contributing.md

Installation

python3.10 -m venv .venv
. .venv/bin/activate
pip install -U pip wheel
pip install -r requirements-dev.txt

Testing

pytest tests --cov=dataframe_api_compat/pandas_standard --cov-fail-under=100 --library pandas-numpy
pytest tests --cov=dataframe_api_compat/pandas_standard --cov-fail-under=100 --library pandas-nullable
pytest tests --cov=dataframe_api_compat/polars_standard --cov-fail-under=100 --library polars-lazy
# prerequisite: `pip install -r requirements-dev-modin.txt`
pytest tests --cov=dataframe_api_compat/modin_standard --cov-fail-under=100 --library modin

100% branch coverage isn't the objective - it's the bare minimum.

Note: there is also requirement about 100% coverage in tests folder, however, to achieve one need to combine coverage from different implementations. How this is done can be seen in the tox.yml CI flow.

Linting

pre-commit run --all-files

Type Checking

First, clone the dataframe_standard to some local path. Then, run:

MYPYPATH=<path to dataframe-api/spec/API_specification> mypy dataframe_api_compat/pandas_standard dataframe_api_compat/polars_standard
# prerequisite: `pip install -r requirements-dev-modin.txt`
MYPYPATH=<path to dataframe-api/spec/API_specification> mypy dataframe_api_compat/modin_standard

For example, if you cloned both repos in the same place, this could be:

MYPYPATH=../dataframe-api/spec/API_specification/ mypy dataframe_api_compat/pandas_standard dataframe_api_compat/polars_standard
MYPYPATH=../dataframe-api/spec/API_specification/ mypy dataframe_api_compat/modin_standard