allow triggerEvent to be called with additional arguments for the listener

[tschaub]

Our triggerEvent method just calls all listeners (optionally bound to some caller). This can be modified easily to call listeners with additional arguments. This makes event stuff more fun.

Pretend a drag feature control wanted to trigger a "predragfeature" event on a layer. Listeners should expect to receive the feature in question and perhaps the start and end pixel (of the drag). In addition, if triggerEvent returns the return from the last listener, then that control can modify its behavior (undo the drag for example) if the arguments (start and end pixel in this case) don't meet some user specified criteria.

Just a mocked up example, but this would give us way more flexibility with our event magic. Basically, I'm asking for triggerEvent to take an optional third argument and return continueChain.

[pgiraud]

I'm totally in favour with this feature. ExtJS behaves in a similar way and it's quite useful.

[tschaub]

send listeners arguments and respect their returns

[tschaub]

Ok, now with tests. And they pass (in FF).

Please review.

[crschmidt]

Can you explain the change from callback.func.call to callback.func.apply? I'm not used to 'call', and I don't get the meanining.

[elemoine]

Replying to crschmidt:

Can you explain the change from callback.func.call to callback.func.apply? I'm not used to 'call', and I don't get the meanining.

You pass apply an array of arguments, while you pass call the arguments directly.

< http://blog.metawrap.com/blog/TheVeryUsefulJavaScriptCallAndApplyFunctionsForOverridingThisForAGivenFunction.aspx>

[tschaub]

yeah - call if you have references to all of the arguments, apply otherwise

[crschmidt]

I'm fine with this if tests pass in IE.

[tschaub]

Yes, tests pass in IE. I will note however that while tests pass in IE, the IE debugger complains about the vector layer tests (without this patch as well). That is, tests "pass" as before in IE.

[pgiraud]

We should be able to commit this since the tests pass for layer_vector in IE.

[tschaub]

(In [5398]) triggerEvent can now be called with additional arguments for the listeners - this means instead of getting news like 'hey, someone won the lottery' listeners now get news like 'hey, you won the lottery' - in addition, the triggerer gets back the return from the listener, so if a listener wants to say 'dont tell anyone else' the triggerer gets that message - thanks for the reviews (closes #1189)