Fix aura flickering when not using DWM.

Fix aura flickering when not using DWM.

By directly invoking DefWindowProc outside of the standard WindowProc / 
message pump flow, the default implementation of WM_NCACTIVATE was causing 
one frame to render using the standard Windows application look-and-feel.

The fix employed here is to use SendMessage, for 2 reasons:

1) It correctly follows Windows' model of 1 message / 1 action
2) It sends the WM_NCACTIVATE through our subclassed window.

I'm TBRing sky@ and beng@ since this is a really Windowsy change, but they do not return for a few weeks.  Also TBRing ananta@ since he wrote the original implementation of this function, but he's also out until Monday the 12th.

BUG=257183, 178600
TBR=sky,beng,ananta

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=216571

[sadrul, 2013-08-08 07:24:57]

lgtm

[sadrul, 2013-08-08 07:25:48]

Didn't mean to lgtm.

Perhaps +scottmg@ can do a proper review.

[scottmg, 2013-08-08 14:44:52]

Yeah, no idea why we were directly DefWindowProc'ing. lgtm.

[commit-bot: I haz the power, 2013-08-08 20:29:42]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/zturner@chromium.org/22353018/1

[commit-bot: I haz the power, 2013-08-09 04:18:33]

Change committed as 216571

[ananta, 2013-08-09 05:07:29]

On 2013/08/09 04:18:33, I haz the power (commit-bot) wrote:
> Change committed as 216571

Please check whether this bug
https://code.google.com/p/chromium/issues/detail?id=229599 does not recur with
this change.

Thanks
Ananta

[scottmg, 2013-08-09 05:23:21]

On 2013/08/09 05:07:29, ananta wrote:
> On 2013/08/09 04:18:33, I haz the power (commit-bot) wrote:
> > Change committed as 216571
> 
> Please check whether this bug
> https://code.google.com/p/chromium/issues/detail?id=229599 does not recur with
> this change.
> 
> Thanks
> Ananta

Yeah, we made sure the frame still draws as activated when the popup takes
focus. Shrikant was a bit concerned that SendMessage rather than DefWindowProc
might cause some strange problems, but we didn't notice any yet at least.

[zturner, 2013-08-09 16:58:43]

On 2013/08/09 05:23:21, scottmg wrote:
> On 2013/08/09 05:07:29, ananta wrote:
> > On 2013/08/09 04:18:33, I haz the power (commit-bot) wrote:
> > > Change committed as 216571
> > 
> > Please check whether this bug
> > https://code.google.com/p/chromium/issues/detail?id=229599 does not recur
with
> > this change.
> > 
> > Thanks
> > Ananta
> 
> Yeah, we made sure the frame still draws as activated when the popup takes
> focus. Shrikant was a bit concerned that SendMessage rather than DefWindowProc
> might cause some strange problems, but we didn't notice any yet at least.

In addition to what scott said, I think it's also worth mentioning that if we do
detect some other strange issues as a result of this change later down the line,
the fix is probably not to revert this change, but to fix them in a different
way.  I think we were invoking undefined behavior by calling DefWindowProc
directly, because it violates the assumption that a single call to WindowProc
handles exactly 1 message: the same one that WindowProc was called with.  We
were handling 2 messages this way, and 1 of them was different than the message
WindowProc was originally invoked with.