Repair CF writer tests failing with libnetcdf >= 4.9.0

[sfinkens]

CF writer tests fail with libnetcdf4 >= 4.9.0, because they use the compression keyword argument, which is ignored by xarray (pydata/xarray#7388). To fix this, tie usage of the compression keyword to the xarray version. There's already an open PR, so we can assume that this will be fixed in the next minor release (xarray >= 2023.03).

My bad, sorry I missed that in #2390 .

[djhoese]

I assume that once the xarray issue is fixed this will start failing? Can we check the netcdf version and the xarray version to have better control of this?

[sfinkens]

I guess they would first issue a deprecation warning before dropping support for the zlib keyword. netcdf4-python currently supports both.

[djhoese]

Could we then assume that this fix will be included in the next xarray release? I do see there is a PR started here: pydata/xarray#7551

If we assume the next xarray will pass all kwargs to libnetcdf, then could we have a check for the current xarray version and use zlib for it and every version before it and then for the next minor/feature release (assume not a bug fix release of xarray) we switch to the new kwarg (still dependent on libnetcdf)?

[codecov]

Codecov Report

Merging #2412 (41226ec) into main (1a785e5) will increase coverage by 0.01%.
The diff coverage is 99.59%.

@@            Coverage Diff             @@
##             main    #2412      +/-   ##
==========================================
+ Coverage   94.73%   94.74%   +0.01%     
==========================================
  Files         329      329              
  Lines       48733    48861     +128     
==========================================
+ Hits        46166    46295     +129     
+ Misses       2567     2566       -1     

Flag	Coverage Δ
behaviourtests	4.42% <4.89%> (+0.01%)	⬆️
unittests	95.37% <99.59%> (+0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
satpy/tests/writer_tests/test_cf.py	99.60% <98.70%> (-0.12%)	⬇️
satpy/composites/__init__.py	91.20% <100.00%> (+0.64%)	⬆️
satpy/tests/enhancement_tests/test_ahi.py	100.00% <100.00%> (ø)
satpy/tests/reader_tests/test_fy4_base.py	100.00% <100.00%> (ø)
satpy/tests/reader_tests/test_ghi_l1.py	100.00% <100.00%> (ø)
satpy/tests/reader_tests/test_msi_safe.py	100.00% <100.00%> (ø)
satpy/tests/reader_tests/test_msu_gsa_l1b.py	100.00% <100.00%> (ø)
...tpy/tests/reader_tests/test_oceancolorcci_l3_nc.py	100.00% <100.00%> (ø)
satpy/tests/test_composites.py	100.00% <100.00%> (ø)
satpy/tests/test_readers.py	98.89% <100.00%> (-0.01%)	⬇️
... and 1 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[sfinkens]

I don't know if that's a valid assumption :D But I think it's a good idea. If that PR doesn't make it into the next xarray release, our tests will remind us.