* Main gyp files are now standardized as platform/<platform>/platform.gyp.
* Each platform gyp file defines appropriate loop_lib and headless_lib variables.
* Each platform gyp file includes mbgl.gypi, which defines base targets which may be useful to all platforms.
* CI targets are consistent across platforms: `make $(PLATFORM) && make test-$(PLATFORM)`.
* Renamed the "linux" test app to "glfw". It's now built in OS X CI.
* Android build flakiness is fixed.
* iOS CI builds the bench and iosapp targets.
* Mesa version is now in one place.
* CI scripts use bash "strict mode" and correct error handling.
* All build output goes to the build directory.
* Removed vestigial iOS/OS X/Android Travis scripts.
These tests are intended to reflect how someone using the custom layer feature might actually write their callbacks. Therefore the callbacks should not use mbgl-specific classes.
This reverts commit 69112a2a09.
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.
* Move asset:// URL handling to DefaultFileSource.
* AssetFileSource implements FileSource interface and follows familiar implementation patterns.
* Move default implementation to platform/default, zip implementation to platform/android.
* Don't bother with modified / expires / etag -- assets are not cached so it doesn't matter.
* Don't bother with interleaving individual IO calls on the implementation thread. That adds a lot of complexity for very little benefit.
It's a premultiplied image. This implies that we were misusing encodePNG
in most cases, as we were passing premultiplied pixels which were then
interpreted as unmultiplied. I changed encodePNG to accept premultipled
pixels, and unpremultiply in the implementations.
First, move style mutation code out of StyleParser and into AnnotationManager,
coalescing it with the mutation code for shape layers.
Second, allow AnnotationManager to keep track of stale tiles entirely
internally. There's no reason to pass sets of TileIDs around.
Third, correct the logic for invalidating the shape source. Since
AnnotationManager does not track shape invalidations on a tile-by-tile basis,
don't try to invalidate the shape source tile-by-tile.
Fixes#1675Fixes#2322Fixes#2095
Annotation tiles may become partially parsed just like regular tiles,
for example if a point annotation is added to the map before the style's
sprite has been loaded. In such cases, they need to be reparsed or the
annotation will not be rendered. Previously, the code path for reparsing
would be short-circuited by a dynamic_cast<VectorTileData*> followed by
a null check. This commit removes that case and adds (back) a virtual
reparse method to the TileData interface.
The callback will be called when any of the resources requested
from the network failed to load.
There is not recovery implemented yet, so the closest thing to
reseting the Map object and reloading all the resources would
be setting a new style.
