I just made a sidecar viewer of output you can run in jupyter console or even a regular ipython terminal to get rich interactives without a notebook. pip install runtimed import runtimed runtimed.sidecar()

Seriously, shall we still teach students how to even program?

"The Synthetic Senior" — my #FOSDEM talk on AI + open source mentorship The 3 C's for finding contributors worth investing in: 💡 Comprehension ⚙️ Context 🔄 Continuity Slides: docs.google.com/presentation...

You might check out some of the resources from Scientific Python if you haven't already: * learn.scientific-python.org/development/... * lectures.scientific-python.org/advanced/int...

Question: I want to practice writing functions in C, then bring those functions into a Python package via ctypes. What is the best way to build the project since I am on a Windows OS and thus my compiler is different from Mac and Linux? #python #c #programming

LLMs are always confident about being able to help you, complete a task or answer a question. They are excellent at hiding their imposter syndrome 😂

Looking for examples: do you know any OSS projects that have thoughtfully updated their contribution process for AI-assisted PRs? Interested in: * PR templates asking for intent/reasoning * Comprehension checks * Other friction-adding methods Would love to learn from communities experimenting here

There is plenty on using AI agents to write code. Curious if anyone has practical patterns for using AI agents to review PRs (including agent‑written code). Got any blog posts, how‑tos, or examples you’d recommend?

I ported a Python library implementing a full HTML5 parser to JavaScript using GPT-5.2 and Codex CLI in 4.5 hours, and decorated for Christmas and watched Knives Out while I was doing it simonwillison.net/2025/Dec/15/...

Do you know if someone has attempted the "meta challenge" of building a repo that will write an implementation in your favourite language? Would be interesting to see how well the idea "use LLM as compiler of specs to super personalised implementation" works today

Using AI in "agent mode" you realise how much of building software is about thinking and how little it is about writing code. When writing code suddenly takes ~zero time I realise how much time I used to have while typing to think about what I was doing

I think GitHub should make an update so when someone positively reviews a PR it adds them as a coauthor. That way they would show up in the git blame. In the age of AI development the reviewers are even more responsible than the committers for the code that gets merged.

In a few years time, will software libraries ship code or inputs to LLMs that users then "build" into specialised libraries by adding constraints for their use-cases?

A new version of scikit-learn has been released 🥳 check out the highlights: scikit-learn.org/stable/auto_... Thanks everyone who contributed to this release! Let me know what you think of the experimental GPU support

Over the last few months the @scikit-learn.org has been spending extra time on improving the open-source supply chain. My take away: don't stray too far from the standard and boring path when setting up CI. Too many ways of getting hacked. Thanks to @github.com blog.scikit-learn.org/press/gh-sosf/

You all know about "Eternal September" - I think I just figured out what I'm calling my AI agent editor: my eternal summer student :D

In the age of AI, LLMs and code assistants what does it mean to be an open-source project that is welcoming to newcomers? As a newcomer, is it enough to use AI to solve a good first issue and submit it as a PR? As a maintainer, what should you do with PRs like that?

Today at #EuroScipy2025, @glemaitre58.bsky.social and I presented a tutorial on pitfalls of machine learning for imbalanced classification problems. We discussed what (not) to do when fitting a classifier and obtaining degenerate precision or recall values. probabl-ai.github.io/calibration-...

I scraped the schedule for Open Sauce 2025 this morning and built an alternative schedule interface with the option to add everything to your calendar (via ICS)... working entirely on my iPhone, using OpenAI Codex and Claude Artifacts Prompts + transcripts: simonwillison.net/2025/Jul/17/...

DID YOU KNOW: Speakers and tutorial or workshop leaders get free registration to #JupyterCon? DID YOU KNOW: The #deadline for submissions is Sunday, July 13 at 11:59 PM PDT (UTC -7) events.linuxfoundation.org/jupytercon/p...

Has someone tried building a "stalebot" (closes stale GitHub issues) that uses a LLM or some such to decide if the issue is stale because the maintainers have not responded or because the creator hasn't followed up?

Use #Python at all? I would like to do a silly thing, which I hope will bring amusement to others as well as me (if it works out). If you wouldn't mind, please fill out this form, and boost for reach: docs.google.com/forms/d/e/1F...

What about using cursor rules files to encode project maintainer knowledge? Put frequent PR review comments in these rules files so that contributors (that use AI tools) can fix/avoid them? Has anyone tried this? Useful? Too brittle?

Harry Frankfurt's "On Bullshit" should be mandatory reading for people who use LLMs to generate text

Any scikit-image fans in Vienna? The skimage team would like to meet in Vienna for a code sprint Aug 14-16 ahead of EuroSciPy. We'd super appreciate it if someone's workplace can host us! All we need is electricity, natural light, and (most important) walking distance to coffee. ☕️😂🙏

<proud advisor> Hot off the arXiv! 🦬 "Appa: Bending Weather Dynamics with Latent Diffusion Models for Global Data Assimilation" 🌍 Appa is our novel 1.5B-parameter probabilistic weather model that unifies reanalysis, filtering, and forecasting in a single framework. A thread 🧵

Bunch of hippies! ;)

Has anyone experimented with making cursor/windsurf/etc rules to help contributors to open-source projects like numpy, scikit-learn, scikit-image and so on? If you could make rules that include some of the knowledge of the maintainers that could be cool

Import tariffs for your Python imports! pypi.org/project/tari...

So war der Winter 2024/25: Schneearm, ausgeprägtes inneralpines Altschneeproblem, wenig tödliche Lawinenunfälle Alle Infos gibts Winterflash: www.slf.ch/de/lawinenbu... 📷 N. Büchi 1/4

What is a nice and short term* for referring to the collection of Python tools and libraries in the data science ecosystem? * should not be trademarked

Announcing a new event initiative: Feminist AI LAN Party! Katharine and I did a pilot event last year and we're now taking it to @pyconde.bsky.social. We've also open-sourced event kits to make it easy to host your own, including: 💣 hacking LLMs 📑 data development ✂️ zine making feministai.party

BREAKING! Guido van Rossum, the legendary creator of #Python, has officially reinstated himself as Benevolent Dictator for Life (BDFL). Feat. Guido van Rossum, @pumpichank.bsky.social, @snarky.ca and @mariatta.ca‬ Stay tuned for the documentary coming this summer! www.youtube.com/watch?v=wgxB...

The one crazy trick open-source maintainers don’t want you to know for getting your PRs reviewed: get a maintainer excited about it before making the PR

Loky 3.5.0 is out! Loky provides an extended version of Python's `concurrent.futures.ProcessPoolExecutor` that leverages cloudpickle to work within interactive Jupyter sessions on all platforms and reuse existing workers to hide the overhead of starting new workers each time.

TIL every fruit is a vegetable, but not every vegetable is a fruit.

The @metoffice.bsky.social uses yellow, amber and red for weather warning severity. They have a nice lookup table to collapse the two dimentions of impact and likeliness. You could do the same for impact and percentage of community affected by a change. www.metoffice.gov.uk/weather/guid...

We're exploring ways to support scientific reproducibility and sustainability with Binder. What if we could reduce the gap between where researchers conduct their computational analyses using open source technologies and bridge them to publishing platforms for dissemination?

📢We are excited to announce the keynote speakers for our 2025 conference: - Alenka Frim from United.Cloud - Ralf Gommers from @quansight.com - Anita Graser from the Austrian Institute of Technology Join us at PyData Paris at Cité des Sciences from Sep 30 to Oct 1, 2025. medium.com/@PyDataParis...

A rising tide lifts all boats. In this case: bet on the right technology and profit from its commodification (cheaper prices). But you do need a long breath, it took nearly a decade of steady progress!

I have a bunch of Python classes that I would like to "wrap" in a class that does $something but to the user looks just like the original class that it wraps (methods, attributes, constructor, etc) How would you do that? Finally a use-case for metaclasses? Class decorator? Something different?

View from last week

Its 2025. Is Keybase still the easiest way to do end to end encrypted got repos?

Don't use decorators, custom dunders, and all the other clever stuff the Python language tempts you with. Just don't do it. A tiny fraction of the code I've read in my lifetime uses these fancy things, and a tiny fraction of that code _should_ have used them. In most cases it is not needed

❄️ The Christmas release is here! ❄️ Introducing scikit-learn 1.6 with: 🟢 2 major features & 34 improvements 🔵 5 efficiency boosts & 21 enhancements 🟡 14 API changes 🔴 30 fixes 👥 160 amazing contributors youtu.be/7wiHChpwJe8

Last week the Python package "Ultralytics" suffered a supply-chain attack on its build and release process. This is a review of the attack from @pypi.org's perspective. There's plenty of advice for how Python projects can increase their #security posture: blog.pypi.org/posts/2024-1...

bsky is to twitter as X is to hackernews? Any ideas what the new HN is? It feels like the original one is not as exciting as it used to be in its early years. Though I can't pinpoint why really, just feels less interesting

Issues and PRs are free, no need to close them!

Have you ever wanted to unpickle a @scikit-learn.bsky.social model you trained with version X while using a newer version X+1? If yes, why? When? How? I'd be interested to hear about your use cases to see if we can make it less painful

Please help us test the first release candidate for scikit-learn 1.6: pip install scikit-learn==1.6.0rc1 Changelog: scikit-learn.org/1.6/whats_ne... In particular, if you maintain a project with a dependency on scikit-learn, please let us know about any regression.