"does that work on your end?"

yes it does. Will remove the xfail and then we can merge once https://github.com/pydata/bottleneck/pull/382 is merged.

I've added xfail , thanks for the link.

Hi @max-sixty , what is a xfail ?

Also, the issue turned out to be bottleneck-bug , ref: https://github.com/pydata/bottleneck/issues/393#issuecomment-978017397 (fix available)

Ok, then it is clearly a bottleneck/numpy issue. I will raise it there and close it here.

Thanks!

I can reproduce it with calling bn.nanmax directly, but I can not reproduce it without the xarray.transpose() function.

• If I call nanmax on the internal data of xarray then nanmax fails with a segfault:

python np_data = xdata['Spec name'].data bn.nanmax(np_data) # Segfault - But if I create a copy of that data and then call nanmax then it works fine.

python np_data = xdata['Spec name'].data new_data = np_data.copy() bn.nanmax(new_data) # works

I suspect that the xarray.transpose function does something with the data-structure (lazy reshuffling of dimensions?) that triggers the fault in bottleneck.

Full code:

```python from collections import OrderedDict import numpy as np import xarray as xr

xr.show_versions()

n_time = 1 # 1 : Fails, 2 : everything is fine

from xarray.core.options import OPTIONS OPTIONS["use_bottleneck"] = True # Set to False for work-around

Build some dataset

dirs = np.linspace(0,360, num=121) freqs = np.linspace(0,4,num=192) spec_data = np.random.random(size=(n_time,192,121))

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

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

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

import bottleneck as bn np_data = xdata['Spec name'].data

new_data = np_data.copy() bn.nanmax(new_data) # works

bn.nanmax(np_data) # Segfault print('direct bn call done') ```

tests on another machine (also win64) with the same result.

Running under WSL/Ubuntu results in a Segmentation Fault

Gets too messy - will clean up and re-open

Minimum conda environment to reproduce:

yml name: ws dependencies: - xarray - numba channels: - defaults - conda-forge resulting in: ```

Name Version Build Channel

blas 1.0 mkl bottleneck 1.3.2 py38h2a96729_1 ca-certificates 2021.10.26 haa95532_2 importlib-metadata 4.8.2 py38haa244fe_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 llvmlite 0.35.0 py38h34b8924_4 mkl 2021.4.0 h0e2418a_729 conda-forge mkl-service 2.4.0 py38h2bbff1b_0 numba 0.52.0 py38hf11a4ad_0 numexpr 2.7.3 py38hb80d3ca_1 numpy 1.21.4 py38h089cfbf_0 conda-forge openssl 1.1.1l h2bbff1b_0 pandas 1.3.4 py38h6214cd6_0 pip 21.3.1 pyhd8ed1ab_0 conda-forge python 3.8.12 h6244533_0 python-dateutil 2.8.2 pyhd3eb1b0_0 python_abi 3.8 2_cp38 conda-forge pytz 2021.3 pyhd3eb1b0_0 setuptools 59.1.1 py38haa244fe_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 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 zlib 1.2.11 h62dcd97_4 ```

So not sure if I should post the issue here on with numba

Thanks for the link to the tutorial!