Is there a problem with codecov and -nightly?



So, nightly is in LightGraphs’ “allowed failures” list, and it’s been failing. We just fixed the last deprecation in LG so that nightly works (a single line change), and our code coverage dropped ~30%:

Reintroducing the nightly failure (and therefore only submitting codecov results from 0.6) brings us back up to 100%.

Is this a known issue? seems to indicate that codecov is trying to test blank lines and triple quotes.


I have submitted an issue:


I see the same problem.

Off topic: I thought the coverage was reported by the latest Julia release. Why is it from unstable nightly build? I think it would be better to use a stable version.


Ideally you would want to merge the coverage from all the test runs, because some code paths may only be active for a specific Julia version or platform.