Created attachment 187169 [details] webgl-composite-modes-repaint-actual.png Check attached expected and actual images.
Created attachment 187170 [details] webgl-composite-modes-repaint-expected.png
It appears that in webgl-composite-modes-repaint test everything works as it implemented. This test uses fast/repaint/resources/repaint.js to invoke repaintTest after initial paint. repaint.js forces page re-layout and uses testRunner.display() before invoking testRunner.js . testRunner.display() paints page contents shadowed by gray surface, so this test produces darker blue tiles with canvas content composited on top. The repaintTest() method contains code which only changes canvas content, so only canvases get updated, but not tiles. From all this it looks that webgl-composite-modes-repaint-actual.png is correct test output. At the same time on chromium canvas update causes full page repaint and final page snapshot appears brighter.
chromium pixel test results don't show the repaint-counter "Grayness".
(In reply to comment #3) > chromium pixel test results don't show the repaint-counter "Grayness". Yes, that's what I'm trying to figure out now. It seems chromium does full screen update instead of canvas update. I need more digging.
(In reply to comment #4) > (In reply to comment #3) > > chromium pixel test results don't show the repaint-counter "Grayness". > > Yes, that's what I'm trying to figure out now. > It seems chromium does full screen update instead of canvas update. I need more digging. That's what happens in chromium in compositing mode.
(In reply to comment #5) > (In reply to comment #4) > > (In reply to comment #3) > > > chromium pixel test results don't show the repaint-counter "Grayness". > > > > Yes, that's what I'm trying to figure out now. > > It seems chromium does full screen update instead of canvas update. I need more digging. > > That's what happens in chromium in compositing mode. Is it by design or is it a known bug?
I think we should close this bug as INVALID, even though I agree that the difference between the "bright" Chromium version and the gray Mac/Qt/EFL version is confusing.
(In reply to comment #7) > I think we should close this bug as INVALID, even though I agree that the difference between the "bright" Chromium version and the gray Mac/Qt/EFL version is confusing. Should I add port specific snapshots for expected.png?
(In reply to comment #8) > (In reply to comment #7) > > Should I add port specific snapshots for expected.png? I would propose to add them. Even though expected.png has the "Grayness" it would be at least possible to track any regression with canvas content.
Created attachment 189989 [details] Patch
Comment on attachment 189989 [details] Patch The image for the expected test result does not seems to be correct. Can you double-check ?
Closing this bug because the EFL port has been removed from trunk. If you feel this bug applies to a different upstream WebKit port and was closed in error, please either update the title and reopen the bug, or leave a comment to request this.