[Piglit] [PATCH 5/9] glx: Convert tests to using epoxy for GLX function pointer management.

[Ian Romanick]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 02/11/2014 12:08 AM, Eric Anholt wrote:
> Ian Romanick <idr at freedesktop.org> writes:
> 
>> How does epoxy handle GetProcAddress returning NULL? 
>> tests/spec/glx_arb_create_context/indirect-es2-profile.c *really*
>> wants to know if it got NULL for a function pointer.  There are a
>> couple other tests that check for NULL to fail a test, but I'm
>> less concerened about those.
> 
> It does just call right down.  I had assumed that this was just
> somebody typing code out without thinking about how glx would
> always return a non-null pointer.  I guess GLX doesn't specify the
> non-NULL behavior we have, but it feels weird to be really careful
> about this one function when we aren't about any others.

If I remember correctly, I was specifically triggering a bug in
NVIDIA's libGL. :)  They allowed you to create an indirect ES2.0
context, but then returned NULL for any ES2.0-specific calls...
because there's no protocol for them... what the heck would it return
anyway?!?

> I could just drop the changes to the file if you want.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (GNU/Linux)

iEYEARECAAYFAlMD81cACgkQX1gOwKyEAw/9HQCeMRns3GunIr/sPFPSZQvfx+I5
nP0AoI5tDft7YBDRLgUAgGeHvjVJ2/8c
=VulW
-----END PGP SIGNATURE-----