Code review comment for lp:~bregma/geis/lp-934207

* Typo:

Fixed.

* reject needs to remove all matching gesture events from the geis queue.

Since it's impossible to guarantee at the API level that no further event for a rejected gesture will ever come through (event generation is asynchronous over possibly several processes), I figured the application needs to handle this anyway. However, since it is not a big effort to remove these events from the internal GEIS queues, I added this and extended the test case accordingly.

* The accept test case should accept on the begin event, and check if an end event for the gesture is received. This will ensure that the gesture is actually accepted and not rejected.

Sure. Done.

* These tests don't check whether a gesture and its touches are really accepted or rejected. To do this, we need a test that has a subwindow that would receive the touch events if the gesture is rejected, and does not receive touch events if the gesture is accepted. I think this is getting into more testing than this specific functionality requires, but if we hit any bugs where geis accept/reject is not being propagated properly we will need to make these tests.

Yes, I think testing how clients interact with the X server is a little beyond the scope of unit tests for the GEIS API.