Key Points

Environments and task runners

  • Virtual environments isolate software
  • Virtual environments solve the update problem
  • A task runner makes it easier to contribute to software

Code to Package

  • Put your code and tests in a standard package structure
  • Use a pyproject.toml file to describe a Python package

Other files that belong with your package

  • Packages should include a README, LICENSE, and CHANGELOG.
  • Choose an existing software license
  • You can also include .gitignore to avoid from committing non-source files.

Metadata

  • Add informational metadata to tell people about your package.
  • Add functional metadata to tell people how to install and use your package.

Versioning

  • Packages should have a version attribute
  • Semantic versioning is an abbreviated changelog, not the solution to all problems
  • You can use packaging tools so that the version number needs updating in one (and only one) place

Publishing package and citation

  • CI can publish Python packages
  • Tagging and GitHub Releases are used to publish versions
  • Zenodo and CITATION.cff are useful for citations

Documentation Overview

  • Sphinx or MkDocs are both good for documentation

Documentation with Sphinx

  • Sphinx is great for documentation

Documentation with MkDocs

  • MkDocs is great for documentation

Checks and tests

  • Run tests and static checks on your codebase.

Task runners

  • A task runner makes it easier to contribute to software

Continuous Integration

  • Set up GitHub Actions on your project
  • Run your tests on multiple platforms and with multiple Python versions