Stuart Anderson wrote:
>
> On Tue, 2 Nov 1999, Jon Leech wrote:
>
> > On Tue, Nov 02, 1999 at 08:03:11PM -0800, Bernd Kreimeier wrote:
> > > Something that I intended to go into earlier - what is the proposed
> > > ABI spec saying about Mesa's "fake" GLX?
> >
> > Nothing, yet. My first reaction was that an ABI group shouldn't say
> > anything about it so long as it behaved according to the GLX API
> > specification, but on reflection I think that an implementation
> > satisfying oglbase should be required to actually support indirect
> > rendering to a remote GLX-capable X server. Network transparency is
> > fundamental to X, and the opportunity to ensure that everyone using
> > Linux will be able to have 3D interoperability seems too good to pass
> > up.
>
> I had assumed that this was a requirment of oglbase. Breaking
> compatability or removing functionallity would have been very bad.
I don't think anyone was aluding to that. The question is whether
Mesa's "fake" GLX is acceptable to the oglbase.
In normal usage, Mesa on Linux does all its rendering with Xlib and
the GLX interface to the X server is simulated. This allows Mesa
to display OpenGL apps on any X server, even if it doesn't support
the GLX extension.
This has turned out to be a very powerful/enabling feature.
However, if a display does support GLX it would be much nicer if
Mesa generated real GLX protocol. The work we're doing at Precision
Insight is going in that direction. But even when Mesa speaks
GLX, it would be nice if the old/existing GLX-emulation still
worked. It would give the most flexibility.
I don't think oldbase would prohibit that.
-Brian
