Environment Management with Pixi

Setup

Installation

Visual Studio Code (VSCode)

Download Link

Pixi

Pixi is available for all major operating systems and provides a consistent way to define, reproduce, and run development environments. It can be installed quickly and works seamlessly with the conda-forge ecosystem.

1. Installation instructions
   1. If you have a windows machine

   winget install prefix-dev.pixi
2. Testing installation

   pixi --version

In this unit, we will learn how to use Pixi to manage packages and environments. Pixi is a modern package manager built on top of Conda ecosystem principles, but with improved performance, reproducibility, and cross-platform support. It uses a pixi.toml file to define environments and supports tasks for automation. This unit includes instructions for installing Pixi and covers core Pixi commands for creating, managing, and sharing environments.

Section 1: Creating and Managing Environments

With Pixi, environments are defined via a pixi.toml file. This ensures that your environment is reproducible and version-controlled. Pixi uses workspaces (directories with a pixi.toml) and automatically activates environments when you enter them.

Code	Description
pixi init	Initialize a new Pixi project with a pixi.toml
pixi add python	Add Python to your environment
pixi run	Run commands inside the environment
pixi shell	Start a shell with the Pixi environment activated
pixi info	Show details about the environment
pixi list	List all packages in the environment

Exercises

1. Initialize a new Pixi environment

   1. Initialize Pixi.

   2. Open the pixi.toml file and inspect its contents.

   3. Add Python to the environment.

      pixi add python

   4. Add pip to the environment

   5. Start a Pixi shell

   6. Verify python is available

      python --version

   7. Verify pip is available

Section 2: Installing Python Packages with Pixi

In this section, you’ll learn how to install Python packages into a Pixi-managed environment. Pixi supports two main ways of adding packages:

• Conda-style installation (via conda-forge): By default.
• PyPI-style installation (via pip): Useful for packages not available on conda-forge.

You can combine both methods in the same environment. This gives you access to a wide range of scientific and general-purpose Python tools.

Code	Description
pixi add numpy	Install numpy from conda-forge
pixi add matplotlib pandas	Install multiple packages from conda-forge
pixi add --pypi package_name	Install package from PyPI
pixi add pip	Add pip support to the environment
pixi shell	Enter the environment to run or install additional packages
pip install seaborn	Use pip inside Pixi shell for PyPI-only packages
pixi remove <package>	Remove a package from the environment
pixi list	View all currently installed packages

Exercises

1. Add numpy, matplotlib
2. Verify they are installed:

   pixi info
3. Install streamlit with --pypi

   pixi add --pypi streamlit
4. Verify streamlit is installed
5. Install pandas with --pypi and verify it is installed
6. Remove pandas and verify that it is gone

   pixi remove --pypi pandas

Section 3: Exporting, Sharing, and Reproducing Environments

Unlike Conda, Pixi manages environments via pixi.toml and pixi.lock files. These files define your environment in a reproducible format. Sharing these files with collaborators ensures they can recreate the exact same environment on any platform.

File	Purpose
pixi.toml	Human-editable file that specifies dependencies
pixi.lock	Auto-generated file that pins exact versions for reproducibility
pixi install	Sync the environment to the lock file

Exercises

1. Simulate sharing an environment

   1. Copy only pixi.toml and pixi.lock to the new location.
   2. Reproduce the environment in your new folder
   3. Confirm that numpy is available and version matches the lock file.