annotate admin/notes/multi-tty @ 102418:b14d67d22033

* nsterm.m (ns_select): Shortcircuit if reentrant call.

-*- coding: utf-8; mode: text; -*-

Copyright (C) 2007, 2008, 2009  Free Software Foundation, Inc.
See the end of the file for license conditions.

From README.multi-tty in the multi-tty branch.
Some of this information may be out of date.


THANKS
------

The following is a (sadly incomplete) list of people who have
contributed to the project by testing, submitting patches, bug
reports, and suggestions.  Thanks!

Bernard Adrian <[e-mail]>
ARISAWA Akihiro <[e-mail]>
Vincent Bernat <[e-mail]>
Han Boetes <[e-mail]>
Francisco Borges <[e-mail]>
Damien Cassou <[e-mail]>
Robert J. Chassell <[e-mail]>
Romain Francoise <[e-mail]>
Ami Fischman <[e-mail]>
Noah Friedman <[e-mail]>
Friedrich Delgado Friedrichs <[e-mail]>
Samium Gromoff <[e-mail]>
Mikhail Gusarov <[e-mail]>
Eric Hanchrow <[e-mail]>
IRIE Tetsuya <[e-mail]>
Yoshiaki Kasahara <[e-mail]>
Bas Kok <[e-mail]>
Jurej Kubelka <[e-mail]>
David Lichteblau <[e-mail]>
Richard Lewis <[e-mail]>
mace <[e-mail]>
Suresh Madhu <[e-mail]>
Xavier Mallard <[e-mail]>
Istvan Marko <[e-mail]>
Ted Morse <[e-mail]>
Gergely Nagy <[e-mail]>
Dan Nicolaescu <[e-mail]>
Kalle Olavi Niemitalo <[e-mail]>
Mark Plaksin <[e-mail]>
Frank Ruell <[e-mail]>
Tom Schutzer-Weissmann <[e-mail]>
Joakim Verona <[e-mail]>
Dan Waber <[e-mail]>
and many others.

Richard Stallman was kind enough to review an earlier version of my
patches.


STATUS
------

It still needs to be ported to Windows/Mac/DOS.  Both multiple
tty device support and simultaneous X and tty frame support works
fine.  Emacsclient has been extended to support opening new tty and X
frames.  It has been changed to open new Emacs frames by default.

Tested on GNU/Linux, Solaris 8, FreeBSD and OpenBSD.

Known problems:

        * GTK support.  If you compile your Emacs with the GTK
          toolkit, some functionality of multi-tty may be lost.  In
          particular, you may get crashes while working on multiple X
          displays at once.  Previous releases of GTK had limitations
          and bugs that prevented full-blown multi-display support in
          Emacs.  (GTK crashed when Emacs tries to disconnect from an
          X server.)  Things are much improved in the current GTK
          version, but if you do experience crashes in libgtk, try
          compiling Emacs with the Lucid toolkit instead.

        * The single-kboard mode.

	  If your multi-tty Emacs session seems to be frozen, you
	  probably have a recursive editing session or a pending
	  minibuffer prompt (which is a kind of recursive editing) on
	  another display.  To unfreeze your session, switch to that
	  display and complete the recursive edit, for example by
	  pressing C-] (`abort-recursive-edit').

	  I am sorry to say that currently there is no way to break
	  out of this "single-kboard mode" from a frozen display.  If
	  you are unable to switch to the display that locks the
	  others (for example because it is on a remote computer),
	  then you can use emacsclient to break out of all recursive
	  editing sessions:

		emacsclient -e '(top-level)'

	  Note that this (perhaps) unintuitive behavior is by design.
	  Single-kboard mode is required because of an intrinsic Emacs
	  limitation that is very hard to eliminate.  (This limitation
	  is related to the single-threaded nature of Emacs.)

	  I plan to implement better user notification and support for
	  breaking out of single-kboard mode from locked displays.

	* Mac and DOS support is broken, doesn't even
  	  compile.  Multiple display support will probably not provide
  	  new Emacs features on these systems, but the multi-tty
  	  branch changed a few low-level interfaces, and the
  	  system-dependent source files need to be adapted
  	  accordingly.  The changes are mostly trivial, so almost
  	  anyone can help, if only by compiling the branch and
  	  reporting the compiler errors.


TESTING
-------

To test the multi-tty feature, start up the Emacs server with the
following commands:

	emacs
	M-x server-start

and then (from a shell prompt on another terminal) start emacsclient
with
	emacsclient -t /optional/file/names...   (for a tty frame)
	emacsclient /optional/file/names...      (for an X frame)

(Make sure both emacs and emacsclient are multi-tty versions.)
You'll hopefully have two fully working, independent frames on
separate terminals. The new frame is closed automatically when you
finish editing the specified files (C-x #), but delete-frame (C-x 5 0)
also works.  Of course, you can create frames on more than two tty
devices.

Creating new frames on the same tty with C-x 5 2 (make-frame-command)
works, and behaves the same way as in previous Emacs versions.  If you
exit emacs, all terminals should be restored to their previous states.

TIPS & TRICKS
-------------

I think the best way to use the new Emacs is to have it running inside
a disconnected GNU screen session, and always use emacsclient for
normal work.  One advantage of this is that not a single keystroke of
your work will be lost if the display device that you are using
crashes, or the network connection times out, or whatever.  (I had an
extremely unstable X server for some time while I was developing these
patches, and running Emacs this way has saved me a number of M-x
recover-session invocations.)

I use the following two bash scripts to handle my Emacs sessions:

-------------------------------------------------------connect-emacs--
#!/bin/bash
# Usage: connect-emacs <name> <args>...
#
# Connects to the Emacs instance called NAME.  Starts up the instance
# if it is not already running.  The rest of the arguments are passed
# to emacsclient.

name="$1"
shift
    
if [ -z "$name" ]; then
    echo "Usage: connect_emacs <name> <args>..." >&2
    exit 1
fi
preload-emacs "$name" wait
/usr/bin/emacsclient.emacs-multi-tty -s "$name" "$@"
----------------------------------------------------------------------

-------------------------------------------------------preload-emacs--
#!/bin/bash
# Usage: preload-emacs <name> [<waitp>]
#
# Preloads the Emacs instance called NAME in a detached screen
# session.  Does nothing if the instance is already running.  If WAITP
# is non-empty, the function waits until the server starts up and
# creates its socket; otherwise it returns immediately.

name="$1"
waitp="$2"
screendir="/var/run/screen/S-$USER"
serverdir="/tmp/emacs$UID"
emacs=/usr/bin/emacs-multi-tty # Or wherever you installed your multi-tty Emacs

if [ -z "$name" ]; then 
    echo "Usage: preload_emacs <name> [<waitp>]" >&2
    exit 1
fi

if [ ! -e "$screendir"/*."$name" ]; then
    if [ -e "$serverdir/$name" ]; then
	# Delete leftover socket (for the wait option)
	rm "$serverdir/$name"
    fi
    screen -dmS "$name" "$emacs" -nw --eval "(setq server-name \"$name\")" -f server-start
fi
if [ ! -z "$waitp" ]; then
    while [ ! -e "$serverdir/$name" ]; do sleep 0.1; done
fi
----------------------------------------------------------------------

I have the following in my profile to have two instances automatically
preloaded for editing and email:

	preload-emacs editor
	preload-emacs gnus

It is useful to set up short aliases for connect-emacs.  I use the
following:

	alias edit="connect-emacs editor"
	alias e=edit
	alias et="connect-emacs editor -t"
	alias gnus="connect-emacs gnus"


THINGS TO DO
------------

** See if `tty-defined-color-alist' needs to be terminal-local.
   Update: Dan says it should be, so convert it.

** Mikhail Gusarov suggest to add a hook akin to
   `after-make-frame-functions' that is called whenever Emacs connects
   to a new terminal.  Good idea!

** emacsclient -t on the console does not work after su.  You have to
   use non-root accounts or start as root to see this.

   	Login: root
	Password:
	# su lorentey
	$ emacsclient -t
	*ERROR*: Could not open file: /dev/tty1

   The tty can be opened as /dev/tty by emacsclient, but not by Emacs.
   This seems to be a serious problem.  Currently my only idea is to
   bring back the ugly pty proxy hack from the initial versions of
   multi-tty.  Suggestions would be appreciated.

   Update: we could change emacsclient to pass its open file
   descriptor to the Emacs process.  Unfortunately, this requires a
   new Lisp-level Emacs API, and as file descriptors are not otherwise
   exported to Lisp, this approach seems at least as ugly as the pty
   proxy idea.

** lisp/vc.el depends on the terminal type during load time.
   `vc-annotate-color-map' is one example that needs to be fixed.

** Understand how `quit_throw_to_read_char' works, and fix any bugs
   that come to light.

** See if getcjmp can be eliminated somehow.  Why does Emacs allow
   asynchronous input processing while it's reading input anyway?

** `delete-frame' events are handled by `special-event-map'
   immediately when read by `read_char'.  This is fine but it prevents
   higher-level keymaps from binding that event to get notified of the
   deleted frame.

   Sometimes it would be useful for Lisp code to be notified of frame
   deletions after they have happened, usually because they want to
   clean up after the deleted frame.  Not all frame-local states can
   be stored as a frame parameter.  For example,
   `display-splash-screen' uses `recursive-edit' with a special keymap
   override to create its buffer---and it leads to all kinds of
   nastiness if Emacs stays in this recursive edit mode after the
   frame containing the splash screen is deleted.  Basically, the
   splash-screen implementation wants to throw out of the recursive
   edit when the frame is deleted; however, it is not legal to throw
   from `delete-frame-functions' because `delete-frame' must not fail.
   (Introducing `delete-frame-after-functions' would not help either
   because `delete-frame' may not fail at that time either.)

   Currently `fancy-splash-screens' installs a
   `delete-frame-functions' hook that sets up a timer to exit the
   recursive edit.  This is an adequate solution, but it would perhaps
   be better to have something like a `frame-deleted' event that could
   be bound in the normal way.

** Trouble: `setenv' doesn't actually set environment variables in the
   Emacs process.  This defeats the purpose of the elaborate
   `server-with-environment' magic around the `tgetent' call in
   `init_tty'.  D'oh.

** (Possibly) create hooks in struct device for creating frames on a
   specific terminal, and eliminate the hackish terminal-related frame
   parameters (display, tty, tty-type).

   	make_terminal_frame
		create_tty_output

** Decide whether to keep the C implementation of terminal parameters,
   or revert to the previous, purely Lisp code.  It turned out that
   local environments do not need terminal parameters after all.

** Move Fsend_string_to_terminal to term.c, and declare get_named_tty
   as static, removing it from dispextern.h.
   Move fatal to emacs.c and declare it somewhere.

** Search for `suspend-emacs' references and replace them with
   `suspend-frame', if necessary.  Ditto for `save-buffers-kill-emacs'
   vs. `save-buffers-kill-display'.

** Emacs crashes when a tty frame is resized so that there is no space
   for all its windows.  (Tom Schutzer-Weissmann)

** Report GTK multi-display problems to GTK maintainers.  For extra
   credit, fix them.

   Currently you can not connect to new X displays when you compile
   Emacs with GTK support.  If you want to play around with GTK
   multidisplay (and don't mind core dumps), you can edit src/config.h
   and define HAVE_GTK_MULTIDISPLAY there by hand.
   
	http://bugzilla.gnome.org/show_bug.cgi?id=85715

   Update: Han reports that GTK+ version 2.8.9 almost gets display
   disconnects right.  GTK will probably be fully fixed by the time
   multi-tty gets into the trunk.

   Update: I am still having problems with GTK+ 2.8.10.  I have the
   impression that the various multidisplay fixes will only get
   released in GTK+ 2.10.

** Audit `face-valid-attribute-values' usage in customize and
   elsewhere.  Its return value depends on the current window system.
   Replace static initializers using it with runtime functions.  For
   example, custom's buttons are broken on non-initial device types.

** Possibly turn off the double C-g feature when there is an X frame.
   C.f. (emacs)Emergency Escape.

** frames-on-display-list should also accept frames.

** Consider the `tty-type' frame parameter and the `display-tty-type'
   function.  They serve the exact same purpose.  I think it may be
   a good idea to eliminate one of them, preferably `tty-type'.

** The handling of lisp/term/*.el, and frame creation in general, is a
   big, big mess.  How come the terminal-specific file is loaded by
   tty-create-frame-with-faces?  I don't think it is necessary to load
   these files for each frame; once per terminal should be enough.
   Update: lisp/term/*.el is not loaded repeatedly anymore, but
   faces.el still needs to be cleaned up.

** Fix frame-set-background-mode in this branch.  It was recently
   changed in CVS, and frame.el in multi-tty has not yet been adapted
   for the changes.  (It needs to look at
   default-frame-background-mode.)  (Update: maybe it is fixed now;
   needs testing.) (Note that the byte compiler has this to say about
   term/rxvt.el:)

	term/rxvt.el:309:17:Warning: assignment to free variable
	    `default-frame-background-mode'

** I think `(set-)terminal-local-value' and the terminal parameter
   mechanism should be integrated into a single framework.

   (Update: `(set-)terminal-local-value' is now eliminated, but the
   terminal-local variables should still be accessible as terminal
   parameters.  This also applies to `display-name' and similar
   functions.)

** Add the following hooks: after-delete-frame-hook (for server.el,
   instead of delete-frame-functions),
   after-delete-terminal-functions, after-create-terminal-functions.

** BULK RENAME: The `display-' prefix of new Lisp-level functions
   conflicts with stuff like `display-time-mode'.  Use `device-'
   or `terminal-' instead.  I think I prefer `terminal-'.

   It turns out that most of the offending Lisp functions were defined
   in the trunk.  Therefore, compatibility aliases should be defined
   for the following names:

    display-color-cells			terminal-color-cells
    display-color-p			terminal-color-p
    display-graphic-p			terminal-graphic-p
    display-grayscale-p			terminal-grayscale-p
    display-images-p			terminal-images-p
    display-mm-height			terminal-mm-height
    display-mm-width			terminal-mm-width
    display-mouse-p			terminal-mouse-p
    display-multi-font-p		terminal-multi-font-p
    display-multi-frame-p		terminal-multi-frame-p
    display-pixel-height		terminal-pixel-height
    display-pixel-width			terminal-pixel-width
    display-pixels-per-inch		terminal-pixels-per-inch
    display-planes			terminal-planes
    display-popup-menus-p		terminal-popup-menus-p
    display-save-under			terminal-save-under
    display-screens			terminal-screens
    display-supports-face-attributes-p	terminal-supports-face-attributes-p
    display-visual-class		terminal-visual-class
    framep-on-display			framep-on-terminal
    frames-on-display-list		frames-on-terminal-list

   The following functions were introduced in the multi-tty branch, and
   were renamed without aliases:

    delete-display			delete-terminal
    display-controlling-tty-p           controlling-tty-p
    display-list                        terminal-list
    display-live-p                      terminal-live-p
    display-name                        terminal-name
    display-tty-type                    tty-type
    frame-display                       frame-terminal
    selected-display			selected-terminal

** The single-keyboard mode of MULTI_KBOARD is extremely confusing
   sometimes; Emacs does not respond to stimuli from other keyboards.
   At least a beep or a message would be important, if the single-mode
   is still required to prevent interference.  (Reported by Dan
   Nicolaescu.)  

   Update: selecting a region with the mouse enables single_kboard
   under X.  This is very confusing.

   Update: After discussions with Richard Stallman, this will be
   resolved by having locked displays warn the user to wait, and
   introducing a complex protocol to remotely bail out of
   single-kboard mode by pressing C-g.

   Update: Warning the user is not trivial to implement, as Emacs has
   only one echo area, shared by all frames.  Ideally the warning
   should not be displayed on the display that is locking the others.
   Perhaps the high probability of user confusion caused by
   single_kboard mode deserves a special case in the display code.
   Alternatively, it might be good enough to signal single_kboard mode
   by changing the modelines or some other frame-local display element
   on the locked out displays.

   Update: In fact struct kboard does have an echo_string slot.

** The session management module is prone to crashes when the X
   connection is closed and then later I try to connect to a new X
   session:

	#0  0xb7ebc806 in SmcGetIceConnection () from /usr/X11R6/lib/libSM.so.6
	#1  0x080e6641 in x_session_check_input (bufp=0xbf86c9c0) at xsmfns.c:144
	#2  0x080d3bbc in XTread_socket (device=0xa722ff8, expected=1, hold_quit=0xbf86ca90) at xterm.c:7037
	#3  0x080fa404 in read_avail_input (expected=1) at keyboard.c:6696
	#4  0x080fa4ca in handle_async_input () at keyboard.c:6900
	#5  0x080d51fa in x_term_init (display_name=162628899, xrm_option=0x0, resource_name=0x857068c "emacs") at xterm.c:10622
	#6  0x080d920e in x_display_info_for_name (name=162628899) at xfns.c:3975
	#7  0x080d92f9 in check_x_display_info (object=1) at xfns.c:274
	#8  0x080d97b8 in Fx_create_frame (parms=151221485) at xfns.c:3016
	#9  0x0815bf72 in Ffuncall (nargs=2, args=0xbf86ceec) at eval.c:2851

   I installed a workaround to prevent this.  The X session manager is
   only contacted when the very first display in the Emacs session is
   an X display.  Also, x_delete_display() on this display aborts
   session management, and XTread_socket only calls
   x_session_check_input when it is called for the display that the
   session was opened on.  While this does not really fix the bug, it
   makes it much less frequent, because session manager support will
   not normally be enabled when Emacs can survive the shutdown of the
   X server.

   See if xsmfns.c should be updated.

** Hunt down display-related functions in frame.el and extend them all
   to accept display ids.

** rif->flush_display_optional (NULL) calls should be replaced by a
   new global function.

** The set-locale-environment hack (adding the DISPLAY option) should
   be replaced with a clean design.

** standard-display-table should be display-local.
   standard-display-european should be display-local.

** With iswitchb-default-method set to 'always-frame, only frames on
   the current display should be considered.  This might involve
   extending `get-buffer-window'.

** Have a look at Vlocale_coding_system.  Seems like it would be a
   tedious job to localize it, although most references use it for
   interfacing with libc and are therefore OK with the global
   definition.

   Exceptions found so far: x-select-text and
   x-cut-buffer-or-selection-value.

** Have a look at fatal_error_hook.

** Have a look at set_frame_matrix_frame.

** Check if we got term-setup-hook right.

** I think tip_frame should be display-local.

** Check display reference count handling in x_create_tip_frame.

** make-frame does not correctly handle extra parameters in its
   argument:

	(frame-parameter (make-frame (list (cons 'foobar 42))) 'foobar)
		=> nil

   (This is likely an error in the CVS trunk.)

** Dan Nicolaescu suggests that -nw should be added as an alias for -t
   in emacsclient.  Good idea.  (Alas, implementing this is not
   trivial, getopt_long does not seem to support two-letter ``short''
   options.  Patches are welcome.)

** Mark Plaksin suggests that emacsclient should accept the same
   X-related command-line arguments as Emacs.  Most of the X-related
   argument-handling is done in Lisp, so this should be quite easy to
   implement.  (For example, Samium Gromoff wants emacsclient to
   support --geometry; implementing this would add that support.)

** Gergely Nagy suggests that C-x # should only kill the current
   frame, not any other emacsclient frame that may have the same file
   opened for editing.  I think I agree with him.

** Very strange bug: visible-bell does not work on secondary
   terminals in xterm and konsole.  The screen does flicker a bit,
   but it's so quick it isn't noticable.

   (Update: This is probably some problem with padding or whatnot on
   the secondary terminals.)

** Move baud_rate to struct display.

** Implement support for starting an interactive Emacs session without
   an initial frame.  (The user would connect to it and open frames
   later, with emacsclient.)

** Fix Mac support (I can't do this entirely myself).  Note that the
   current state of Mac-specific source files in the multi-tty tree
   are not useful; before starting work on Mac support, revert to
   pristine, pre-multi-tty versions.

** Fix DOS support (I can't do this entirely myself).  Note that the
   current state of DOS-specific source files in the multi-tty tree
   are not useful; before starting work on DOS support, revert to
   pristine, pre-multi-tty versions.

** Fix Windows support. Currently bootstraping works on w32, but Emacs
   crashes on startup and none of the multi-tty features are
   implemented. Many XXX comments mark things that probably need
   updating, ChangeLogs will help in spotting changes to X specific
   files that may need porting.

** Do a grep on XXX and ?? for more issues.

** flow-ctrl.el must be updated.

** Fix stuff_char for multi-tty.  Doesn't seem to be of high priority.

DIARY OF CHANGES
----------------

(ex-TODO items with explanations.)

-- Introduce a new struct for terminal devices.

   (Done, see struct tty_output.  The list of members is not yet
   complete.)

-- Change the bootstrap procedure to initialize tty_list.

   (Done, but needs review.)

-- Change make-terminal-frame to support specifying another tty.

   (Done, new frame parameters: `tty' and `tty-type'.)

-- Implement support for reading from multiple terminals.

   (Done, read_avail_input tries to read from each terminal, until one
   succeeds.  MULTI_KBOARD is not used.  Secondary terminals don't send
   SIGIO!)

   (Update: They do, now.)

   (Update2: After enabling X, they don't.)

-- other-frame should cycle through the frames on the `current'
   terminal only.

   (Done, by trivially modifiying next_frame and prev_frame.)

-- Support different terminal sizes.

   (Done, no problem.)

-- Make sure terminal resizes are handled gracefully.  (Could be
   problematic.)

   (Done.  We don't get automatic SIGWINCH for additional ttys,
   though.)

-- Extend emacsclient to automatically open a new tty when it connects
   to Emacs.

   (Done.  It's an ugly hack, needs more work.)

-- Redisplay must refresh the topmost frame on *all* terminals, not
   just the initial terminal.

   (Done, but introduced an ugly redisplay problems.  Ugh.)

-- Fix redisplay problems.

   (Done; it turned out that the entire Wcm structure must be moved
   inside tty_output.  Why didn't I catch this earlier?)

-- Provide a way for emacsclient to tell Emacs that the tty has been
   resized.

   (Done, simply forward the SIGWINCH signal.)

-- Each keypress should automatically select the frame corresponding
   to the terminal that it was coming from.  This means that Emacs
   must know from which terminal the last keyboard event came from.

   (Done, it was quite simple, the input event system already
   supported multiple frames.)

-- Fix SIGIO issue with secondary terminals.

   (Done, emacsclient signals Emacs after writing to the proxy pseudo
   terminal.  Note that this means that multi-tty does not work with
   raw ttys!)

   (Update: This is bullshit.  There is a read_input_waiting function,
   extend that somehow.)

   (Update of update: The first update was not right either, extending
   read_input_waiting was not necessary.  Secondary ttys do seem to
   send signals on input.)

   (Update^3: Not any more.)

-- Make make-terminal-frame look up the `tty' and `tty-type' frame
   parameters from the currently selected terminal before the global
   default.

   (Done.)

-- Put all cached terminal escape sequences into struct tty_output.
   Currently, they are still stored in global variables, so we don't
   really support multiple terminal types.

   (Done.  It was not fun.)

-- Implement sane error handling after initialization.  (Currently
   emacs exits if you specify a bad terminal type.)  The helpful error
   messages must still be provided when Emacs starts.

   (Done.)

-- Implement terminal deletion, i.e., deleting local frames, closing
   the tty device and restoring its previous state without exiting
   Emacs.

   (Done, but at the moment only called when an error happens during
   initialization.  There is a memory corruption error around this
   somewhere.)  (Update: now it is fully enabled.)

-- Implement automatic deletion of terminals when the last frame on
   that terminal is closed.

   (Done.)

-- Restore tty screen after closing the terminal.

   (Done, we do the same as Emacs 21.2 for all terminals.)

-- 'TERM=dumb src/emacs' does not restore the terminal state.

   (Done.)

-- C-g should work on secondary terminals.

   (Done, but the binding is not configurable.)

-- Deal with SIGHUP in Emacs and in emacsclient.  (After this, the
   server-frames may be removed from server.el.)

   (Done, nothing to do.  It seems that Emacs does not receive SIGHUP
   from secondary ttys, which is actually a good thing.)  (Update: I
   think it would be a bad idea to remove server-frames.)

-- Change emacsclient/server.el to support the -t argument better,
   i.e. automatically close the socket when the frame is closed.

   (Seems to be working OK.)

-- Fix mysterious memory corruption error with tty deletion.  To
   trigger it, try the following shell command:

	while true; do TERM=no-such-terminal-definition emacsclient -h; done

   Emacs usually dumps core after a few dozen iterations.  (The bug
   seems to be related to the xfreeing or bzeroing of
   tty_output.Wcm.  Maybe there are outside references to struct Wcm?
   Why were these vars collected into a struct before multi-tty
   support?)

   (Done.  Whew.  It turned out that the problem had nothing to do
   with hypothetical external references to Wcm, or any other
   tty_output component; it was simply that delete_tty closed the
   filehandles of secondary ttys twice, resulting in fclose doubly
   freeing memory.  Utterly trivial matter.  I love the C's memory
   management, it puts hair on your chest.)

-- Support raw secondary terminals.  (Note that SIGIO works only on
   the controlling terminal.) Hint: extend read_input_waiting for
   multiple ttys and hopefully this will be fixed.

   (Done, it seems to have been working already for some time.  It
   seems F_SETOWN does work, after all.  Not sure what made it fail
   earlier, but it seems to be fixed (there were several changes
   around request_sigio, maybe one of them did it).
   read_input_waiting is only used in sys_select, don't change
   it.)  (Update: After adding X support, it's broken again.)
   (Update^2: No it isn't.) :-)

-- Find out why does Emacs abort when it wants to close its
   controlling tty.  Hint: chan_process[] array.  Hey, maybe
   noninterrupt-IO would work, too?  Update: no, there is no process
   for stdin/out.

   (Done.  Added add/delete_keyboard_wait_descriptor to
   term_init/delete_tty.  The hint was right, in a way.)

-- Issue with SIGIO: it needs to be disabled during redisplay.  See if
   fcntl kernel behavior could be emulated by emacsclient.

   (Done.  Simply disabled the SIGIO emulation hack in emacsclient.)
   (Update: it was added back.)  (Update^2: and removed again.)

-- server.el: There are issues with saving files in buffers of closed
   clients.  Try editing a file with emacsclient -f, and (without
   saving it) do a delete-frame.  The frame is closed without
   question, and a surprising confirmation prompt appears in another
   frame.

   (Done.  delete-frame now asks for confirmation if it still has
   pending buffers, and modified buffers don't seem to be deleted.)

-- emacsclient.el, server.el: Handle eval or file open errors when
   doing -t.

   (Done.)

-- Make parts of struct tty_output accessible from Lisp.  The device
   name and the type is sufficient.

   (Done, see frame-tty-name and frame-tty-type.)

-- Export delete_tty to the Lisp environment, for emacsclient.

   (Done, see delete-tty.)

-- Get rid of the accessor macros in termchar.h, or define macros for
   all members.

   (Done.)

-- Move device-specific parameters (like costs) commonly used by
   device backends to a common, device-dependent structure.

   (Done.  See struct display_method in termhooks.h.)

-- Fix X support.

   (Done.  Well, it seems to be working.)

-- Allow simultaneous X and tty frames.  (Handling input could be
   tricky.  Or maybe not.)

   (Done.  Allowed, that is.  It is currently extremely unstable, to
   the point of being unusable.  The rif variable causes constant
   core dumps.  Handling input is indeed tricky.)

-- Rewrite multi-tty input in terms of MULTI_KBOARD.

   (Done.  In fact, there was no need to rewrite anything, I just
   added a kboard member to tty_display_info, and initialized the
   frame's kboard from there.)

-- Fix rif issue with X-tty combo sessions.  IMHO the best thing to do
   is to get rid of that global variable (and use the value value in
   display_method, which is guaranteed to be correct).

   (Done, did exactly that.  Core dumps during combo sessions became
   much rarer.  In fact, I have not yet met a single one.)

-- Add multi-tty support to talk.el.

   (Done.)

-- Clean up the source of emacsclient.  It is a mess.

   (Done, eliminated stupid proxy-pty kludge.)

-- Fix faces on tty frames during X-tty combo sessions.  There is an
   init_frame_faces call in init_sys_modes, see if there is a problem
   with it.

   (Done, there was a stupid mistake in
   Ftty_supports_face_attributes_p. Colors are broken, though.)

-- C-x 5 2, C-x 5 o, C-x 5 0 on an emacsclient frame unexpectedly
   exits emacsclient.  This is a result of trying to be clever with
   delete-frame-functions.

   (Fixed, added delete-tty-after-functions, and changed server.el to
   use it.)

-- Something with (maybe) multi-keyboard support broke function keys
   and arrows on ttys during X+tty combo sessions.  Debug this.

   (I can't reproduce it, maybe the terminal type was wrong.)

-- Fix input from raw ttys (again).

   (Now it seems to work all right.)

-- During an X-tty combo session, a (message "Hello") from a tty frame
   goes to the X frame.  Fix this.

   (Done.  There was a safeguard against writing to the initial
   terminal frame during bootstrap which prevented echo_area_display
   from working correctly on a tty frame during a combo session.)

-- If there are no frames on its controlling terminal, Emacs should
   exit if the user presses C-c there.

   (Done, as far as possible.  See the SIGTERM comment in
   interrupt_signal on why this seems to be impossible to solve this
   in general.)

-- During an X session, Emacs seems to read from stdin.  Also, Emacs
   fails to start without a controlling tty.

   (Fixed by replacing the troublesome termcap display with a dummy
   bootstrap display during bootstrap.

-- Do tty output through struct display, like graphical display
   backends.

   (Done.)

-- Define an output_initial value for output_method for the initial
   frame that is dumped with Emacs.  Checking for this frame (e.g. in
   cmd_error_internal) is ugly.

   (Done, breaking interactive temacs.)

-- The command `emacsclient -t -e '(delete-frame)'' fails to exit.

   (Fixed.)

-- frame-creation-function should always create a frame that is on the
   same display as the selected frame.  Maybe frame-creation-function
   should simply be removed and make-frame changed to do the right
   thing.

   (Done, with a nice hack.  frame-creation-function is now frame-local.)

-- Fix C-g on raw ttys.

   (Done.  I disabled the interrupt/quit keys on all secondary
   terminals, so Emacs sees C-g as normal input.  This looks like an
   overkill, because emacsclient has extra code to pass SIGINT to
   Emacs, so C-g should remain the interrupt/quit key on emacsclient
   frames.  See the next entry why implementing this distinction would
   be a bad idea.)

-- Make sure C-g goes to the right frame with ttys.  This is hard, as
   SIGINT doesn't have a tty parameter. :-(

   (Done, the previous change fixes this as a pleasant side effect.)

-- I have seen a case when Emacs with multiple ttys fell into a loop
   eating 100% of CPU time.  Strace showed this loop:

	getpid()                                = 30284
	kill(30284, SIGIO)                      = 0
	--- SIGIO (I/O possible) @ 0 (0) ---
	ioctl(6, FIONREAD, [0])                 = -1 EIO (Input/output error)
	ioctl(5, FIONREAD, [0])                 = -1 EIO (Input/output error)
	ioctl(0, FIONREAD, [0])                 = 0
	sigreturn()                             = ? (mask now [])
	gettimeofday({1072842297, 747760}, NULL) = 0
	gettimeofday({1072842297, 747806}, NULL) = 0
	select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0})
	select(9, [0 3 5 6], NULL, NULL, {0, 0}) = 2 (in [5 6], left {0, 0})
	gettimeofday({1072842297, 748245}, NULL) = 0

   I have seen something similar with a single X frame, but have not
   been able to reproduce it for debugging.

   Update: This may have been caused by checking for nread != 0
   instead of nread > 0 after calling read_socket_hook in
   read_avail_input.

   (Fixed.  This was caused by unconditionally including stdin in
   input_wait_mask in init_process.  The select call in
   wait_reading_process_input always returned immediately, indicating
   that there is pending input from stdin, which nobody read.

   Note that the above strace output seems to be an unrelated but
   similar bug.  I think that is now fixed.)

-- Exiting Emacs while there are emacsclient frames doesn't restore the
   ttys to their default states.

   (This seems to be fixed by some previous change.)

-- Allow opening an X session after -nw.

   (Done.)

-- Fix color handling during tty+X combo sessions.  (It seems that tty
   sessions automatically convert the face colors to terminal colors
   when the face is loaded.  This conversion must happen instead on
   the fly in write_glyphs, which might be problematic, as color
   approximation is currently done in lisp (term/tty-colors.el).)
   (Update: hm, colors seem to work fine if I start emacs with -nw and
   then create an X frame.  Maybe it's just a small buglet somewhere.)

   (Seems to be fixed.  The problem was in startup.el, it did not
   initialize tty colors when the initial window system was
   graphical.)

-- emacs -nw --eval '(y-or-n-p "Foobar")' segfaults.  (Reported by
   Romain Francoise)

   (Fixed, there was a keyboard initialization problem.)

-- Fix interactive use of temacs.  There are face-related SEGVs, most
   likely because of changes in realize_default_face, realize_face.

   (Fixed.)

-- Don't exit Emacs when the last X connection fails during a
   multi-display session.

   (Fixed.)

-- Dan Nicolaescu noticed that starting emacsclient on the same
   terminal device that is the controlling tty of the Emacs process
   gives unexpected results.

   (Fixed.)

-- Istvan Marko reported that Emacs hang on ttys if it was started
   from a shell script.

   (Fixed.  There was a bug in the multi-tty version of
   narrow_foreground_group.  tcsetpgrp blocks if it is called from a
   process that is not in the same process group as the tty.)

-- emacsclient -t from an Emacs term buffer does not work, complains
   about face problems.  This can even lock up Emacs (if the recursive
   frame sets single_kboard).  Update: the face problems are caused by
   bugs in term.el, not in multi-tty.  The lockup is caused by
   single_kboard mode, and is not easily resolvable.  The best thing to
   do is to simply refuse to create a tty frame of type `eterm'.

   (Fixed, changed emacsclient to check for TERM=eterm.  The face
   complaints seem to be caused by bugs in term.el; they are not
   related to multi-tty.)

-- Find out the best way to support suspending Emacs with multiple
   ttys.  My guess: disable it on the controlling tty, but from other
   ttys pass it on to emacsclient somehow.  (It is (I hope) trivial to
   extend emacsclient to handle suspend/resume.  A `kill -STOP' almost
   works right now.)

   (Done.  I needed to play with signal handling and the server
   protocol a bit to make emacsclient behave as a normal UNIX program
   wrt foreground/background process groups.)

-- There is a flicker during the startup of `emacs -nw'; it's as if
   the terminal is initialized, reset and then initialialized again.
   Debug this.  (Hint: narrow_foreground_group is called twice during
   startup.)

   (This is gone.)

-- Robert Chassell has found serious copy-paste bugs with the
   multi-tty branch.  There seem to be redisplay bugs while copying
   from X to a terminal frame.  Copying accented characters do not
   work for me.

   (Patch-124 should fix this, by changing the interprogram-*-function
   variables to be frame-local, as suggested by Mark Plaksin
   (thanks!).  I think that the redisplay bugs are in fact not bugs,
   but delays caused by single_kboard --> perhaps MULTI_KBOARD should
   be removed.)

-- frame-creation-function was removed, which might be a bad idea.
   Think up a compatible solution.

   (It was an internal interface that may be changed when necessary.)

-- Change Lisp code not to (getenv "TERM"); use the `tty-type' frame
   parameter or the frame-tty-type function instead.  (M-x tags-search
   "TERM" helps with this.)  Update: Actually, all getenv invocations
   should be checked for multi-tty compatibility, and an interface
   must be implemented to get the remote client's environment.

   (Done.  Only getenv calls in lisp/term/*.el were changed; other
   calls should be mostly left as they are.)

-- Add an elaborate mechanism for display-local variables.  (There are
   already a few of these; search for `terminal-local' in the Elisp
   manual.)

   (Not needed.  Display-local variables could be emulated by
   frame-local variables.)

-- Emacs assumes that all terminal frames have the same locale
   settings as Emacs itself.  This may lead to bogus results in a
   multi-locale setup. (E.g., while logging in from a remote client
   with a different locale.)
   (Update after new bugreport by Friedrich Delgado Friedrichs: 
   (at least) the structs terminal_coding and keyboard_coding in
   coding.c must be moved to struct display, and the Lisp interface
   [set-]keyboard-coding-system must be adapted for the change.)

   (Fixed.  Emacs now uses the locale settings as seen by the
   emacsclient process for server tty frames.)
   (Update: Not really; Vlocale_coding_system is still global.)

-- Make `struct display' accessible to Lisp programs.  Accessor functions:

	(displayp OBJECT):  Returns t if OBJECT is a display.
		=> Implemented as display-live-p.

	(display-list):  Returns list of currently active displays.
		=> Implemented.

	(selected-display):  Returns the display object of the selected frame.
		=> Not strictly necessary, but implemented anyway.

	(frame-display FRAME):  Returns the display object of FRAME.
	        => Implemented.

	(display-frames DISPLAY):  Returns a list of frames on DISPLAY.
		=> Already implemented, see frames-on-display-list.

	(display-type DISPLAY):  Returns the type of DISPLAY, as a
		symbol.  (See `framep'.)
		=> Implemented as display-live-p.

	(display-device DISPLAY): Returns the name of the device that
		DISPLAY uses, as a string.  (E.g: "/dev/pts/16", or
		":0.0")
		=> Implemented as display-name.

	etc.

   See next issue why this is necessary.

   (Update: The consensus on emacs-devel seems to be to do this via
   integer identifiers.  That's fine by me.)

   (Done.)

-- The following needs to be supported:

	$ emacsclient -t
		C-z
	$ emacsclient -t
		(This fails now.)

   The cleanest way to solve this is to allow multiple displays on the
   same terminal device; each new emacsclient process should create
   its own display.  As displays are currently identified by their
   device names, this is not possible until struct display becomes
   accessible as a Lisp-level object.

   (Done.)

-- Miles Bader suggests that C-x C-c on an emacsclient frame should
   only close the frame, not exit the entire Emacs session.  Update:
   see above for a function that does this.  Maybe this should be the
   new default?  

   (Done.  This is the new default.  No complaints so far.)

-- Clean up the frame-local variable system.  I think it's ugly and
   error-prone.  But maybe I just haven't yet fully understood it.

   (Nothing to do.  It doesn't seem ugly any more.  It's rather clever.)

-- Support multiple character locales.  A version of
   `set-locale-environment' needs to be written for setting up
   display-local settings on ttys.  I think calling
   set-display-table-and-terminal-coding-system and
   set-keyboard-coding-system would be enough.  The language
   environment itself should remain a global setting.

   (Done, by an ugly hack.)

-- The terminal customization files in term/*.el tend to change global
   parameters, which may confuse Emacs with multiple displays.  Change
   them to tweak only frame-local settings, if possible.  (They tend
   to call define-key to set function key sequences a lot.)

   (Done, by making `function-key-map' terminal-local (i.e., part of
   struct kboard).  This has probably covered all the remaining problems.)

-- Make `function-key-map' and `key-translation-map' terminal-local.

   (Done.)

-- Implement `terminal-local-value' and `set-terminal-local-value' to
   allow deterministic access to terminal local bindings.  The
   encode-kb package can not set up `key-translation-map' without
   these.  The terminal-local bindings seem to be independent of what
   frame is selected.

   (Done.)

-- xt-mouse.el needs to be adapted for multi-tty.  It currently
   signals an error on kill-emacs under X, which prevents the user
   from exiting Emacs. (Reported by Mnemonikk on freenode.)

   (Done, I hope.)
   

-- Having {reset,init}_all_sys_modes in set-input-mode breaks arrow
   keys on non-selected terminals under screen, and sometimes on other
   terminal types as well.  The other function keys continue to work
   fine.  Sometimes faces on these screens become garbled.

   This only seems to affect displays that are of the same terminfo
   type as the selected one. Interestingly, in screen Emacs normally
   reports the up arrow key as `M-o A', but after the above SNAFU, it
   complains about `M-[ a'.  UNIX ttys are a complete mystery to me,
   but it seems the reset-reinitialize cycle somehow leaves the
   non-selected terminals in a different state than usual.  I have no
   idea how this could happen.

   Currently set-input-mode resets only the currently selected
   terminal, which seems to somehow work around the problem.

   Update:

	Dan Nicolaescu <[e-mail]> writes:
	> Some terminals have 2 modes for cursor keys: Application Mode where
	> the cursor keys transmit the codes defined in the terminfo entry, and
	> Cursor mode. Applications have to send the smkx and rmkx terminfo
	> strings to switch between the 2 modes. So Emacs (and emacsclient) have
	> to send smkx when initializing and rmkx when quitting (or on
	> suspend). 

   (I think patch-370 fixed this.)

-- This long-standing bug (first reported by Han Boetes) seems to come
   and go all the time.  It is time to track it down and fix it.

	emacs
		M-x server-start
	
	# From another xterm:
	emacsclient -e '(y-or-n-p "Do you want me to crash? ")'
		# Notice how the answer ends up in the *scratch* buffer
		M-x garbage-collect
		SIGSEGV

   (Fixed in patch-414 after detailed analysis by Kalle Olavi Niemitalo.)

-- normal-erase-is-backspace-mode in simple.el needs to be updated for
   multi-tty (rep. by Dan Waber).  (The Delete key is broken on X
   because of this.)

   (Fixed in patch-427.)

-- I think keyboard-translate-table should be made terminal-local.

   (Done in patch-431.)

-- The semantics of terminal-local variables are confusing; it is not
   clear what binding is in effect in any given time.  See if
   current_kboard (or at least the terminal-local bindings exported to
   Lisp) might be changed to be tied to the selected frame instead.
   Currently, `function-key-map' and `key-translation-map' may be
   accessed reliably only using the hackish
   `(set-)terminal-local-value' functions.

   Perhaps there should be a difference between `last-command' &co.
   and these more conventional configuration variables.
   (E.g. `symbol-value' would use current_kboard to access
   `last-command', but SELECTED_FRAME()->display->kboard to get the
   value of `function-key-map'.

   (Fixed in patch-434.)

-- If the first key pressed on a new tty terminal is a function key,
   it is not recognized correctly.  May be related to the bug below.

   (Seems to have been fixed as a side effect of patch-434.  "The bug
   below" was the set-input-mode madness.)  

   (Update: this bug was fixed for good in patch-449.  It was tracked
   down to a bug in `read_key_sequence': it failed to reinitialize its
   local function-key-map/key-translation-map references when it
   switched keyboards.  I don't understand why did this bug only
   appear on brand new frames, though!)

-- Disable connecting to a new X display when we use the GTK toolkit.

   (Disabled in patch-450.)

-- Implement automatic forwarding of client environment variables to
   forked processes, as discussed on the multi-tty list.  Terminal
   parameters are now accessible in C code, so the biggest obstacle is
   gone.  The `getenv_internal' and `child_setup' functions in
   callproc.c must be changed to support the following variable:

	terminal-local-environment-variables is a variable defined in ...

	Enable or disable terminal-local environment variables.

	If set to t, `getenv', `setenv' and subprocess creation
	functions use the environment variables of the emacsclient
	process that created the selected frame, ignoring
	`process-environment'.

	If set to nil, Emacs uses `process-environment' and ignores
	the client environment.

	Otherwise, `terminal-local-environment-variables' should be a
	list of variable names (represented by Lisp strings) to look
	up in the client environment.  The rest will come from
	`process-environment'.

   (Implemented in patch-461; `terminal-getenv', `terminal-setenv' and
   `with-terminal-environment' are now replaced by extensions to
   `getenv' and `setenv', and the new `local-environment-variables'
   facility.  Yay!)

   (Updated in patch-465 to fix the semantics of let-binding
   `process-environment'.  `process-environment' was changed to
   override all local/global environment variables, and a new variable
   `global-environment' was introduced to have `process-environment's
   old meaning.)

   (Updated in patch-466 to fix the case when two emacsclient sessions
   share the same terminal, but have different environment.  The local
   environment lists are now stored as frame parameters, so the
   C-level terminal parameters are not strictly necessary any more.)

-- `Fdelete_frame' is called from various critical places where it is
   not acceptable for the frame deletion to fail, e.g. from
   x_connection_closed after an X error.  `Fdelete_frame' now protects
   against `delete-frame-functions' throwing an error and preventing a
   frame delete. (patch-475)

-- Fix set-input-mode for multi-tty.  It's a truly horrible interface;
   what if we'd blow it up into several separate functions (with a
   compatibility definition)?

   (Done.  See `set-input-interrupt-mode', `set-output-flow-control',
   `set-input-meta-mode' and `set-quit-char'.) (patch-457)

-- Let-binding `overriding-terminal-local-map' on a brand new frame
   does not seem to work correctly.  (See `fancy-splash-screens'.)
   The keymap seems to be set up right, but events go to another
   terminal.  Or is it `unread-command-events' that gets Emacs
   confused?  Investigate.

   (Emacs was confused because a process filter entered
   `recursive-edit' while Emacs was reading input.  I added support
   for this in the input system.) (patch-489)

-- I smell something funny around pop_kboard's "deleted kboard" case.
   Determine what are the circumstances of this case, and fix any
   bug that comes to light.

   (It happens simply because single_kboard's terminal is sometimes
   deleted while executing a command on it, for example the one that
   kills the terminal.  There was no bug here, but I rewrote the whole
   single_kboard mess anyway.) (patch-489)

-- Understand Emacs's low-level input system (it's black magic) :-)
   What exactly does interrupt_input do?  I tried to disable it for
   raw secondary tty support, but it does not seem to do anything
   useful.  (Update: Look again. X unconditionally enables this, maybe
   that's why raw terminal support is broken again.  I really do need
   to understand input.)
   (Update: I am starting to understand the read_key_sequence->read-char
   ->kbd_buffer_get_event->read_avail_input->read_socket_hook path.  Yay!)

   (Update: OK, it all seems so easy now (NOT).  Input could be done
   synchronously (with wait_reading_process_input), or asynchronously
   by SIGIO or polling (SIGALRM).  C-g either sets the Vquit_flag,
   signals a 'quit condition (when immediate_quit), or throws to
   `getcjmp' when Emacs was waiting for input when the C-g event
   arrived.)

-- Replace wrong_kboard_jmpbuf with a special return value of
   read_char.  It is absurd that we use setjmp/longjmp just to return
   to the immediate caller.

   (Done in patch-500.)

-- `tool-bar-mode', `scroll-bar-mode', `menu-bar-mode' and
   'fringe-mode' are modes global to the entire Emacs session, not
   just a single frame or a single terminal.  This means that their
   status sometimes differs from what's actually displayed on the
   screen.  As a consequence, the Options | Show/Hide menu sometimes
   shows incorrect status, and you have to select an option twice for
   it to have any visible effect on the current frame.

   Change Emacs so that the status of the items in the Options |
   Show/Hide menu correspond to the current frame.

   (Done in patch-537.)

-- The `default-directory' variable should somehow be set to the
   cwd of the emacsclient process when the user runs emacsclient
   without file arguments.  Perhaps it is OK to just override the
   directory of the *scratch* buffer.

   (Done in patch-539.)

-- The borders on tooltip frames on X are messed up.  More
   specifically, the frame's internal border (internal-border-width
   frame parameter) is not filled with the correct background color.

   It seems the border contents are drawn onto by the
   update_single_window call in `x-show-tip'.  After some debugging, I
   think the window's background color is not set up
   correctly---calling `x_clear_area' fills the specified area with
   black, not light yellow.

   (Done in patch-544.  A background_pixel field was defined both in
   struct frame and struct x_output, and Emacs got confused between
   them.)


This file is part of GNU Emacs.

GNU Emacs is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

GNU Emacs is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.

;;; arch-tag: 8da1619e-2e79-41a8-9ac9-a0485daad17d