The canvas test 2d.composite.operation.clear fails because webkit supports a vendor specific composite op named "clear" that doesn't follow the conventions of the spec - should be named "webkit-clear" per the spec. However, for compatibility reasons, it doesn't currently make sense to drop the op (or drop by renaming it). Thus, I leave the test to remain on the skip list and make a second version of the test called 2d.composite.operation.clear2 that tests for a bogus op to make sure that we ignore unrecognized ops.
Created attachment 63019 [details] Patch
Comment on attachment 63019 [details] Patch Can we support "clear" and "webkit-clear"?
After understanding this better, that doesn't seem like that good an idea. (In reply to comment #2) > (From update of attachment 63019 [details]) > Can we support "clear" and "webkit-clear"?
Created attachment 63099 [details] Patch
Created attachment 64137 [details] Updated name of test Just updated the name of the new test to avoid any confusion as to the nature of the test.
FWIW, this LGTM.
Created attachment 65654 [details] Patch
Comment on attachment 65654 [details] Patch r=me
Comment on attachment 65654 [details] Patch Clearing flags on attachment: 65654 Committed r66294: <http://trac.webkit.org/changeset/66294>
All reviewed patches have been landed. Closing bug.