Pandas ENH: Should pandas.CategoricalDtype use pandas.NA as the sentinel value instead of float('nan') ?

Almost all of the the new nullable data types (which are incredibly useful btw.) seem to use pd.NA as the null-value. However, I noticed that pandas.CategoricalDtype still uses float('nan'), which I suppose is for historical reasons.

Are there any plans - for the sake of consistency - to switch CategoricalDtype's sentinel value to pandas.NA in the future?

Comment From: jbrockmendel

Are there any plans - for the sake of consistency - to switch CategoricalDtype's sentinel value to pandas.NA in the future?

No. I think it's more likely we'll go the opposite direction with #29962

Comment From: phofl

closing in favour of #50711