Aggregation 'first' is only implemented for dask arrays when method='blockwise' but I can't change the method

[iuryt]

Hi,

Awesome package.
Maybe I didn't understand all the details about the different parallel algorithms, but I am trying to use func="first" with ,method="map-reduce" as in the following code

import xarray as xr
import numpy as np
from flox.xarray import xarray_reduce

n = int(10000)
dsi = xr.Dataset(
    dict(
        x = ("obs",np.random.randn(n)),
        y = ("obs",np.random.randint(1,20+1,n))
    )
).chunk(obs=100)

expected = np.unique(dsi.y)
group_min = xarray_reduce(dsi,dsi.y,expected_groups=expected,func="first",method="map-reduce").load()

And it returns:

NotImplementedError: Aggregation 'first' is only implemented for dask arrays when method='blockwise'.

Is this the expected behavior?

[dcherian]

Would nanfirst work for you?

Check the results though, first, last, nanfirst and nanlast are lightly tested

[iuryt]

Returns AttributeError: module 'numpy' has no attribute 'nanfirst'