- static rendering now also runs in a separate thread; you have to start it with map.start(Map::Mode::Static) and join the thread with map.stop() before destructing the Map object
- map.renderStill() takes a callback with will be invoked on the *map* thread, so you'll have to figure out your own method of dispatching back to the main thread.
`Map` should be able to deal with having no access token or JSON even while it’s running. Most of `Map` accounts for this situation, but `reloadStyle()` incorrectly assumes that one or the other is set. This change corrects the assumption in `reloadStyle()`.
Also expose the access token and style name to clients as ordinary KVO-compliant properties. Actually, they’re not so ordinary because they’re inspectable!
Ref #1070, #1147
This brings the easing transition code a bit closer to how easings work
in gl-js. Instead of having an array of individual transitions for scale,
rotate, and pan, there is a single transition function that does all the
required calculations. This permits us to:
* Eliminate the "timeout" transition. (Fixes#126)
* Replace start/stopPanning() et al with setGestureInProgress(). Apps or
SDKs are expected to make paired calls to setGestureInProgress(). This
state will be ORed with the active easing state to determine when to
use texture interpolation. (Fixes#79)
* Run style recalculations only when an ease transition that affects the
zoom is in progress. (Fixes#1155)
- we're now always loading and parsing styles in the map thread
- adds an atomic enum that stores the actual updates we need to perform
- adds a MapData object that mutexes access to variables used from both threads
On iOS, the main thread does framebuffer operations just before it requests a rerender of the view contents. This means that GL operations on the Map thread will interfere with that and cause a crash. While most GL operations happen throughout rendering (which is synchronized with the main thread), deletion of shared_ptrs release in turn their OpenGL objects. This may happen at any time and isn't synced with rendering.
This patch changes this so instead of deleting all objects immediately, we are collecting all abandoned IDs and delete them in bulk once the main thread has given us permission to use OpenGL calls (i.e. during render()).
