reduce() by multiple dims on groupby object

[tlogan2000]

MCVE Code Sample

# Your code here
import xarray as xr
import numpy as np
url = 'https://data.nodc.noaa.gov/thredds/dodsC/GCOS/monthly_five_degree/19810101-NODC-L3_GHRSST-SSTblend-GLOB_HadSST2-Monthly_FiveDeg_DayNitAvg_19810101_20071231-v01.7-fv01.0.nc'

ds = xr.open_dataset(url, chunks=dict(time=12))
# reduce() directly on dataArray - THIS IS OK
ds.analysed_sst.reduce(np.percentile, dim=('lat','lon'), q=0.5) # ok

# Group by example
rr = ds.analysed_sst.rolling(min_periods=1, center=True, time=5).construct("window")

g = rr.groupby("time.dayofyear")
print(g.dims)
test1d = g.reduce(np.percentile, dim=('time'), q=0.5) # ok
testall = g.reduce(np.percentile, dim=xr.ALL_DIMS, q=0.5) # ok

# .reduce() w/ 2dims  on grouby obj not working 
test2d = g.reduce(np.nanpercentile, dim=('time','window'), q=0.5)

Expected Output

reduced output performed over multiple dimensions (but not xr.ALL_DIMS) on a groupby object

Problem Description

Using .reduce() on a groupby object is only successful when given a single dimensions or by using xr.ALL_DIMS. I wish to apply a reduce on a subset of dims (last line of code above) but gives folowing error:

Traceback (most recent call last):
  File "/home/travis/.PyCharmCE2019.2/config/scratches/scratch_20.py", line 13, in <module>
    test = g.reduce(np.percentile, dim=('time','window'), q=0.5)
  File "/home/travis/.conda/envs/Xarray/lib/python3.7/site-packages/xarray/core/groupby.py", line 800, in reduce
    % (dim, self.dims)
ValueError: cannot reduce over dimension ('time', 'window'). expected either xarray.ALL_DIMS to reduce over all dimensions or one or more of ('time', 'lat', 'lon', 'window').

Note: Using reduce() on a subset of dims directly on a xr.DataArray seems fine (line 7).

Output of xr.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.7.4 (default, Aug 13 2019, 20:35:49) [GCC 7.3.0] python-bits: 64 OS: Linux OS-release: 4.15.0-65-generic machine: x86_64 processor: x86_64 byteorder: little LC_ALL: None LANG: en_CA.UTF-8 LOCALE: en_CA.UTF-8 libhdf5: 1.10.2 libnetcdf: 4.6.3 xarray: 0.14.0 pandas: 0.25.1 numpy: 1.17.2 scipy: 1.3.1 netCDF4: 1.5.2 pydap: None h5netcdf: None h5py: None Nio: None zarr: None cftime: 1.0.3.4 nc_time_axis: None PseudoNetCDF: None rasterio: None cfgrib: None iris: None bottleneck: 1.2.1 dask: 2.4.0 distributed: 2.4.0 matplotlib: 3.1.1 cartopy: None seaborn: None numbagg: None setuptools: 41.0.1 pip: 19.2.2 conda: None pytest: 5.0.1 IPython: None sphinx: None

[dcherian]

🤦‍♂️ I can't believe we don't test for this. Fix is incoming.

[tlogan2000]

Thx for the quick reply. For info this was working fine with v.0.13.0... it only is occuring since the 0.14.0 upgrade