[gdm-list] Status of GDM 2.22?

[Brian Cameron <Brian Cameron Sun COM>]

GDM community:

Now that we have passed the GNOME 2.21 API and feature freezes (and the
UI freeze is just one week away), I think it is time to assess where
we are in terms of the D-Bus rewrite and our roadmap.

I am a bit concerned about the energy level, which seems to have dwindled
a bit.  The last major update about the rewrite was posted to this list
in October, the GDM Wiki hasn't been updated since November, and there
has been little discussion at all since before the holidays.  Jon/Ray/etc,
do you think you could provide an update and refresh the TODO list on the
Wiki?

  http://live.gnome.org/GDM/NewDesign
  http://live.gnome.org/GDM/ToDo

At any rate, I think it makes sense to get an update on where we are,
what  people are doing, and our expectations regarding when we think the
GDM rewrite will be ready for a stable release.

When we first discussed whether to fork or whether the rewrite should
replace the existing GDM, we seemed to agree on a few options[1]:

- The rewrite would become GDM 2.22 if we got everything done in time.

- If we didn't get everything done in time, then we could skip the
  2.22 release and continue working in the development branch.  Then
  we would try to release the rewrite in a later release cycle.

- Although we decided forking was a bad option six months ago, now we
  have an opportunity to rethink.

----

I do have a number of concerns about the rewrite, and I want to see what
people think about them.  Forgive me if I mis-speak about any of these
issues.  I have not tested the latest code, so I might be off on a
bullet item or two.

  + Audit logic (both SELinux and Solaris) is not yet ported.  This is
    probably a hard requirement for some users.

  + Lack of configuration support or migration of configuration from older
    versions of GDM.  I am especially concerned since the current GDM
    documentation claims the configuration is stable.

    Many people have argued that it makes sense to deprecate and/or drop
    some configuration options.  To this I agree.  However, it seems that
    the new rewrite supports very few of the old "stable" configuration
    options.  This seems likely to cause frustrated amongst users who
    depend on the ability to configure GDM.  I have, for example, special
    concerns about people who depend on the ability to configure the
    [server] and [server-foo] sections of the configuration file.

  + Shutdown/Reboot/Suspend not working.

  + Lack of Xinerama support (not sure how well multihead support is
    looking).

  + Ability to select language in the login GUI.  There has been some
    discussion that this feature may not be so important, but I suspect
    this will be a big issue for some users.

  + Accessibility support in the login GUI.

  + Failsafe login in the login GUI.  (ditto on the "may not be so
    important but likely a big issue for some users")

  + How well is XDMCP choosing working in the rewrite?

Are there other important issues people want to highlight?  Probably
no need to again highlight Sun Ray issues since we have already talked
about those at length.

----

I am interested to hear what the various distros plan to do about GDM
in the 2.22 release cycle.  Do distros feel that GDM is ready for a
stable release or does it make more sense for GDM to punt for another
release cycle?

I know that Sun cannot switch to the new login program until it meets
various requirements (Sun Ray support, audit, etc.), so Solaris will
continue to use GDM 2.20 until those issues are addressed.  However, I
know other distros may not be so concerned about interface stability
and may want to switch to the rewrite sooner.

Or does it make sense to fork the rewrite into a separate project?  As
the rewrite has developed, it seems that there is little interest in
supporting the look & feel, themeability, the configuration, or other
aspects that make GDM "GDM".  As Jon has stated, there are now only a
handful of functions in the rewrite which have remained from the old
code-base.

Also, forking would allow distros to install both GDM and the new
login program in parallel.  This would make it easier for distros to
support both login programs while the rewrite still does not support
all functionality that users depend upon.

Lastly, there are a number of patches/fixes that didn't go into the
GDM 2.20 branch since they changed strings, etc.  If we fork, then
the old GDM can be maintained a bit better.

It might be a good idea to raise this discussion to a more general
list, such as desktop-devel-list to get a wider perspective on what
people think.

Thoughts?

Brian

[1] http://mail.gnome.org/archives/gdm-list/2007-October/msg00037.html