We want to remove Element::ensureShadowRoot. Let's try to stop calling Element::ensureShadowRoot in Internals first, because changing Internals often causes build error.
Created attachment 125092 [details] Test For Building
Comment on attachment 125092 [details] Test For Building Attachment 125092 [details] did not pass gtk-ews (gtk): Output: http://queues.webkit.org/results/11396858
Created attachment 125230 [details] Patch
Comment on attachment 125230 [details] Patch Attachment 125230 [details] did not pass gtk-ews (gtk): Output: http://queues.webkit.org/results/11424020
Created attachment 125251 [details] Patch
Comment on attachment 125251 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=125251&action=review > Source/WebCore/ChangeLog:8 > + Instead of ensureShadowRoot, we check the existence of shadow root and create one if there is not. Could you explain why we need this change? > Source/WebCore/testing/Internals.cpp:199 > + return ShadowRoot::create(host, ec).get(); This is actually safe. But its safety isn't obvious at a glance. Could you change the return type to RefPtr, to make this clearly safe?
> > Source/WebCore/testing/Internals.cpp:199 > > + return ShadowRoot::create(host, ec).get(); > > This is actually safe. But its safety isn't obvious at a glance. > Could you change the return type to RefPtr, to make this clearly safe? Changing PassRefPtr breaks tests actually. So let me leave this as is.
Created attachment 125262 [details] Patch
Comment on attachment 125262 [details] Patch > Changing PassRefPtr breaks tests actually. > So let me leave this as is. Wow.
Comment on attachment 125262 [details] Patch Clearing flags on attachment: 125262 Committed r106635: <http://trac.webkit.org/changeset/106635>
All reviewed patches have been landed. Closing bug.