BUG: fillna() doesn't work with `value=None` without method specified, but method is deprecated about pandas HOT 6 OPEN

I believe this issue existed prior to the deprecation: it was never possible to just .fillna(value=None). And doing .fillna(value=None, method="ffill") would ignore the value parameter (since forward filling does not utilize a value). When the deprecation is enforced, we can make the value parameter required (instead of having a default), and then enable users to use .fillna(value=None).

from pandas.

Ref: #47639

from pandas.

I believe this issue existed prior to the deprecation: it was never possible to just .fillna(value=None). And doing .fillna(value=None, method="ffill") would ignore the value parameter (since forward filling does not utilize a value). When the deprecation is enforced, we can make the value parameter required (instead of having a default), and then enable users to use .fillna(value=None).

So should we update the docs now to reflect that Series.fillna(value=None) doesn't work? Because the way it reads now, it appears that it should work. I think that is what led to the confusion of the OP in pandas-dev/pandas-stubs#884

from pandas.

So should we update the docs now to reflect that Series.fillna(value=None) doesn't work?

I would vote no, but not opposed if someone wants to do that. This deficiency in the docs has existed for a long time, and I expect it to work come 3.0, so it doesn't seem worth the effort to change now.

from pandas.

agree with @rhshadrach

from pandas.

I have a PR in the works for enforcing said deprecation, I'll add a test for this and use that PR to close this issue.

from pandas.