Kai Mühlbauer

How is this handled in the light of this comment https://github.com/pydata/xarray/issues/9282#issuecomment-2310573129 by @dcherian?

Hi Justus @keewis , thanks for raising this. This looks like some improper encoding in the file: The string is encoded as `ascii` as we see from the `h5dump` ```...

@keewis As it seems the data is already converted from some netcdf3 format to netcdf4. If it is possible for the data provider to switch to `UTF-8` for the string...

Thanks @bnlawrence for raising this and sorry for the delay. This isn't raised in the current testsuite. Is this still an issue for you or can you give more details...

we are testing: ``` matrix: python-version: ["3.10", "3.12"] h5py-version: ["3.10", "3.11"] numpy-version: ["1"] ``` ``` matrix: python-version: ["3.11", "3.12"] h5py-version: ["3.11"] numpy-version: ["2"] ```

@drew-parsons Please have a look at: https://github.com/h5netcdf/h5netcdf/actions/runs/11743663241/job/32717024125?pr=242 There is a fresh run with the current matrix. No issue there.

Thanks @drew-parsons. I'm having a hunch that these testcontext construction might not work as expected. I'm afk, but will check this over the weekend. If you find the time you...

@drew-parsons Thanks for sharing the stack trace. Here is the stacktrace I'm getting: ```python tmp_local_or_remote_netcdf = '/tmp/pytest-of-kai/pytest-13/test_user_type_errors_new_api_0/testfile.nc' def test_user_type_errors_new_api(tmp_local_or_remote_netcdf): enum_dict1 = dict(one=1, two=2, three=3, missing=254) enum_dict2 = dict(one=0, two=2, three=3,...

Closed by #244. Thanks for sticking to this until now, @drew-parsons.

@bnlawrence This was introduced quite some time ago in #17 in connection with lazy evaluation. In consequence this has been tripping up people, see #195 and #196, #197, #221. But...