"Making an open source package - lessons learned"
summary
Concrete lessons from building the open source package scores focus on what actually changed outcomes: using pyproject.toml for config, choosing hatch (but only using parts of it), and getting PyPI publishing and release tagging right after painful versioning mistakes (dropping a leading v and missing patch numbers). Automated testing and 100% coverage catch downstream breakage, Sphinx + Napoleon drives usable API docs and runnable tutorials, contributor guidelines and a JOSS paper improve adoption, and the overall effort proves worthwhile despite costly early missteps.
AI-written summary · about
chapters (19)
- 0:00Open source journey
- 1:07About this talk
- 3:35Package managers like Hatch
- 5:01Disk layout and installs
- 7:05Configuration with pyproject.toml
- 8:10Versioning pitfalls
- 10:17Branching strategies
- 12:23Automated testing
- 14:56Testing GUI and CLI
- 17:07Decimal precision in tests
- 18:13Type hinting challenges
- 20:44Linting and static analysis
- 21:47Documentation types
- 23:32Maintaining tutorials
- 25:18Docstrings and comments
- 26:20Documentation tech stack
- 27:45Rendering docs across platforms
- 29:11Release notes and PyPI
- 30:13Community and academic publishing
related talks
26mOpen Source on Easy Mode
28mMaking a Splash with your Open Source Project
46mThe art of building and sustaining successful OSS tools and infrastructure
38mTools for Maintaining an Open Source Project
22mPublishing well formed Python packages
27mDesign Your Open Source Project
25mLessons Learned from Cultivating Open Source Projects and Communities
30mFriends don't let friends package alone