v0.5.2
Aurora v0.5.2 Release Notes
Release date: 2025-08-03
This release represents a major modernization and feature expansion of the Aurora MT processing pipeline, following the JOSS release v0.3.13. The focus is on robust, flexible, and reproducible transfer function estimation, with a particular emphasis on feature-based channel weighting and improved metadata integration.
Major Features and Changes
Feature-Based Channel Weighting
- Aurora now supports feature-based channel weighting, enabling more robust, data-driven weighting strategies for transfer function estimation.
- Integration with
mt_metadata.features.weightsallows users to define, load, and apply channel weights based on features such as striding window coherence. - The new test suite (
test_feature_weighting.py) demonstrates how feature weights can be injected into the processing pipeline and compared to standard processing. - This lays the foundation for future expansion to additional features and more flexible weighting schemes.
Packaging and Dependency Modernization
- Migration to PEP 621-compliant
pyproject.tomlfor all metadata and dependencies. - Minimal
setup.pyretained for legacy compatibility. - All dependencies, including
mth5(from GitHub master), are now managed in TOML. - License files and package data are now included automatically.
CI/CD and Testing Improvements
- Continuous integration workflows updated to use
uvfor fast dependency installation and reproducible builds. - All dev dependencies are now installed from
pyproject.toml. - Test coverage improved, including new tests for utility functions and feature weighting.
- Automated notebook execution for documentation.
Documentation and Usability
- Improved docstrings and code comments, especially around feature weighting and metadata integration.
- Release notes and PR summaries now included for transparency and reproducibility.
Other Notable Changes
- General codebase cleanup and modernization.
- Alphabetized and cleaned up dependencies in TOML files.
- Improved error handling and logging.
Thanks to all contributors and users for feedback and support!
-- The Aurora Development Team