On Tue, May 8, 2012 at 7:42 AM, Paul Berry <[email protected]> wrote: > This patch series contains an alternative proposal to the patch "Split > accuracy test to allow new multisample tests utilize this code" from > 5/4, along the lines Anuj and I discussed over email yesterday. Anuj, > let me know if you think this alternative is reasonable. I went ahead > and left you as the author of the patch, but I would be equally happy > taking over authorship if you'd prefer. > > Patch 2/2 contains the turn-on-off test, with the necessary > modifications to work with the new refactor. I have the impression > that this test is still a work in progress--is that right, Anuj? Test is complete only for color buffer testing.
>I got that impression because (a) the test ignores its <test_type> > command line argument (it always tests in "color" mode), (b) the > "small" variant of the test fails on my nVidia reference platform, and > (c) the patch doesn't add the test to all.tests. I kept the <test_type> option to accommodate depth and stencil testing in future. "small" variant is untested and irrelevant for this test case. So, can be removed. I missed adding it to all.tests. I'll make these changes along with few updated comments in turn-on-off.c once you push [PATCH v2 1/2]. OR I can push both the patches together along with suggested changes. > For what it's worth, I think it would be ok to remove the command-line > options for the turn-on-off test (hardcoding to num_samples=4, > test_type=color, small=false, and depthstencil=true) since varying > those options is unlikely to expose any bugs that wouldn't already be > caught by the accuracy test. I agree. command line options are not required at the moment. > [PATCH v2 1/2] Split accuracy test to allow new multisample tests utilize > this code > [PATCH v2 2/2] Add test to turn on/off MSAA in a FBO Thanks for accommodating the requested features while re-factoring. I am fine with both the patches. _______________________________________________ Piglit mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/piglit
