-
Notifications
You must be signed in to change notification settings - Fork 1.3k
Remove mips and armeabi as supported ABIs #11458
Conversation
Yeah, we should never throw in destructors; it's undefined behavior. |
The
@lbud do you recall what the function of this comparison is intended to be? |
I'm putting this PR on hold til official release of NDK17 + CI image update mapbox/mbgl-ci-images#16. |
Working back on this after updating the mbgl-ci-image, addressed previous comments but now hitting:
These errors are similar to #11458 (comment) but are found in boost that we fetch from Mason. Any guidance how I can fix this? |
@tobrun To find out if there are any other issues with the new NDK, you can remove -Werror from CMAKE_CXX_FLAGS in CMakeLists.txt. This may unblock you from making progress. Some mapbox projects like vtzero allow pasisng -DWERROR=OFF for precisely these kind of experiments. Eventually we should add something similar that we can use in MapboxGLAndroidSDK/build.gradle. |
@tobrun we're using this pattern to silence the warnings from external libraries:
Technically, we should include them with |
platform/android/src/bitmap.cpp
Outdated
if (result != ANDROID_BITMAP_RESULT_SUCCESS) { | ||
throw std::runtime_error("bitmap decoding: could not unlock pixels"); | ||
} | ||
AndroidBitmap_unlockPixels(&env, jni::Unwrap(*bitmap)); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we replace this with a warning?
platform/android/src/run_loop.cpp
Outdated
if (close(fds[PIPE_IN]) || close(fds[PIPE_OUT])) { | ||
throw std::runtime_error("Failed to close file descriptor."); | ||
} | ||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why are these calls no longer necessary?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
they still are, replaced the throw std::runtime_error();
with logging an error.
@@ -102,9 +102,6 @@ void FillExtrusionBucket::addFeature(const GeometryTileFeature& feature, | |||
|
|||
const Point<double> perp = util::unit(util::perp(d1 - d2)); | |||
const auto dist = util::dist<int16_t>(d1, d2); | |||
if (dist > std::numeric_limits<int16_t>::max()) { | |||
edgeDistance = 0; | |||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Same here?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the compiler warned that this check was always false, didn't see an issue removing it.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Hmm yeah, but I think this means that we should change the underlying code that leads to dist
being a uint16_t
, since the overflow will be hidden there.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Restored the check and added silenced the warning for now to make it compile:
56ff8b1#diff-f07899604a0493652b354b7d538040b1R107
How would you like to proceed? ticketing this in a follow up ticket or fixing it in this PR?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
followup ticket 👍
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
3c525e4
to
78517b9
Compare
@kkaefer this PR is ready for another round of 👀, follow up ticket in #11909. |
compileSdkVersion: 25, | ||
buildToolsVersion: '26.0.3' | ||
targetSdkVersion : 26, | ||
compileSdkVersion: 26, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we bump compileSdkVersion
to the 27 while we are at it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good on my end 👍
[android] - bump CI image to NDK 17 compatible [core] - remove setting edgeDistance to 0, comparison 'const short' > 32767 is always false [android] - remove throwing in desructor, undefined behaviour [android] - bump dependency versions of project
Is there a way that I don't have to wait for this PR and still use NDK 17.x? |
Is there an ETA for this PR? |
Capturing from ndk-guides that:
This PR adds the required changes to build with NDK 17.
cc @mapbox/maps-android