[Piglit] Testing for GL_ARB_draw_instanced

[Kenneth Graunke]

On 02/28/2012 01:32 PM, Eric Anholt wrote:
> Here's testing for almost everything else I came up with from the spec.
> The weaknesses I'd note are:
>
> * We don't test whether one of the required shader extensions or GL3
>    is actually present.
>
> * We don't test for the non-presence of variables if optional
>    extensions are not present.
>
> * We don't test very large instance counts.
>
> * We don't test interaction with non-GL3 (EXT_gpu_shader4 or
>    NV_vertex_program4).
>
> * We don't test other implied behavior of text like:
>
>     "The command
>
>          void DrawElementsInstancedARB(enum mode, sizei count, enum type,
>                  const void *indices, sizei primcount);
>
>      behaves identically to DrawElements except that<primcount>
>      instances of the set of elements are executed, and the value of
>      <instanceID>  advances for each iteration.  It has the same effect
>      as:
>
>          if (mode, count, or type is invalid )
>              generate appropriate error
>          else {
>              for (int i = 0; i<  primcount; i++) {
>                  instanceID = i;
>                  DrawElements(mode, count, type, indices, i);
>              }
>              instanceID = 0;
>          }"
>
>    We test that basic instancing-related rendering works, and we test
>    the explicitly mentioned error behavior for "type" in the spec, but
>    for example checking whether count or mode are invalid is not done.

This series is:
Reviewed-by: Kenneth Graunke <kenneth at whitecape.org>

I agree with your list of missing tests though.  It's basically what I 
came up with.