It seems r121192 caused svg text layout regressions. Do we need rebaseline? http://build.chromium.org/p/chromium.webkit/builders/Webkit%20Win/builds/15352 Regressions: Unexpected text diff mismatch : (3) svg/W3C-SVG-1.1-SE/text-intro-05-t.svg = TEXT svg/W3C-SVG-1.1/text-intro-05-t.svg = TEXT svg/text/text-intro-05-t.svg = TEXT Regressions: Unexpected image mismatch : (7) compositing/shadows/shadow-drawing.html = IMAGE fast/text/atsui-kerning-and-ligatures.html = IMAGE svg/W3C-SVG-1.1/animate-elem-52-t.svg = IMAGE svg/W3C-SVG-1.1/filters-conv-01-f.svg = IMAGE svg/W3C-SVG-1.1/masking-intro-01-f.svg = IMAGE svg/custom/js-late-gradient-and-object-creation.svg = IMAGE svg/zoom/page/zoom-foreignObject.svg = IMAGE Regressions: Unexpected image and text mismatch : (1) fast/css/text-rendering.html = IMAGE+TEXT
That change shouldn't have had any effect on SVG layouts at all. macpherson@ is preparing a new patch to fix the underlying issue that I worked around in r121192. CCing him on this bug just in case he knows of any interaction there.
Committed r121471: <http://trac.webkit.org/changeset/121471>
I have rebaselined the SVG content associated with this bug, but that still leaves three files that seem to have changed due to kerning.
compositing/shadows/shadow-drawing.html looks basically correct, something imperceptible about the shadow rendering changed. I'll rebaseline that. The other two tests definitely look wrong -- they should be kerning and aren't. It would be nice if someone could track down a regression range.