Signal input flush when all flings have terminated

Signal input flush when all flings have terminated

Synthetic gestures rely on a flush signal from the InputRouter to
indicate when a particular gesture has finished. Currently, this fails
to capture fling animations, as they are hosted in the renderer.

Wire up the fling end notification to the InputRouter, allowing it to
gate flush signalling based on fling animation existence. This will
allow proper execution of synthetic fling gestures with telemetry.

BUG=321141
Committed: https://crrev.com/4216439c7d9c0f0796e9aff6639164d077546960
Cr-Commit-Position: refs/heads/master@{#321504}

[jdduke (slow), 2015-03-13 01:01:29]

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

[jdduke (slow), 2015-03-13 01:01:30]

miletus@: Could you take a look? The key change here is in input_router_impl
with respect to active fling bookkeeping. Everything else is just minor
tweaks/renaming.

Once this lands, I'll add some fling tests using the existing "swipe" gesture
with a few tweaks to guarantee more consistent fling velocity generation.
Thanks.

[Yufeng Shen (Slow to review), 2015-03-13 15:53:42]

https://codereview.chromium.org/1003023002/diff/1/content/browser/renderer_ho...
File content/browser/renderer_host/input/input_router_impl.cc (right):

https://codereview.chromium.org/1003023002/diff/1/content/browser/renderer_ho...
content/browser/renderer_host/input/input_router_impl.cc:612: }
I saw that gesture event queue also maintains a fling_in_progress bit. Can it
and active_fling_count_ somehow be merged into one ?

[jdduke (slow), 2015-03-13 16:08:04]

https://codereview.chromium.org/1003023002/diff/1/content/browser/renderer_ho...
File content/browser/renderer_host/input/input_router_impl.cc (right):

https://codereview.chromium.org/1003023002/diff/1/content/browser/renderer_ho...
content/browser/renderer_host/input/input_router_impl.cc:612: }
On 2015/03/13 15:53:42, Yufeng Shen wrote:
> I saw that gesture event queue also maintains a fling_in_progress bit. Can it
> and active_fling_count_ somehow be merged into one ?

Probably, although the gesture queue fling in progress bit is a little
different, as it simply tracks whether we should discard a GestureFlingCancel at
the *end* of the queue. So if a fling is active and a GestureFlingCancel is
enqueued but not sent, it will think there's no fling in progress. It should
really be named "needs_fling_cancel_".

I'll see if I can merge the two, it might mean making the GEQ filter events when
they hit the head of the queue, rather than the tail. Will post back here.

[jdduke (slow), 2015-03-13 21:56:23]

OK, moved the logic to the GEQ, definitely the right thing to do. I'd still like
to refactor the GEQ to only filter events at the head of the queue, but I'll
leave that for a follow-up.

[Yufeng Shen (Slow to review), 2015-03-16 16:21:34]

On 2015/03/13 21:56:23, jdduke wrote:
> OK, moved the logic to the GEQ, definitely the right thing to do. I'd still
like
> to refactor the GEQ to only filter events at the head of the queue, but I'll
> leave that for a follow-up.

l-g-t-m

[jdduke (slow), 2015-03-16 17:37:34]

jdduke@chromium.org changed reviewers:
+ dcheng@chromium.org, sadrul@chromium.org

[jdduke (slow), 2015-03-16 17:37:35]

+sadrul for content/browser/renderer_host/render_widget* changes.
+dcheng for trivial content/common security review.

[dcheng, 2015-03-17 01:50:25]

rs lgtm for moving IPC message definition from view_messages.h to
input_messages.h.

[sadrul, 2015-03-17 16:36:02]

The trybot failures look relevant. Mind looking at those?

[jdduke (slow), 2015-03-17 16:59:45]

Patchset #4 (id:60001) has been deleted

[jdduke (slow), 2015-03-17 17:00:03]

On 2015/03/17 16:36:02, sadrul wrote:
> The trybot failures look relevant. Mind looking at those?

Compile issues fixed.

[jdduke (slow), 2015-03-19 15:30:11]

On 2015/03/17 17:00:03, jdduke wrote:
> On 2015/03/17 16:36:02, sadrul wrote:
> > The trybot failures look relevant. Mind looking at those?
> 
> Compile issues fixed.

Sadrul, anything else you'd like to see here?

[sadrul, 2015-03-19 22:51:04]

lgtm

[jdduke (slow), 2015-03-19 23:12:03]

The CQ bit was checked by jdduke@chromium.org

[jdduke (slow), 2015-03-19 23:12:04]

The patchset sent to the CQ was uploaded after l-g-t-m from dcheng@chromium.org
Link to the patchset: https://codereview.chromium.org/1003023002/#ps80001
(title: "Fix test compile")

[commit-bot: I haz the power, 2015-03-19 23:12:28]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1003023002/80001

[commit-bot: I haz the power, 2015-03-20 03:09:46]

Committed patchset #4 (id:80001)

[commit-bot: I haz the power, 2015-03-20 03:10:55]

Patchset 4 (id:??) landed as
https://crrev.com/4216439c7d9c0f0796e9aff6639164d077546960
Cr-Commit-Position: refs/heads/master@{#321504}