dfsc Release Checklist

Use this checklist before tagging a public release.

Package

  • [ ] python -B tools/pre_release_audit.py reports at least 95% internal readiness.
  • [ ] pyproject.toml version matches the release tag.
  • [ ] pip install -e .[test] succeeds in a clean environment.
  • [ ] python -m build creates both wheel and source distribution.
  • [ ] python -m twine check dist/* passes.
  • [ ] python -m unittest discover -s tests passes.
  • [ ] python tools/mlsl_doctor.py passes on CPU.
  • [ ] CUDA smoke test is recorded when a GPU is available.

Documentation

  • [ ] README.md quick start runs without modification.
  • [ ] docs/API.md lists all public dfsc entry points.
  • [ ] docs/ROADMAP.md separates implemented, experimental, and planned work.
  • [ ] The hosted documentation site builds from mkdocs.yml.
  • [ ] mkdocs build --strict passes in CI.

Reproducibility

  • [ ] Experiment outputs are regenerated or explicitly marked as archived.
  • [ ] Paper tables and figures are regenerated from current results.
  • [ ] The software artifact audit reports all mandatory checks as passing.

Publication Boundary

  • [ ] Claims avoid presenting dfsc as a universal fractional solver.
  • [ ] Experimental wrappers are labeled experimental in code, docs, and paper.
  • [ ] Public benchmark data provenance is recorded when external data are used.

Public Release

  • [ ] Create the public Git repository and protect the release branch.
  • [ ] Publish the signed version tag and attach CI-built distributions.
  • [ ] Publish the package to PyPI from a trusted release workflow.
  • [ ] Archive the tagged source on Zenodo and add the assigned DOI to CITATION.cff.
  • [ ] Replace all pre-release availability wording only after the public URLs resolve.