Remove Android-specific modifications to non-client-based Geolocation
Geolocation::suspend()/resume() and GeolocationService::suspend()/resume()were first added in http://trac.webkit.org/changeset/38018 Geolocation::suspend()/resume() were removed in http://trac.webkit.org/changeset/56188 but reinstated for Android in http://trac.webkit.org/changeset/58061. When GeolocationClient implementations for GTK (http://trac.webkit.org/changeset/38765) and EFL (http://trac.webkit.org/changeset/81556) were added, they added empty stubs for suspend() and resume() even though these methods are only called on Android. The Android port of WebKit is being removed in Bug 66688, so it should be safe to remove the methods on Geolocation and on the clients.
Created attachment 104823 [details] Patch
Committed r93596: <http://trac.webkit.org/changeset/93596>
Well this code is used by Qt on our branch and it is used by iOS as well.
> Well this code is used by Qt on our branch and it is used by iOS as well. Sorry, I wasn't aware of this. Wasn't it you that reviewed http://trac.webkit.org/changeset/56188 which removed them previously? Do these methods need to be present upstream? If so, we should add comments to make this clear so they're not deleted again in the future.
We are going to upstream all of our suspend/resume code, so they will be needed yes. We will readd them as part of that.
> We will readd them as part of that. OK Do you have plans to switch to the client-based Geolocation implementation? There's consensus that it's the right approach and that we should remove the non-client-based implementation. It would also remove the need for add()/resume() as all this logic can be handled by the client. See Bug 40373.
(In reply to comment #7) > > We will readd them as part of that. > OK > > Do you have plans to switch to the client-based Geolocation implementation? There's consensus that it's the right approach and that we should remove the non-client-based implementation. It would also remove the need for add()/resume() as all this logic can be handled by the client. See Bug 40373. Yes, we will be using that, sure.