Re: New module proposal: LightDM

[Brian Cameron <brian cameron oracle com>]

Bastien:

> On Fri, 2011-05-13 at 13:43 -0500, Brian Cameron wrote:
> > GDM has evolved into a display manager that is most focused on tight
> > integration with GNOME.  This is perfect for GNOME users and distros
> > that focus on GNOME users.  However, GDM is not always a good choice
> > for other desktop systems, distros that perhaps want to provide
> > multiple desktop choices and be more desktop neutral about display
> > management, or distros that need to support devices that may not support
> > things light OpenGL.
>
> Since when did GDM require OpenGL? I must have missed the point when we
> started using gnome-shell in the login screen...

Yes, you are right.  GDM does not currently use OpenGL.

My comment was meant to be understood as an example of how GDM may be
moving in a direction that requires certain hardware or only works on
certain operating systems.  Sorry if I was not clear.  For example, I
also raised concerns that the GDM/ConsoleKit evolution may be moving in
a Linux-focused direction.

If GDM is evolving into a display manager with tight GNOME integration
that works only with specific hardware and/or operating systems, then
an alternative display manager may be needed by some users.  It is not
really clear what the future GDM/ConsoleKit plans are in this regards,
and nobody seems to clarify.

> <snip>
> > I think it is obviously important to Oracle to have display management
> > options that are not too tightly bound to things that are not supported
> > on Solaris like systemd, DeviceKit, PolicyKit, etc.  Also, Oracle's Sun
> > Ray products work best with a display manager that supports a non
> > OpenGL GUI.  I could imagine GDM becoming more tightly focused on
> > OpenGL in the future, like GNOME Shell.  Thus, perhaps LightDM could be
> > considered a "fallback" display manager for the GNOME community.

Again, I was just trying to highlight that some things that display
managers need to do can be different across different distros.  There
is already code in upstream GDM to make it work well with Solaris RBAC
instead of PolicyKit, for example.

> I'll repeat what I said in the past. Solaris developers will need to
> write some code at some point, or just give up. We can't stand around
> waiting for them to make a move when we want to better the functionality
> of GNOME as a Desktop.

I have personally written & committed over 100 patches to GDM since the
2.21 rewrite timeframe.  Work I have done has improved GDM usability,
XDMCP code, how GDM works when managing multiple displays at the same
time, etc.  Other Solaris developers have also helped, such as Halton
Huo.  This has bee time consuming since (as Jos pointed out), getting
changes accepted into GDM can be very time consuming.  So, developers
working on Solaris have been supportive of the new GDM rewrite, I
think.  Your comment seems to be rather dismissive.

Anyway, is there a real need or benefit to talk about Solaris GDM
participation in a discussion about whether to accept LightDM as a
module?

Brian