.item() on a DataArray with dtype='datetime64[ns]' returns int

[IvoCrnkovic]

MCVE Code Sample

import datetime
import xarray as xr

test_da = xr.DataArray(datetime.datetime(2019, 1, 1, 1, 1))

test_da
# <xarray.DataArray ()>
# array('2019-01-01T01:01:00.000000000', dtype='datetime64[ns]')

test_da.item()
# 1546304460000000000

Expected Output

I would think it would be nice to get a datetime out of the .item() call then the nanosecond representation.

Output of xr.show_versions()

When I call xr.show_versions() i get an error but im running xarray 0.12.3

[shoyer]

This is basically a direct manifestation of this NumPy bug: numpy/numpy#12550

We could potentially had a workaround in xarray but it would be good to figure out the NumPy fix first, so we can be consistent.

[dopplershift]

Well it'd be nice to have something, because what we have to do right now is:

times[0].values.astype('datetime64[ms]').astype('O')

which is an abomination and terrible for teaching.

[jthielen]

Would it make sense to add something to the dt accessor, such as_datetime? @dopplershift's example would still be somewhat verbose, but it seems more sensible:

times[0].dt.as_datetime().item()

[dopplershift]

Yeah, it's not much shorter, but it's a much easier concept for new users to grasp, IMO.