After the code change for bug 81737, the new markup is using shorthand notations as desired, however, it may generate a string like "border: initial solid red", which is an invalid declaration because the "initial" keyword is now combined with other component values in the same declaration. "border: initial solid red" should be "border: solid red"
Created attachment 134129 [details] Fixes the bug
This is causing a regression in editing because copy & pasted markup now has literal "initial" in some cases, and CSS parser can't handle that.
Comment on attachment 134129 [details] Fixes the bug Looks ok to me as a workaround. We really need to nail this problem better in the future. Happy to see that my proposed solution worked.
Just curious, how about StylePropertySet::get4Values(), can it generate a string contains "initial"
(In reply to comment #4) > Just curious, how about StylePropertySet::get4Values(), can it generate a string contains "initial" Possibly according to the code. Maybe we should fix it there as in theory getPropertyValue could return something buggy. Needs to be tested though.
(In reply to comment #5) > (In reply to comment #4) > > Just curious, how about StylePropertySet::get4Values(), can it generate a string contains "initial" > > Possibly according to the code. Maybe we should fix it there as in theory getPropertyValue could return something buggy. Needs to be tested though. But it will return only initial once.
Comment on attachment 134129 [details] Fixes the bug This probably should differentiate between implicit and explicit initial. r=me in any case as this fixes a regression.
Committed r112321: <http://trac.webkit.org/changeset/112321>