GNOME Bugzilla – Bug 332410
should not ship gnome-wm
Last modified: 2011-02-10 00:40:43 UTC
It is of dubious usefulness, and its documentation is partially wrong and misleading. Just remove it. Originally filed here: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=182698
This would also obsolete bug #350848 and bug #110722.
We have to keep a gnome-wm executable since it's part of our "ABI" (it could break some sessions if we removed it). However, we can decide to make it a link to metacity. Or we can keep it as it is, which is not a big issue, IMHO.
It seems it's getting used more and more, to decide what should be launched (mainly a choice between compiz or metacity).
We need to solve this in the new code base (which has a different way of defining the default session). One option would be to have a gnome-wm.desktop which runs gnome-wm and have it in the default session. Not sure if this is a clean solution though...
(In reply to comment #3) > It seems it's getting used more and more, to decide what should be launched > (mainly a choice between compiz or metacity). See also bug 461730, which covers the capplet side of picking compiz vs metacity, although they don't seem to be worrying about how to communicate that decision to gnome-session yet. (I suggested that we probably want a dbus interface so it can say "remove metacity from the session", "add compiz to the session", etc.) (In reply to comment #4) > One option would be to have a gnome-wm.desktop which runs gnome-wm and have it > in the default session. Not sure if this is a clean solution though... If we do that, we need to make it clear that it's just for (a) backward compatibility with old saved sessions, and (b) picking between officially GNOME-blessed WMs as in the bug 461730 case. We definitely want to get out of the "please update gnome-wm for yet another change to upstream compiz/beryl/whatever-the-kids-are-calling-it-this-week", etc. business Hm... actually: (In reply to comment #2) > We have to keep a gnome-wm executable since it's part of our "ABI" (it could > break some sessions if we removed it). This is not actually true with new-gnome-session; if their saved session tries to run gnome-wm, it will fail, and then they'll have no wm, and so we'll run metacity anyway. Or for that matter, the importing-old-saved-session code could just translate gnome-wm to metacity automatically.