[BUG] Fatal error in xarray.max() function after installing bottleneck on Win64

[RubendeBruin]

Describe the bug
When working with xarray, I get a fatal error when running code after installing bottleneck. Without bottleneck the code runs fine.

To Reproduce

from collections import OrderedDict

import numpy as np
import xarray as xr

# Build some dataset
dirs = np.linspace(0,360, num=121)
freqs = np.linspace(0,4,num=192)
spec_data = np.random.random(size=(192,121))
data = [spec_data]

dims = ('time', 'freq', 'dir')
coords = OrderedDict()
coords['time'] = [0,]
coords['freq'] = freqs
coords['dir'] = dirs

print('constructing data-array')

xdata = xr.DataArray(
            data=data, coords=coords, dims=dims, name='Spec name',
        ).to_dataset()

print('transposing data-array')

print('getting max')

print(xdata.max())  # works fine

tdata = xdata.transpose(..., "freq")

print('getting max')

print(tdata.max())  # <==== Process finished with exit code -1073741819 (0xC0000005)
print('done!')

Using win64, this code runs fine in an environment without bottleneck:

name: ws
dependencies:
  - python=3.9
  - pandas=1.2.4   # pinned to older version without bottleneck requirement
  - xarray
channels:
  - defaults
  - conda-forge

resulting in:

# Name                    Version                   Build  Channel
ca-certificates           2021.10.26           haa95532_2
importlib-metadata        4.8.2            py39hcbf5309_0    conda-forge
importlib_metadata        4.8.2                hd8ed1ab_0    conda-forge
intel-openmp              2021.4.0          haa95532_3556
libblas                   3.9.0              12_win64_mkl    conda-forge
libcblas                  3.9.0              12_win64_mkl    conda-forge
liblapack                 3.9.0              12_win64_mkl    conda-forge
mkl                       2021.4.0           h0e2418a_729    conda-forge
numpy                     1.21.4           py39h6635163_0    conda-forge
openssl                   1.1.1l               h2bbff1b_0
pandas                    1.2.4            py39hf11a4ad_0
pip                       21.3.1             pyhd8ed1ab_0    conda-forge
python                    3.9.7                h6244533_1
python-dateutil           2.8.2              pyhd3eb1b0_0
python_abi                3.9                      2_cp39    conda-forge
pytz                      2021.3             pyhd3eb1b0_0
setuptools                59.1.1           py39hcbf5309_0    conda-forge
six                       1.16.0             pyhd3eb1b0_0
sqlite                    3.36.0               h2bbff1b_0
tbb                       2021.4.0             h59b6b97_0
typing_extensions         4.0.0              pyha770c72_0    conda-forge
tzdata                    2021e                hda174b7_0
ucrt                      10.0.20348.0         h57928b3_0    conda-forge
vc                        14.2                 h21ff451_1
vs2015_runtime            14.29.30037          h902a5da_5    conda-forge
wheel                     0.37.0             pyhd3eb1b0_1
xarray                    0.20.1             pyhd8ed1ab_0    conda-forge
zipp                      3.6.0              pyhd3eb1b0_0

When adding bottleneck to the environment, either by doing mamba install bottleneck or by adding it to the yml file the script terminates on getting max().

name: ws
dependencies:
  - python=3.9
  - pandas=1.2.4
  - xarray
  - bottleneck
channels:
  - defaults
  - conda-forge

resulting in:

# Name                    Version                   Build  Channel
bottleneck                1.3.2            py39h7cc1a96_1
ca-certificates           2021.10.26           haa95532_2
importlib-metadata        4.8.2            py39hcbf5309_0    conda-forge
importlib_metadata        4.8.2                hd8ed1ab_0    conda-forge
intel-openmp              2021.4.0          haa95532_3556
libblas                   3.9.0              12_win64_mkl    conda-forge
libcblas                  3.9.0              12_win64_mkl    conda-forge
liblapack                 3.9.0              12_win64_mkl    conda-forge
mkl                       2021.4.0           h0e2418a_729    conda-forge
numpy                     1.21.4           py39h6635163_0    conda-forge
openssl                   1.1.1l               h2bbff1b_0
pandas                    1.2.4            py39hf11a4ad_0
pip                       21.3.1             pyhd8ed1ab_0    conda-forge
python                    3.9.7                h6244533_1
python-dateutil           2.8.2              pyhd3eb1b0_0
python_abi                3.9                      2_cp39    conda-forge
pytz                      2021.3             pyhd3eb1b0_0
setuptools                59.1.1           py39hcbf5309_0    conda-forge
six                       1.16.0             pyhd3eb1b0_0
sqlite                    3.36.0               h2bbff1b_0
tbb                       2021.4.0             h59b6b97_0
typing_extensions         4.0.0              pyha770c72_0    conda-forge
tzdata                    2021e                hda174b7_0
ucrt                      10.0.20348.0         h57928b3_0    conda-forge
vc                        14.2                 h21ff451_1
vs2015_runtime            14.29.30037          h902a5da_5    conda-forge
wheel                     0.37.0             pyhd3eb1b0_1
xarray                    0.20.1             pyhd8ed1ab_0    conda-forge
zipp                      3.6.0              pyhd3eb1b0_0

So really the only change is

bottleneck                1.3.2            py39h7cc1a96_1

Expected behavior
Script should not crash

[RubendeBruin]

Not sure if this is a bottleneck or a xarray error. Created a somewhat more detailed issue here: pydata/xarray#6002

[RubendeBruin]

Superseeded by #393