[Android] Thoroughly reset gesture detection upon page navigation

[Android] Thoroughly reset gesture detection upon page navigation

The current gesture reset logic on Android uses a cancellation event
synthesized from the active touch sequence. However, this fails to reset
detection for timeout-based events, e.g. delayed tap and double-tap,
just after the pointer has just been released. Expose an explicit gesture
detection reset hook on the GestureProvider, called when the main
frame navigates or the window loses focus.

BUG=418402
Committed: https://crrev.com/d47c1e4b6bb7e9b485f6dfede0236ddd5c823878
Cr-Commit-Position: refs/heads/master@{#303749}

[jdduke (slow), 2014-11-11 17:49:52]

jdduke@chromium.org changed reviewers:
+ tdresser@chromium.org

[jdduke (slow), 2014-11-11 17:49:53]

tdresser@: PTAL. I've been through about 50 version of this patch, and this is
just the beginning (eventually we'll be resetting the input router, but that's
trick with things like event validation, etc...).

The simplest version here would be to always feed a cancel to
FilteredGestureProvider, but because of the change to GestureProvider that
allows redundant TouchCancel events, we run into problems quickly with the
TouchDispositionGestureFilter (sigh...). As Aura doesn't have a tap delay, I'm
much less concerned here, but the next step will be hard flushing and resetting
the router after navigation.

[tdresser, 2014-11-11 19:06:39]

On 2014/11/11 17:49:53, jdduke wrote:
> tdresser@: PTAL. I've been through about 50 version of this patch, and this is
> just the beginning (eventually we'll be resetting the input router, but that's
> trick with things like event validation, etc...).
> 
> The simplest version here would be to always feed a cancel to
> FilteredGestureProvider, but because of the change to GestureProvider that
> allows redundant TouchCancel events, we run into problems quickly with the
> TouchDispositionGestureFilter (sigh...). As Aura doesn't have a tap delay, I'm
> much less concerned here, but the next step will be hard flushing and
resetting
> the router after navigation.

Low level LGTM (with nit).

Naive question: is all the renderer state completely reset on page navigation?
Do we need to worry about ensuring the renderer gets touch cancel events?

If we navigate with fingers down, we'll end up getting some extraneous touch end
events. I guess at this point, we'll just ignore them, which is fine. It may
make input event validation a bit trickier though.

[tdresser, 2014-11-11 19:06:47]

https://codereview.chromium.org/717573004/diff/20001/ui/events/gesture_detect...
File ui/events/gesture_detection/gesture_provider_unittest.cc (right):

https://codereview.chromium.org/717573004/diff/20001/ui/events/gesture_detect...
ui/events/gesture_detection/gesture_provider_unittest.cc:626: // The long press
triggers window focus loss by opening a context menu.
Update comment.

[jdduke (slow), 2014-11-11 20:34:16]

jdduke@chromium.org changed reviewers:
+ sievers@chromium.org

[jdduke (slow), 2014-11-11 20:34:16]

sievers@: Owner review for
content/browser/renderer_host/render_widget_host_view_android.cc? Thanks.

[no sievers, 2014-11-11 21:55:48]

On 2014/11/11 20:34:16, jdduke wrote:
> sievers@: Owner review for
> content/browser/renderer_host/render_widget_host_view_android.cc? Thanks.

lgtm

[jdduke (slow), 2014-11-11 21:56:16]

The CQ bit was checked by jdduke@chromium.org

[commit-bot: I haz the power, 2014-11-11 21:57:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/717573004/20001

[jdduke (slow), 2014-11-11 22:19:35]

The CQ bit was unchecked by jdduke@chromium.org

[jdduke (slow), 2014-11-11 22:21:53]

The CQ bit was checked by jdduke@chromium.org

[jdduke (slow), 2014-11-11 22:21:56]

https://codereview.chromium.org/717573004/diff/20001/ui/events/gesture_detect...
File ui/events/gesture_detection/gesture_provider_unittest.cc (right):

https://codereview.chromium.org/717573004/diff/20001/ui/events/gesture_detect...
ui/events/gesture_detection/gesture_provider_unittest.cc:626: // The long press
triggers window focus loss by opening a context menu.
On 2014/11/11 19:06:47, tdresser wrote:
> Update comment.

Done.

[commit-bot: I haz the power, 2014-11-11 22:22:52]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/717573004/40001

[commit-bot: I haz the power, 2014-11-11 23:47:27]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2014-11-11 23:48:53]

Patchset 3 (id:??) landed as
https://crrev.com/d47c1e4b6bb7e9b485f6dfede0236ddd5c823878
Cr-Commit-Position: refs/heads/master@{#303749}