the patch is forthcoming.
Created attachment 40669 [details] Proposed fix.
Comment on attachment 40669 [details] Proposed fix. It would be even better if we had some test that was failing because of this backward code. This was harmless. Why? r=me
the only place it was called from was destructor of ScriptCachedFramedata - so the RetainedPtr was destructed anyways and a window was released. The only difference was that gcController().garbageCollectSoon(); was not invoked, as WebKit usually tries to do after releasing a DOMWindow.
sorry, should be 'after releasing JSDOMWindow'.
landed: http://trac.webkit.org/changeset/49134