Make Simulate.mouseEnter/Leave use direct dispatch

[sophiebits]

Fixes #1297.

onMouseEnter and onMouseLeave shouldn't actually use direct dispatch, but doing so is more useful than doing nothing (and I don't think it precludes adding proper enter/leave dispatching later, either).

Test Plan: grunt test

[jordwalke]

Did you see how enter/leave actually has its own dispatch path?
Standard bubbling to target t goes from root to t back to root.

A leave/enter from tstart to tend, dispatches leaves from tstart to the firstCommonAncestor but stops just one short of that common ancestor. enters are dispatched similarly, from the firstCommonAncestor to tend, but starting just one after the first common ancestor on the path to `tend . I think I completely pulled this behavior out of thin air. But it made sense to me and we're at liberty to reimagine mouse events since this is not a standard DOM event :D

[sophiebits]

Yeah, I saw. Enter/leave are now in the spec and I think they work the same way you did?

I didn't have great ideas for a simulation API for that though -- see #1297. I figured direct dispatch is simple and is 99% of the time what you need in a test.

[facebook-github-bot]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Facebook open source project. Thanks!

[zpao]

Yay or nay?

[jordwalke]

@spicyj Do you recall our conversation about this and what my thoughts were? I seem to recall saying that the test cases should operate exactly like the actual code that runs in applications.

[ryanzec]

Any update on this? This is a pretty big issue for me since I do commonly use mouseenter/mouseleave and right now I can write any unit tests for those pieces of code.

[ryanzec]

It looks like the latest code (0.12.1) uses ReactBrowserEventEmitter instead of ReactEventEmitter

[ryanzec]

Any reason why this is not being merged?

[ryanzec]

Another month has gone by, any update on this (this would help greatly with my unit tests)?

[sophiebits]

@ryanzec Thanks for the nudge – we had some concern over whether this is the exact final API that we want, but this is an improvement so I agree we should probably merge it.

@zpao r?

[ryanzec]

@spicyj Ok, its been about 2 more months so I figure I would give another nudge. Personally I think it would be better to have something that works than something that does not work. I would rather be able to test now and then have to update my tests if the way to test this event changes then not being able to test this event at all.

[sophiebits]

@ryanzec Thanks for the nudge. :)

[vsiao]

Will this be a 0.14 fix?

[sophiebits]

Yes.

[vsiao]

Thanks! :]