Re: Test suite issues

Hi Steve,

I haven't seen any objections to these changes so go ahead and update 
the files in SVN directly.

Thanks,
Lars Erik

On 12.02.2011 03:15, Steve Block wrote:
> Andrei and I have taken a look through all of the tests and determined
> which are valid - those which test features within the scope of the
> spec and which do not rely on mocking. Many of the tests are not
> valid, mostly because they rely on mocking, which is inappropriate as
> discussed previously.
>
> Below is the list of valid tests. Some need significant modification,
> generally due to invalid assumptions about when permissions are
> requested and all need superfluous calls to setupDevice() removed.
>
> Next I'll make the required changes to the tests and update the
> documentation. Lars Erik, is it OK if I update the published version
> in SVN directly, or would you like to see a zip first?
>
> Steve
>
>
> Test 00001: user denies access, check that no success callback is
> called and error callback is returned with correct error (and in due
> time) (with modification)
> Test 00002: user allows access, check that success callback is called
> (with modification)
> Test 00011: call getCurrentPosition with null success callback, check
> if exception
> Test 00141 (optional): user asked to approve/remember, then asked to
> revoke, then reload. The permission dialogue should reappear. (with
> modification)
> Test 00142: (optional): user asked to deny/remember, then ask to
> revoke, then reload. The permission dialogue should reappear (with
> modification)
> Test 00026: check for window.* [Navigator object] has a 'geolocation'
> member which is an instance of class NavigatorGeolocation.
> Test 00027: call getCurrentPosition() without arguments. Exception expected.
> Test 00028: call getCurrentPosition() with wrong type for first
> argument. Exception expected.
> Test 00029: call getCurrentPosition() with wrong type for second
> argument. Exception expected.
> Test 00130: call watchPosition() without arguments. Exception expected.
> Test 00131: call watchPosition() with wrong type for first argument.
> Exception expected.
> Test 00132: call watchPosition() with wrong type for second argument.
> Exception expected.
> Test 00031: pass if getCurrentPosition returns withing 100ms (with modification)
> Test 00099: set the device to return a preset latitude, longitude and
> accuracy, and check that the API returns same values (with
> modification)
> Test 00058: call watchPosition() without arguments
> Test 00059: call watchPosition() with wrong type for first argument.
> Test 00060: call watchPosition() with wrong type for second argument.
> Test 00062: check that watchPosition returns a long, immediately.
> Test 00080: Test clearWatch with wrong identifier, fail if more than 1
> call to the same clearWatch
> Test 00123: call getCurrentPosition with wrong type for
> enableHighAccuracy. No error expected. (can expand)
> Test 00086: set timeout and maximumAge to 0, check that timeout error
> raised (getCurrentPosition)
> Test 00088: set timeout and maximumAge to 0, check that timeout error
> raised (watchPosition)
> Test 00091: check that a negative timeout value is equivalent to a 0
> timeout value (getCurrentLocation)
> Test 00092: check that a negative timeout value is equivalent to a 0
> timeout value (watchPosition)
> Test 00095: check existence of timestamp attribute, and correct type
> Test 00143: check that heading is NaN when speed is 0
> Test 00108: check that the error type exists and that its attributes
> have the right numeric value
> Test 00116 fire many getCurrentPosition, count callbacks
> Test 00118 uniqueness of watchPosition id values (call function many
> times, check collisions)
> Test 00119 callbacks recursively calling getCurrentPosition
> Test 00120 callbacks recursively calling watchPosition
>

Received on Wednesday, 16 February 2011 13:49:39 UTC