Brian Paul wrote:
>
> Thomas Tanner wrote:
>
> > I don't see why you're objecting to this idea.
> > I just want to implement support for dynamically loadable
> > drivers which requires replacing the old "kludges"
> > with another - in your opinion - "kludge".
>
> I see the point you're trying to make. I guess the problems, as I
> see them, are:
>
> 1. Linux (Unix) and Windows are, by far, the most popular platforms for
> Mesa/OpenGL. The GLX and WGL interfaces are very well established.
> Nobody will use a Mesa-specific interface instead of GLX/WGL since
> they'd be limiting themselves to Mesa versus many other OpenGL
> implementations.
Untrue. If the new generic interface can be implemented using a
dynamic library based system, it would not necessarily have to be
Mesa-specific. It could be implemented directly or as a wrapper layer
around the GLX/WGL/etc interfaces, whether those interfaces are
themselves implemented by Mesa or another OpenGL-type library/driver.
> 2. The X/Mesa is not intended to be directly used by application
> developers. It's an intermediate layer under Mesa's pseudo GLX.
> It helps to manage complexity in the system. Similarly, more people
> take advantage of 3D hardware via the GLX layer than the FX/Mesa
> interface.
If GLX were to be implemented as a wrapper layer around the generic
interface (reverse of the above), People could use GLX directly or by
wrapping it around another interface (generic or otherwise).
> 3. The OS/Mesa interface is somewhat popular. I can't hardly change
> or replace it now.
No need. Those who like to use it can wrap it around the generic
interface. That practice will die off PDQ if a genuinely good generic
interface is created. If this happens, the generic interface will take
over as the standard, if not it will join the legion of other interfaces
which no one uses anymore (DOSMesa, SVGAMesa, etc). Win/win.
> 4. The other interfaces (SVGA, DOS, GGI, etc) aren't used by too
> many people compared to GLX/WGL (judging from the absense of any
> email regarding them).
Perhaps this is because we currently do not have a generic interface,
so people have to code directly to GLX, etc and it takes too much time
to support them all. For example, right now I am having to fake the
3Dfx interface on top of the GGIMesa interface in order to use GGIMesa
with QuakeIII. GLX has been held up as the one true API which will free
us from this mess, but it does so at the cost of tying all Mesa use into
the X windows environment. That is unacceptable to those of us who want
to use Mesa in embedded environments where the rather large overhead of
X is a showstopper.
> If you can design a uniform layer on top of some of these interfaces
> you should consider how many people will actually make use of it
> before you expend a lot of time on it.
This generic interface idea is a godsend, IMO. I see no reason why
this generic interface should not immediately be written. I volunteer
to write (or help write) the generic-to-GGIMesa and generic-to-GLX
wrappers, in both directions. If the generic interface works out, I'll
see about trying to get some game developers to support it. They should
be _very_ happy to have a platform-independent way to use OpenGL without
having to use GLUT, and also not tying them to X windows the way GLX
does....
> In any case, you're welcome to design a new interface layer. We
> certainly can't stop you. It could be fun to do and you'd probably
> learn something in the process. If you write a detailed spec we'd
> be happy to critique it. :)
I really want to do this. Thomas and anyone else who is interested,
please e-mail me and let's get something going.
Jon
_______________________________________________
Mesa-dev maillist - [EMAIL PROTECTED]
http://lists.mesa3d.org/mailman/listinfo/mesa-dev
