Re: [gdm-list] GDM status.

[Brian Cameron <Brian Cameron Sun COM>]

Jon & Release Team:

> > > I think the following issues should be considered:
> > >
> > > + GDM doesn't support multi-display.  This means that distros that
> > >   want to support users with terminal-server, thin client, or other
> > >   novel setups will not be able to migrate to the new GDM rewrite until
> > >   this is addressed.
> > >
> > > + The new GDM rewrite doesn't contain any documentation.  GUI
> > >   configuration, integration hooks like the Init/PreSession/PostSession
> > >   script, how to configure ConsoleKit so it works properly with GDM, how
> > >   accessibility works, etc. are probably topics that should be
> > >   discussed.
> > >
> > > + The XDMCP chooser is not yet available from the login screen.
> > >
> > > + gdmsetup is not yet supported
> > >
> > > + gdmgreeter XML style themes are not supported

Another issue is that there has not yet been a 2.23 release of GDM (the
2.22.0 release was a bit late and came out on May 5th - a week before
the 2.23.2 release).  I think this makes it difficult for distros to
test the latest features.  Will there be a 2.23 release soon?  Also, I
think the latest GDM depends on some features in ConsoleKit that have
not yet been released, so a new release of ConsoleKit would also be
useful.

> > Some of those items look like real blockers. What is the plan for 2.24
> > in the GDM team?
>
> No, none of these are blockers.

If the release team feels that the lack of multi-display support is an
acceptable regression, then this will obviously cause problems for any
distro needing to support users who require this feature, such as any
terminal server-like setups.

Personally I also think being unable to access the XDMCP chooser from
the login screen will be a fairly serious regression for users who still
depend on it.  That said, XDMCP is old, crufty, and not very secure; so
I could understand if this would be considered an acceptable regression
for now.

I will leave it up to the release team to decide if these regressions
are acceptable.  Obviously any distros who need to support such features
can choose to hold off on updating to the new GDM if these regressions
will cause serious issues for their users.

> Documentation would have been nice
> for 2.22.  I think Brian is still committed to leading that effort
> [1].

Yes, I can help with this.  I should hopefully be able to get some
reasonable documentation together well before the September 1st string
freeze date.  Though, since the schedule is tight, there is some risk
that we might need to ask the release and l10n teams to allow some late
string changes.

> Brian was mistaken about the daemon not having configuration.

Yes, that is great to hear.  I am glad this fix made it into the 2.24
timeframe.

> "Multi-terminal" stuff will have to wait until we get multi-seat
> support in ConsoleKit.  I talked about some of these plans at GUADEC.
> [2]  Hopefully this will land for 2.26.

Great.

> Some of the things that were in gdmsetup have moved to other control
> center tools.  Hopefully, we'll have a page in the forthcoming user
> account tool to customize the login screen as well.

This sounds reasonable to me.  Will the user account tool also be ready
in the 2.24 timeframe?

> Someone who feels very strongly that XML greeter skins should continue
> to be supported will have to step up and do the work for this.  I
> think it is best to wait until a) we have a better idea which canvas
> will prevail b) we have a better idea how a new authentication system
> will work.

Personally I agree that this is not a serious issue, though I expect it
will cause stress for some users.

> No doubt there is still work to do - same as any module.  But there
> should be no doubt that the new GDM is ready to ship.

I agree that it is ready to ship for distros that only need to support
users who have a single display managed at-a-time.  However, it is
obviously not ready for any users in "Multi-terminal" environments.  It
sounds like we will need to wait until GDM 2.26 to support such users.

Obviously the advantage of accepting the GDM rewrite now is that this
would likely accelerate its development, and give users access to its
useful new features (such as better integration with a11y, keyboard
switching, power management, etc.).  From a more conservative
perspective, another six months would likely be enough time to sort out
such rough edges and make for a smoother transition.  There are pros and
cons either way.  I think the release team has been provided with enough
information to make a decision, at any rate.

At the very least, though, I think the GNOME release notes should
highlight such significant non-blocker regressions if the new GDM
rewrite is approved.  This way distros and users can decide if such
regressions are acceptable or if they want to hold off until they are
addressed.

Brian