Removes the redundant IMB::OnFocus call in DNWA::HandleNativeFocus().

Removes the redundant IMB::OnFocus call in DNWA::HandleNativeFocus().

Currently HWNDMessageHandler::OnSetFocus happens before WM_ACTIVATE is handled in HWNDMessageHandler::OnWndProc().
Root cause: when WM_ACTIVATE comes, in HWNDMessageHandler::OnWndProc(), it will call DefWindowProc() first, which may result in generating WM_SETFOCUS message and call OnWndProc() recursively and therefore OnSetFocus() is called.
After the recursive OnWndProc() returns, PostProcessActivateMessage() is called to handle WM_ACTIVATE.

So DWTHW::HandleNativeFocus() happens before DNWA::HandleActivationChanged().
Therefore, IMB::OnFocus() will get NULL for host_->GetTextInputClient(), which will result in the IMB instance set to the host InputMethod as the focused TextInputClient (which should be the RWHVA).

It would be risky to fix that order in HWNDMessageHandler.
Instead, we can remove the IMB::OnFocus call in DWTHW::HandleNativeFocus() which seems to be redundant, because DNWA::OnWindowFocused() can do the thing, which happens after the top window is activated.

TBR=ananta@chromium.org,sky@chromium.org
BUG=486604
TEST=Verified on local win build.

Committed: https://crrev.com/4719aba37c66ae6f677977107819119494d70748
Cr-Commit-Position: refs/heads/master@{#330707}

[Shu Chen, 2015-05-15 06:58:46]

shuchen@chromium.org changed reviewers:
+ sadrul@chromium.org, yukawa@chromium.org

[Shu Chen, 2015-05-15 06:58:47]

Hi yukawa@ & sadrul@, can you please review this? Thanks

Yukawa and I have done offline discussions and investigations.

It is worth a try to remove this problematic redundant call and verify the
canary builds.

I've done some tests locally and didn't found regressions.

[yukawa, 2015-05-15 07:24:31]

LGTM from the viewpoint of Windows IME message handling.

[sadrul, 2015-05-16 02:24:25]

sadrul@chromium.org changed reviewers:
+ ananta@chromium.org, sky@chromium.org
- sadrul@chromium.org

[sadrul, 2015-05-16 02:24:26]

--> sky@/ananta@

I am not really a good reviewer for Windows code, but left a comment about the
corresponding X11 code.

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
File ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc (right):

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc:827: }
The X11 version seems to still do this (e.g.
DesktopWindowTreeHostX11::OnNativeWidgetFocus(), OnNativeWidgetBlur()). Should
those be removed too?

[Shu Chen, 2015-05-20 06:28:48]

TBR'ed ananta@ & sky@ as it is not easy to identify issues by reviewing without
testing/verifications. I'm going to commit this and verify in M45 canary.

[Shu Chen, 2015-05-20 06:28:57]

The CQ bit was checked by shuchen@chromium.org

[commit-bot: I haz the power, 2015-05-20 06:29:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1139623003/20001

[Shu Chen, 2015-05-20 06:30:52]

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
File ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc (right):

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc:827: }
On 2015/05/16 02:24:25, sadrul wrote:
> The X11 version seems to still do this (e.g.
> DesktopWindowTreeHostX11::OnNativeWidgetFocus(), OnNativeWidgetBlur()). Should
> those be removed too?

Let's limit the scope of this change in Windows platform. Could change for X11
separately if we confirm this fix is correct and DWTHX11::OnNativeWidgetFocus()
can be removed.

[commit-bot: I haz the power, 2015-05-20 07:11:27]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-05-20 07:12:06]

Patchset 2 (id:??) landed as
https://crrev.com/4719aba37c66ae6f677977107819119494d70748
Cr-Commit-Position: refs/heads/master@{#330707}

[sky, 2015-05-22 17:29:03]

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
File ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc (right):

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc:822: // TODO(beng):
inform the native_widget_delegate_.
I think it's worth a comment here as to why you don't need to call to
InputMethod.

[Shu Chen, 2015-05-25 01:14:25]

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
File ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc (right):

https://codereview.chromium.org/1139623003/diff/20001/ui/views/widget/desktop...
ui/views/widget/desktop_aura/desktop_window_tree_host_win.cc:822: // TODO(beng):
inform the native_widget_delegate_.
On 2015/05/22 17:29:02, sky wrote:
> I think it's worth a comment here as to why you don't need to call to
> InputMethod.

Thanks for your comment. My later cl soon to refactor the IMF (ownership change
for InputMethod) will make window tree host to call OnFocus/OnBlur of
InputMethod.