beginSmoothScroll should send wheel ticks at constant velocity.

beginSmoothScroll should send wheel ticks at constant velocity.

There are two fixes here:
1. Compute the wheel tick amount based on time elapsed rather than our post
   message rate.
2. Try to trigger a new forward when needed on the inputAck, instead of using
   postTask. This helps when the main thread is busy doing swaps, which happens
   on OSX.

BUG=149408
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=161038

[nduca, 2012-10-06 02:15:26]

This (I think) should make the scrolls a lot smoother in our benchmarks.

[piman, 2012-10-06 06:18:57]

LGTM+nits

At a high level, if this functionality is going to continue to grow, it would be
nice to separate it into another class/file.

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
File content/browser/renderer_host/render_widget_host_impl.cc (right):

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
content/browser/renderer_host/render_widget_host_impl.cc:1599:
active_smooth_scroll_gestures_.size())
nit: ! .empty() instead of .size()

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
content/browser/renderer_host/render_widget_host_impl.cc:1626: if
(in_process_event_types_.size())
nit: ! .empty()

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
content/browser/renderer_host/render_widget_host_impl.cc:1636: if
(active_smooth_scroll_gestures_.size() == 0) {
nit: empty()

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
content/browser/renderer_host/render_widget_host_impl.cc:1643:
base::TimeDelta::FromMilliseconds(kSyntheticScrollMessageIntervalMs);
nit: line continuations at +4

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
content/browser/renderer_host/render_widget_host_impl.cc:1645:
(last_smooth_scroll_gestures_tick_time_ + preferred_interval) -
+4

http://codereview.chromium.org/11013043/diff/3/content/browser/renderer_host/...
content/browser/renderer_host/render_widget_host_impl.cc:1687: if
(in_process_event_types_.size() != 0)
!.empty()

[nduca, 2012-10-10 02:12:14]

> At a high level, if this functionality is going to continue to grow, it would
be
> nice to separate it into another class/file.
Agreed.

[commit-bot: I haz the power, 2012-10-10 02:18:31]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nduca@chromium.org/11013043/11001

[commit-bot: I haz the power, 2012-10-10 04:19:20]

Change committed as 161038