Feature or enhancement
Proposal:
Currently I am trying to add Android support for aiohttp and I am using cibuildwheel to build an test the wheel. For simplicity I am using the default managed Android emulator.
One missing piece, however, is measuring the code coverage of the tests. It is possible to track coverage using pytest-cov, but I haven’t yet found a way to copy the generated .coverage file from the managed Android emulator back to the host PC for further processing. Using adb pull doesn’t work, since the managed Android emulators shut down automatically after the tests. Of course, I could work with a connected Android emulator where the emulator stays on after the tests, allowing me to copy the necessary data using adb pull, but then I would have to manually start an Android emulator before calling cibuildwheel. I would prefer to avoid this extra effort.
Therefore, I suggest modifying the android.py script and the Android testbed to provide a generic way to copy data from the Android emulator back to the host. Here are a few ideas I have:
On idea is to copy the encode files as base64 and output them via logcat. This would be a simple but probably not very practical approach, especially if the files are large.
A propably better solution would be to use the (pretty bad documented) folders build/outputs/managed_device_android_test_additional_output and build/outputs/connected_android_test_additional_output to copy files from the emulator to the host. See for example here or here.
As far as I can tell, this feature would actually only be relevant for third-party packages built using cibuildwheel, for example. However, since cibuildwheel uses the android.py script and the Android testbed project from CPython, such a feature would need to be added to CPython and cannot be implemented within cibuildwheel itself.
I already have a working draft that uses managed_device_android_test_additional_output, but I wanted to discuss this first to see if it’s a useful addition or if it’s actually out of scope for CPython?
@mhsmith What do you think?
Has this already been discussed elsewhere?
No response given
Links to previous discussion of this feature:
No response
Feature or enhancement
Proposal:
Currently I am trying to add Android support for aiohttp and I am using cibuildwheel to build an test the wheel. For simplicity I am using the default managed Android emulator.
One missing piece, however, is measuring the code coverage of the tests. It is possible to track coverage using
pytest-cov, but I haven’t yet found a way to copy the generated.coveragefile from the managed Android emulator back to the host PC for further processing. Usingadb pulldoesn’t work, since the managed Android emulators shut down automatically after the tests. Of course, I could work with a connected Android emulator where the emulator stays on after the tests, allowing me to copy the necessary data usingadb pull, but then I would have to manually start an Android emulator before callingcibuildwheel. I would prefer to avoid this extra effort.Therefore, I suggest modifying the android.py script and the Android testbed to provide a generic way to copy data from the Android emulator back to the host. Here are a few ideas I have:
On idea is to copy the encode files as base64 and output them via logcat. This would be a simple but probably not very practical approach, especially if the files are large.
A propably better solution would be to use the (pretty bad documented) folders
build/outputs/managed_device_android_test_additional_outputandbuild/outputs/connected_android_test_additional_outputto copy files from the emulator to the host. See for example here or here.As far as I can tell, this feature would actually only be relevant for third-party packages built using cibuildwheel, for example. However, since cibuildwheel uses the
android.pyscript and the Android testbed project from CPython, such a feature would need to be added to CPython and cannot be implemented within cibuildwheel itself.I already have a working draft that uses
managed_device_android_test_additional_output, but I wanted to discuss this first to see if it’s a useful addition or if it’s actually out of scope for CPython?@mhsmith What do you think?
Has this already been discussed elsewhere?
No response given
Links to previous discussion of this feature:
No response