On 01/22/2013 02:29 PM, Eric Anholt wrote:
Ian Romanick <[email protected]> writes:
From: Ian Romanick <[email protected]>
OpenGL ES 3.0 (and presumably other versions of OpenGL) have two
requirements that these tests expect to be violated:
1. Members of a uniform block that are not used will not be marked
active.
2. It is an error to apply a row_major or column_major layout qualifier
to a non-matirx type.
These changes fix the UBO tests to actually use all of the members of
each block and not use row_major and column_major layout qualifiers on
non-matrix types.
Without these fixes, a patch series that is going out to the mesa-dev
list very soon that fixes many gles3conform failures will regress these
tests.
This commit is two commits plus some other garbage all smashed together.
Yes, yes it is. It was almost 3am when I sent it. git commit -a for
the lose. I also missed a case. I'll re-send.
I thought what I did was a pretty obvious interpretation of the
ARB_uniform_buffer_object spec, but there's room for both of these
changes I guess. I don't know of any UBO apps to test with to see if
anybody regresses from throwing a new error on inappropriate
row_major/column_major, so we're probably safe.
The conformance tests are forcing us to do this, so we don't have a lot
of choice.
_______________________________________________
Piglit mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/piglit
