On Friday, March 24, 2017 6:05:36 PM PDT Roland Scheidegger wrote: > Am 24.03.2017 um 22:51 schrieb Kenneth Graunke: > > This patch makes glCopyImageSubData require mipmap completeness when the > > texture object's built-in sampler object has a mipmapping MinFilter. > > > > Fixes (on i965): > > dEQP-GLES31.functional.debug.negative_coverage.*.buffer.copy_image_sub_data > > > > Signed-off-by: Kenneth Graunke <kenn...@whitecape.org> > > --- > > src/mesa/main/copyimage.c | 25 +++++++++++++++++++++++-- > > 1 file changed, 23 insertions(+), 2 deletions(-) > > > > Resending the same patch as before. The Khronos OpenGL ARB and ES working > > group both decided that although this behavior is unfortunate and strange, > > Android's CTS requires it and so everybody else's implementations already > > work this way. > > Sounds like a rather lame justification IMHO - especially since it > apparently breaks other CTS tests.
I agree...someone made the comment of "if we start allowing this, and apps use it, then those apps won't run on Android devices", which, amusingly, is the flipside of the normal portability argument: "if we start disallowing this we might break existing applications". > But ah well if everybody prefers simple hacks in their driver to make it > conformant to a pointless error behavior, so be it. Yeah. *shrug*. There are a lot of failures on the road here: - non-immutable textures being a thing - copyimage allowing non-immutable textures - texture objects including a sampler object - default filtering for integer textures making them incomplete... I could go on, but...hindsight is 20/20, I suppose. > Reviewed-by: Roland Scheidegger <srol...@vmware.com> Thanks!
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
