Resolwe

===================

Added
-----
- Add support for ``Python`` 3.10
- Add health checks support for deploy in Kubernetes
- Add ``collecttools_kubernetes`` management command
- Add ``COMMUNICATION_CONTAINER_LISTENER_CONNECTION`` to separate settings for
listener (where to bind to) and containers (where to connect to)
- Support ``docker_volume`` setting in connector config
- Support use of named volumes as processing or input volume in Docker executor
- Support SSL connection to Redis

Changed
-------
- **BACKWARD INCOMPATIBLE:** Require ``Django`` 3.2
- **BACKWARD INCOMPATIBLE:** Require ``Django Priority Batch`` version 4
- Do not prepare tools configmaps in Kubernetes workload connector
- Enable Docker containers to connect to the custom network
- Auto-delete completed jobs in Kubernetes after 5 minutes
- Optionally add affinity to the Kubernetes job
- Remove support for setting permissions using old syntax


===================

===================

Added
-----
- Add MD5 checksum to ``DataBrowseView`` view


===================

===================

Added
-----
- Support ``range`` parameter in fields of Python proces

Changed
-------
- Flush stdout/stderr on Python processes on every write

Fixed
-----
- Add attribute as a field to a ``resolwe.process.fields.GroupField`` in Python
process only if it is an instance of ``resolwe.process.fields.Field``.


===================

===================

Changed
-------
- Do not fail in case of missing files in ``UriResolverView``

Fixed
-----
- Remove references to temporary export files from the database and make sure
they are not created anymore
- Wrap ``move_to_collection`` in transaction and only call method if collection
has changed


===================

===================

Added
-----
- **BACKWARD INCOMPATIBLE:** New permission architecture: it is not based on
Guardian anymore. The main benefits of new architecture are speed gains in
common operations, such as setting a permission and retrieving objects with
the given permission.
- Allow overriding process resources in data object

Changed
-------
- Allow mounting connectors into pods as persistent volume claim instead of
volume of type ``hostPath``

Fixed
-----
- use the same connector inside pod to handle files and directories
- When data object was deleted listener did not receive the terminate message
and pod did not terminate immediatelly


===================

===================

Added
-----
- Add ``compare_models_and_csv`` management script to check if all
``ReferencedPath``s point to a valid file in the aws database
- Add method ``get_latest`` to ``Process`` class in Python Processes returning
the latest version of the process with the given slug
- Support assuming role in S3 connector

Changed
-------
- Set hashes during upload to avoid creating multiple versions of the object
in S3 bucket with enabled versioning


===================