Contributing to Pint

Pint uses (and thanks):

You can contribute in different ways:

Report issues

You can report any issues with the package, the documentation to the Pint issue tracker. Also feel free to submit feature requests, comments or questions.

Contribute code

To contribute fixes, code or documentation to Pint, fork Pint in github and submit the changes using a pull request against the master branch.

  • If you are submitting new code, add tests (see below) and documentation.

  • Write “Closes #<bug number>” in the PR description or a comment, as described in the github docs.

  • Log the change in the CHANGES file.

  • Execute pre-commit run --all-files and resolve any issues.

In any case, feel free to use the issue tracker to discuss ideas for new features or improvements.

Notice that we will not merge a PR if tests are failing. In certain cases tests pass in your machine but not in travis. There might be multiple reasons for this but these are some of the most common

  • Your new code does not work for other Python or Numpy versions.

  • The documentation is not being built properly or the examples in the docs are not working.

  • linters are reporting that the code does no adhere to the standards.

Setting up your environment

If you’re contributing to this project for the fist time, you can set up your environment on Linux or OSX with the following commands:

$ git clone
$ cd pint
$ python -m virtualenv venv
$ source venv/bin/activate
$ pip install -e .
$ pip install -r requirements_docs.txt
$ pip install pre-commit # This step and the next are optional but recommended.
$ pre-commit install

Writing tests

We use pytest for testing. If you contribute code you need to add tests:

  • If you are fixing a bug, add a test to, or amend/enrich the general test suite to cover the use case.

  • If you are adding a new feature, add a test in the appropiate place. There is usually a for each file. There are some other test files that deal with individual/specific features. If in doubt, ask.

  • Prefer functions to classes.

  • When using classes, derive from QuantityTestCase.

  • Use parametrize as much as possible.

  • Use fixtures (see instead of instantiating the registry yourself.

  • Checkout for some convenience functions before reinventing the wheel.

  • When your test does not modify the registry, use sess_registry fixture.

Running tests and building documentation

To run the test suite, invoke pytest from the pint directory:

$ cd pint
$ pytest

To run the doctests, invoke Sphinx’s doctest module from the docs directory:

$ cd docs
$ make doctest

To build the documentation, invoke Sphinx from the docs directory:

$ cd docs
$ make html

Extension Packages

Pint naturally integrates with other libraries in the scientific Python ecosystem, and a small number of `extension/compatibility packages<numpy.html#Compatibility-Packages>`_ have arisen to aid in compatibility between certain packages. Pint’s rule of thumb for integration features that work best as an extension package versus direct inclusion in Pint is:

  • Extension (separate packages)

  • Integration (built in to Pint)

    • Duck array types wrapped by Pint (below Pint in the type casting hierarchy)

    • Intermingling of APIs occurs

    • Examples: Dask