Re: [proximity] some more tests...

On Friday, May 25, 2012 at 5:31 PM, Anssi Kostiainen wrote:

> Hi Marcos,
>  
> On 25.5.2012, at 14.36, ext Marcos Caceres wrote:
>  
> This looks even better, great work!
>  
> One thing I noticed is that the polyfill is firing events even if there are no listeners registered (I remember you commented about this earlier so I checked how your polyfill behaves). This is really a QoI issue, and the spec is intentionally silent on this.
> If you wish, you could augment addEventListener and start firing only when there are listeners registered, something like (using device proximity as an example):
>  
> var p = Window.prototype;
> p._addEventListener = p.addEventListener;
> p.addEventListener = function (type, callback, capture) {
> if (type === 'deviceproximity') {
> // start firing events, if not already
> }
> this._addEventListener(type, callback, capture);
> };
>  
> ... and stop firing when all the listeners are removed (and handle .ondeviceproximity similarly). That said, augmenting built-in objects can be risky, so this may not be worth it.
>  
> Btw. All this relates to the popular topic of addEventListener side effects ...
>  
Yeah… I didn't want to do that because of the world famous side effects - so I just let it run (not great for battery, cpu, I imagine). I can control the window.onwhatever because I can start "sensor.sense()" and the yet to be implemented "sensor.stop()" for those through the [[Get]] method (but that's again a side effect, d'oh!); but like you coded above, the window.addEventListener() will have side effects if they turn the sensor on (I don't think we should pretend that it won't) :( Not sure how to proceed without having some explicit means of turning on the sensor through JS:

window.navigator.deviceproximity.enable() … or something that no one will like :)   
>  
> Anyway, thanks for the cool polyfill and the test suite!
>  

Thank you for taking a looksy, and thanks for updating spec :)

Received on Friday, 25 May 2012 16:51:43 UTC