[Webkit-unassigned] [Bug 64970] [Gtk] Support for client-based geolocation

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sun Aug 7 09:45:14 PDT 2011


https://bugs.webkit.org/show_bug.cgi?id=64970





--- Comment #9 from Zan Dobersek <zandobersek at gmail.com>  2011-08-07 09:45:14 PST ---
(In reply to comment #7)
> (From update of attachment 101634 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=101634&action=review
> 
> This patch looks pretty good in general, and unskips a ton of tests!  I don't quite understand why we want to implement client-based geolocation without exposing an API. Is  it necessary just to unskip the tests?

Using client-based geolocation eases the use of the mock geolocation client which is needed for _many_ tests and I've started this task pretty much because of that. But if there's an interest for an API I guess it should be made (though, to be honest, I don't have much idea on what it would look like).

> Could we somehow create mock locations by interfacing directly with geoclue? If so that would be great, because we could test the geoclue stuff with layout tests.

Great you've brought that up. It bugs me that the mock clients are used to only test behavior of WebCore-related functionality of a feature but not the functionality of an external library that we use (in this case GeoClue). But we could test GeoClue behavior by creating a fake provider for testing purposes that would then be used on buildbots. This would then remove the necessity of using GeolocationClientMock. I'll look how the provider could be updated from DumpRenderTreeSupportGtk class and if this would be functional at all.

But before that I'll upload the updated patch. And thanks for the review!

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.



More information about the webkit-unassigned mailing list