Developer notes

These notes complement the contribution guidelines with details for developers working on qpbenchmark itself.

Development environment

The repository uses pixi to manage environments and tasks:

Command	Description
pixi run test	Run unit tests
pixi run coverage	Run unit tests with a coverage report
pixi run lint	Check formatting, lint with ruff and type-check with mypy
pixi run format	Format source code with ruff
pixi run -e py39 test	Run unit tests on a given Python version (py39 to py312)
pixi run qpbenchmark --test-set my_test_set.py run	Run the benchmark with all solvers of the solvers environment
pixi run docs-serve	Serve this website locally, with live reload

Working on qpbenchmark and a test set together

Test sets are separate repositories that depend on the qpbenchmark package. Their pixi.toml gets it from PyPI:

[pypi-dependencies]
qpbenchmark = ">=3.0.0"

When a change spans both repositories, for instance a new benchmark feature and the test-set results that use it, point the test set to your development version of qpbenchmark instead. Either use a local clone, installed in editable mode so that your changes apply without reinstalling:

[pypi-dependencies]
qpbenchmark = { path = "../qpbenchmark", editable = true }

Or use a branch from a Git repository, for instance from your fork:

[pypi-dependencies]
qpbenchmark = { git = "https://github.com/your-username/qpbenchmark.git", branch = "your-branch" }

Note that we use qpbenchmark as a Python package from PyPI, with its source going under [pypi-dependencies] rather than under [dependencies].

Don't forget to revert to a released version, e.g. qpbenchmark = ">=2.8.0", before merging changes to the test set.