Xegl lives!

[Lars Knoll]

On Tuesday 24 May 2005 18:52, Jon Smirl wrote:
> On 5/24/05, Zack Rusin <zrusin at trolltech.com> wrote:
> > Jon, please don't be getting frustrated but I don't think that just
> > abondoning Xorg and moving over to Xegl is a viable alternative.
> > Trust me, personally, I'd consider working on Xegl much more fun, but I
> > don't think we can expect from the users that they will start switching
> > anytime in the near future. It will take a while. So lets do it
> > incrementally. There's just lots and lots of things to do.
>
> I have no expectations that users can switch now. But we have a very
> small pool of developers and splitting them between projects really
> slows things down. Xorg is working just fine, can't we just leave it
> be and work on getting Xgl ready?

Well, it's working just fine in a way that the implementation is correct. If 
you ask the toolkit people (mainly cairo and Qt), you'd get a different 
answer. For lots of things we would like to do Xorg is really slow.

I agree (and I think most other people here do as well), that Xgl is the way 
to go. But XAA based drivers will probably live longer than one would like 
them to, and it seems to me that at least Xorg 7.0 will still have XAA as the 
default driver architecture.

Adding one or two hooks to the existing drivers could give us a lot in the 
short term (i.e for Xorg 7.0) without too much work. KDE4/Qt4 and the next 
Gtk/Gnome versions will need this now in the next few months.

> There is no technical reason we couldn't have build Xgl several years
> ago, the problem is that an insufficient number of developers will
> help work on the project and bring it to completion. If everyone just
> came together and worked toward a switchover we could probably have
> something ready by the end of the year.

Zack is paid for working most of his time on X, so you do have another person 
that will help with the effort. If my time permits (which I don't know 
currently) I would like to help out here and there as well.

Cheers,
Lars

> For example, right now I am trying to glue Keith's input subsystem
> from kdrive into Xgl. I have never worked on input systems for X
> before and I have no clue as to what I am doing. I suspect a better
> approach might be to instead build Xgl as a dix into the xorg server.
> But since I am trying to get a working server for OLS I need to get an
> input system from somewhere. This is a prime example of where someone
> with the right experience set helping could make things go a lot
> smoother.