The prefix.dev blog
Where we talk about software package and environment management.
)
)
Virtual Packages in the Conda ecosystem
Virtual packages are a neat trick to inject system requirements into the SAT solver and resolve for compatible packages automatically. In this blog post we talk about how they are used in the Conda ecosystem to support complex cross-platform package distributions.
)
What is a Conda package, actually?
At its core, a conda package really is just a "glorified" tarball—a compressed archive of files with some metadata attached.
)
S3 support in Pixi and rattler-build
We're excited to announce S3 support across our entire toolchain – rattler, pixi, and rattler-build now support the most common cloud storage standard, thanks to contributions from our friends at QuantCo. This vendor-agnostic approach to distributing Conda packages represents a major step forward for the ecosystem, offering teams the flexibility to host packages on any S3-compatible provider (AWS, Cloudflare R2, Backblaze B2, Hetzner, and more) with built-in authentication and minimal vendor lock-in.
)
Less Boilerplate, More Logic: Parameterising Pixi Tasks
Introducing powerful extensions to the existing task system
)
Enhancing the Conda Ecosystem in 2025
We are working on some exciting Conda Enhancement Proposals (CEP) in 2025. Read more about them here.
)
Trusted publishing to conda channels
Never forget your API keys again—with trusted publishing to prefix's channels
)
Pixi Global: Declarative Tool Installation
Think Homebrew, but cross-platform and easy to share with collaborators
)
Building CPU optimized packages for conda-forge
Did you know your packages could be even faster by enabling optimized CPU instructions? Learn how to build optimized packages for conda-forge.
)
rattler-build in conda-forge
rattler-build — the revolutionary build tool for the conda ecosystem (almost) available in conda-forge
)
Towards a Vendor-Lock-In-Free conda Experience
Why conda-forge is expected to remain free and open-source in the foreseeable future
)
WASM + Conda: Revolutionizing Scientific Computing in the Browser
Did you know that Conda packages can be built for WASM? Read more on how to build and use them with pixi!
)
Pixi - reproducible, scientific software workflows!
As scientists, your focus should be on research, not wrestling with software environments. At this year's SciPy conference, we're excited to show Pixi.
)
50x faster solves with sharded repodata
Repodata fetching can be painfully slow. We propose an alternative method that can speed things up drastically.
)
The joy of building conda packages with rattler-build
Learn how to integrate rattler-build into the conda-forge ecosystem, covering the essentials of conda-build and rattler-build, and how to create modern conda packages.
)
Using Python Projects With Pixi
pixi 0.18.0 ships with git, path and editable PyPI dependencies, improved PyPI <-> Conda mapping and experimental pyproject.toml support.
)
Introducing Pixi's Multiple Environments
Unleash the power of multiple environments — for the real power user.
)
7 Reasons to Switch from Conda to Pixi
Pixi is conda-compatible and comes with: more speed, lockfiles and tasks. It is the next-gen pacage manager for Python, R and more!
)
Adopting uv in pixi
tl;dr we're integrating uv into pixi
)
Building a C++ package with rattler-build
Conda packages are not only for Python! Learn how to build a C++ pacakge with conda-froge and rattler-build
)
Reproducible Notebooks with Pixi
Data scientists and researchers love to work with Jupyter Notebooks. Pixi makes it easy and more reproducible than ever!
)
rattler-build: A new parser
We have rewritten the rattler-build recipe parser with great developer experience in mind.
)
Prefix.dev — End of 2023
Follow what we have created this year, and plan for the next year.
)
Unleashing PyPI support in pixi
We've deeply integrated PyPI packages into pixi.toml's — read more...
)
Introducing Py-Rattler
Py-Rattler the Python partner to the blazingly fast rattler library.
)
Introducing rip — the fast & barebones pip implementation
`rip` resolves and installs packages from PyPI from pure Rust
)
Let Roboticists Stress About Boxes not Packages
Learn why `pixi` and the conda ecosystem are perfect of robotics.
)
Let's stop dependency hell
Pixi is a new package manager for the conda ecosystem, written in Rust. It's fast multi-platform, and language-agnostic.
)
The new Rattler Resolver
We wrote a new, fast SAT solver for conda packages in Rust – a port of the libsolv C library.
)
Why we support PackagingCon
prefix.dev is going to be at PackagingCon 2023, the software packaging conference
)
How we implemented API keys for prefix!
It's easy to overlook how important security is when implementing API keys
)
Introducing Channels: Host Your Own Packages
You can now host your own packages on prefix.dev
)
The new rattler-build
We are releasing a Rust based conda-build replacement that builds cross-platform binary packages much faster.
)
Repodata patching: how conda-forge keeps compatible
Keeping packages in a large repository compatible over time is tricky. conda-forge has a highly useful mechanism for that: repodata-patching.
)
Introducing Rattler: Conda from Rust
At prefix.dev we love Rust, but using conda/mamba from Rust is hard, so we started building Rattler, a set of crates that offer clean, compartmentalized building blocks for package management.
)
Introducing Online Environment Solving
We've released our first version of online environment solving.
To upper bound or not – the Python packaging debates
A hot debate is happening in the Python packaging world – should you use upper bounds to constrain your dependencies or not? We think there must be a better way.
)
Mamba 1.2 Release
We've released mamba 1.2 and here are all the improvements
)
Launching prefix.dev
We are excited to tell you more about what we are up to!
)
We're hiring
We are hiring engineers to work on developer tools (C++, Rust), our platform (Rust, TypeScript, React) and conda-forge (Python, CI pipelines)...