On 10/05/2012 03:27 AM, Neil Roberts wrote:
Eric Anholt <[email protected]> writes:
If this is testing GL_EXT_unpack_subimage, it should go in
tests/spec/gl_ext_unpack_subimage/
It's not really testing GL_EXT_unpack_subimage but is testing a specific
bug in the Intel driver when uploading a subregion of a BGR image. There
is already a test for GL_EXT_unpack_subimage and this test case borrows
To add a bit more background, this is OpenGL 1.0 functionality that was
removed in GLES. It was added back by GL_EXT_unpack_subimage. The
generic texturing tests currently live in tests/texturing. Is that
maybe a better place? I agree that tests/bugs was a mistake.
some code from it. Maybe it would make more sense to convert this test
to use regular GL. Then it wouldn't need a shader and could be quite a
bit simpler. However, the bug was found via Wayland which is using GLES2
so it seemed like a good way to replicate the bug.
One of our dreams and goals is for tests like this to be able to run in
both desktop GL and GLES[123] mode. The Waffle work is getting us
closer to that goal.
+static GLboolean pass;
I'm assuming Brian's review covered most everything, but as a note:
We're trying to move to reasonable C types like bool for everything that
isn't specified by GL API.
I think Rob is adapting the test case for Brian's changes so hopefully
he can incorporate this suggestion too.
Regards,
- Neil
_______________________________________________
Piglit mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/piglit
_______________________________________________
Piglit mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/piglit
