Skip to content

1.1.0

Choose a tag to compare

View all tags
@pont-us pont-us released this 21 May 13:34
· 132 commits to main since this release
v1.1.0
6935103

  • Ensure attributes are maintained when missing dimensions are added to
    concatenation dimension variable. (#32)

  • Added some basic batch utilities that help spawning multiple concurrent
    nc2zarr jobs, see new module nc2zarr.batch. (#19)

  • Local input and output paths may now include tilde '~' which will expand
    to the current user's home, and '~username' for the home directory
    of a specified user. (#26)

  • Fixed a problem when input/sort_by was "name" and one of input/paths
    ended with "/". In these cases sorting did not work. (#29)

  • Fixed problem where appending unnecessarily required a coordinate
    variable (#27)

  • Input path that were no wildcards have been ignored if the input path did
    not exist. Now an error is raised in such cases. (#25)

  • Fixed exception TypeError: 'int' object is not iterable
    raised with some process/rechunk configurations. (#23)

  • You can now provide your Python code for customization
    of the datasets read and written. (#16)

    • input/custom_preprocessor: "module:function" is called on
      each input after optional variable selection.
    • process/custom_processor: "module:function" is called after
      optional renaming and before optional rechunking.
    • output/custom_postprocessor: "module:function" is called before
      the final dataset is written. (#21)

    All three functions are expected to receive an xarray.Dataset object
    as only argument and return the same or modified xarray.Dataset object.
    Note: to let Python import "module" that is not a user package,
    you can extend the PYTHONPATH environment variable before
    calling nc2zarr:

    export PYTHONPATH=${PYTHONPATH}:path/to/my/modules
nc2zarr ...

  • Add more example configurations and scripts.

Assets 2
Loading