Bug 97188
Summary: | [GTK] Layout Test fast/text/word-space-with-kerning.html is failing | ||
---|---|---|---|
Product: | WebKit | Reporter: | Zan Dobersek <zan> |
Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
Status: | RESOLVED FIXED | ||
Severity: | Normal | CC: | d-r, mrobinson |
Priority: | P2 | Keywords: | Gtk, LayoutTestFailure |
Version: | 528+ (Nightly build) | ||
Hardware: | Unspecified | ||
OS: | Unspecified |
Zan Dobersek
The following layout test is failing on GTK
fast/text/word-space-with-kerning.html
The test was introduced in r129068
http://trac.webkit.org/changeset/129068
Attachments | ||
---|---|---|
Add attachment proposed patch, testcase, etc. |
Martin Robinson
This test appears to be passing now, perhaps with the switch to Harfbuzz for complex text rendering. I'll unskip it.
Dominik Röttsches (drott)
Really? I thought this would be blocked by bug 102603, at least the test name suggests that.
Martin Robinson
(In reply to comment #2)
> Really? I thought this would be blocked by bug 102603, at least the test name suggests that.
Interesting. The bots were reporting this test as passing. Perhaps we are inadvertently passing?
Dominik Röttsches (drott)
(In reply to comment #3)
> Interesting. The bots were reporting this test as passing. Perhaps we are inadvertently passing?
So okay, I think it's correct that this passes now, but I am not sure it had something to do with switching to Harfbuzz. From the introduction to this test case (r129068):
" (WebCore::RenderText::computePreferredLogicalWidths): When kerning is enabled, words are
measured with their trailing space, if there is one, then the width of a space is subtracted.
Changed that width, stored in the local variable wordTrailingSpaceWidth, to include the
word spacing, since it is included in the result of widthFromCache()."
So this failed only when kerning was enabled. Since the -webkit-font-kerning: normal; in this test currently does not have any effect with the FreeType backend, the test passes.