Tags

2025.09.1

- Writers now support filename templates
- Bug fixes to support bounds on `xarray` datasets

2025.09.0

- various bug fixes for metadata, ordering of dag nodes, and reuse

2025.06.1

- support for python 3.13
- use `h5netcdf` engine by default when available
- improve consistency between netcdf engines between read/write operations

2025.06.0

Features
---------
- Custom loaders can now be specified in pipelines
- Pressure interpolation method added to processors (requires numba and scipy)
- `cell_area` method has been added
- simplified `dag` syntax in pipelines
- Pipelines now supports lists in `reuse` keyword
- Pipelines outputs can now be merged into a single file

Bug Fixes
----------
- `open_mfdataset` now logs and raises errors
- `metadata` fixes
- `chunks` fixes to support `xr.chunk` application
- Fixes for `name` issues when used in `dag` in pipelines

2025.04.1

- Add ability to specify input file names and directory structure

2025.04.0

- add distributed runner interface to CLI
- add metadata field to pipelines

2025.03.1

- add `cycle` interface to pipelines for computing yearly climatologies
- add `area_mean` to pipeline stages

2025.03.0

- ability to call arbitrary xarray functions using simplified syntax
- ability to set netCDF4 encoding in yaml configuration
- bug fix to ensure proper sequencing of variables when they appear in multiple stages

2025.02.0

- add 3 and 6 hour resampling stages
- add custom resampling for stages
- fix bug where renaming and some other operations were causing reuse from wrong stage

2025.01.3

- Improved merging of dask `chunks` when variable appears in multiple stages
- Ability to set constants in `setup` that can be reused in pipelines

2025.01.2

- fix bug where variable present in multiple pipelines can cause problems if `destination: null`
- fix bug CLI bug where `canproc-pipeline` was crashing on directory creation
- fix bug where `reuse` was not working on computed variables
- fix example pipeline in docs

2025.01.1

- simplify syntax in the yaml file remove computed tag
- fix `destination` keyword so it actually works
- allow `compute` and `dag` in the same stage
- store node for output when write occurs to avoid removal during dag optimization
- fix in DaskRunner to support using `dask.distributed`

2025.01.0

pipelines: allow mathematical operations in transform operations (add, scale, ...)

2024.12