✅ Contribution validation#

A Bookshelf contribution often requires some additional modifications other than the code itself. To ensure the different modifications are correctly contributed, we have set up different validation steps.

⚙️ Automatic validation#

First, we have defined some scripts that are automatically executed when proposing a pull request (and after each commit inside the pull request). These scripts verify:

  • If the metadata of the features, modules, and datapacks are defined and up to date (see metadata page).

  • If there is no regression in the code, by executing the different unit tests (see debug page).

A Bookshelf contributor can also manually and locally execute these checks using the all.ipynb notebook in the scripts folder (requires Python, see scripts/pyproject.toml file).

Some checks about the metadata can fail because the specified metadata are not up to date. Indeed, some module/feature metadata such as the dependencies are computed, and the Bookshelf contributor needs to manually run the generators to enable their computation. These generators are available in the scripts/all.ipynb notebook and are not executed by the CI/CD.

📄 Pull request template#

When a Bookshelf contributor opens a pull request on the Bookshelf repository, the message of the pull request is filled with a checklist. This checklist helps the Bookshelf contributor to ensure that they have not forgotten any changes we cannot automatically verify, such as the modification of the changelog file. Here is a non-exhaustive list of what a feature contribution requires:

  • Update the changelog with the content of the contribution and the ID of the pull request.

  • Document the new features in the documentation and update the documentation for existing features (if needed).

  • Provide some unit tests if possible.

  • Link the pull request to an existing issue.

🔎 Manual review#

Finally, other Bookshelf contributors manually review the pull request before approving it.