This logic has no render tree dependency. Also by moving the logic to RenderTreeUpdate, we can remove RenderNamedFlowThread::m_flowThreadChildList -> more secure.
Created attachment 294129 [details] Patch
Comment on attachment 294129 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=294129&action=review > Source/WebCore/ChangeLog:8 > + Reviewed by NOBODY (OOPS!). > + > + No change in functionality. Moar words here pls.
Comment on attachment 294129 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=294129&action=review > Source/WebCore/style/RenderTreeUpdater.cpp:335 > renderTreePosition().computeNextSibling(element); > - > - RenderTreePosition insertionPosition = parentFlowRenderer > - ? RenderTreePosition(*parentFlowRenderer, parentFlowRenderer->nextRendererForElement(element)) > - : renderTreePosition(); > - > + RenderTreePosition insertionPosition = !parentFlowRenderer ? renderTreePosition() : RenderTreePosition(*parentFlowRenderer, > + renderTreePosition().nextSiblingRenderer(element, m_styleUpdate->elementStyle(element))); Wouldn't just using 'style' here work instead of 'm_styleUpdate->elementStyle(element)'? Could we just provide style or Style::Update& to computeNextSibling() above and eliminate the parentFlowRenderer special case completely?
Created attachment 294217 [details] Patch
Created attachment 294252 [details] Patch
Comment on attachment 294252 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=294252&action=review > Source/WebCore/style/RenderTreeUpdater.cpp:324 > + bool willMoveToFlowThread = element.shouldMoveToFlowThread(style); > + > + if (willMoveToFlowThread) Doesn't really need boolean here.
Created attachment 294255 [details] Patch
Created attachment 294256 [details] Patch
Comment on attachment 294256 [details] Patch Clearing flags on attachment: 294256 Committed r208470: <http://trac.webkit.org/changeset/208470>
All reviewed patches have been landed. Closing bug.