Skip to content

8. Publish

We are ready to publish your package. Users don't want to install your package through a git hosting platform, but through a dedicated package manager. In the Python ecosystem, there are two main platforms for making your package available.

PyPI

PyPI is the Python Package Index. It is a central repository for Python packages and is used by pip to install packages. PyPI is a good place to publish your package as it is the most widely used package manager for Python. There are several ways to publish your package to PyPI. A simple CI/CD integration is discussed in the CI/CD section. PyPI also has a test instance which you can use to test your workflow. You could also use it to publish each commit to a test instance and then use the main instance for releases. But this is probably overkill for most projects.

twine

If you'd like to play around with this process locally, you can use the twine command line tool. Twine is a command line tool for uploading packages directly to PyPI. It is very simple:

Using twine:

  1. Create some distributions in the normal way: 
    python -m build
  2. Upload to Test PyPI and verify things look right: 
    twine upload --repository testpypi dist/*
  3. Verify the upload: 
    pip install --index-url https://test.pypi.org/simple/ <your-package>
    Or to upload to the main PyPI: 
    twine upload dist/*

Note

This process expects that you have the package setup correctly, as discussed in the build section.

Conda

Conda is a package manager for Python and other languages. It is used to manage packages and environments and has the advantage of being able to handle non-Python packages. While you can install a pip package in a conda environment, it is recommended to use conda packages for conda environments. Conda-forge is a community-driven collection of conda packages. It is the usual place to publish your package. You could of course use a different channel. You will need to set up a conda feed for your package once. See the conda-forge documentation for more information. The process is a bit more complicated than PyPI. But then condaforge can be triggered by any new PyPI release, so you only need to trigger it once.

Package manager

While pip is the most common package manager for PyPI, there are others worth mentioning:

Also for conda there are other package managers:

Here are additional resources you could add to the "Resources" section of your guide:

Resources