Xgl/Xegl future?

[Michel Dänzer]

On Sun, 2005-08-21 at 22:32 -0400, Adam Jackson wrote:
> On Sunday 21 August 2005 22:02, Michel Dänzer wrote:
> > On Sun, 2005-08-21 at 20:53 -0400, Adam Jackson wrote:
> 
> > > The only challenge is deciding what channel to do that
> > > over, SAREA or DRI protocol or maybe GLX protocol + DRM interface.
> >
> > In which case this wouldn't be completely abstracted in the GL
> > implementation, and Xgl wouldn't be vendor neutral.
> 
> Like any interesting GLX extension, you'll have a frontend in the GLX engine 
> and a backend in the DDX driver (or DRI driver, someday).  My reference to 
> which channel to go through is just about how the client and server 
> communicate the device-specific info.

Okay, so let's see if I'm getting this right:

      * The GL implementation used by Xgl will have to provide an
        extension that allows getting information about where given
        renderbuffers are located in 'some form'.
      * Xgl will then forward this information in a (GL)X extension
        which provides this information about given drawables in 'some
        form'.
      * The client side GL implementation will use this (GL)X extension
        to find out where to do direct rendering to.

With a scheme like that, I guess 'some form' could actually be opaque to
Xgl and GL implementation specific at the same time, as long as the GL
implementations on the server and client side use the same 'some form',
which is pretty much given. That's the part that wasn't clear to me.

Assuming this makes sense and/or is what you were thinking of, thanks
for pointing me in the right direction. :)


-- 
Earthling Michel Dänzer      |     Debian (powerpc), X and DRI developer
Libre software enthusiast    |   http://svcs.affero.net/rm.php?r=daenzer