incubator-callback-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Filip Maj <>
Subject Re: Hack to compass.js to support iOS watchHeadingFilter
Date Tue, 03 Apr 2012 20:48:18 GMT
Hey Becky,

I fixed up the tests from your fork and pushed to my fork:

Basically, from a black box perspective, if you pass no object (i.e.
undefined) into the compass methods as options, then you can expect exec()
to call into native with undefined.

I'm not sure if it's "purer" but, to make it consistent with use of exec()
across all of our APIs, we would not use an options object to pass
parameters into native but split the parameters out as individual
variables in the parameter array going into exec.

The above not necessary for this revision of compass API, though, as iOS
is the only one grabbing any parameters anyways. For 1.7 I can help
refactor the iOS implementation to clean this up.

That being said, I would like to refactor/change the parameter-passing
convention in exec(), and I think passing a single object with named
parameters is nicer and clearer than an array of values. But that is for
another release :)

On 4/3/12 1:25 PM, "Becky Gibson" <> wrote:

>As discussed on this list,  I added back support for the iOS watch compass
>heading by filter method by "extending" the unified JS Compass.watchFilter
>and getCurrentHeading() methods.   Frankly, I think this is a bit of a
>but it works.  If the call to watchHeading contains an options.filter
>parameter and it is > 0, I call getCurrentHeading with that options object
>and do not set up a interval to repeatedly call getCurrentHeading.  I hack
>a "special" value into the timers array so that when a clearWatch is
>called, I know to call Compass.stopHeading rather than just clearing the
>The issue is that sometimes getCurrentHeading is called with an options
>object and other times options is undefined.  I don't see any issues on
>calling exec with the undefined options param.  Need someone to verify
>on Android.  Also, this breaks the unit tests, and not being familiar with
>Jasmine, wasn't sure how to deal with an "optional" parameter?   So, could
>use some advice on how to handle that.  I could also make it a bit more
>"pure" and only pass the options parameter when it is legitimate or pass
>null with it was not passed in.  Looking for suggestions as to how best to
>fix this and modify the tests to pass before I check in.

View raw message