To release a new version of bqplot on PyPI:

conda create -c conda-forge --override-channels -y -n bqplotrelease jupyterlab nodejs twine ipywidgets pip jupyter_packaging
conda activate bqplotrelease

We check out a fresh copy in a release/ subdirectory to make sure we do not overwrite our development repo.

mkdir -p release
cd release
# Delete any previous checkout we have here.
rm -rf bqplot


git clone git@github.com:bloomberg/bqplot.git
cd bqplot

cd js/
npm version [Major/minor/patch]
npm install
npm publish
cd ..

Update _version.py (bump both the python package version, and if needed, the Javascript version)

git clean -dfx
python setup.py sdist
python setup.py bdist_wheel
twine upload dist/*

Check to make sure the only changes are to the package.json, package-lock.json, and _version.py files:

git diff

Commit the changes in git

git commit -sa

Tag the release

git tag [version, like 0.12.4]

Push your change to a new PR and ask for a review to merge the PR.

Update the conda-forge feedstock.