[PATCH v2][xdm] Implement ConsoleKit support.

[Takashi Iwai]

At Sat, 23 Jul 2011 12:49:29 +0200,
Julien Cristau wrote:
> 
> On Sat, Jul 23, 2011 at 12:35:11 +0200, Takashi Iwai wrote:
> 
> > At Sat, 23 Jul 2011 10:49:53 +0200,
> > Julien Cristau wrote:
> > > 
> > > On Sat, Jul 23, 2011 at 09:31:41 +0200, Takashi Iwai wrote:
> > > 
> > > > At Fri, 22 Jul 2011 21:49:08 +0200,
> > > > Julien Cristau wrote:
> > > > > 
> > > > > Is "ck-daemon isn't running" a good reason to prevent the user from
> > > > > logging in?  (Why?)
> > > > 
> > > > It's reached only when you set use_consolekit is set.  You say you
> > > > are using CK but it's not detected, thus it's an error.
> > > > 
> > > use_consolekit defaults to yes, so that's not really true.
> > 
> > Feel free to set it off as default.
> > 
> > But note that this is set true only when the consolekit support is
> > found and enabled via configure script, and most modern distros will
> > make it enabled as default anyway.
> > 
> There's a difference between a build-time setting and a run-time
> setting.

Yes, but there is no difference: most distros will do enable it as
default for runtime setting, too.

> > >  It doesn't
> > > seem to be flagged as an error by pam-ck-connector, in any case (it
> > > returns PAM_IGNORE), I'm not sure why it's fatal for xdm.
> > 
> > It could be handled as non-fatal when no error code is set, too.
> > 
> > But usually you'll face problems sooner or later in such a situation
> > because it means that you logged in without proper device
> > permissions.
> > 
> "I don't have access to the audio device" is a better failure mode than
> "I can't even log in", in my book.

Its your book, and others may think differently like me.  I'd like to
see the obvious error beforehand than too late.

That said, this is a kind of problem for which no 100% correct answer
exists.  And I personally don't care at all how the default behavior
would be, as this is about a handling of exceptional error case.  So
feel free to cook the code.


thanks,

Takashi