Bug#1071752: satpy: arm64 fails netCDF4 tests with scipy 1.12

[Antonio Valentino]

Dear Drew,

On Fri, 24 May 2024 17:16:07 +0200 Drew Parsons <dparsons at debian.org> wrote:
> Source: satpy
> Version: 0.48.0-2
> Severity: normal
> 
> scipy 1.12 is triggering test failure on arm64
> 
> https://ci.debian.net/packages/s/satpy/unstable/arm64/46960923/
> https://ci.debian.net/packages/s/satpy/unstable/arm64/
> 
> amd64 is passing without error.
> 
> The errors seem to come from _netCDF4.pyx
> Does it just need a rebuild?
> 
> e.g.
> 1148s _____________________ ERROR at setup of test_bad_area_name _____________________
> 1148s 
> 1148s tmp_path_factory = TempPathFactory(_given_basetemp=None, _trace=<pluggy._tracing.TagTracerSub object at 0xffff6d774b50>, _basetemp=PosixPath('/tmp/pytest-of-debci/pytest-1'), _retention_count=3, _retention_policy='all')
> 1148s 
> 1148s     @pytest.fixture(scope="session")
> 1148s     def himl2_filename_bad(tmp_path_factory):
> 1148s         """Create a fake himawari l2 file."""
> 1148s         fname = f'{tmp_path_factory.mktemp("data")}/AHI-CMSK_v1r1_h09_s202308240540213_e202308240549407_c202308240557548.nc'
> 1148s         ds = xr.Dataset({"CloudMask": (["Rows", "Columns"], clmk_data)},
> 1148s                         coords={"Latitude": (["Rows", "Columns"], lat_data),
> 1148s                                 "Longitude": (["Rows", "Columns"], lon_data)},
> 1148s                         attrs=badarea_attrs)
> 1148s >       ds.to_netcdf(fname)
> 1148s 
> 1148s /usr/lib/python3/dist-packages/satpy/tests/reader_tests/test_ahi_l2_nc.py:64: 
> 1148s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
> 1148s /usr/lib/python3/dist-packages/xarray/core/dataset.py:2298: in to_netcdf
> 1148s     return to_netcdf(  # type: ignore  # mypy cannot resolve the overloads:(
> 1148s /usr/lib/python3/dist-packages/xarray/backends/api.py:1339: in to_netcdf
> 1148s     dump_to_store(
> 1148s /usr/lib/python3/dist-packages/xarray/backends/api.py:1386: in dump_to_store
> 1148s     store.store(variables, attrs, check_encoding, writer, unlimited_dims=unlimited_dims)
> 1148s /usr/lib/python3/dist-packages/xarray/backends/common.py:397: in store
> 1148s     self.set_variables(
> 1148s /usr/lib/python3/dist-packages/xarray/backends/common.py:439: in set_variables
> 1148s     writer.add(source, target)
> 1148s /usr/lib/python3/dist-packages/xarray/backends/common.py:284: in add
> 1148s     target[...] = source
> 1148s /usr/lib/python3/dist-packages/xarray/backends/netCDF4_.py:80: in __setitem__
> 1148s     data[key] = value
> 1148s src/netCDF4/_netCDF4.pyx:5519: in netCDF4._netCDF4.Variable.__setitem__
> 1148s     ???
> 1148s src/netCDF4/_netCDF4.pyx:5802: in netCDF4._netCDF4.Variable._put
> 1148s     ???
> 1148s _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
> 1148s 
> 1148s >   ???
> 1148s E   RuntimeError: NetCDF: HDF error
> 1148s 
> 1148s src/netCDF4/_netCDF4.pyx:2034: RuntimeError
> 
> 
> Later there are other errors, "OSError: [Errno 28] No space left on device"
> 


I'm sorry for the very late reply.
Apparently it was a transitory issue because I see that now we have 
scipy 1.13 in unstable and testing and, apparently, debci for arm64 is 
green.

I suggest to close the bug.
We can reopen if the issue pops up again.

kind regards
-- 
Antonio Valentino