createVariable: KeyError: 't' (works with netCDF4)

[nschloe]

MWE:

import h5netcdf.legacyapi as netCDF4
# import netCDF4  # works


with netCDF4.Dataset("out.e", "w") as rootgrp:
    rootgrp.createDimension("time_step", None)
    data = rootgrp.createVariable("time_whole", "f4", "time_step")
    data[:] = 0.0

Traceback (most recent call last):
  File "b.py", line 7, in <module>
    data = rootgrp.createVariable("time_whole", "f4", "time_step")
  File "/home/nschloe/.local/lib/python3.8/site-packages/h5netcdf/legacyapi.py", line 91, in createVariable
    return super(Group, self).create_variable(
  File "/home/nschloe/.local/lib/python3.8/site-packages/h5netcdf/core.py", line 500, in create_variable
    return group._create_child_variable(keys[-1], dimensions, dtype, data,
  File "/home/nschloe/.local/lib/python3.8/site-packages/h5netcdf/core.py", line 463, in _create_child_variable
    shape = tuple(self._current_dim_sizes[d] for d in dimensions)
  File "/home/nschloe/.local/lib/python3.8/site-packages/h5netcdf/core.py", line 463, in <genexpr>
    shape = tuple(self._current_dim_sizes[d] for d in dimensions)
  File "/usr/lib/python3.8/collections/__init__.py", line 891, in __getitem__
    return self.__missing__(key)            # support subclasses that define __missing__
  File "/usr/lib/python3.8/collections/__init__.py", line 883, in __missing__
    raise KeyError(key)
KeyError: 't'

[shoyer]

You’re setting dimensions=“time_step”. NetCDF4-Python apparently interprets this as a list of one string whereas h5netcdf interprets it as a sequence of characters.

I would argue that this is probably a programming error, and it would be best to require users to explicitly pass a list or tuple of strings.

If you or anyone else is interested, I would be happy to accept a pull request making this behavior raise TypeError.

[nschloe]

I would argue that this is probably a programming error,

Agreed. netcdf should be stricter here.