Skip to content

Pixi Package Manager#

TL;DR

Pixi is a modern, performant package manager designed as an alternative to Conda, Mamba, and even PyPI. It simplifies dependency management across operating systems, making project reproducibility easier across systems.

What is Pixi?#

Pixi is a package manager built to improve upon and provide an alternative to Conda and Mamba (or environments in general). Specifically, Pixi aids projects that require managing complex dependencies and ensuring reproducibility across different operating systems.

Pixi excels in ease of use, speed, and robustness, positioning itself as a compelling replacement for Conda when managing project-level dependencies, especially in multi-language (e.g., Python and R) data science projects.

Python-only projects

For projects focused exclusively on Python, you might also consider using uv instead, a modern alternative to pip and/or pyenv.

For extra context on why the developers built Pixi in the first place, see their Vision page.

Why Consider Pixi?#

Reproducibility and Cross-platform Compatibility#

Pixi excels in dependency management and reproducibility, particularly across different operating systems. With Pixi, sharing your environment setup and ensuring your project runs identically on other machines is straightforward:

  • Simply include a pixi.toml file in your project repository.
  • Another user runs pixi install and instantly replicates your environment.

Tasks#

Instead of describing ways to run pipelines or commands to do repeated actions, you can define tasks in the pixi.toml file. This way, instead of running long commands with convoluted arguments and flags, you can alias it with a simple name. You can find more details on tasks on their task documentation.

Enhanced Performance and Usability#

Pixi enhances the performance and usability experience compared to traditional tools like Conda or Anaconda, providing:

  • Faster dependency resolution
  • Simplified, intuitive command-line interface (CLI)
  • Robust management of dependencies for multiple languages within a single environment

Getting Started#

Installation instructions can be found directly on the Pixi installation guide.

Once installed, a great starting point and introduction can be found on their dedicated Getting Started page.

Here's a cheatsheet on how you might use Pixi for basic project management:

Command Description
pixi init Start a pixi project in the current working dir
pixi install Install all dependencies from pixi.toml
pixi add <pkg> Add a new dependency and update environment
pixi run <cmd> Run a command (or task) within Pixi environment
pixi shell Activate a shell with Pixi-managed environment

Example Usage#

Working in Python with Pixi#

Instead of pip installing Python packages globally, or using a Conda environment, you can define per-project dependencies like so:

pixi add python@3.10 # specify an exact package version!
pixi add pandas

# or add multiple at once:
pixi add python@3.10 pandas

Working in R with Pixi#

Because Pixi utilizes the Conda package ecosystem, packages you can install with Conda are also available in Pixi. This includes R packages, which can be installed easily:

pixi add r-base
pixi add r-ggplot2
pixi add bioconductor-limma -c bioconda

Installing R and Bioconductor Packages

R packages are often prepended with r- in the package name, and depend on the base R installation r-base. Bioconductor packages are also available if you add bioconda as a channel, and are often prepended with bioconductor-. The easiest way to find the exact package names is to search for them on the Anaconda Cloud.

Bioconductor packages on ARM-based Macs

Some Bioconductor packages have not yet been built for ARM-based Macs (Apple Silicon) and thus might raise an error:

pixi-add-pgx

If you encounter issues, you can try falling back to osx-64 which uses rosetta to run x86_64 binaries on ARM-based Macs:

[workspace]
channels = ["conda-forge", "bioconda"]
name = "my_project"
- platforms = ["osx-arm64"]
+ platforms = ["osx-64"]

VS Code Integration#

You can install the VS Code extension to help manage Pixi environments within the IDE. You can find it on the VS Code Marketplace:

Additional Resources#