| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
* [ios] The new location manager API provides two new protocols MGLLocationManager and MGLLocationManagerDelegate to handle the location cycle and updates respectively. This enables developers to chose the appropriate location provider according to their needs, or transition between outdoors/indoors location updates.
It does provide a default implementation based on CLLocationManager.
|
| |
|
|
| |
We can't run the checks without potentially exposing the Github App private key to malicious PRs.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Removes mgl_expressionByReplacingTokensWithKeyPaths and associated code. Converting on output is no longer necessary: from the prior commit, core converts token strings to expressions at parse time; all that's necessary is to ensure that the runtime styling API does so as well.
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
unmodified
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
| |
Fixed a crash that occurred when the user started a gesture before the drift animation for a previous gesture was complete.
|
| |
|
|
|
|
|
| |
`runnable->iter = runnables.end()` assumes that `::end()` is a
`static` or `constxpr` and thread-safe. This is likely to be true
for most of the implementations. That said we cannot rely on
undefined behavior and we should assume the worse case scenario.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
all point arrays as rings. Simplify the logic to only compare against
the next pt in the bound.
|
| |
|
|
| |
It is already wrapped with #ifdef __cplusplus extern "C", so we can safely compile it as C++ but retain C linkage
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Save a fresh ccache on every build by including (a checksum of) the current revision in the cache key. In the worst case, saving the cache appears to take ~1:30 for android-release; other builds appear to take a small amount of time compared to the time saved by having a fresh cache, versus a cache that's up to a week old.
Restore caches in the following priority order (all scoped to current architecture and job):
1. The cache for the current branch and revision.
2. If that doesn't exist, the most recent cache for the current branch.
3. If that doesn't exist, the cache for the parent branch and merge base, where "parent branch" means the base branch of the pull request specified by $CIRCLE_PULL_REQUEST (obtained from the GitHub API), and "merge base" means the revision returned by `git merge-base` when given the parent branch and $CIRCLE_BRANCH (obtained by executing that shell command). The intent here is to start a newly created branch off with the most relevant cache from its parent branch.
4. If that doesn't exist, the most recent cache for the parent branch.
5. If that doesn't exist, the most recent cache for master.
6. If that doesn't exist, use the most recent cache for any build.
We cannot use the branch names or revisions themselves as key segments, because CircleCI lacks a way to interpolate arbitrary environment variables into the cache key name. Instead, we write the desired value into a file and use the `{{ checksum "filename" }}` syntax.
|
| | |
|
| |
|
|
| |
So its checksum can be used in a CI cache key name.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
canceled
Reverts d5aff71 (#11948) and implements an alternative solution.
Even with the previous fix applied, we were calling `renderFinished()` directly when the render was canceled. This led to the following scenario (all on the main thread):
- `Map::renderStill()` callback is called, `uv_async_send` schedules a call to `renderFinished()`.
- `NodeMap::cancel()` destroys the map object, and calls `renderFinished()` directly, which cleans up the request
- `NodeMap::startRender()` is called to schedule a _new_ render, which means that a new callback will be set
- The `uv_async_t` fires and calls `renderFinished` again:
- We don't return early as intended by the fix in #11948 since a new callback/`req` object is now available.
- Since neither error nor image has been set, we'll see a `Didn't get an image` error message.
- We call the callback with the error. In scenario where we're reusing Map objects, we will start a new render, which will now trigger the `Map is currently rendering an image` error in `Map::renderStill`, and the map object enters a circle in which it will now fail repeatedly with this error.
This commit fixes the issue by completely removing the existing `uv_async_t` handle that could have a scheduled call, and replacing it with a fresh instance of an `uv_async_t` handle.
|
| |
|
|
|
|
| |
Use a single template function for each of {to,from}ExpressionValue, always dispatching to ValueConverter.
Remove unused ValueConverter::ExpressionType typedefs.
|
| |
|
|
| |
used to determine if an asynchrous result should be set on map.
|
| | |
|
| |
|
|
|
| |
CircleCI’s cron jobs run on UTC, while their containers are set to US-Pacific. This change pushes back the `nightly` workflow from 12AM UTC (5PM PDT) to 5AM UTC (10PM PDT).
/cc @julianrex @fabian-guerra
|
