Thanks for helping us build Rubicon!
Cloning the Repository#
Make a fork of the rubicon-ml repo and clone the fork:
git clone https://github.com/<your-github-username>/rubicon-ml cd rubicon-ml
You may want to add
as an upstream remote:
git remote add upstream https://github.com/capitalone/rubicon-ml
Creating a Development Environment#
We have a
conda environment YAML file with all the necessary dependencies
at the root of the repository.
conda env create -f environment.yml
After you’ve cloned the repository, use
pip to install
python -m pip install -e .[all]
black rubicon_ml tests flake8 rubicon_ml tests isort -rc rubicon_ml tests
Install and configure pre-commit to automatically run
each of these tools before committing. Once installed, run
to set up the git hooks.
rubicon_ml uses pytest for testing.
Run the tests from the root
rubicon_ml directory as follows:
python -m pytest
We follow numpydoc
formatting for our docstrings. This lets us use
sphinx to automatically
generate a library reference.
We also use nbsphinx to render our
example notebooks directly from our repo. To reduce the complexity of our
documentation builds, we only commit executed notebooks to the repo so
nbsphinx doesn’t have to spend time executing them itself.
To build the documentation locally, use the environment YAML file in the
docs directory to create a new
conda environment with all necessary
cd docs/ conda env create -f docs-environment.yml
Activate the new environment, install a local copy of
make html command from the
docs directory to build the
conda activate rubicon-ml-docs pip install --no-deps -e ../ make html
The newly built documentation can be opened in a browser.
Never commit built documentation code directly, only the source.
.gitignore should handle keeping built docs out of the repo, and
our CICD handles deploying newly committed documentation.