Currently, we don't clear the drawing buffer after compositing while an FBO is bound to the context. This is normally ok, as the drawing buffer will be cleared when it's bound and something happens to it, but it will cause any texImage2D/drawImage/toDataURL from the canvas to see the old contents.
Created attachment 90302 [details] Patch
Comment on attachment 90302 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=90302&action=review Looks good. > Source/WebCore/html/canvas/WebGLRenderingContext.h:454 > + bool m_depthMask; Please use GC3Dboolean instead of bool.
Comment on attachment 90302 [details] Patch Looks good overall. Could you please submit a revised patch addressing Mo's feedback?
Created attachment 90632 [details] Patch
Comment on attachment 90632 [details] Patch Looks good.
The commit-queue encountered the following flaky tests while processing attachment 90632 [details]: http/tests/xmlhttprequest/encode-request-url-2.html bug 51765 (author: ap@webkit.org) The commit-queue is continuing to process your patch.
Comment on attachment 90632 [details] Patch Clearing flags on attachment: 90632 Committed r84602: <http://trac.webkit.org/changeset/84602>
All reviewed patches have been landed. Closing bug.