If a CSS rule that uses the :hover pseudo-class tries to change the value of the flow-from property, that rule seems to be effectively ignored. Removing the flow-from declaration from the rule lets all other property changes to apply. In the attached test-case, try moving the mouse over the green square. EXPECTED: the green square should turn red (border color changes) and the black square should turn green (content is not flowed anymore and background color - green - is revealed) ACTUAL: nothing happens If the -webkit-flow-from: none; declaration is dropped from the #region:hover rule, the green square will turn red (border color changes).
Created attachment 191758 [details] HTML file highlighting the problem
However, the rule is not actually disabled, since forcing the :hover state on the region (via the Web Inspector for instance), produces the expected results.
This is not a bug that affects just regions. Basically, changing any style in the :hover class that requires the element to be detached and reattached (such as changing it's display property) causes the :hover class to be ignored.
Linked issue: https://bugs.webkit.org/show_bug.cgi?id=7555
Created attachment 204014 [details] Patch The actual problem was fixed by https://bugs.webkit.org/show_bug.cgi?id=7555 This patch only adds tests for this particular problem (changing flow-from and flow-into inside :hover).
Comment on attachment 204014 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=204014&action=review > LayoutTests/fast/regions/hover-single-flow-from-none.html:47 > + testRunner.notifyDone(); is this really needed? If you remove the calls to waitUntilDone and notifyDone, will the test still work? > LayoutTests/fast/regions/hover-single-flow-from-other.html:54 > + testRunner.notifyDone(); the previous comment applies to this test too.
Comment on attachment 204014 [details] Patch Clearing flags on attachment: 204014 Committed r151311: <http://trac.webkit.org/changeset/151311>
All reviewed patches have been landed. Closing bug.