Layout milestones aren't synchronized with UI-side compositing transactions
Created attachment 261550 [details] Patch
Comment on attachment 261550 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=261550&action=review > Source/WebKit2/Shared/mac/RemoteLayerTreeTransaction.h:263 > + WebCore::LayoutMilestones m_newlyReachedLayoutMilestones; Can this be initializer here… > Source/WebKit2/Shared/mac/RemoteLayerTreeTransaction.mm:493 > + , m_newlyReachedLayoutMilestones(0) …instead of here? > Source/WebKit2/WebProcess/WebPage/WebPage.cpp:5041 > + // Clients should not set userData for this message, and it won't be passed through. > + ASSERT(!userData); Doesn’t make sense to have an assertion about the behavior of client code. Does make sense to rev the client interface to remove the user data.
(In reply to comment #2) > Comment on attachment 261550 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=261550&action=review > > > Source/WebKit2/Shared/mac/RemoteLayerTreeTransaction.h:263 > > + WebCore::LayoutMilestones m_newlyReachedLayoutMilestones; > > Can this be initializer here… > > > Source/WebKit2/Shared/mac/RemoteLayerTreeTransaction.mm:493 > > + , m_newlyReachedLayoutMilestones(0) > > …instead of here? Definitely. > > Source/WebKit2/WebProcess/WebPage/WebPage.cpp:5041 > > + // Clients should not set userData for this message, and it won't be passed through. > > + ASSERT(!userData); > > Doesn’t make sense to have an assertion about the behavior of client code. Hmm, that was an Anders plan. But ok! > Does make sense to rev the client interface to remove the user data. OK.
Created attachment 261669 [details] Patch
Created attachment 261671 [details] Patch
Comment on attachment 261671 [details] Patch Clearing flags on attachment: 261671 Committed r190064: <http://trac.webkit.org/changeset/190064>
All reviewed patches have been landed. Closing bug.