The enclosing block is not becoming the correct size because when first-line styles change that should cause a layout to happen, nothing actually happens.
Created attachment 55810 [details] Fix the issue
Comment on attachment 55810 [details] Fix the issue r=me
Fixed in r59211.
How come pixel results were always correct?
This caused a test failure on Qt. I assume the Qt results just need an update? http://build.webkit.org/results/Qt%20Linux%20Release/r59212%20(11625)/fast/css/pseudo-cache-stale-pretty-diff.html
(In reply to comment #6) > This caused a test failure on Qt. I assume the Qt results just need an update? > http://build.webkit.org/results/Qt%20Linux%20Release/r59212%20(11625)/fast/css/pseudo-cache-stale-pretty-diff.html Yes, it needs an update. I'll commit it immediately.
(In reply to comment #6) Qt specific expected file updated: http://trac.webkit.org/changeset/59214
The pixel test results were not previously correct, we just had the old (incorrect) behavior baked into the expectations. Currently this test fails pixel tests. It just needs a new baseline. I'll upload one momentarily.
Created attachment 55869 [details] Patch
D'oh, webkit-patch marked the old one as obsolete. Anyway, can I get a rubber-stamp on the new pixel baselines?
Comment on attachment 55869 [details] Patch LGTM. You can always pass --no-obsolete in that case. :) pseudo-cache-stale-expected.png doesn't look very good with teh form on top of the second sentence. Is that expected?
I think you're looking at the before. The next expected.png does not have the form control on top of the text (that's what Hyatt's patch fixed).
Ah yes, thanks.
Comment on attachment 55869 [details] Patch Landed http://trac.webkit.org/changeset/59251 (the bad ChangeLog changes fixed in the next rev).
http://trac.webkit.org/changeset/59252 might have broken GTK Linux 64-bit Release The following changes are on the blame list: http://trac.webkit.org/changeset/59252 http://trac.webkit.org/changeset/59253