Because the shader has access to the whole image, repainting a filter chain with a FECustomFilter needs to paint the whole RenderLayer instead of just the dirty areas.
Created attachment 138643 [details] Patch V1
Comment on attachment 138643 [details] Patch V1 View in context: https://bugs.webkit.org/attachment.cgi?id=138643&action=review One small comment > LayoutTests/css3/filters/resources/color-add.fs:8 > + // Offset the color value with 0.3 on each color channel. Comment isn't quite correct (and you also only add 0.1 in the test :)
(In reply to comment #2) > (From update of attachment 138643 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=138643&action=review > > One small comment > > > LayoutTests/css3/filters/resources/color-add.fs:8 > > + // Offset the color value with 0.3 on each color channel. > > Comment isn't quite correct (and you also only add 0.1 in the test :) Good catch. Will fix. Thanks for the r+.
(In reply to comment #3) > (In reply to comment #2) > > (From update of attachment 138643 [details] [details]) > > View in context: https://bugs.webkit.org/attachment.cgi?id=138643&action=review > > > > One small comment > > > > > LayoutTests/css3/filters/resources/color-add.fs:8 > > > + // Offset the color value with 0.3 on each color channel. > > > > Comment isn't quite correct (and you also only add 0.1 in the test :) > > Good catch. Will fix. > > Thanks for the r+. I will also add the tests in chromium's expected results as requiring rebaselines. I want the cr-linux to finish to grab the generated ones on linux.
It seems like the ews takes too long. I've committed with only mac results for now. Landed in http://trac.webkit.org/changeset/115138.