The canvas/philip/tests/2d.drawImage.animated.poster.html layout test is failing on GTK+ after r181553 (https://trac.webkit.org/changeset/181553). This revision implements APNG support. The diff is the following: --- /home/ch01/jhbuild/src/WebKit/WebKitBuild/Release/layout-test-results/canvas/philip/tests/2d.drawImage.animated.poster-expected.txt +++ /home/ch01/jhbuild/src/WebKit/WebKitBuild/Release/layout-test-results/canvas/philip/tests/2d.drawImage.animated.poster-actual.txt @@ -1,2 +1,2 @@ -Passed +Failed assertion: got pixel [255,0,0,255] at (50,25), expected [0,255,0,255] +/- 2
According to APNG specs, the first frame can be marked hidden. This test: Frame 0: green (hidden) Frame 1: red This test assumes canvas drawImage() is sophisticated enough be able to return hidden apng frame (green). Instead the image decoders in both Firefox and WebKit prefer not to keep the useless hidden frame in memory, so both return the first *visible* frame. You can see Firefox draws red frame too: http://w3c-test.org/2dcontext/drawing-images-to-the-canvas/2d.drawImage.animated.poster.html So I think we should simply mark this test accordingly.
There are no references to this bug in any TestExpectations. It's probable this bug was solved at some point but it wasn't marked as closed. I'm closing this bug now. If you think this bug report is still valid, please reopen it and add an entry to TestExpectations.
<rdar://problem/104678760>