[Webkit-unassigned] [Bug 108960] [GTK] Remove subresource leaks from WebKit1 and WebKit2

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=108960





--- Comment #16 from Carlos Garcia Campos <cgarcia at igalia.com>  2013-02-11 08:59:57 PST ---
(In reply to comment #14)
> (In reply to comment #12)
> 
> > If the document loader is caching the subresources, the problem is still there, no?
> 
> I haven't confirmed this, but I suspect that the DocumentLoader has an idea of when a subresource is no longer used. In WebKit we are just saving all of them without any eviction policy.

In that case get_subresources might not work, if some subresources has been deleted by the document loader.

> > I'm not sure I understand this, we would create different WebKitWebResource objects, one to wrap the resource while loading to pass it to the signals and a different one to return in get_subresources. The former would be owned by webkit and released when the resource load finishes while the latter would be owned by the caller and not cached by webkit.
> 
> That's pretty close, but it seems like that would cause the subresources to continue to leak. Applications still aren't expecting to receive a transfer=full reference. What we could do is to deref any resources we create in a timeout though.

Well, I was thinking of deprecating the old api and adding a new async api to get the subresources that would transfer the ownership to the caller.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.