Typed Settings uses Hatch for environment management, building and publishing. However, you can also use pip and virtualenv, if you like.

It uses nox to run the linters and tests against a matrix of different dependency and Python versions. Nox is similar to tox but uses Python to describe all tasks.

It also uses pre-commit to lint the code you’re going to commit.

Setting up a Development Environment#

  1. Clone the project and change into its directory:

    $ git clone
    $ cd typed-settings
  2. Create a virtual environment in your preferred ways:

    • Using Hatch:

      $ hatch shell

      This not only creates and activates an environment but also installs/updates all development dependencies and pre-commit.

    • Using virtualenvwrapper:

      $ mkvirtualenv typed-settings
    • Using virtualenv:

      $ virtualenv .env
      $ source .env/bin/activate
    • Using venv:

      $ python -m venv .env
      $ source .env/bin/activate
  3. If you did not use Hatch, install all development requirements and Typed Settings itself in development mode:

    (typed-settings)$ pip install -e .[dev]  # Not needed with hatch
    (typed-settings)$ pre-commit install --install-hooks


Typed Settings uses flake8 with a few plug-ins (e.g., bandit) and mypy for linting:

(typed-settings)$ flake8 PATH...
(typed-settings)$ mypy PATH...
(typed-settings)$ # or
(typed-settings)$ hatch run lint

Black and Isort are used for code formatting:

(typed-settings)$ black PATH...
(typed-settings)$ isort PATH...
(typed-settings)$ # or
(typed-settings)$ hatch run fix [PATH...]

Pre-commit also runs all linters and formatters with all changed files every time you want to commit something.


You run the tests with pytest. It is configured to also run doctests in src/ and docs/ and to test the examples in that directory, so do not only run it on tests/.

(typed-settings)$ pytest
(typed-settings)$ # or
(typed-settings)$ hatch run test

Hatch provides a shortcut for quickly running the tests and measure the coverage:

(typed-settings)$ hatch run cov

You will not get to 100% with this though, since some compatibilty code will not be executed.

You can also use nox to run tests for all supported Python versions at the same time. This should get you to 100% coverage.

Just run nox to build a package, test it, and lint it:

(typed-settings)$ nox


Sphinx is used to build the documentation. The documentation is formatted using reStructuredText (maybe we’ll switch to Markdown with the MyST parser at some time). There’s a makefile that you can invoke to build the documentation:

(typed-settings)$ make -C docs html
(typed-settings)$ # or
(typed-settings)$ hatch run docs
(typed-settings)$ make -C docs clean html  # Clean rebuild
(typed-settings)$ # or
(typed-settings)$ hatch run clean-docs
(typed-settings)$ open docs/_build/html/index.html  # Use "xdg-open" on Linux


When you commit something, take your time to write a precise, meaningful commit message. In short:

  • Use the imperative: Fix issue with XY.

  • If your change is non-trivial, describe why your change was needed and how it works. Separate this from the title with an empty line.

  • Add references to issues, e.g. See: #123 or Fixes: #123.

When any of the linters run by Pre-commit finds an issue or if a formatter changes a file, the commit is aborted. In that case, you need to review the changes, add the files and try again:

(typed-settings)$ git status
(typed-settings)$ git diff
(typed-settings)$ git add src/typed_settings/...

Releasing New Versions#

Releases are created and uploaded by the CI/CD pipeline. The release steps are only executed in tag pipelines.

To prepare a release:

  1. Update the CHANGELOG.rst. Use an emoji for each line. The changelog contains a legend at the bottom where you can look-up the proper emoji.

  2. Update the version in pyproject.toml.

  3. Commit using the message Bump version from a.b.c to x.y.z.

  4. Create an annotated tag: git tag -am 'Release x.y.z' x.y.z.

  5. Push everything: git push --atomic origin main x.y.z.

  6. The CI/CD pipeline automatically creates a release on the testing PyPI. Check if everything is okay.

  7. Manually trigger the final release step.