dataClay 3.x
============

3.0 (May 2023)
--------------

Highlights
~~~~~~~~~~

- Support to Python 3 built-in type annotations.
- New, simpler, and more efficient class registration process.
- Enabling multiple client sessions.

Other changes
~~~~~~~~~~~~~

- Complete refactoring featuring metadata distribution to avoid performance bottlenecks.
- Migrated Java services and runtime to Python codebase.
- Optimized object serialization performance.
- New :meth:`~dataclay.DataClayObject.sync` method to synchronize object metdata.
- Support to installation with Spack.
- Support to orchestration with Ansible.
- Support to tracing with OpenTelemetry.
- Other minor optimizations and bug fixes.

3.1 (November 2023)
-------------------

Highlights
~~~~~~~~~~

- Support to Python versions from 3.9 to 3.12, with the corresponding images available.
- Support to metadata annotations through ``typing.Annotated``.
- Annotation :class:`~dataclay.contrib.nvm.InNVM` to store the values of annotated attributes in Non-Volatile Memory devices.
- Annotation :class:`~dataclay.annotated.LocalOnly` to avoid serialization of annotated attributes.

Other changes
~~~~~~~~~~~~~

- Support for Redis Cluster infrastructure for metadata.
- Improved user-defined serialization of objects.
- Fixes and improvements for multi-threaded and/or distributed applications.
- Improved settings management and sanity checks during initialization (backend and client).
- Other minor optimizations and bug fixes.