hmacht at suse.de
Fri Mar 30 03:52:19 PDT 2007
On Fri 30. Mar - 11:05:19, Richard Hughes wrote:
> On Fri, 2007-03-30 at 11:31 +0200, Holger Macht wrote:
> > On Thu 29. Mar - 23:43:22, Richard Hughes wrote:
> > > On Fri, 2007-03-30 at 00:38 +0200, Holger Macht wrote:
> > > >
> > > > Simply speaking, for me, Shutdown() involves two
> > > > basic tasks. o.f.Desktop.Session.Save() and o.f.Hal.Shutdown().
> > > > Suspend() doesn't include either of the two.
> > >
> > > Yes, I agree. But we don't have o.f.Desktop.Session.Save right now, and
> > > probably won't for quite a while yet.
> > Why not? That should be the aim and the desktop people should get their
> > ass up ;-)
> > Specs should be defined from the bottom to the top IMHO. Get a standard
> > general desktop interface and build the pm interface above it.
> Sure, it's okay to put this in a desktop API, but then we have to
> duplicate the Inhibit stuff and provide some complex interaction.
Is there a common restriction that one interface (o.f.pm) can't be served
by two different applications? And those two applications can't use the
same set of functionality? We shouldn't do things because of limited
functionality. Fix the functionality then.
I like to see something like this in future:
And only the o.f.D.PM.* methods are served by the power management
> > > I'll do you a deal. I'll deprecate Shutdown() when we have a o.f.Desktop
> > > spec implemented by at least one desktop. ;-)
> > What consumers do you actually have for your Shutdown() method in
> > gnome-power-manager right now?
> That's a good point; not many.
> - [x] shutdown the computer when finished converting
> - The logout dialogue
So in current GNOME, you can't logout without g-p-m running?
> - Any random desktop shortcut icon
> - The GDM login screen
> > So if you like to have a spec as fast as possible, go for it, but you
> > won't get my agreement because it will create more problems in the long
> > run as it will solve.
> What about inhibit? We can do some cool integration stuff just by
> keeping this on the same service.
Again, if inhibit is important for Shutdown() and Reboot() in the same way
as it is for Suspend() and friends, inhibit functionality should be in
> Also, think twm. We now are asking them to add *two* interfaces into the
> desktop base install. It's much simpler for them to install a shortcut
> setuid to call /sbin/shutdown. Ick.
Just do it only one time, and do it right.
And if you're arguing that way, you can just make all other methods except
Shutdown() and Reboot() optional. That's not what we're trying to reach.
More information about the xdg