Bernd Kreimeier wrote:
>
> Ralph Giles writes:
> > I'm not sure we can come up with something sufficiently broad to make
> > everyone happy though,
>
> We can try. Where's the daring spirit that brought us
> context-independend entry points?
Ugh, I don't think if any of us want to go through something like
that again. :)
> Seriously: it's worth it, and easily abandoned if we can't even
> agree on a list of mandatory tokens and their order.
As I said on the UtahGLX list, I don't want to over-enginneer this.
We just need something a little better than the hodge-podge seen
so far. We've got deadlines coming up and don't have much time left
for this.
I guess I still don't see the problem with YYYYMMDD versions.
My thinking was this number would be updated when either:
a) a significant bug is fixed or feature is added, or
b) the sources are labeled and an official release is made.
Whether the driver is build from a tarball or CVS shouldn't matter that
much. If you think it does matter, then you might as well also worry
about what OS/Linux distro it was built on, who compiled it, with
which compiler flags and on what system it was compiled (ala the startup
messages in emacs).
Also remember that there are GL_VENDOR and GL_VERSION strings to
help identify implementations.
If you want to get really picky, you could imbed in your driver binary
a very detailed string prefixed with a well-known token then extract it
with the Unix 'strings' program.
-Brian
