dbus-1.2.3 and gnome-settings-daemon-2.24.1
hp at pobox.com
Wed Mar 11 15:59:10 PDT 2009
On Wed, Mar 11, 2009 at 12:35 PM, Colin Walters <walters at verbum.org> wrote:
> I see the gnome-session patch as in a similar vein as the current
> autospawning code - a last ditch attempt for "legacy" situations.
> Here's a random example:
> The tutorial tells you to just run "gnome-session" in the script,
> which *used* to work. Should VNC run dbus-session? Well, it's
> debatable and gets into the whole mess of multiple logins for a
> particular uid. Should all things be updated to run "dbus-launch
> gnome-session". Probably. But the gnome-session patch does that
> automatically, at the cost of a small speed hit.
Another cost to this is that if people run anything else *besides*
gnome-session (e.g. if they have a .Xclients like:
) then that will break. In fact in the link you posted above, it says
to run an xterm before gnome-session, and then any app launched from
that xterm would not find the dbus-daemon. (Or, would only find it via
X property anyway, maybe OK.)
the advantage to gnome-session breaking instead of trying to recover
is that people at least get an explanation of what's wrong and
distributions are forced to fix their stuff.
But, maybe the best of all worlds is for gnome-session to recover, but
print a loud warning explaining the problem.
More information about the dbus