Development

uv sync                          # Python environment with rumoca (built from its git commit) and CasADi
uv run pytest                    # structure, bundle and rumoca tests
uv run pytest -m openmodelica    # simulations of the examples in OpenModelica (Docker)
uv run ruff check . && uv run ruff format --check . && uv run mypy

The OpenModelica tests run the examples in the official image openmodelica/openmodelica:v1.26.9-ompython and compare them with FastSmoothBuildings/Resources/ReferenceResults/OpenModelica; uv run pytest -m openmodelica --update-references rewrites the references after a deliberate change. The rumoca tests integrate the CasADi models and compare them with the same references, without Docker.

Releasing

A change merged into main is released once CI passes on it when it changes what users install: the library, the Python package, pyproject.toml, README.md or LICENSE (changes to the tests, the study, the documentation site or CI wait for the next such change, as semantic versioning has it). The release workflow tags the version of FastSmoothBuildings/package.mo when it has no tag yet (the first release, or a version raised by hand in a pull request), and otherwise raises the minor version and commits it (python scripts/bump_version.py minor). The tag publishes the wheel and the sdist on PyPI and creates the GitHub release with the library as a zip for OpenModelica and Dymola; its notes are the section of the version in CHANGELOG.md, or the commits since the previous release. The workflow uses the repository secrets ACCESS_TOKEN (a personal access token that may push to the repository) and PYPI_API_TOKEN (a PyPI API token), as trano does.

The study

The study against Buildings (study/) runs the models trano generates from the YAML descriptions of BESTEST cases and two houses, with Buildings and with FastSmoothBuildings, in OpenModelica:

uv sync --group study                                   # trano, trano-simulate, trano-optimization
uv run python -m study.buildings run 950FF              # both libraries, every variant of a case (Docker)
uv run python -m study.buildings report --docs          # the page "Against Buildings" from the committed results
uv run python -m study.replay compare residential_after # FastSmoothBuildings replaying the controls of Buildings
uv run python -m study.analysis.scripts.study_responses # the numbers of a report of the analysis

The simulations need Docker and the Modelica libraries of trano (TRANO_MODELICA_LIBRARIES); the results are committed (study/buildings/results, study/analysis/results), so the reports and the pages render without them. The workflow study.yaml runs the study on demand, or on a push whose commit message carries [study], and commits its results.

The documentation

uv sync --group docs
uv run mkdocs serve

The pages of the library are generated from the Documentation annotations of its classes (docs/gen_pages.py), the pages of the analysis from study/analysis. The workflow docs.yaml publishes the site on GitHub Pages.