Right now, FrameLoaderClient::didChangeScrollOffset() doesn't get called for web pages like http://instagram.com/p/UrCCIVBQX3/ because they have overflow on the body. We can easily address this by calling it at the end of RenderLayer::scrollTo(). <rdar://problem/13258596>
Created attachment 189884 [details] Patch
Comment on attachment 189884 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=189884&action=review > Source/WebCore/ChangeLog:10 > + FrameLoaderClient::didChangeScrollOffset() doesn't get called for web pages like I have removed the "like" in my local copy.
Comment on attachment 189884 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=189884&action=review > Source/WebCore/ChangeLog:6 > + RenderLayer::scrollTo() should call FrameLoaderClient::didChangeScrollOffset() > + https://bugs.webkit.org/show_bug.cgi?id=110673 > + -and corresponding- > + <rdar://problem/13258596> This is perfect. >> Source/WebCore/ChangeLog:10 >> + FrameLoaderClient::didChangeScrollOffset() doesn't get called for web pages like > > I have removed the "like" in my local copy. Like totally.
http://trac.webkit.org/changeset/143825
This patch might have caused <https://code.google.com/p/chromium/issues/detail?id=193555>. We're investigating.
The discussion continues on Bug 111278. The folks there confirm that this patch caused the regression.
This patch also lacks a regression test.
I rolled out this patch in http://trac.webkit.org/changeset/146185. Please feel free to re-land it once it doesn't cause a regression in the Chromium port.
Created attachment 193917 [details] Patch This patch should not break Chromium.
Thanks, Simon! http://trac.webkit.org/changeset/146260