Closed jarq6c closed 1 year ago
Thanks for opening and documenting this, @jarq6c. I should have some bandwidth late into the afternoon to look into this and also checkout the PR's that i've neglected for longer that I would like to admit.
This looks like a pretty easy fix. We should just swap df.loc
for setting with df[]
. We need to set by replacement (copy) rather than in-place because are changing the data type of the underlying array when we downcast.
I'm glad to see that pandas is improving the consistency of it's copy vs view semantics.
When using the nwis iv client, I receive this warning: