In the current implementation of applyInlineStyleToNodeRange, we do not start a run from a node that contains child nodes. We also fail to remove redundant styled elements in a run.
Created attachment 67338 [details] fixes the bug
Comment on attachment 67338 [details] fixes the bug View in context: https://bugs.webkit.org/attachment.cgi?id=67338&action=prettypatch > WebCore/editing/ApplyStyleCommand.cpp:1170 > - if (mode == RemoveAttributesAndElements) > + if (mode == RemoveIfNeeded) > removeNodePreservingChildren(element); Should mode == RemoveAlways also cause us to removeNodePreservingChildren?
(In reply to comment #2) > (From update of attachment 67338 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=67338&action=prettypatch > > > WebCore/editing/ApplyStyleCommand.cpp:1170 > > - if (mode == RemoveAttributesAndElements) > > + if (mode == RemoveIfNeeded) > > removeNodePreservingChildren(element); > Should mode == RemoveAlways also cause us to removeNodePreservingChildren? That's a very valid point. Let me fix that.
Created attachment 67471 [details] fixed per Tony's comment
Committed r67449: <http://trac.webkit.org/changeset/67449>
http://trac.webkit.org/changeset/67449 might have broken Qt Linux Release minimal The following changes are on the blame list: http://trac.webkit.org/changeset/67449 http://trac.webkit.org/changeset/67450 http://trac.webkit.org/changeset/67451
http://trac.webkit.org/changeset/67459 rolled out this patch due the failure of /fast/events/event-input-contentEditable.html Will rebaseline this patch and commit again.
I can't tell who was to blame, I thought bug 45071 was what caused that test to regress?
Committed r67490: <http://trac.webkit.org/changeset/67490>