This reverts a previous change that used only the contentInset
when setting the center coordinate of the map view.
Instead of contentInset, edgePaddingForFollowing is used. This fixes
a regression that caused the user dot annotation to be misplaced
when the map was configured to show the annotation in any place other
than in the center of the map.
In addition, edgePaddingForFollowing is updated to account for
contentInset when it calculates its UIEdgeInsets. This fixes a bug
where the user dot would move to the center of the content frame
(when it was adjusted for content insets) but the map would not scroll
along with it yielding the effect of the user dot being misplaced.
- Ensure that sqlite3_config() is always called before db init.
- Should only be called once, now.
- Move sqliteVersionCheck into mapbox::sqlite namespace.
Previously, we declared MGLStyleLayer initializers in the MGLStyleLayer
protocol as optional methods. This made it easy to opt in and out
of initializers that did or or did not make sense for the subclass.
However, this approach was dangerous since it was easy for an
application developer to initialize an MGLStyleLayer subclass with an
init method that was actually not implemented in that class causing
an exception at runtime.
This commit moves the init methods that each subclass supports to
each subclass so xcode (and the compiler) help the developer avoid
the previously possible runtime exception.
In addition, a new init method is added that takes the source layer
and passes that on to `mbgl::style:Layer` (but only on classes
where this is possible in core). This allows an application developer
to style a specific source layer (i.e. the contour lines of the mapbox
terrain vector source).
Finally, this refactors MGLStyleLayer classes to use an MGLSource instead of a
string identifier for the source when initializing the style.
For (hopefully) better performance than creating a new v8::Context to
wrap each callback while still avoiding leaking memory with v8::FunctionTemplate.
Adds a JavaScript shim in front of module.exports.Map to wrap the
req.respond API internally and preserve the public callback-passing
API, while still exporting the correct prototype.
Until Style::recalculate() is called to check that there are no visible layers using the source, we have to assume there are. Otherwise, Style::isLoaded() can return a false positive.
This paves the way for updates to filter and layout properties to trigger a source reload, without each SDK having to participate in the implementation.
* Renamed {Source,Tile}Observer::onNeedsRepaint to onTileUpdated. Messages should be in terms of what happened to the observed object, not in terms of what the observer needs to do. This also removes a confusing overlap of virtual methods on StyleObserver.
* Added style::Observer::onUpdate(Update). This is also a violation of the above rule, but I'm hopeful that it will disappear when update batching is implemented.
