Switch Sky to pointer events

Switch Sky to pointer events

This CL switches Sky to a pointer events model that unifies mouse and touch
with a common event model. This implementation is just enough to make the
example widgets work. The rest of the design is in the specs, which we hope to
converge with over time.

R=ojan@chromium.org

Committed: https://chromium.googlesource.com/external/mojo/+/0f185a7b8d22b52b921fd4c399256a83d5639981

[ojan, 2015-01-23 10:23:54]

ojan@chromium.org changed reviewers:
+ ojan@chromium.org

[ojan, 2015-01-23 10:23:55]

lgtm

https://codereview.chromium.org/866213004/diff/20001/sky/engine/core/frame/Ne...
File sky/engine/core/frame/NewEventHandler.h (right):

https://codereview.chromium.org/866213004/diff/20001/sky/engine/core/frame/Ne...
sky/engine/core/frame/NewEventHandler.h:40: typedef std::map<int, RefPtr<Node>>
PointerTargetMap;
Why didn't HashMap work? stl containers are terribad, no?

https://codereview.chromium.org/866213004/diff/20001/sky/engine/public/platfo...
File sky/engine/public/platform/WebInputEvent.h (right):

https://codereview.chromium.org/866213004/diff/20001/sky/engine/public/platfo...
sky/engine/public/platform/WebInputEvent.h:192: return type == PointerDown ||
Probably premature optimization, but we could save two branches here by doing
range checks type >= PointerDown || type <= PointerCancel.

[abarth-chromium, 2015-01-23 16:51:33]

https://codereview.chromium.org/866213004/diff/20001/sky/engine/core/frame/Ne...
File sky/engine/core/frame/NewEventHandler.h (right):

https://codereview.chromium.org/866213004/diff/20001/sky/engine/core/frame/Ne...
sky/engine/core/frame/NewEventHandler.h:40: typedef std::map<int, RefPtr<Node>>
PointerTargetMap;
On 2015/01/23 at 10:23:55, ojan wrote:
> Why didn't HashMap work? stl containers are terribad, no?

I kept having trouble with HashMap asserting in checkKey.  HashMap steals some
of the key values to mean special things (e.g., 0 and -1).  I tried some of the
other traits, but I failed to find something that worked.

In reality, the number of entries in this map are equal to the number of fingers
(or mice) that are down at once.  In practice, that's like one or two.  We could
use an array instead, which would probably be slightly more efficient.

https://codereview.chromium.org/866213004/diff/20001/sky/engine/public/platfo...
File sky/engine/public/platform/WebInputEvent.h (right):

https://codereview.chromium.org/866213004/diff/20001/sky/engine/public/platfo...
sky/engine/public/platform/WebInputEvent.h:192: return type == PointerDown ||
On 2015/01/23 at 10:23:55, ojan wrote:
> Probably premature optimization, but we could save two branches here by doing
range checks type >= PointerDown || type <= PointerCancel.

Ok.  That's what the other similar functions do.

[abarth-chromium, 2015-01-23 17:38:50]

Committed patchset #2 (id:20001) manually as
0f185a7b8d22b52b921fd4c399256a83d5639981 (presubmit successful).