Releases: cedadev/CFAPyX
Releases · cedadev/CFAPyX
2024.11.6: Bug Fixes
CFA: Create Functionality
The October 11th (2024.10.11) release includes the following feature additions:
General:
- Name change: The package
CFAPyX
has now officially becomecfapyx
in every instance of the python package. The documentation and github repository will continue to use the capitalised version, but all installations and code examples should now usecfapyx
on importing. In future the package may again change name to reflect the established use, and will most likely later becomecf-aggregate
- the official CF Aggregations package for CF 1.12. - Logging: A standardised logger has been added to the
cfapyx
package, with limited logging messages for the xarray engine but an extensive collection within thecreate
module. - Mutable Defaults:
cfapyx
is now free from the evils of mutable defaults in all instances. - Support for multiple locations: Fixed the method of determining trailing dimensions for the
location
parameter in CFA-netCDF files. - Poetry:
cfapyx
now uses Poetry to handle project dependencies within the pyproject.toml file. The requirements files are still present for the current version, but will be removed with the next minor update.
CFA Create:
- CFANetCDF class for constructing CF-compliant Aggregation files. This is in alpha release for the 2024.10.11 version, but should perform well with expected NetCDF structures including scalar dimensions and variables, and identical variables across multiple files.
2024.9.2: Chunks and Optimisations
- Added Create partitions method for FragmentArrayWrapper. Applies chunk structure to all fragments, instead of overlapping.
- Added Optimise chunks routine when provided special kwarg 'optimised' for a chunkable dimension.
Version 2024.8.13: Functional Update
Release Notes:
- Multiple Aggregations per CFA-netCDF file.
- Base substitutions in CFA-netCDF files now handled correctly.
- Groups within CFA-netCDF files handled using
GroupedDatasetWrapper
- Multiple trailing locations in
location
are handled properly. (Issue #3) - Partial documentation including all code samples and docstrings now live!
- Basic testing suite now added to main branch.
- Size-1 dimension handling in
partition
version 1.1 - Dask Compute now fully flexible - all sub-dask objects return a copy of themselves when requested (Issue #6)