Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
71 lines (53 loc) · 2.89 KB

CONTRIBUTING.md

File metadata and controls

71 lines (53 loc) · 2.89 KB
Raw

Contributing to EROTICA

Thanks for your interest. EROTICA is a research package for Gaia open-cluster analysis, so contributions are judged on scientific correctness as much as on code quality.

Reporting bugs

Open a GitHub issue with:

  • what you ran (the call, the columns you clustered on, the data source),
  • what you expected and what happened,
  • the traceback if there is one, and your Python + EROTICA versions (python -c "import erotica; print(erotica.__version__)").

For a scientific discrepancy (membership, ages, radii), say which numbers you expected and why — a literature value, another tool, a previous run.

Asking questions

Use GitHub Discussions for usage questions and methodology questions ("should I cluster in 5D for my cluster?"). The membership guide covers the main trade-offs.

Development setup

git clone https://github.com/notluquis/erotica.git
cd erotica
pip install -e ".[dev]"
pytest -q

Submitting changes

  1. Branch off dev (git checkout -b fix/short-description).

  2. Make the change and add a test that fails without it. For anything touching the science — membership, isochrones, structure, dynamics — the test should assert on behaviour, not just that the code runs.

  3. Keep the style consistent with ruff, which replaced black + isort + flake8 and is what .pre-commit-config.yaml actually runs (line length 100, configured in pyproject.toml):

    ruff check --fix erotica/ tests/
ruff format erotica/ tests/

    black and isort are not installed by the dev extra. This step named them until 2026-08-04, so following it produced "command not found".

  4. Run the suite: pytest -q.

  5. Open a pull request against dev describing what changed and, for science changes, what you validated it against.

Scientific conventions

These are house rules; a PR that breaks them will be asked to change:

  • Don't silently change a published default. The NGC 6383 results in data/test/ are a reproducibility artifact; if a change moves those numbers, say so in the PR.
  • Options with disclosure. When adding a method that can fail on some cluster morphology, add it as an option and document the failure mode — do not make it the default. See the membership guide for the pattern.
  • State what is estimated vs assumed. Priors, fixed parameters, and the model grid an age is conditioned on belong in the docstring.
  • Uncertainties are not decoration. Prefer returning a distribution or an interval over a bare point estimate where the method supports it.

Code of conduct

Be respectful and constructive. Harassment or personal attacks are not tolerated; report concerns to lescobar2019@udec.cl.