[chromium] Fix ownership semantics for LayerChromium/CCLayerImpl
Created attachment 87787 [details] Patch
Comment on attachment 87787 [details] Patch You knew I was gonna ask - test?
If I knew how to reproduce this bug reliably, I would happily check in a test. I just don't want to check in a test with setTimeout('window.location.reload()', 100) in it. That would go directly into the FLAKY bin. Do you have any suggestions on how to test this better? However, I did run it through the LayoutTests/compositing suite (which caught a correctness issue that I fixed).
Committed r82624: <http://trac.webkit.org/changeset/82624>