The way I propose in the following to reproduce this is depend on the patch in #66283. Run "Tools/Scripts/run-webkit-tests -p --tolerance 2 -2 LayoutTests/animations/missing-values-first-keyframe.html" Result: expectation failure with a diff image that looks like full black. Expected result: pass
Created attachment 104035 [details] Patch
Comment on attachment 104035 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=104035&action=review > Tools/DumpRenderTree/qt/ImageDiff.cpp:128 > + if (!difference) > + fprintf(stdout, "diff: %01.2f%% passed\n", 0.0); > + else { Good catch, thanks for the fix. - I think (difference==0) would be more talkative. - Modifying difference to 0.0 is unnecessary. All ports use this copy/paste code. :) r=me with these changes.
(In reply to comment #2) > (From update of attachment 104035 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=104035&action=review > > > Tools/DumpRenderTree/qt/ImageDiff.cpp:128 > > + if (!difference) > > + fprintf(stdout, "diff: %01.2f%% passed\n", 0.0); > > + else { > > Good catch, thanks for the fix. > > - I think (difference==0) would be more talkative. Personally I agree but '== 0' is not WebKit style. > - Modifying difference to 0.0 is unnecessary. All ports use this copy/paste code. :) But than how could I test it in the if? > > r=me with these changes.
(In reply to comment #3) >> - I think (difference==0) would be more talkative. > Personally I agree but '== 0' is not WebKit style. Ooops, you're right. Negating a double value is strange for me, but it works. :) So use !difference to make style bot happy. ;) > But than how could I test it in the if? I meant the printf line.
Committed r93509: <http://trac.webkit.org/changeset/93509>