On desktop GL, if they don't match, no error will be generated; but in GLES, "the same constraints and errors apply to the TexSubImage commands' argument format and the internalformat of the texture array being respecified as apply to the format and internalformat arguments of its TexImage counterparts".
Created attachment 80375 [details]
Created attachment 80376 [details]
Comment on attachment 80376 [details]
Looks good. Please add the test to the WebGL conformance suite.
This test is almost a duplication from one of the more/function/ test. Should we still add it?
(In reply to comment #4)
> This test is almost a duplication from one of the more/function/ test. Should we still add it?
It would be better from a maintenance perspective if we tried to keep the WebGL layout tests a subset of the WebGL conformance tests in the Khronos repository. If this test is mostly a copy of one of the more/function/ tests, what about moving it into the conformance suite proper upstream?
You mean delete the more/ test and add this one?
(In reply to comment #6)
> You mean delete the more/ test and add this one?
Committed r76874: <http://trac.webkit.org/changeset/76874>