Version 0.9.0 (9 July 2026)

Supported versions

The minimum supported version of Dask and distributed is now 2025.11.0. Although this does not cover the full SPEC0 support window, this release added two features that the simulator makes use of.

The util module was renamed to utils and split into sub-modules. The quaternion utilities now need to be imported from openstb.simulator.utils.quaternion. Attempting to import them directly from util will fail.
The abc.SimulationConfig class was renamed to abc.ControllerConfig to match the naming of the Controller plugin it configures.

Added the ScatteringModel plugin type to model acoustic scattering from an interface. Plugins for constant scattering and Lambertian scattering are included.
A circular trajectory plugin was added. This is registered under the name circular.
The openstb.simulator.utils.reduction.DaskReductionTree class was added to implement a generic reduction tree. This submits tasks to a Dask cluster to run user-provided reduction functions, and passes final results to a given output function.
A openstb.simulator.controller.base.LoopController base class was created to perform common actions for a simulation controller which loops over a set of variables.
Environment plugins now provide a density method. The base class provides a default implementation which estimates the density using the Gibbs SeaWater toolbox so any existing plugins inheriting from the base class will work without modification.

The simple point controller plugin was restructured to make the code cleaner. It now uses the new LoopController base class and DaskReductionTree utility added in this release.
Clarified that the salinity method of an Environment plugin should give the value in g/kg.
Some of the static type hinting was simplified using features available in Python 3.12+.

The type hinting in the tests was fixed so the tests can run in a Python 3.12 environment.

Various continuous integration workflows were added to check code quality and run the unit tests on push and pull request were added. This includes testing in an environment with the oldest supported versions of the dependencies.
A CI workflow to publish released versions to PyPI using trusted publishing was added.