Currently Canvas3D (WebGL) isn't supported in chromium when doing accelerated compositing.
Created attachment 59841 [details] Proposed patch This code requires: http://codereview.chromium.org/2864021/show to be landed in the chromium repository first.
Attachment 59841 [details] did not build on chromium: Build output: http://webkit-commit-queue.appspot.com/results/3297828
Comment on attachment 59841 [details] Proposed patch WebKit/chromium/public/WebGraphicsContext3D.h:116 + virtual unsigned int getPlatformTextureId() = 0; nit: "unsigned int" -> "unsigned" in webkit style WebKit/chromium/src/GraphicsContext3D.cpp:363 + if (!webContext->initialize(webAttributes, webView)) { instead of initializing a WebGraphicsContext3D with a WebView, we should instead ask the WebViewClient to create a WebGraphicsContext3D for us. things that you request from a WebKitClient should not be WebView or WebFrame dependent. looks fine otherwise.
Created attachment 59951 [details] Proposed patch. Fixed linux compile issue.
(In reply to comment #3) > (From update of attachment 59841 [details]) > WebKit/chromium/public/WebGraphicsContext3D.h:116 > + virtual unsigned int getPlatformTextureId() = 0; > nit: "unsigned int" -> "unsigned" in webkit style Done. > > WebKit/chromium/src/GraphicsContext3D.cpp:363 > + if (!webContext->initialize(webAttributes, webView)) { > instead of initializing a WebGraphicsContext3D with a WebView, we > should instead ask the WebViewClient to create a WebGraphicsContext3D > for us. things that you request from a WebKitClient should not be > WebView or WebFrame dependent. If it's ok, I'd like to make this change in a separate CL. I've filed a bug for it: https://bugs.webkit.org/show_bug.cgi?id=41315 > > looks fine otherwise.
Comment on attachment 59951 [details] Proposed patch. Fixed linux compile issue. Looks good. Very exciting.
Landed as http://trac.webkit.org/changeset/62130