esa-esdl/esdl-core

loading data via Cube.data.dataset() can fail

Closed this issue · 2 comments

dzelge commented

When loading a cube into an xarray Dataset using Cube.data.dataset(), the following error may happen:

OSError: Unable to read attribute (no appropriate function for conversion path)

I believe this is due to calling Xarray.open_mfdataset using the parameter engine='h5netcdf'. removing this parameter will fix the error. It seems that we ran into a version issue betweeb h5py and netCDF4.

kmuehlbauer commented

@dzelge Could you please retest with h5py 3.7.0 and h5netcdf 1.0.1? The underlying issue should be fixed upstream.

forman commented

Thanks @kmuehlbauer for the info!

As this issue is almost 4 years old, and the conda env does limit neither h5py nor h5netcdf versions, and @dzelge is no longer working for ESDL, I am closing this issue.

  • 👍1