Release Process

Build a version

  1. Pick a release name for the next version

    Releases are versioned using the major.minor.bugfix numbering system.

  2. Update the release notes on the What’s New page

    Check the git log for changes since the last release.

  3. Check that Travis and readthedocs are passing for the latest commit

    Make sure that the tests have finished running!

  4. Tag the branch

    Use the format of datacube-major.minor.bugfix.

    git tag datacube-1.6.0
    git push --tags
  5. Draft a new release on the Datacube releases GitHub page

    Include the items added to the release notes in step 2.

Marking it stable

Once a built version has been tested, found to be stable, and the team agrees, we make it the new stable version.

  1. Merge changes leading up to the release into the stable branch

    This will also update the stable docs.

  2. Upload the build to PyPi

    This step is done by GitHub Actions when tag is pushed.

    Manually it looks something like this:

    python sdist bdist_wheel
    twine upload dist/*

    This should upload the project to

  3. Update conda-forge recipe

    Follow the instrucions under Updating datacube-feedstock in the Datacube Feedstock repository.

    It should involve modifying the version number in the recipe and updating the SHA hash. The hash should be generated from the .tar.gz mentioned in the source of the recipe.

    openssl sha256 <downloaded-datacube-source.tar.gz>