Great to see the community supporting the `pixi` package manager expanding!
Posts by
We're thrilled to announce our second non-prefix core maintainer of Pixi!
Welcome Lucas Colley! He's been an ESOC 2025 student and since then kept being active, both in the Pixi repo, our Discord community and by adding Pixi to Open-source projects like CPython, SciPy & Numpy!
Pixi has a new subcommand: `publish`! It is now one step to build and upload a new package.
Shortens your CI and improves the `pixi build` experience. Enjoy!
We are happy to release DuckLake v1.0, a production-ready lakehouse format specification. Its reference implementation, the ducklake DuckDB extension, is available as of today in DuckDB v1.5.2.
For more details, read the announcement blog post:
ducklake.select/2026/04/13/d...
Now on @sciam.bsky.social: See the Artemis II mission in 12 unforgettable photos. Can't say I disagree with any of these selections!
www.scientificamerican.com/article/see-...
I'm still figuring things out myself.
I also don't love a `pixi run post-install` task or similar, I'm not sure there's a way to avoid it (at present) but I'm looking into it...
I'm quite bullish on this approach:
pavel.pink/blog/pixi-sk...
Package and version your skills as conda packages which allows you to specify their dependencies, including binary deps.
A `pixi-skills` CLI can then symink the installed skills into the well-known location for your agent of choice.
ArviZ now has built-in tools for prior & likelihood sensitivity analysis via power-scaling!
Instead of fitting multiple models with different priors, you fit once and use importance sampling to approximate the effect of perturbing the prior or likelihood.
Oh this is great: there’s now an #Artemis II Lunar Flyby Gallery on the NASA images site: www.nasa.gov/gallery/luna... (thx @dpmanchee.bsky.social for pointing it out)
Your optimization model is only as good as the forecast you feed it.
We partnered with FICO to show what happens when you replace point estimates with full distributions. The results changed the decisions, not just the forecasts.
Full technical dive here: dub.link/zrDyPCF
#PyMC #FICO
Apache Arrow is a core data interchange format, so it's good to see the ecosystem evolving and expanding.
Are you dreaming of your own specialized *-forge? Our new build tool "rattler-build" makes it simple. Check out our new docs on how to create a custom "r-forge", "rust-forge", "ruby-forge", ... rattler-build.prefix.dev/latest/tips...
This jibes pretty well with my own experience of using AI tools.
There's a nice #Artemis II mission tracker here: artemis.cdnspace.ca
LOTS of data about its current position, crew activities, and whether or not the toilet is currently "Go"
DuckLake 1.0 will ship with the “data inlining” feature: small updates are staged in the catalog instead of creating many small files.
In today's blog post, Pedro Holanda shows how DuckLake implements inlining and shows an almost 1000× speedup on certain operations.
ducklake.select/2026/04/02/d...
Here is the recording of the webinar I gave last week on GPU support in @scikit-learn.org and comparison of a scikit-learn pipeline vs the TabICLv2 foundational model on a non-linear heteroscedastic quantile regression task.
app.livestorm.co/probabl/webi...
@me checks calendar...
The notebook.link team is building "serverless scientific computing" on top of the modern Conda ecosystem & the prefix.dev package server. Read the guest blog on how they shipped interactive WASM for Python, R, C++, Fortran and more: prefix.dev/blog/server...
"Building From Source Shouldn't Be This Hard"
That's our motto, read our latest blogpost to understand what we're building and how Pixi already helps you with today!
🗞️ Read it here: prefix.dev/blog/buildi...
#pacakgemanagement #pixi #ROS #condaforge
I put together a visual LLM Architecture Gallery that collects (~50) recent open-weight model designs in one place.
Architecture diagrams, config links, tech reports, explainers... you name it!
Hopefully useful as a reference & learning resource:
sebastianraschka.com/llm-architec...
No more duplicate dependencies in pyproject.toml!
Pixi Python packages can now set `ignore-pypi-mapping = false` to automatically convert `pyproject.toml` dependencies into conda package requirements.
The goal is no configuration, this is a step in that direction!
Lettermatic ✕ GitHub Next: the inside story of Monaspace, a truly groundbreaking advancement for how we display code. We knew that Lettermatic would go deep, but we were still amazed by their level of knowledge and craft.
See their case study below, and get the fonts at monaspace.githubnext.com
Do you need more control over the packages that will be installed in your Pixi workspaces?
You can now define [constraints], which allow you to specify which versions are allowed when a package is installed.
✨ skrub version 0.8.0 has been released ✨
This version includes several new features, including multiple improvements to the functionality and performance of the Data Ops, along with a few bug fixes and improvements to the docs.
Changelog:
skrub-data.org/stable/CHANG...
Highlights below ⤵️
Not currently using it at work as we have to support wheel users :/
Getting pixi-build FOMO!
I'll have to start using it for my own projects.
Wow! That's nearly the worst possible scenario! 😨
Seems a super powerful capability for debugging builds!