bug#7552: 23.2; Emacs doesn't die when told to by X session manager

[Chong Yidong <cyd@stupidchicken.com>]

Robin Green <greenrd@greenrd.org> writes:

> Until now, this would probably not matter, since normally when an X
> session is stopped, all clients will be killed anyway. However, KDE 4.6
> beta 1 allows the user to stop and start subsessions called "activities"
> without logging out. When the KDE session manager tells emacs to die,
> it should do so - but it doesn't, and as a consequence, the following
> happens:
>
> When you stop the activity in KDE, because the stopped activity isn't
> currently visible, emacs is not visible on the display - but it still
> remains running. Then if you restart the same activity, another copy of
> emacs opens up, and the frame of the original emacs process reappears -
> so you now have two emacs processes running, which is not what is
> supposed to happen.
>
> The X session management specification does anticipate subsessions
> being used - indeed, it gives an example of them - but as far as I know,
> they have not been widely used until now, which may explain why this bug
> has gone unnoticed for many years.

AFAICT, there is code for Emacs to die when told to by the session
manager.  Could you check if Emacs is compiled with xsm support?  You
should have something like this in your configure log:

  checking X11/SM/SMlib.h usability... yes
  checking X11/SM/SMlib.h presence... yes
  checking for X11/SM/SMlib.h... yes
  checking for SmcOpenConnection in -lSM... yes