[Piglit] [PATCH] New testsuite for ARB_color_buffer_float (v2)

[Luca Barbieri]

Your test results and success of the MRT test should imply that nVidia
is not clamping *any* color buffer on FIXED_ONLY as long as at least
one is floating point (since that's what the test checks for right
now).

Since ATI doesn't seem to support such FBO configurations, both nVidia
and ATI would be compatible with such an implementation.

Not sure if it's what the OpenGL specification intended, but it's
significantly easier to implement (since we can resolve FIXED_ONLY
immediately, producing a single bit), so perhaps it makes sense to
stick to this, which is how my Mesa/Gallium implementation currently
behaves.

> Summary results from an NVIDIA GTX 260 are at:
> http://people.freedesktop.org/~idr/NVIDIA-ARB_cbf/.  It looks like
> clear, drawpixels, getteximage, mrt, and render all pass.  probepixel,
> queries, and readpixels all fail.

Does the render test output the following line:
Unclear specification on GL_ARB_fog_*

If that line is printed, it means that they don't switch shader
clamping like in fixed function mode, otherwise they do.

Unfortunately piglit seems to cut output after a fixed length, so I
can't tell from the summary you posted.

Thanks a lot for the feedback!