[webkit-reviews] review denied: [Bug 78404] [chromium] Replace RefPtr with OwnPtr for CCLayerImpl tree structure : [Attachment 126837] Patch

[bugzilla-daemon at webkit.org]

James Robinson <jamesr at chromium.org> has denied Tien-Ren Chen
<trchen at chromium.org>'s request for review:
Bug 78404: [chromium] Replace RefPtr with OwnPtr for CCLayerImpl tree structure
https://bugs.webkit.org/show_bug.cgi?id=78404

Attachment 126837: Patch
https://bugs.webkit.org/attachment.cgi?id=126837&action=review

------- Additional Comments from James Robinson <jamesr at chromium.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=126837&action=review


Unfortunately we can't change the main thread LayerList type away from RefPtr<>
at this time.  This may be a show-stopper, at the least I can't think of a
simple way to resolve this with the way we share templated code.  Perhaps there
is a way with more template indirection but at that point the whole thing
becomes less appealing.

> Source/WebCore/platform/graphics/chromium/TreeSynchronizer.h:49
> +    typedef HashMap<int, OwnPtr<CCLayerImpl> > CCLayerImplMap;
>  
> -    typedef HashMap<int, RefPtr<CCLayerImpl> > CCLayerImplMap;
> +    TreeSynchronizer(); // Not instantiable.

no reason to switch the order of these two lines

> Source/WebCore/platform/graphics/chromium/TreeSynchronizer.h:-50
> -    // Declared as static member functions so they can access functions on
LayerChromium as a friend class.

I found this comment helpful - did you not?

> Source/WebCore/platform/graphics/chromium/cc/CCLayerTreeHost.h:206
> +    typedef Vector<LayerChromium*> LayerList;

Woah - I don't think we can do this at all.  We need the LayerLists on the main
thread side (such as CCLayerTreeHost::m_updateList) to retain references to
their layers because calling into WebKit to paint will sometimes cause the
layer tree to get mutated and we can't risk a dangling pointer.