No test was added in bug 76498 (for the reasons given there) but it allowed color correction for WebGL image textures in more cases (JPEG, opaque PNG images). The fix for bug 78239 regressed that because the alpha state of the frame was set _after_ calling ImageFrame::setStatus() to mark the image frame complete.
Created attachment 134223 [details] Patch
Comment on attachment 134223 [details] Patch Code looks fine. Is this covered by existing tests, either layout tests or WebGL conformance tests? If not, is it possible to add one? r=me
No existing tests that I know of. My ChangeLog mentioned "No new tests. The application of the color profile is not guaranteed for images used as textures in WebGL. Have things improved since your wrote that on bug 76498?
(In reply to comment #3) > No existing tests that I know of. My ChangeLog mentioned "No new tests. The application of the color profile is not guaranteed for images used as textures in WebGL. Have things improved since your wrote that on bug 76498? Note that in bug 75999 (as you reminded me over chat), a test case was added which verifies that application of a color profile doesn't completely destroy the contents of a texture. This test couldn't be added to the WebGL conformance suite, because some browsers might not apply the color profile -- but it was fine to add to the WebKit layout tests, and provided some value in the form of a regression test. The tolerance was increased to the point where the test passed even on platforms that didn't apply the color profile. I wonder whether something similar could be done here. But regardless, you have an r+, so you can determine whether you want to just commit the fix for the known bug now or try to add a test case for it.
I'd rather not add a test that is flaky/sketchy/unreliable.
Comment on attachment 134223 [details] Patch Clearing flags on attachment: 134223 Committed r112493: <http://trac.webkit.org/changeset/112493>
All reviewed patches have been landed. Closing bug.