Re: does anyone use gtkglextmm successfully?

[Daniel Elstner <daniel kitta googlemail com>]

Am Montag, den 18.05.2009, 09:17 +0800 schrieb 明覺:
> On Sun, May 17, 2009 at 9:43 PM, Daniel Elstner
> <daniel kitta googlemail com> wrote:
> > Am Sonntag, den 17.05.2009, 09:20 -0400 schrieb Mitchell Laks:
> >> shi minjue gmail com wrote
> >> >
> >> > no, i install the non-free nvidia driver manually, not even by a
> >> > debian package, the driver contains opengl implementation, and mesa is
> >> > the opengl implementation of debian, so I think they are conflict.
> >>
> >> Unfortunately, I think you are wrong.
> >>
> >> Not because I understand it :)
> >>
> >> Perhaps a bigger debian guru could explain how it works.
> >
> > It's using a so-called diversion to replace the Mesa GL libraries with
> > its own.
> >
> > However, personally I'd rather use the Debian packages.  At least in
> > Ubuntu, the Nvidia drivers are fairly recent.  Also, the driver is
> > unlikely to be the cause of the problem anyway.
> 
> If the driver is a different thing with opengl library, then I would
> also think it's not the driver caused the problem, but the opengl
> library, that's mesa. :)

No, you were right about the Nvidia driver shipping its own OpenGL
implementation.  It's complicated by the fact that the normal Debian
packages of the Nvidia driver do not replace the Mesa header files with
the Nvidia ones, but nonetheless the Nvidia binary driver contains a
complete OpenGL implementation (and GLX for that matter).

Still, I don't think it's a good idea to replace the Debian packages of
the Nvidia driver unless absolutely necessary.

By the way, if you are looking for an example of a C++ project that is
using gtkmm and gtkglext (but not gtkglextmm), you may want to have a
look at Somato:  http://github.com/danielkitta/somato/tree/master

Good luck,
--Daniel