This is also reported in crbug: http://code.google.com/p/chromium/issues/detail?id=74636 After a discussion with Adam, it seems like using ActiveDOMObject for RequestAnimationFrameCallback is the proper way, however, delete callbacks on Document::detach also fixes this memory holding. Let's get the quick/simple fix in now and merge back to chrome M10 and M11, then we can do the proper ActiveDOMObject thing on the top of the tree.
Created attachment 86107 [details] Patch
Comment on attachment 86107 [details] Patch Can you add a FIXME comment to both these locations saying that we should consider using ActiveDOMObject in the future?
Will do upon landing.
Committed r81407: <http://trac.webkit.org/changeset/81407>
As a nit, I personally like .clear() instead of = 0 to null out an OwnPtr<>'s reference. As for ActiveDOMObject you are all welcome to debate ap@ on that point but I'm not inclined to jump on that grenade again.
(In reply to comment #5) > As a nit, I personally like .clear() instead of = 0 to null out an OwnPtr<>'s reference. > > As for ActiveDOMObject you are all welcome to debate ap@ on that point but I'm not inclined to jump on that grenade again. Feel free to change it, and thanks for the warning of the grenade. :)