Open jd-foster opened 2 years ago
How does this PR differ from #53? If the difference is small, why not add to #53?
Patch coverage: 100.00
% and project coverage change: +1.17
:tada:
Comparison is base (
65a5928
) 97.02% compared to head (b3f3f12
) 98.20%.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Do you have feedback about the report comment? Let us know in this issue.
The main difference is it doesn't use the mapfoldl
machinery; this is a style choice in some ways, since it wasn't clear to me how that worked, and might also be a challenge to other contributors.
Happy to make this a pull request to #53 instead, but want to run the check on this separately first.
This is an attempt at building from #53 (and addressing issue #31) that aims to follow the existing style of code and docstrings of NaNMath. I'm going for clarity in both rather than generality or performance; in fact, deliberately restricting dispatch to AbstractFloat types. A lot of credit due on this to @sethaxen since I've adapted his PR tests and docstrings.