[Piglit] [PATCH 1/2] glslparsertest: Allow ! in extension list to require non-support.

[Kenneth Graunke]

On 03/12/2012 12:39 PM, Eric Anholt wrote:
> On Mon, 12 Mar 2012 00:10:16 -0700, Kenneth Graunke<kenneth at whitecape.org>  wrote:
>> For example:
>> // [config]
>> // required_extensions: !GL_ARB_draw_instanced
>> // [end config]
>> will cause the test to run only if the GL_ARB_draw_instanced is _not_
>> advertised.
>
> Seems like a reasonable feature.
>
> Reviewed-by: Eric Anholt<eric at anholt.net>

Cool, thanks!

> For the feature macro testing, I'm thinking we should probably have a C
> test that encodes what we understand about how the feature macros work
> and how to test them using command-line args, instead of writing tests
> per-extension (or, in the current case, just for the one extension we
> bothered with).
>
> I haven't reviewed the second patch, because while it looks sane to me,
> I seem to recall that idr's interpretation was that the macros didn't
> exist unless you #extensioned.  But I may be totally wrong.

idr was mistaken.  The macros always exist...and in fact, that's 
precisely what makes them useful.  That lets you do things like:

#ifdef GL_ARB_draw_instanced
#define instanceID gl_InstanceIDARB
#else
uniform int instanceID;
#endif

and then have one set of shaders which work regardless of whether the 
extension is present.