This bug oversees the cleanup of tests that already pass but are skipped when testing as well as points to bugs for tests that need fixing to pass and offers minor patches that enable more tests.
Using revision 78552 and cairo trunk, test canvas/philip/tests/2d.path.isPointInPath.edge succeeds on my machine too. It apparently only fails on bots, so we should check if it still does and if so, check whether it's actually cairo that affects this test.
Created attachment 82449 [details] Unskip already passing tests This patch unskips about 25 tests that already pass.
Comment on attachment 82449 [details] Unskip already passing tests r=me
Comment on attachment 82449 [details] Unskip already passing tests Clearing flags on attachment: 82449 Committed r78570: <http://trac.webkit.org/changeset/78570>
All reviewed patches have been landed. Closing bug.
Reopening this bug, will leave commit bot out of it in the future.
This is a part of the Skipped file covering canvas tests that I've modified to clarify why these tests fail. Only tests that are failing because of Gtk/Cairo implementation are listed, the others are (probably) failing because of incomplete implementation of the canvas element in WebCore. # These tests are failing for us, but not for Mac. This likely # indicates platform specific problems. # When drawing canvas into itself, WebCore draws canvas's buffer into the same buffer. This # results in recursive drawing into cairo surface. We should either use groups when drawing # with cairo or copy the buffer every time the HTMLCanvasElement gets repainted. canvas/philip/tests/2d.drawImage.self.2.html # https://bugs.webkit.org/show_bug.cgi?id=54471 # http://cgit.freedesktop.org/cairo/commit/?id=4d4056872db94573183473610ad1d81d5439fdc6 canvas/philip/tests/2d.path.arc.selfintersect.1.html canvas/philip/tests/2d.path.arc.shape.3.html # Succeeds with Cairo trunk, should be unskipped when Cairo 1.11.3/1.10.3 is released # Cairo commit http://cgit.freedesktop.org/cairo/commit/?id=e46dc9e17a6f6737d5624dbcdd3c42c2353a72f1 # Cairo test http://cgit.freedesktop.org/cairo/commit/?id=7386960d063e4a3642fb281aaae071404d837bdb canvas/philip/tests/2d.path.arc.zeroradius.html # http://cgit.freedesktop.org/cairo/commit/?id=303f56a2d34a23d933a02fa625873211fb51eb3b canvas/philip/tests/2d.path.rect.selfintersect.html # Strangely set miter limit - should be 1.414, or there's a bug in maths! canvas/philip/tests/2d.path.rect.zero.6.html # Repeat-x/-y doesn't work - not supported in Cairo, workaround needed # https://bugs.webkit.org/show_bug.cgi?id=53085 canvas/philip/tests/2d.pattern.paint.repeatx.coord1.html canvas/philip/tests/2d.pattern.paint.repeatx.outside.html canvas/philip/tests/2d.pattern.paint.repeaty.coord1.html canvas/philip/tests/2d.pattern.paint.repeaty.outside.html # Quality should be in range [0, 100] # https://bugs.webkit.org/show_bug.cgi?id=55878 canvas/philip/tests/toDataURL.jpeg.quality.basic.html
(In reply to comment #7) > # When drawing canvas into itself, WebCore draws canvas's buffer into the same buffer. This > # results in recursive drawing into cairo surface. We should either use groups when drawing > # with cairo or copy the buffer every time the HTMLCanvasElement gets repainted. > canvas/philip/tests/2d.drawImage.self.2.html You have my approval to put all of this information directly into the Skipped file! Thanks for looking into these.
Created attachment 103226 [details] This patch updates the list of skipped canvas tests along with explanation about Cairo-specific failures
Comment on attachment 103226 [details] This patch updates the list of skipped canvas tests along with explanation about Cairo-specific failures Clearing flags on attachment: 103226 Committed r92605: <http://trac.webkit.org/changeset/92605>