There is no reason we should ever have to call Document::updateStyleForAllDocuments(), yet JSCallbackData::invokeCallback() and JSCustomXPathNSResolver::lookupNamespaceURI() both do so.
I tried this. The only new test failure is editing/selection/leak-document-with-selection-inside.html which makes bogus assumptions about how long documents live.
(In reply to comment #1) > I tried this. The only new test failure is editing/selection/leak-document-with-selection-inside.html which makes bogus assumptions about how long documents live. Actually it's just flakey. It fails without any change.
Created attachment 207687 [details] Patch
This patch removes the callers. If it passes EWS I'll remove the function too.
Comment on attachment 207687 [details] Patch r=me
See also: bug 46761. Good to hear that tests are no longer failing, although EWS is in such shape that it probably won’t provide a trustworthy answer. Looks like we still have a caller in Element::setActive()?
See also discussion in bug 32580.
http://trac.webkit.org/changeset/42384 is related.
https://trac.webkit.org/r153872