As reported in http://crbug.com/145082. See attached file for repro.
Created attachment 161961 [details] repro
Created attachment 161963 [details] Patch
Comment on attachment 161963 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=161963&action=review > LayoutTests/css3/calc/zoom-with-em.html:19 > + shouldEvaluateTo(calc.offsetWidth + "", nocalc.offsetWidth + ""); Can I do something better here? This feels a wee bit dodgy :(
Comment on attachment 161963 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=161963&action=review >> LayoutTests/css3/calc/zoom-with-em.html:19 >> + shouldEvaluateTo(calc.offsetWidth + "", nocalc.offsetWidth + ""); > > Can I do something better here? This feels a wee bit dodgy :( You could toString or String() them. But, meh, doesn't really matter.
Comment on attachment 161963 [details] Patch Clearing flags on attachment: 161963 Committed r127557: <http://trac.webkit.org/changeset/127557>
All reviewed patches have been landed. Closing bug.
(In reply to comment #5) > (From update of attachment 161963 [details]) > Clearing flags on attachment: 161963 > > Committed r127557: <http://trac.webkit.org/changeset/127557> The test added in this revision, css3/calc/zoom-with-em.html, has been failing on Mountain Lion: <http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK1%20(Tests)/r127558%20(589)/css3/calc/zoom-with-em-pretty-diff.html>.
(In reply to comment #7) > The test added in this revision, css3/calc/zoom-with-em.html, has been failing on Mountain Lion: <http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK1%20(Tests)/r127558%20(589)/css3/calc/zoom-with-em-pretty-diff.html>. It is also failing on Lion.
(In reply to comment #7) > (In reply to comment #5) > > (From update of attachment 161963 [details] [details]) > > Clearing flags on attachment: 161963 > > > > Committed r127557: <http://trac.webkit.org/changeset/127557> > > The test added in this revision, css3/calc/zoom-with-em.html, has been failing on Mountain Lion: <http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK1%20(Tests)/r127558%20(589)/css3/calc/zoom-with-em-pretty-diff.html>. Landed Mac-specific expected results in <http://trac.webkit.org/r127560>.
(In reply to comment #7) > The test added in this revision, css3/calc/zoom-with-em.html, has been failing on Mountain Lion: <http://build.webkit.org/results/Apple%20MountainLion%20Release%20WK1%20(Tests)/r127558%20(589)/css3/calc/zoom-with-em-pretty-diff.html>. It is also failing on Qt: <http://build.webkit.org/results/Qt%20Linux%20Release/r127558%20(51650)/css3/calc/zoom-with-em-pretty-diff.html>. The generic expected results appear to assume subpixel layout, meaning every port that doesn’t use subpixel layout needs platform-specific expected results checked in. I don’t know whether subpixel layout is the rule or the exception among ports. If it’s an exceptional case, then perhaps the generic results should be replaced and subpixel-layout-based results should be placed in the appropriate ports’ platform directories.
Oops. Thanks for checking in the rebaselines Dan! Ideally I'll rewrite the test so it no longer reports the actual pixel values - I only care that they are equal, not what they are.
(In reply to comment #11) > Oops. Thanks for checking in the rebaselines Dan! > > Ideally I'll rewrite the test so it no longer reports the actual pixel values - I only care that they are equal, not what they are. r=me to do that. Or to change the test to use a few zoom factors that yield integer pixel amounts.
I updated the expectations in http://trac.webkit.org/changeset/127563 - moving the results you kindly landed for Mac into the generic location, and moved the results I generated with the chromium port into the chromium directory. Hopefully this should fix the bots, and I'll rewrite the test in a bit.
Test rewritten at https://bugs.webkit.org/show_bug.cgi?id=95922