masked median of empty dimension fails

a = np.ma.masked_array(np.array([], dtype=float, ndmin=3))
np.ma.median(a, axis=2)

fails with

IndexError: index -1 is out of bounds for axis 2 with size 0

As integer indexing of empty slices does not work.
Normal median slices instead and returns the mean of an empty slice, so nan.

What should masked median return? a masked nan to somewhat match the normal median or an empty array again?

Provide feedback