-
Notifications
You must be signed in to change notification settings - Fork 2
Description
Describe the bug
When generating a dataset using the csg we can use different categories for different variables. This leads to None values in the processing info for category - variable combinations that are not used in the csg. ds.pr.to_netcdf can't handle this partial processing info as the ds.pr.to_netcdf routine expects that all objects have a serialize method. The code fails in
_data_format.py
Failing Test
I added a failing test in the processing_info branch. A real life example is the regression test for the csg where currently processing info has to be removed for the test to work (branch csg_regression_test).
Expected behavior
Saving netcdf files should also be possible with processing info that contains None values
Screenshots
If applicable, add screenshots to help explain your problem.
System (please complete the following information):
- OS: Linux
- Python version 3.12
Additional context
Solve together with #280