Closed jrycw closed 2 months ago
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 81.68%. Comparing base (
14ce2c5
) to head (4b377b8
).
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
I've added an issue describing removing numpy, in case you want to tackle it. But no worries if not!
I've added an issue describing removing numpy, in case you want to tackle it. But no worries if not!
296
May I inquire about the possibility of handling or representing something like nan
in the future without using numpy
?
Thanks for submitting this PR! I think we're likely going to try and remove numpy in the near future, so aren't looking to add it to functions like
_gt_min
. I flagged one block of code that would be nice to merge, though!If you're up for it, we'd gladly accept a separate PR that removes numpy, but I realize it's a fairly gnarly task. No worries if it's outside of what you're looking to do. I really appreciate all this time you've put into cleaning things up 😅.
Removing numpy
as a dependency is not a trivial task, and it may be beyond the scope of this PR. I've modified the PR to retain the block you marked. Additionally, there's a return None
statement in _check_any_na_in_list
, which I believe should be removed. As always, feel free to make further updates to the PR based on the team's judgment.
May I inquire about the possibility of handling or representing something like nan in the future without using numpy?
Yeah, we can use a combo of None
for nan and math.inf
for np.inf
(similar to how polars converts types in to_list()
)
This PR primarily addresses the following three issues:
numpy
._flatten_list
into a function that can recursively flatten the list. This feature may be unwanted, so feel free to reject this change._gt_first
and_gt_last
more general so they can accept any iterable as the parameter.