Gracefully support Navigator Gamepad methods in a detached state.

Gracefully support Navigator Gamepad methods in a detached state.

If the frame and window of a Navigator has been detached (closed), the
Navigator Gamepad methods must still continue to be available. Arrange
for that to happen by allowing the Gamepad Navigator supplement to
be creatable without a frame.

R=
BUG=355783
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=172920

[sof, 2014-04-27 18:18:21]

Please take a look.

(A set of clusterfuzz crashes being reported against Gamepad, I noticed.)

[eseidel, 2014-04-28 15:49:54]

Adam is your best reviewer here.  My vote is for whatever is simplest.  If it's
possible (compatible?) for navigator.gamepad to just disappear when the window
detaches, that's also fine with me.  These neuter-on-detach objects have been a
source of crashers in the past.

[sof, 2014-04-28 16:25:55]

On 2014/04/28 15:49:54, eseidel wrote:
> Adam is your best reviewer here.  My vote is for whatever is simplest.  If
it's
> possible (compatible?) for navigator.gamepad to just disappear when the window
> detaches, that's also fine with me.  These neuter-on-detach objects have been
a
> source of crashers in the past.

Yes, DNT was at least one. I wouldn't mind not supporting supplements in this
state, but I don't think there is a way around having objects and methods
continue to return non-null values on lookup when detached.

(Cc:ed Gamepad API folks for input.)

[abarth-chromium, 2014-04-28 17:58:13]

https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
File Source/core/frame/DeviceSensorEventController.cpp (right):

https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
Source/core/frame/DeviceSensorEventController.cpp:37: :
PageLifecycleObserver(document ? document->page() : 0)
Folks would then class in the Page* rather than the document.

https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
File Source/core/frame/DeviceSensorEventController.h (right):

https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
Source/core/frame/DeviceSensorEventController.h:65: Document* m_document;
Maybe rather than storing this pointer, this class should get its document from
an abstract virtual function?  It looks like folks use this class by subclassing
and each have different ways of watching for the document's destruction.

[sof, 2014-04-28 18:58:41]

https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
File Source/core/frame/DeviceSensorEventController.h (right):

https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
Source/core/frame/DeviceSensorEventController.h:65: Document* m_document;
On 2014/04/28 17:58:14, abarth wrote:
> Maybe rather than storing this pointer, this class should get its document
from
> an abstract virtual function?  It looks like folks use this class by
subclassing
> and each have different ways of watching for the document's destruction.

Thanks, I like that suggestion, will try to make it work out.

[sof, 2014-04-29 10:39:21]

On 2014/04/28 18:58:41, sof wrote:
>
https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
> File Source/core/frame/DeviceSensorEventController.h (right):
> 
>
https://codereview.chromium.org/256593010/diff/1/Source/core/frame/DeviceSens...
> Source/core/frame/DeviceSensorEventController.h:65: Document* m_document;
> On 2014/04/28 17:58:14, abarth wrote:
> > Maybe rather than storing this pointer, this class should get its document
> from
> > an abstract virtual function?  It looks like folks use this class by
> subclassing
> > and each have different ways of watching for the document's destruction.
> 
> Thanks, I like that suggestion, will try to make it work out.

Now done.

[abarth-chromium, 2014-04-29 18:21:41]

lgtm

[commit-bot: I haz the power, 2014-04-29 18:22:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sigbjornf@opera.com/256593010/30001

[commit-bot: I haz the power, 2014-04-30 04:30:16]

Change committed as 172920