Instead of exposing these methods in a private header that winds up triggering warnings publicly, just redeclare the methods in a category. Since this is a test app, we’ll catch any issues that arise from refactoring simply by testing.
Fixes#1579.
`-pauseMetricsCollection` and `-resumeMetricsCollection` were originally introduced to mitigate concerns around battery usage, but `MGLMapboxEvents` has since evolved to tread much lighter on the battery. There’s no longer any need for the client to call these methods directly. The iOS test project now includes MGLMapboxEvents.h explicitly to avoid having to package a header that will go unused.
A typo in a selector literal meant that `-[MGLMapViewDelegate mapViewDidFailLoadingMap:withError:]` would never be sent to delegates even if `MGLMapView` had called it. Along for the ride, create a proper `NSError` to pass into that callback, and an error domain for that error.
Deprecated `-mapID` and `-setMapID:` to accurately reflect the APIs the ID is used for. Marked the property unavailable but reimplemented its getter and setter to assert with helpful messages.
Fixes#1500.
Made `+[MGLAccountManager setMapboxMetricsEnabledSettingShownInApp:]` unavailable, with a message explaining what to do instead. Removed a commented-out call to that method. Only the environment for an access token if one hasn’t already been set in Info.plist.
That way there’s no ambiguity about when you should call things like `+[MGLAccountManager setMapboxMetricsEnabledSettingShownInApp:]`. In fact, that method is now deprecated because it’s so easy to call in the wrong place.
Fixes#1535.
`MGLMapboxEvents` automatically recreates its location manager when the host app is reauthorized for Location Services. Added some redundancy to ensure that there is no location manager when either Location Services or Mapbox Metrics is off.
The shared `MGLMapboxEvents` now owns the only instance of `MGLMetricsLocationManager`. It also owns a `MGLMapboxEventsData` object, to which all user-specific data has been relegated. This object can be built up and torn down on the fly, whenever the opt-out preference is toggled. `MGLMapboxEvents` has also given up knowledge of the access token and now queries `MGLAccountManager` for that information, which `MGLMapView` keeps up-to-date.
This still has the unwanted side effect off blocking the Map thread
waiting for the workers to return (because they hold a reference
to the Map thread main loop).
Maybe the only way to solve this would be having a I/O thread for
dispatching messages.
The Request is not bound anymore to the invoker thread because the workers
will request for glyphs, but the reply will arrive at the Map thread.
