[chromium] NOT FOR REVIEW Drop root layer tiles on platforms that cache the front buffer
Created attachment 116220 [details] Patch
Created attachment 116222 [details] Patch
nduca: Can you PTAL?
Chromium side here: http://codereview.chromium.org/8637014/ Because of the extension mechanism, they can land independently.
Created attachment 116394 [details] Patch
Created attachment 116396 [details] Patch
Comment on attachment 116396 [details] Patch This looks good to me.
Comment on attachment 116396 [details] Patch Thanks enne. r=me
Comment on attachment 116396 [details] Patch cq+'ing on backer's offline request.
Comment on attachment 116396 [details] Patch Clearing flags on attachment: 116396 Committed r101104: <http://trac.webkit.org/changeset/101104>
All reviewed patches have been landed. Closing bug.
This patch stashes a bool discardAllTextures on the CCSettings struct, which is a misuse of the CCSettings struct, or i'm misunderstanding its purpose. Afaict, the discardAllTextures really is just a cached value of LayerRendererCapabilities::usingFrontBufferCached. CCSettings is a struct that exposes "compositor features" to embedders of the compositor API. This is not a public feature. LayerRendererCapabilities is available on both the impl thread an the main thread. From CCLayerTreeHost, you can efficiently obtain the capabilities with m_proxy->layerRendererCapabilities(). Also, the variable names here are massively inconsistent. I would love the capabilities bit to be contextHasCachedFrontBuffer or contextSupportsFrontBufferCached. Sorry I didn't catch this in the previous codereview. /me was out sick.