memory corruption error around this
779 somewhere.) (Update: now it is fully enabled.)
781 -- Implement automatic deletion of terminals when the last frame on
782 that terminal is closed.
786 -- Restore tty screen after closing the terminal.
788 (Done, we do the same as Emacs 21.2 for all terminals.)
790 -- 'TERM=dumb src/emacs' does not restore the terminal state.
794 -- C-g should work on secondary terminals.
796 (Done, but the binding is not configurable.)
798 -- Deal with SIGHUP in Emacs and in emacsclient. (After this, the
799 server-frames may be removed from server.el.)
801 (Done, nothing to do. It seems that Emacs does not receive SIGHUP
802 from secondary ttys, which is actually a good thing.) (Update: I
803 think it would be a bad idea to remove server-frames.)
805 -- Change emacsclient/server.el to support the -t argument better,
806 i.e. automatically close the socket when the frame is closed.
808 (Seems to be working OK.)
810 -- Fix mysterious memory corruption error with tty deletion. To
811 trigger it, try the following shell command:
813 while true; do TERM=no-such-terminal-definition emacsclient -h; done
815 Emacs usually dumps core after a few dozen iterations. (The bug
816 seems to be related to the xfreeing or bzeroing of
817 tty_output.Wcm. Maybe there are outside references to struct Wcm?
818 Why were these vars collected into a struct before multi-tty
821 (Done. Whew. It turned out that the problem had nothing to do
822 with hypothetical external references to Wcm, or any other
823 tty_output component; it was simply that delete_tty closed the
824 filehandles of secondary ttys twice, resulting in fclose doubly
825 freeing memory. Utterly trivial matter. I love the C's memory
826 management, it puts hair on your chest.)
828 -- Support raw secondary terminals. (Note that SIGIO works only on
829 the controlling terminal.) Hint: extend read_input_waiting for
830 multiple ttys and hopefully this will be fixed.
832 (Done, it seems to have been working already for some time. It
833 seems F_SETOWN does work, after all. Not sure what made it fail
834 earlier, but it seems to be fixed (there were several changes
835 around request_sigio, maybe one of them did it).
836 read_input_waiting is only used in sys_select, don't change
837 it.) (Update: After adding X support, it's broken again.)
838 (Update^2: No it isn't.) :-)
840 -- Find out why does Emacs abort when it wants to close its
841 controlling tty. Hint: chan_process[] array. Hey, maybe
842 noninterrupt-IO would work, too? Update: no, there is no process
845 (Done. Added add/delete_keyboard_wait_descriptor to
846 term_init/delete_tty. The hint was right, in a way.)
848 -- Issue with SIGIO: it needs to be disabled during redisplay. See if
849 fcntl kernel behaviour could be emulated by emacsclient.
851 (Done. Simply disabled the SIGIO emulation hack in emacsclient.)
852 (Update: it was added back.) (Update^2: and removed again.)
854 -- server.el: There are issues with saving files in buffers of closed
855 clients. Try editing a file with emacsclient -f, and (without
856 saving it) do a delete-frame. The frame is closed without
857 question, and a surprising confirmation prompt appears in another
860 (Done. delete-frame now asks for confirmation if it still has
861 pending buffers, and modified buffers don't seem to be deleted.)
863 -- emacsclient.el, server.el: Handle eval or file open errors when
868 -- Make parts of struct tty_output accessible from Lisp. The device
869 name and the type is sufficient.
871 (Done, see frame-tty-name and frame-tty-type.)
873 -- Export delete_tty to the Lisp environment, for emacsclient.
875 (Done, see delete-tty.)
877 -- Get rid of the accessor macros in termchar.h, or define macros for
882 -- Move device-specific parameters (like costs) commonly used by
883 device backends to a common, device-dependent structure.
885 (Done. See struct display_method in termhooks.h.)
889 (Done. Well, it seems to be working.)
891 -- Allow simultaneous X and tty frames. (Handling input could be
892 tricky. Or maybe not.)
894 (Done. Allowed, that is. It is currently extremely unstable, to
895 the point of being unusable. The rif variable causes constant
896 core dumps. Handling input is indeed tricky.)
898 -- Rewrite multi-tty input in terms of MULTI_KBOARD.
900 (Done. In fact, there was no need to rewrite anything, I just
901 added a kboard member to tty_display_info, and initialized the
902 frame's kboard from there.)
904 -- Fix rif issue with X-tty combo sessions. IMHO the best thing to do
905 is to get rid of that global variable (and use the value value in
906 display_method, which is guaranteed to be correct).
908 (Done, did exactly that. Core dumps during combo sessions became
909 much rarer. In fact, I have not yet met a single one.)
911 -- Add multi-tty support to talk.el.
915 -- Clean up the source of emacsclient. It is a mess.
917 (Done, eliminated stupid proxy-pty kludge.)
919 -- Fix faces on tty frames during X-tty combo sessions. There is an
920 init_frame_faces call in init_sys_modes, see if there is a problem
923 (Done, there was a stupid mistake in
924 Ftty_supports_face_attributes_p. Colors are broken, though.)
926 -- C-x 5 2, C-x 5 o, C-x 5 0 on an emacsclient frame unexpectedly
927 exits emacsclient. This is a result of trying to be clever with
928 delete-frame-functions.
930 (Fixed, added delete-tty-after-functions, and changed server.el to
933 -- Something with (maybe) multi-keyboard support broke function keys
934 and arrows on ttys during X+tty combo sessions. Debug this.
936 (I can't reproduce it, maybe the terminal type was wrong.)
938 -- Fix input from raw ttys (again).
940 (Now it seems to work all right.)
942 -- During an X-tty combo session, a (message "Hello") from a tty frame
943 goes to the X frame. Fix this.
945 (Done. There was a safeguard against writing to the initial
946 terminal frame during bootstrap which prevented echo_area_display
947 from working correctly on a tty frame during a combo session.)
949 -- If there are no frames on its controlling terminal, Emacs should
950 exit if the user presses C-c there.
952 (Done, as far as possible. See the SIGTERM comment in
953 interrupt_signal on why this seems to be impossible to solve this
956 -- During an X session, Emacs seems to read from stdin. Also, Emacs
957 fails to start without a controlling tty.
959 (Fixed by replacing the troublesome termcap display with a dummy
960 bootstrap display during bootstrap.
962 -- Do tty output through struct display, like graphical display
967 -- Define an output_initial value for output_method for the initial
968 frame that is dumped with Emacs. Checking for this frame (e.g. in
969 cmd_error_internal) is ugly.
971 (Done, breaking interactive temacs.)
973 -- The command `emacsclient -t -e '(delete-frame)'' fails to exit.
977 -- frame-creation-function should always create a frame that is on the
978 same display as the selected frame. Maybe frame-creation-function
979 should simply be removed and make-frame changed to do the right
982 (Done, with a nice hack. frame-creation-function is now frame-local.)
984 -- Fix C-g on raw ttys.
986 (Done. I disabled the interrupt/quit keys on all secondary
987 terminals, so Emacs sees C-g as normal input. This looks like an
988 overkill, because emacsclient has extra code to pass SIGINT to
989 Emacs, so C-g should remain the interrupt/quit key on emacsclient
990 frames. See the next entry why implementing this distinction would
993 -- Make sure C-g goes to the right frame with ttys. This is hard, as
994 SIGINT doesn't have a tty parameter. :-(
996 (Done, the previous change fixes this as a pleasant side effect.)
998 -- I have seen a case when Emacs with multiple ttys fell into a loop
999 eating 100% of CPU time. Strace showed this loop:
1002 kill(30284, SIGIO) = 0
1003 --- SIGIO (I/O possible) @ 0 (0) ---
1004 ioctl(6, FIONREAD, [0]) = -1 EIO (Input/output error)
1005 ioctl(5, FIONREAD, [0]) = -1 EIO (Input/output error)
1006 ioctl(0, FIONREAD, [0]) = 0
1007 sigreturn() = ? (mask now [])
1008 gettimeofday({1072842297, 747760}, NULL) = 0
1009 gettimeofday({1072842297, 747806}, NULL) = 0
1010 select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0})
1011 select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0})
1012 gettimeofday({1072842297, 748245}, NULL) = 0
1014 I have seen something similar with a single X frame, but have not
1015 been able to reproduce it for debugging.
1017 Update: This may have been caused by checking for nread != 0
1018 instead of nread > 0 after calling read_socket_hook in
1021 (Fixed. This was caused by unconditionally including stdin in
1022 input_wait_mask in init_process. The select call in
1023 wait_reading_process_input always returned immediately, indicating
1024 that there is pending input from stdin, which nobody read.
1026 Note that the above strace output seems to be an unrelated but
1027 similar bug. I think that is now fixed.)
1029 -- Exiting Emacs while there are emacsclient frames doesn't restore the
1030 ttys to their default states.
1032 (This seems to be fixed by some previous change.)
1034 -- Allow opening an X session after -nw.
1038 -- Fix color handling during tty+X combo sessions. (It seems that tty
1039 sessions automatically convert the face colors to terminal colors
1040 when the face is loaded. This conversion must happen instead on
1041 the fly in write_glyphs, which might be problematic, as color
1042 approximation is currently done in lisp (term/tty-colors.el).)
1043 (Update: hm, colors seem to work fine if I start emacs with -nw and
1044 then create an X frame. Maybe it's just a small buglet somewhere.)
1046 (Seems to be fixed. The problem was in startup.el, it did not
1047 initialize tty colors when the initial window system was
1050 -- emacs -nw --eval '(y-or-n-p "Foobar")' segfaults. (Reported by
1053 (Fixed, there was a keyboard initialization problem.)
1055 -- Fix interactive use of temacs. There are face-related SEGVs, most
1056 likely because of changes in realize_default_face, realize_face.
1060 -- Don't exit Emacs when the last X connection fails during a
1061 multi-display session.
1065 -- Dan Nicolaescu noticed that starting emacsclient on the same
1066 terminal device that is the controlling tty of the Emacs process
1067 gives unexpected results.
1071 -- Istvan Marko reported that Emacs hang on ttys if it was started
1072 from a shell script.
1074 (Fixed. There was a bug in the multi-tty version of
1075 narrow_foreground_group. tcsetpgrp blocks if it is called from a
1076 process that is not in the same process group as the tty.)
1078 -- emacsclient -t from an Emacs term buffer does not work, complains
1079 about face problems. This can even lock up Emacs (if the recursive
1080 frame sets single_kboard). Update: the face problems are caused by
1081 bugs in term.el, not in multi-tty. The lockup is caused by
1082 single_kboard mode, and is not easily resolvable. The best thing to
1083 do is to simply refuse to create a tty frame of type `eterm'.
1085 (Fixed, changed emacsclient to check for TERM=eterm. The face
1086 complaints seem to be caused by bugs in term.el; they are not
1087 related to multi-tty.)
1089 -- Find out the best way to support suspending Emacs with multiple
1090 ttys. My guess: disable it on the controlling tty, but from other
1091 ttys pass it on to emacsclient somehow. (It is (I hope) trivial to
1092 extend emacsclient to handle suspend/resume. A `kill -STOP' almost
1095 (Done. I needed to play with signal handling and the server
1096 protocol a bit to make emacsclient behave as a normal UNIX program
1097 wrt foreground/background process groups.)
1099 -- There is a flicker during the startup of `emacs -nw'; it's as if
1100 the terminal is initialized, reset and then initialialized again.
1101 Debug this. (Hint: narrow_foreground_group is called twice during
1106 -- Robert Chassell has found serious copy-paste bugs with the
1107 multi-tty branch. There seem to be redisplay bugs while copying
1108 from X to a terminal frame. Copying accented characters do not
1111 (Patch-124 should fix this, by changing the interprogram-*-function
1112 variables to be frame-local, as suggested by Mark Plaksin
1113 (thanks!). I think that the redisplay bugs are in fact not bugs,
1114 but delays caused by single_kboard --> perhaps MULTI_KBOARD should
1117 -- frame-creation-function was removed, which might be a bad idea.
1118 Think up a compatible solution.
1120 (It was an internal interface that may be changed when necessary.)
1122 -- Change Lisp code not to (getenv "TERM"); use the `tty-type' frame
1123 parameter or the frame-tty-type function instead. (M-x tags-search
1124 "TERM" helps with this.) Update: Actually, all getenv invocations
1125 should be checked for multi-tty compatibility, and an interface
1126 must be implemented to get the remote client's environment.
1128 (Done. Only getenv calls in lisp/term/*.el were changed; other
1129 calls should be mostly left as they are.)
1131 -- Add an elaborate mechanism for display-local variables. (There are
1132 already a few of these; search for `terminal-local' in the Elisp
1135 (Not needed. Display-local variables could be emulated by
1136 frame-local variables.)
1138 -- Emacs assumes that all terminal frames have the same locale
1139 settings as Emacs itself. This may lead to bogus results in a
1140 multi-locale setup. (E.g., while logging in from a remote client
1141 with a different locale.)
1142 (Update after new bugreport by Friedrich Delgado Friedrichs:
1143 (at least) the structs terminal_coding and keyboard_coding in
1144 coding.c must be moved to struct display, and the Lisp interface
1145 [set-]keyboard-coding-system must be adapted for the change.)
1147 (Fixed. Emacs now uses the locale settings as seen by the
1148 emacsclient process for server tty frames.)
1149 (Update: Not really; Vlocale_coding_system is still global.)
1151 -- Make `struct display' accessible to Lisp programs. Accessor functions:
1153 (displayp OBJECT): Returns t if OBJECT is a display.
1154 => Implemented as display-live-p.
1156 (display-list): Returns list of currently active displays.
1159 (selected-display): Returns the display object of the selected frame.
1160 => Not strictly necessary, but implemented anyway.
1162 (frame-display FRAME): Returns the display object of FRAME.
1165 (display-frames DISPLAY): Returns a list of frames on DISPLAY.
1166 => Already implemented, see frames-on-display-list.
1168 (display-type DISPLAY): Returns the type of DISPLAY, as a
1169 symbol. (See `framep'.)
1170 => Implemented as display-live-p.
1172 (display-device DISPLAY): Returns the name of the device that
1173 DISPLAY uses, as a string. (E.g: "/dev/pts/16", or
1175 => Implemented as display-name.
1179 See next issue why this is necessary.
1181 (Update: The consensus on emacs-devel seems to be to do this via
1182 integer identifiers. That's fine by me.)
1186 -- The following needs to be supported:
1193 The cleanest way to solve this is to allow multiple displays on the
1194 same terminal device; each new emacsclient process should create
1195 its own display. As displays are currently identified by their
1196 device names, this is not possible until struct display becomes
1197 accessible as a Lisp-level object.
1201 -- Miles Bader suggests that C-x C-c on an emacsclient frame should
1202 only close the frame, not exit the entire Emacs session. Update:
1203 see above for a function that does this. Maybe this should be the
1206 (Done. This is the new default. No complaints so far.)
1208 -- Clean up the frame-local variable system. I think it's ugly and
1209 error-prone. But maybe I just haven't yet fully understood it.
1211 (Nothing to do. It doesn't seem ugly any more. It's rather clever.)
1213 -- Support multiple character locales. A version of
1214 `set-locale-environment' needs to be written for setting up
1215 display-local settings on ttys. I think calling
1216 set-display-table-and-terminal-coding-system and
1217 set-keyboard-coding-system would be enough. The language
1218 environment itself should remain a global setting.
1220 (Done, by an ugly hack.)
1222 -- The terminal customization files in term/*.el tend to change global
1223 parameters, which may confuse Emacs with multiple displays. Change
1224 them to tweak only frame-local settings, if possible. (They tend
1225 to call define-key to set function key sequences a lot.)
1227 (Done, by making `function-key-map' terminal-local (i.e., part of
1228 struct kboard). This has probably covered all the remaining problems.)
1230 -- Make `function-key-map' and `key-translation-map' terminal-local.
1234 -- Implement `terminal-local-value' and `set-terminal-local-value' to
1235 allow deterministic access to terminal local bindings. The
1236 encode-kb package can not set up `key-translation-map' without
1237 these. The terminal-local bindings seem to be independent of what
1242 -- xt-mouse.el needs to be adapted for multi-tty. It currently
1243 signals an error on kill-emacs under X, which prevents the user
1244 from exiting Emacs. (Reported by Mnemonikk on freenode.)
1249 -- Having {reset,init}_all_sys_modes in set-input-mode breaks arrow
1250 keys on non-selected terminals under screen, and sometimes on other
1251 terminal types as well. The other function keys continue to work
1252 fine. Sometimes faces on these screens become garbled.
1254 This only seems to affect displays that are of the same terminfo
1255 type as the selected one. Interestingly, in screen Emacs normally
1256 reports the up arrow key as `M-o A', but after the above SNAFU, it
1257 complains about `M-[ a'. UNIX ttys are a complete mystery to me,
1258 but it seems the reset-reinitialize cycle somehow leaves the
1259 non-selected terminals in a different state than usual. I have no
1260 idea how this could happen.
1262 Currently set-input-mode resets only the currently selected
1263 terminal, which seems to somehow work around the problem.
1267 Dan Nicolaescu <dann@ics.uci.edu> writes:
1268 > Some terminals have 2 modes for cursor keys: Application Mode where
1269 > the cursor keys transmit the codes defined in the terminfo entry, and
1270 > Cursor mode. Applications have to send the smkx and rmkx terminfo
1271 > strings to switch between the 2 modes. So Emacs (and emacsclient) have
1272 > to send smkx when initializing and rmkx when quitting (or on
1275 (I think patch-370 fixed this.)
1277 -- This long-standing bug (first reported by Han Boetes) seems to come
1278 and go all the time. It is time to track it down and fix it.
1283 # From another xterm:
1284 emacsclient -e '(y-or-n-p "Do you want me to crash? ")'
1285 # Notice how the answer ends up in the *scratch* buffer
1289 (Fixed in patch-414 after detailed analysis by Kalle Olavi Niemitalo.)
1291 -- normal-erase-is-backspace-mode in simple.el needs to be updated for
1292 multi-tty (rep. by Dan Waber). (The Delete key is broken on X
1295 (Fixed in patch-427.)
1297 -- I think keyboard-translate-table should be made terminal-local.
1299 (Done in patch-431.)
1301 -- The semantics of terminal-local variables are confusing; it is not
1302 clear what binding is in effect in any given time. See if
1303 current_kboard (or at least the terminal-local bindings exported to
1304 Lisp) might be changed to be tied to the selected frame instead.
1305 Currently, `function-key-map' and `key-translation-map' may be
1306 accessed reliably only using the hackish
1307 `(set-)terminal-local-value' functions.
1309 Perhaps there should be a difference between `last-command' &co.
1310 and these more conventional configuration variables.
1311 (E.g. `symbol-value' would use current_kboard to access
1312 `last-command', but SELECTED_FRAME()->display->kboard to get the
1313 value of `function-key-map'.
1315 (Fixed in patch-434.)
1317 -- If the first key pressed on a new tty terminal is a function key,
1318 it is not recognized correctly. May be related to the bug below.
1320 (Seems to have been fixed as a side effect of patch-434. "The bug
1321 below" was the set-input-mode madness.)
1323 (Update: this bug was fixed for good in patch-449. It was tracked
1324 down to a bug in `read_key_sequence': it failed to reinitialize its
1325 local function-key-map/key-translation-map references when it
1326 switched keyboards. I don't understand why did this bug only
1327 appear on brand new frames, though!)
1329 -- Disable connecting to a new X display when we use the GTK toolkit.
1331 (Disabled in patch-450.)
1333 -- Implement automatic forwarding of client environment variables to
1334 forked processes, as discussed on the multi-tty list. Terminal
1335 parameters are now accessible in C code, so the biggest obstacle is
1336 gone. The `getenv_internal' and `child_setup' functions in
1337 callproc.c must be changed to support the following variable:
1339 terminal-local-environment-variables is a variable defined in ...
1341 Enable or disable terminal-local environment variables.
1343 If set to t, `getenv', `setenv' and subprocess creation
1344 functions use the environment variables of the emacsclient
1345 process that created the selected frame, ignoring
1346 `process-environment'.
1348 If set to nil, Emacs uses `process-environment' and ignores
1349 the client environment.
1351 Otherwise, `terminal-local-environment-variables' should be a
1352 list of variable names (represented by Lisp strings) to look
1353 up in the client environment. The rest will come from
1354 `process-environment'.
1356 (Implemented in patch-461; `terminal-getenv', `terminal-setenv' and
1357 `with-terminal-environment' are now replaced by extensions to
1358 `getenv' and `setenv', and the new `local-environment-variables'
1361 ;;; arch-tag: 8da1619e-2e79-41a8-9ac9-a0485daad17d