Closed geordie666 closed 1 year ago
Before we merge this I should try to work out why coverage has dropped so much. I'm not sure how I could have reduced coverage by 11% with only a dozen or so changes, but I'll investigate.
I'm confused about the drop in coverage. Running pytest --cov
on Cori for this branch and for main
create coverage reports that differ by only three lines, corresponding to the three lines of code I removed from randoms.py. So, at NERSC I obtain coverage results that are completely aligned with my expectation.
I'm therefore going to assume that the drop in coverage from 56.222% to 44.728% was a glitch in coveralls.
@sbailey: I think I'm ready to merge this branch, unless you have further feedback regarding my comments about why nside_chunk
can't be array-like.
@geordie666, I would not get into the habit of blaming coveralls on this. I have found them to be highly reliable. The more likely explanation is that tests are being skipped in the GitHub Actions environment.
@weaverba137: Sure, agreed. I used sloppy language. I should have stated "a glitch or change in something leading up to and including the coveralls step."
This PR addresses #790 (as well as fixing another couple of minor issues).
Sorry this took so long, I was expecting to need to update many occurrences of
np.int
andnp.float
and hadn't realized there were only a few.TMI: The code I ran to check my casting choices were correct for the specific examples mentioned in #790 was:
and