0.10.2

Changes in 0.10.2

Enhancements

• Added new module xcube.core.subsampling for function
  subsample_dataset(dataset, step) that is now used by default
  to generate the datasets level of multi-level datasets.

• Added new setting Authentication.IsRequired to the xcube serve
  configuration. If set to true, xcube Server will reject unauthorized
  dataset requests by returning HTTP code 401.

• For authorized clients, the xcube Web API provided by xcube serve
  now allows granted scopes to contain wildcard characters *, **,
  and ?. This is useful to give access to groups of datasets, e.g.
  the scope read:dataset:*/S2-*.zarr permits access to any Zarr
  dataset in a subdirectory of the configured data stores and
  whose name starts with "S2-". (#632)

• xcube serve used to shut down with an error message
  if it encountered datasets it could not open. New behaviour
  is to emit a warning and ignore such datasets. (#630)

• Introduced helper function add_spatial_ref()
  of package xcube.core.gridmapping.cfconv that allows
  adding a spatial coordinate reference system to an existing
  Zarr dataset. (#629)

• Support for multi-level datasets has been improved:

  • Introduced new parameters for writing multi-level datasets with the
    "file", "s3", and "memory" data stores (#617). They are
    • base_dataset_id: If given, the base dataset will be linked only
      with the value of base_dataset_id, instead of being copied as-is.
      This can save large amounts of storage space.
    • tile_size: If given, it forces the spatial dimensions to be
      chunked accordingly. tile_size can be a positive integer
      or a pair of positive integers.
    • num_levels: If given, restricts the number of resolution levels
      to the given value. Must be a positive integer to be effective.
  • Added a new example notebook
    5_multi_level_datasets.ipynb
    that demonstrates writing and opening multi-level datasets with the
    xcube filesystem data stores.
  • Specified xcube Multi-Resolution Datasets
    definition and format.

• xcube gen2 returns more expressive error messages.

Fixes

• Fixed problem where the dataset levels of multi-level datasets were
  written without spatial coordinate reference system. In fact,
  only spatial variables were written. (#646)

• Fixed problem where xcube Server instances that required
  user authentication published datasets and variables for
  unauthorised users.

• Fixed FsDataAccessor.write_data() implementations,
  which now always return the passed in data_id. (#623)

• Fixes an issue where some datasets seemed to be shifted in the
  y-(latitude-) direction and were misplaced on maps whose tiles
  are served by xcube serve. Images with ascending y-values are
  now tiled correctly. (#626)

Other

• The xcube level CLI tool has been rewritten from scratch to make use
  of xcube filesystem data stores. (#617)

• Deprecated numerous classes and functions around multi-level datasets.
  The non-deprecated functions and classes of xcube.core.mldataset should
  be used instead along with the xcube filesystem data stores for
  multi-level dataset i/o. (#516)

  • Deprecated all functions of the xcube.core.level module
    • compute_levels()
    • read_levels()
    • write_levels()
  • Deprecated numerous classes and functions of the xcube.core.mldataset
    module
    • FileStorageMultiLevelDataset
    • ObjectStorageMultiLevelDataset
    • open_ml_dataset()
    • open_ml_dataset_from_object_storage()
    • open_ml_dataset_from_local_fs()
    • write_levels()

• Added packages python-blosc and lz4 to the xcube Python environment
  for better support of Dask distributed and the Dask service
  Coiled.

• Replace the dependency on the rfc3339-validator PyPI package with a
  dependency on its recently created conda-forge package.

• Remove unneeded dependency on the no longer used strict-rfc3339 package.

Full Changelog: v0.10.1...v0.10.2