Re: GNOME Usability Improvements - Fix the window manager!

[Michael Fulbright <msf redhat com>]

Ok crew lets read the whole message next time before replying.

I did not say to throw away the GNOME "holy grail" of supporting all
window managers.

What I did say is (sorry old-timers, I know replying to your own
message is taboo):

> Note I have advocated two things:
>
>  1) Get the spec worked out in common for both the GNOME and KDE environments.
>     A window manager author will then have a single target and it is
>     much more likely window managers will properly work with both.
>
>  2) For the much larger class of potential new users (compared to the small
>     current user base of Linux) we should have a window manager we have
>     designed that works closely with GNOME (via CORBA) that completely
>     hides the window manager from the user. Themes will be shared with
>     gtk+ themes, the issue of who owns the background will be gone, the
>     pager and task list applets won't have to use broken X11 tricks to
>     get window lists, etc. 

I put in #1 because yes, hackers love to hack and I want GNOME to be
flexible enough to keep them interested. If we're going to have that option 
we should at least do it in a way compatible with KDE.  I understand at least 
one window manager isn't working with GNOME because they saw they had to
support both KDE and GNOME hints. Working with the KDE team to iron out 
a single spec will be a win for both of us. Complying with this spec means
that you use a legacy window manager with GNOME/KDE and it will act
reasonably.

I put in #2 because so far it seems a great deal of people I speak with
see #1 as some sort of "black and white" issue. You either are either
window manager independent, or you aren't.  Of course this is not true at
all.  Thats why at the end of my email I advocated _at_least_ two levels
of GNOME compliance.

If it were not for us "encouraging" raster to make enlightenment work
with GNOME, I don't think we'd have a user friendly window manager that
worked cleanly with GNOME now. He did a great job - the enlightenment-conf
configuration tool makes extremely easy to configure enlightenment.
The window manager spec he helped finalize does work and we have a working
pager/task list to boot.  Of course it was a first try and we know it can
be improved upon now we have experience with it.

However, I do not think we should expect window manager authors to twist their
existing window managers into being completely built for GNOME. They
work on window managers for other reasons.  That is exactly why I've come
to realize we have to have a single reference window manager written
and engineered to be a "window manager" and no more.  We can make it
fit GNOME perfectly - this is one area I think KDE got right from the
beginning.  If this window manager is properly engineered then it will
be possible for other window managers, if they so choose, to work at
this higher level of compliance. 

So to recap - I say we shoudl work with KDE and get a wm spec so people can 
use legacy window managers with our glorious new desktop environments, 
and simultaneously we need a reference window manager that works in a much
more intimate fashion with GNOME. The later is going to be the one I bet
that new users are going to find the most intuitive to use.  We've tried
for 18 months to find the "holy grail" and we're no closer today than
when we started. We need to explore new options.

Dr Mike