Change WebRTCInternals::GetInstance() to use a leaky LazyInstance.

Change WebRTCInternals::GetInstance() to use a leaky LazyInstance.

WebRTCInternals uses NotificationRegistrar which must be deleted on the UI thread. In cases where the UI thread is not the same as the main thread (like in Chromium Embedded Framework) the destruction of WebRTCInternals via AtExitManager at shutdown will cause a CalledOnValidThread() check failure. There's no particular reason why the object needs to be deleted at shutdown so just leak it instead.

BUG=none
TEST=none

Committed: https://crrev.com/8ad274252338620049d06e533ba6a671c7847825
Cr-Commit-Position: refs/heads/master@{#295751}

[Marshall, 2014-09-18 19:37:58]

marshall@chromium.org changed reviewers:
+ acolwell@chromium.org

[Marshall, 2014-09-18 19:39:08]

Please review.

[Marshall, 2014-09-18 19:39:27]

marshall@chromium.org changed reviewers:
+ dalecurtis@chromium.org
- acolwell@chromium.org

[Marshall, 2014-09-18 19:39:49]

Please review (changing review since acolwell@ is leaving)

[DaleCurtis, 2014-09-19 01:05:55]

dalecurtis@chromium.org changed reviewers:
+ tommi@chromium.org

[DaleCurtis, 2014-09-19 01:05:55]

+tommi from WebRTC who may be able to answer my thought.

https://codereview.chromium.org/575283004/diff/1/content/browser/media/webrtc...
File content/browser/media/webrtc_internals.cc (right):

https://codereview.chromium.org/575283004/diff/1/content/browser/media/webrtc...
content/browser/media/webrtc_internals.cc:64: return Singleton<WebRTCInternals,
Hmm, this seems like it should be a leaky LazyInstance instead.

[tommi (sloooow) - chröme, 2014-09-19 01:28:42]

https://codereview.chromium.org/575283004/diff/1/content/browser/media/webrtc...
File content/browser/media/webrtc_internals.cc (right):

https://codereview.chromium.org/575283004/diff/1/content/browser/media/webrtc...
content/browser/media/webrtc_internals.cc:64: return Singleton<WebRTCInternals,
On 2014/09/19 01:05:55, DaleCurtis wrote:
> Hmm, this seems like it should be a leaky LazyInstance instead.

Do you mean to create a static variable instead?
(I'm a bit slow at the moment so sorry if this is obvious - perhaps you just
need to spell it out to me :))

[DaleCurtis, 2014-09-19 02:27:57]

https://codereview.chromium.org/575283004/diff/1/content/browser/media/webrtc...
File content/browser/media/webrtc_internals.cc (right):

https://codereview.chromium.org/575283004/diff/1/content/browser/media/webrtc...
content/browser/media/webrtc_internals.cc:64: return Singleton<WebRTCInternals,
On 2014/09/19 01:28:42, tommi wrote:
> On 2014/09/19 01:05:55, DaleCurtis wrote:
> > Hmm, this seems like it should be a leaky LazyInstance instead.
> 
> Do you mean to create a static variable instead?
> (I'm a bit slow at the moment so sorry if this is obvious - perhaps you just
> need to spell it out to me :))

We typically  prefer LazyInstance to Singleton. Or so says  the Singleton header
:)

[DaleCurtis, 2014-09-19 17:44:39]

Hmm, looking at the headers again it looks like the Singleton header isn't as
explicit about not using it as it used to be.

I still think it should be a LazyInstance though, Marshall can you switch it
over to a leaky instance like done in MediaInternals:

https://code.google.com/p/chromium/codesearch#chromium/src/content/browser/me...

[Marshall, 2014-09-19 17:55:53]

Changed to use LazyInstance. Please review.

[DaleCurtis, 2014-09-19 17:56:34]

lgtm

[Marshall, 2014-09-19 18:03:11]

The CQ bit was checked by marshall@chromium.org

[commit-bot: I haz the power, 2014-09-19 18:05:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/575283004/20001

[commit-bot: I haz the power, 2014-09-19 18:53:08]

Committed patchset #2 (id:20001) as 61ef196c43771e6e06c522b0fa70ec3ff2aba201

[commit-bot: I haz the power, 2014-09-19 18:53:42]

Patchset 2 (id:??) landed as
https://crrev.com/8ad274252338620049d06e533ba6a671c7847825
Cr-Commit-Position: refs/heads/master@{#295751}

[tommi (sloooow) - chröme, 2014-09-19 18:59:01]

lgtm