Move FrameDestructionObserver into LocalFrame.

Move FrameDestructionObserver into LocalFrame.

Prepratory cleanup for making recursive frame detach work for all
frames. FrameDestructionObserver can only observe a LocalFrame.
Similarly willDetachFrameHost() and detachFromFrameHost() can only be
called on LocalFrames, since these methods are only called by
FrameLoader.

BUG=none

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=179178

[dcheng, 2014-07-29 06:51:56]

Ken: you don't have any plans for using these callbacks do you? The way I see
it, if we need to specialize the behavior for RemoteFrame, we can just use the
Frame::detach virtual I'll be adding in a followup CL.

[kenrb, 2014-07-29 19:07:52]

These mostly look like they can live on LocalFrame without any problem, but why
detachFrameHost()? It is only called by FrameLoader, but it happens as a part of
removing the frame from the frame tree. Would there not be an analogous call
when removing a RemoteFrame from the tree?

[dcheng, 2014-07-29 19:09:27]

On 2014/07/29 at 19:07:52, kenrb wrote:
> These mostly look like they can live on LocalFrame without any problem, but
why detachFrameHost()? It is only called by FrameLoader, but it happens as a
part of removing the frame from the frame tree. Would there not be an analogous
call when removing a RemoteFrame from the tree?

Since I'm exposing a virtual detach method on Frame anyway, I'd likely just
override detach for RemoteFrame to handle that.

[kenrb, 2014-07-29 19:20:15]

On 2014/07/29 19:09:27, dcheng (OOO) wrote:
> Since I'm exposing a virtual detach method on Frame anyway, I'd likely just
> override detach for RemoteFrame to handle that.

lgtm, although I would prefer to keep RemoteFrame and LocalFrame as similar as
possible, for functionality that is common to both such as FrameTree and
FrameHost interaction.

[dcheng, 2014-07-29 19:28:55]

On 2014/07/29 at 19:20:15, kenrb wrote:
> On 2014/07/29 19:09:27, dcheng (OOO) wrote:
> > Since I'm exposing a virtual detach method on Frame anyway, I'd likely just
> > override detach for RemoteFrame to handle that.
> 
> lgtm, although I would prefer to keep RemoteFrame and LocalFrame as similar as
possible, for functionality that is common to both such as FrameTree and
FrameHost interaction.

Problem is, I just don't think it's worth exposing a virtual for that. We only
get a gain out of that if we can end up calling the virtual from a code path
common to both local and remote frames, but that's actually pretty hard to
do--detachFromFrameHost() is only called from FrameLoader. We could try to
refactor the call out of FrameLoader, but I don't think it's worth it--it's
intertwined with so much other code.

[dcheng, 2014-07-29 21:34:38]

The CQ bit was checked by dcheng@chromium.org

[commit-bot: I haz the power, 2014-07-29 21:35:29]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/dcheng@chromium.org/427673008/1

[commit-bot: I haz the power, 2014-07-29 22:06:52]

Change committed as 179178