When a SymbolLayer doesn't have a text-font defined, we automatically add Open Sans/Arial Unicode MS. However, when the SymbolLayer is only used for rendering icons, it doesn't have text-field defined either. In those cases, we still force downloading Open Sans/Arial Unicode MS during offline pack creation. If the user doesn't use this font, this change should save ~15MB and a few seconds in download time.
Core test uses stubbed "glyph.pbf" without Chinese glyphs
Darwin test relies on locally available "PingFang" font.
Android test relies on locally available "Droid" font. 'expected.png' is NOT correct b/c I haven't figured out how to run unit tests on Android yet.
This adds a `convertJSON` template, to be used like:
Error error
optional<Foo> foo = convertJSON<Foo>(string, error);
Internally, it parses the string with RapidJSON and then calls `convert<Foo>(parsed, error)`.
While here, rationalize GeoJSON converters and fix error handling for Tileset conversion in OfflineDownload.
This reverses #3095. Rationale:
* We're now exposing source attributes as a public API. Making those attributes unavailable at certain times complicates that API.
* We're preparing to split RenderSource out of Source. Removing this removes a point of coupling between the two.
Provide Image::copy, which handles copying rectangles from a source to a destination, with thorough bounds checking.
Also fixes an indexing error in SpriteAtlas, where the top row of pixels in a wrapped image was copied from the wrong source row.
* [core] Add interface to get image from sprite atlas
* [tests] Add tests for Map::getImage
* [ios, macos] WIP: get MGLImage for name from style
* [ios, macos] Fixed -imageForName:
Convert from sprite images to platform images using the existing encodePNG() function, which is also used for printing. Allow -imageForName: to return nil without an assertion failure. Added a basic test.
So far, we didn't properly disable layers that are outside the zoom range. This means that we rendered layers that should not have been rendered, albeit we didn't make any attempt to load tiles for those layers. However, when zooming in/out, existing tiles might already have been loaded in the source which continued to be rendered. In most cases they weren't actually visible because either the matrices weren't updated, or the clip IDs weren't set so that they would be "rendered" off-screen and clipped completely. In any case, we did way too much work.
