The build output’s readme now explains in full detail the installation process, rather than linking to an online resource that may not match the current version. Also cleaned up the tag-finding code to better handle double-digit situations.
package.sh now recognizes several environment variables through make that influence the build settings and targets. Info.plist variables are substituted with the correct values in both the static and dynamic frameworks.
Fixes#3656.
The library is packaged as a static framework, so there isn’t a separate resource bundle anymore. Instead, look for the static framework inside the Frameworks folder.
Copied strip-frameworks.sh from realm/realm-cocoa@7cc31db631 for realm/realm-cocoa#2759. This script, which is embedded in the dynamic framework, strips out content for invalid architectures from any embedded framework and specifically strips out Simulator content when archiving to work around an App Store bug.
Rewrote the iOS setup documentation. In particular, the MAPBOX_ACCESS_TOKEN environment variable is only for use with the iosapp demo application and doesn’t work in third-party applications.
make iproj now produces a target that pulls together static libraries like core and platform-ios into a real dynamic framework. iosapp is pretty much just a regular iOS application that links Mapbox.framework (except for the inclusion of default_styles.hpp). iosapp runs fine in the Simulator and on a device, and the same is true for any application linking against Mapbox.framework.
The ipackage target produces both a Bitcode-disabled static framework and a Bitcode-enabled dynamic framework, eliminating the need for a separate framework.sh. It disables code signing, since that happens on copy when the framework is embedded inside the application bundle. It also merges the device and simulator builds into a single fat framework.
Also bumped itest minimum deployment target to iOS 8.0, the first version that supports linking frameworks.
Fixes#828.
Per MapKit behavior, zooming or rotating with a gesture can no longer kick the user out of user tracking mode. In user tracking mode, the zoom animation is always anchored on the user dot, wherever it may be.
The user dot or user puck’s position is now configurable via a new alignment property in any user tracking mode. Also, -setUserTrackingMode:animated: is now exposed publicly, and setting its animated parameter to NO skips the initial animation.
In user tracking mode, keep the user dot in a fixed location on screen, instead easing the map view so that the fixed location corresponds to the new user location. There is one exception: MGLMapView reacts to the first location update the same way as before, so that it appears to move to where the user dot has been all along.
In course tracking mode, weight the user dot down towards the bottom of the view, since it’s more important to see the road ahead than the road behind.
Fixes#2600, and #1041 to some extent.
The zoom level is already clamped by Transform. -currentMinimumZoom appears to be useful only as an unrotation threshold; it doesn’t correspond to the minimum allowed zoom level.
Map and Transform methods assume an origin at the top-left corner of the view, like iOS, Android, and GLFW but unlike OS X. Transform is responsible for flipping coordinates between the top-left origin of Map and the bottom-left origin of TransformState.
Fixes#3574.
Avoid unrotating due to programmatic rotation, even programmatic rotation that might expose empty portions of the map. Only snap to north due to a gesture.
All Transform methods that take a PrecisionPoint now assume a “flipped” origin at the upper-left corner of the view. Previously, some methods assumed an origin at the lower-left corner.
I regenerated assets.zip so that all file paths have an `assets/` prefix, as the Android AssetFileSource implementation asserts, and removed `TEST_DATA` from the paths.
- Deprecate and no-op `+[MGLAccountManager mapboxMetricsEnabledSettingShownInApp]`
- Check for attribution button hiding and make sure ToS are understood
CADisplayLink holds a strong reference to its target, forming a cycle that must be broken with -[CADisplayLink invalidate] when the animation is complete. I don’t yet have enough faith that will-change and did-change notifications are always coming from mbgl in pairs, so this change limits CADisplayLink to when MGLMapView is in the view hierarchy. It also pauses the CADisplayLink when the view is hidden or the application is in the background. Finally, -[MGLMapView invalidate] has been renamed because that term tends not to mean “redraw” in Cocoa but is rather tied to timers.
Fixes#3130.
[ios] Also invalidate CADisplayLink on removal from window
[ios] Also shut down CADisplayLink when view is hidden
Renamed MGLCalloutView to MGLCompactCalloutView and MGLCalloutViewProtocol to MGLCalloutView to avoid an awkward workaround for a Swift name collision. Replaced individual title and subtitle properties with a single representedObject property that lets you use custom annotation properties in the custom callout view. Overrode a problematic SMCalloutView method. Added lots more documentation.
