Type of DataArray.data for 0D different than nD

[BaptisteC137]

Is it the expected behavior to have a different type for DataArray.data depending on the dimension of the data ? For the moment if DataArray.data is of dimension 0 is returns a pint.Quantity and otherwise a numpy.array.

Small example:

test_data_array = xr.Dataset({'test': (( 'x', 'y'), np.random.random(10).reshape(10, 1))}, 
                              {'x': np.arange(10), 'y': np.arange(1)})['test']
test_data_array['y'].attrs['units'] = 'MHz'

dt_quantify_sq = test_data_array.squeeze().pint.quantify()
dt_quantify = test_data_array.pint.quantify()
print('squeezed y:', type(dt_quantify_sq.y.data), ' with shape:',dt_quantify_sq.y.shape)  # class 'pint.Quantity'
print('squeezed x:',type(dt_quantify_sq.x.data), ' with shape:', dt_quantify_sq.x.shape)  # class 'numpy.ndarray'
print('not squeezed y:', type(dt_quantify.y.data), ' with shape:', dt_quantify.y.shape)    # class 'numpy.ndarray'
print('not squeezed x:', type(dt_quantify.x.data), ' with shape:', dt_quantify.x.shape)    # class 'numpy.ndarray'

Could it be a good idea to have a DataArray.pint.values function that always returns a pint.Quantity ?

[keewis]

yeah, that's been traditionally an issue. However, this has changed on main to allow you to put units on "dimension coordinates" (or rather, indexed coordinates), we're just missing a release.

I've been delaying the release for quite a while now because I didn't get any feedback, but I think I should just cut the release now (i.e. sometime this week).