For distinguishing services of the same category in the configuration file.

Still need to support separate IDs for multiple publisher configurations in
the same file, as we...

By using the loaded configuration in the argument parser defaults, so that
command line options ...

And have the broker use that feature.

And have the broker use that feature.

It uses either the builtin `tomllib` or the external `tomli` according to the
Python version, so...

Since project configuration points to that function.

This removes the partial downloads feature. Partial downloads can have different bad effects:

...

This adds the `caterva2.tools.hdf5_to_cat2` script to help export an existing
HDF5 file to a new...

Decouple deps

The old code failed to get a proper dtype from metadata to build the
uninitialized dataset.

All...

But without strict type checking, so that we can handle subclasses. Then set
the packed result ...

There may be attributes of other types which Blosc2 should be able to
serialize with msgpack int...

This release fixes the lack of optimization support when iterating over
`b2h5py.B2Dataset` insta...

And adapt the related tests. Also remove the readme notice about the
operation not being optimi...

Only bytes values are supported for the moment.

Add dependency on blosc2 for the tools extra.

And depend on it for the tools extra.

Accessing the HDF5 file depends on h5py, added tool dependency.

Although this may be completely dropped later on.

So as to decouple reporting from the function itself.

And define tqdm-based reporter right in c...

Clients are the only users, and it avoids some extra dependencies in services.

As many will be used both by services and clients.

By explicitly checking for boolean in JSON object without going through schema
validation.

Which should be moved to services.

This release adds a new way to use Blosc2 optimized slicing explicitly on a
single dataset. You...