Having two separate visible rectangles is confusing. We should compute one on higher level.
Created attachment 228873 [details] patch
Comment on attachment 228873 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=228873&action=review > Source/WebCore/platform/graphics/TiledBacking.h:53 > + virtual void setTiledScrollingIndicatorPosition(const FloatPoint&) = 0; Why doesn't this happen automatically when tiles are updated? Don't we have enough data in TileController to choose a good position?
Comment on attachment 228873 [details] patch View in context: https://bugs.webkit.org/attachment.cgi?id=228873&action=review > Source/WebKit2/WebProcess/WebPage/mac/TiledCoreAnimationDrawingArea.mm:454 > + for (const auto& layer : m_pageOverlayLayers.values()) > + layer->flushCompositingState(m_scrolledExposedRect); Do we want to flush these right away? Seems like we should just do this at flush time. In any case, I'm upending the overlay code in https://bugs.webkit.org/show_bug.cgi?id=131353 so I will sort it out.
(In reply to comment #2) > (From update of attachment 228873 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=228873&action=review > > > Source/WebCore/platform/graphics/TiledBacking.h:53 > > + virtual void setTiledScrollingIndicatorPosition(const FloatPoint&) = 0; > > Why doesn't this happen automatically when tiles are updated? Don't we have enough data in TileController to choose a good position? I really just punted figuring it out in this patch.
http://trac.webkit.org/changeset/166958