Make load events in iframe elements work with OOPIF (Blink side).

Make load events in iframe elements work with OOPIF (Blink side).

Today, when the parent frame defines a load event handler in the
iframe element of an out-of-process child frame, the event does not
fire in the parent process.  This CL adds the plumbing to forward the
event from the child frame's process to the parent frame's process.

Chromium side: https://codereview.chromium.org/954793002

BUG=453690
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=190783

[alexmos, 2015-02-24 20:02:24]

alexmos@chromium.org changed reviewers:
+ dcheng@chromium.org

[alexmos, 2015-02-24 20:02:25]

Daniel, could you please take a look?  This should fix one of the two major
blockers for running security layout tests with OOPIF.  You mentioned you may
want to tag something along to this -- let me know if that's still the case.

[dcheng, 2015-02-24 20:11:05]

https://codereview.chromium.org/937203003/diff/1/public/web/WebRemoteFrame.h
File public/web/WebRemoteFrame.h (right):

https://codereview.chromium.org/937203003/diff/1/public/web/WebRemoteFrame.h#...
public/web/WebRemoteFrame.h:38: virtual void DispatchFrameOwnerLoadEvent() const
= 0;
dispatchLoadEventForFrameOwner() might be a bit more consistent with how things
in the Blink API layer are named.

As an aside, it feels like it should be possible to integrate this somehow with
the didStart/didStopLoading stuff, but the current structure is fine.

[alexmos, 2015-02-24 20:56:23]

https://codereview.chromium.org/937203003/diff/1/public/web/WebRemoteFrame.h
File public/web/WebRemoteFrame.h (right):

https://codereview.chromium.org/937203003/diff/1/public/web/WebRemoteFrame.h#...
public/web/WebRemoteFrame.h:38: virtual void DispatchFrameOwnerLoadEvent() const
= 0;
On 2015/02/24 20:11:05, dcheng wrote:
> dispatchLoadEventForFrameOwner() might be a bit more consistent with how
things
> in the Blink API layer are named.
> 
> As an aside, it feels like it should be possible to integrate this somehow
with
> the didStart/didStopLoading stuff, but the current structure is fine.

Done.

I can double-check with Nasko about didStart/didStopLoading.  I noticed that
FrameOwner::dispatchLoad is also dispatched when the load is canceled because of
XFO or CSP, so it may not be exactly the same.

[dcheng, 2015-02-24 21:24:27]

On 2015/02/24 at 20:56:23, alexmos wrote:
> https://codereview.chromium.org/937203003/diff/1/public/web/WebRemoteFrame.h
> File public/web/WebRemoteFrame.h (right):
> 
>
https://codereview.chromium.org/937203003/diff/1/public/web/WebRemoteFrame.h#...
> public/web/WebRemoteFrame.h:38: virtual void DispatchFrameOwnerLoadEvent()
const = 0;
> On 2015/02/24 20:11:05, dcheng wrote:
> > dispatchLoadEventForFrameOwner() might be a bit more consistent with how
things
> > in the Blink API layer are named.
> > 
> > As an aside, it feels like it should be possible to integrate this somehow
with
> > the didStart/didStopLoading stuff, but the current structure is fine.
> 
> Done.
> 
> I can double-check with Nasko about didStart/didStopLoading.  I noticed that
FrameOwner::dispatchLoad is also dispatched when the load is canceled because of
XFO or CSP, so it may not be exactly the same.

Yeah, it's a little different. But I'm still cautiously optimistic that we can
make this a bit organized in the future. Maybe I'm being too optimistic =)

Regardless, LGTM

[alexmos, 2015-02-24 23:36:31]

The CQ bit was checked by alexmos@chromium.org

[commit-bot: I haz the power, 2015-02-24 23:36:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/937203003/20001

[commit-bot: I haz the power, 2015-02-24 23:52:09]

Committed patchset #2 (id:20001) as
https://src.chromium.org/viewvc/blink?view=rev&revision=190783