Re: [gdm-list] face-browser proposal for new gdm

[Brian Cameron <Brian Cameron Sun COM>]

Matthias/Frederic/Others:

> > > But finding this
> > > "Theme-support was based on a XML-file in the old gdm. This should be
> > > taken over to the new gdm."
> > > in the text has me concerned. I hope you don't mean that the new greeter should
> > > read the same xml format that the current greeter was using for themes ?
> > > Getting rid of that format is one of my personal hopes for a new greeter...
> > Well, I must say I seriously hope the xml format will be kept.
> >
> > It is one of the few things which allows us to have almost the same
> > greeter theme file between kdm and gdm. It would be a pity to loose it
> > (and even more to not be able to have a graphically appealing and
> > themable greeter).
>
> I'd propose to have a little poll among the people who have to tweak
> the various distros gdm themes regularly. I can only say that from my 
> own experience, and from sitting next to Ray, the format is not 
well-loved...

It can be frustrating to create a good theme, and there are bugs.
However, I suspect many users do enjoy using the various themes that
are available - and the ability to (reasonably) easily customize themes
to their linking.

> The semantics are poorly documented, hard to understand and difficult
> to use.

The documentation could be better, I agree.  However, I'm not sure its
really as bad as all that.  The most difficult part is defining a good
theme that works well at all display sizes, which I think would be a bit
of work regardless of how you go about customizing the appearance of
GDM.

> The format also has limitations that make many interesting designs
> impossible to implement (e.g. the fact that size-changes of elements
> at runtime are very poorly handled, leading to broken situations when
> translated strings vary in size). The implementation on top of
> GnomeCanvas has serious drawbacks, e.g. for accessibility.

True, there has been talk of moving to something more sophisticated such
as cairo.  I'm sure there are other bugs also.

> I believe the current thinking wrt to themable greeters is to offer a
> sane dbus interface that facilitates writing alternative greeters 
> rather than offering explicit theming support.

For some situations, it perhaps makes sense to "write your own greeter"
to make it look the way you want (if you wanted to depend on new
interfaces, such as an OpenGL greeter, for example).

However, suggesting that they need to write a new greeter application to
change the appearance isn't a great solution for theming/skin changing.
While the XML format may not be perfect, it works for most situations
reasonably and there are plenty of themes that work with it.

Also, keep in mind the the older and "less sane" stdin/stdout protocol
for the daemon/slave talking to each other was not really a barrier to
writing your own "custom" greeter.  It would have been fairly trivial
to write your own greeter using the old code.  I'm not sure that the
fact we now use D-Bus will really mean that people will suddenly develop
a serious interest in writing greeter programs.  Time will tell.

> If you really insist on keeping compatibility with the old xml format,
> your best bet may be to port the old greeter to the greeter dbus 
> interface - though I have no idea how feasible that is.

It shouldn't be too difficult, since the main thing that changed is
just the protocol for talking to the daemon.  I guess we will find out
if anybody loves gdmgreeter if anybody thinks it is worth the time to
do this work.

It would be more work to enhance gdmgreeter to work with something
better than GnomeCanvas, which might be a good idea if someone thinks
it is worth the effort to make work with the new D-Bus protocols.

Brian