Is there a good way to measure how much test coverage test/run_test.py provides?
It seems someone asked something similar and got no response here: How to get a full coverage report of tests?
Is there a good way to measure how much test coverage test/run_test.py provides?
It seems someone asked something similar and got no response here: How to get a full coverage report of tests?
@ptrblck, you seem to be the go-to person for this – do you know where I can start?
I don’t know a good answer right now, but the idea sounds interesting.
Could you point me to some references, how the coverage is tested in other projects?
I know, that torchvision uses codecov for the PRs, but I’m unsure if you are looking for this approach or another one.
I tried to use coverage tool to get test coverage for torch source. And got 19% coverage which seems to be very low. Just asked question: PyTorch low test coverage
@ptrblck Do you know who might be able to help with this?
Let’s add @ezyang to it, as he might be able to give more insight or knows whom to ping.