Part of a two-sided patch: https://chromiumcodereview.appspot.com/10496016 defines WEBPImportPictureRGBX|BGRX for importing picture data ignoring the alpha channel. WebKit should use these methods to import data for encoding when creating a image/webp dataURL. The methods WebKit currently uses, namely WEBPImportPictureRGBA|BGRA, will include the alpha channel during import in a future release of libwebp so we should stop using these methods to future-proof the code.
Created attachment 145546 [details] Patch
Waiting on DEPS http://src.chromium.org/viewvc/chrome?view=rev&revision=140264 to roll into webkit.
lgtm, but one need to wait for the new functions to be in the official libwebp tree (release 0.1.4?) before using them... btw: there's a missing checksum reported for the second PNG
(In reply to comment #3) > lgtm, but one need to wait for the new functions to be in the official libwebp tree (release 0.1.4?) before using them... Actually, I just need chromium DEPS to rev 140264 to use the functions since they are in the chromium repository. > btw: there's a missing checksum reported for the second PNG I generated the png result with run-webkit-tests, that should create a checksum, and the error report says "create the png with run-webkit-tests" :) Dunno what's going on there.
http://trac.webkit.org/changeset/119468 chromium DEPS rolled.
Created attachment 145918 [details] Patch
Since the linux EWS are backed up, built chromium-linux and ran the new test. Works fines.
Comment on attachment 145918 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=145918&action=review > LayoutTests/ChangeLog:13 > + * fast/canvas/canvas-toDataURL-webp-no-alpha-expected.txt: Added. > + * fast/canvas/canvas-toDataURL-webp-no-alpha.html: Added. Can you make this a reftest?
Maybe, prefer to see the composite onto black pixels here.
Discussed the existing test fast/canvas/canvas-toDataURL-webp.html with tkent, and we agree to replace it with the new test.
Created attachment 146291 [details] Patch
Comment on attachment 146291 [details] Patch ok
Comment on attachment 146291 [details] Patch Rejecting attachment 146291 [details] from commit-queue. Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 1 ERROR: /mnt/git/webkit-commit-queue/LayoutTests/ChangeLog neither lists a valid reviewer nor contains the string "Unreviewed" or "Rubber stamp" (case insensitive). Full output: http://queues.webkit.org/results/12908507
Created attachment 146437 [details] Patch for landing
Comment on attachment 146437 [details] Patch for landing Clearing flags on attachment: 146437 Committed r119787: <http://trac.webkit.org/changeset/119787>
All reviewed patches have been landed. Closing bug.
(In reply to comment #4) > > > btw: there's a missing checksum reported for the second PNG > > I generated the png result with run-webkit-tests, that should create a checksum, and the error report says "create the png with run-webkit-tests" :) > > Dunno what's going on there. Bug 88368.
Added rebaselines on http://trac.webkit.org/changeset/119801
FYI, as per garden-o-matic, the MISSING line in TestExpectations needs to include Lion and you need a line for the equivalent virtual gpu test platform/chromium/virtual/gpu/fast/canvas/canvas-toDataURL-webp.html. http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=fast%2Fcanvas%2Fcanvas-toDataURL-webp.html
http://trac.webkit.org/changeset/119804 for the virtual GPU.
(In reply to comment #19) > FYI, as per garden-o-matic, the MISSING line in TestExpectations needs to include Lion Want lion: so optimize http://trac.webkit.org/changeset/119812
(In reply to comment #19) > http://test-results.appspot.com/dashboards/flakiness_dashboard.html#tests=fast%2Fcanvas%2Fcanvas-toDataURL-webp.html All look good now, thanks for fyi.