Tutorial - Packaging: Share your code for pip and Conda 

Python packages are reusable code that can be fetched, installed, and then imported by another program. Their contents are exactly the same as any Python project, only contains some extra configuration, so they can be bundled and uploaded to a shared place, for users to download from. This tutorial will teach you how to generate said metadata, so you can bundle your code into a "distributable," for others to install with either pip or Conda.


Attendees will be able to produce a Python package using Setuptools's PEP 517 backend, ready to be published to both a Python Package Index (PyPI) and a community Conda repository (conda-forge). The package will contain both importable Python files (pure modules and packages) and non-code data files.

Attendees will also be able to familiarize themselves with commonly-used project management tools, and be ready to understand workflows of popular packages, and to make contributions for them.

Target audience 

Python users that already understand Python modules, and want to package such code so they can use them with pip install or conda install.

Attendees are encouraged to bring their own modules to package, and explore how they can be converted into an installable format.

Speaker introduction 

Tzu-Ping Chung (@uranusjr)

TP builds his career around open source software, and enjoys committing his efforts to help make the world better. He builds all kinds of software for a living, from embedded system to single-page web applications, and contributes to the community when he can.

He served as chairperson during 2017–2018, and performed various other roles for PyCon Taiwan. These days, he spends most of his free time contributing to core Python packaging tools such as pip, and involving in packaging-related architectural design discussions as a member of the Python Packaging Authority (PyPA).

Detail description 

We will be following the Packaging Python Projects tutorial from the Python Packaging User Guide for setup.py and PyPI, and Building conda packages from scratch from the Conda-build documentation for conda-build and conda-forge. Some additional topics not able to fit into the conference schedule will be covered in slides for future reference.

  • 13:30–13:50: Overview and setup troubleshooting.
  • 13:50–14:15: Create a Python package with Setuptools and setup.py.
  • 14:15–14:45: Upload to the package index (PyPI).
  • 14:45–15:30: Excersise and Q&A.
  • 14:45–15:00: Break.
  • 15:00–15:15: Package for Conda with conda-build.
  • 15:15–15:45: Excersise and Q&A.
  • 15:45–16:00: Upload to conda-forge.
  • 16:00–16:30: Miscellaneous topics: binary packages, continuous integration, and other useful tools (as time permits).


Attendees should know how to write Python modules (files that can be imported by another file). They should have basic understandings to the command line, and know how to create, rename, and edit files in a text editor or the command line, on their own computers.

Please have the followings ready before the tutorial:

  • Any Python installation with pip, and either virtualenv or venv available.
  • Conda from either Miniconda or Anaconda.
  • (Optional) Some Python code you want to package and publish as a Python and/or Conda package.