PR to pandas

[tswast]

pandas devs are open to an experimental date/time dtype.

We'd still keep this package around to backport the types to older pandas.

[jorisvandenbossche]

For date, the most relevant issue discussing this on the pandas side is pandas-dev/pandas#32473

Now, I think there are still some aspects to discuss / decide upon on how we would like to approach the actual storage. For example, what you are doing (using a datetime64 with midnight at zero) also has advantages versus storing the number of days since epoch (as using datetime64[D] would be. Eg the conversion date <-> timestamp becomes much cheaper.

[tswast]

We did try datetime64[D] indeed it made the conversion to timestamp much more difficult. Also, I suspect adding time to date to get a timestamp will be a common operation, so datetime64[ns] was a better fit for that too.

[chalmerlowe]

Closing. Looking to deprecate this library at some point in the future by relying upon existing datatypes in pandas and python.