konsti

@ryantm @charliermarsh (We currently don't support multiple package outputs, but I'm curious about your use case)

@ryantm @charliermarsh What's the difference between the packages that you're looking for?

The uv build backend is on the front page of HN✌️

@KevinKaichuang uv venvs should work just like with a regular remote venv, though pycharm doesn't run uv commands on the remove (yet). anything specific that's not working with it?

is it possible to use a venv set up with uv as an ssh python interpreter in pycharm?

you can solve sudokus in python packaging not python code, python packaging [project] name = "sudoku" version = "1.0.0" dependencies = [ "sudoku_3_1 == 2", "sudoku_5_7 == 6", "sudoku_0_7 == 5" ... ] github.com/konstin/sudoku…

Announcing the Astral OSS Fund. We're giving > $3,000 per Astral team member per year to open source projects, maintainers, and foundations, inspired by @getsentry's OSS Pledge. astral.sh/blog/astral-os…

@HenrySchreiner3 @peteronyisi1 @HEPfeickert @prefix_dev What do you think about supporting overrides for requirements at the application and maybe even library level as a remedy?

@peteronyisi1 @HEPfeickert @prefix_dev It's basically a best-effort mechanism in conda-forge, and it really helps that conda-forge is centrally maintained. PyPI is up to the package authors. (Regardless, this would probably be supported, though, if someone put in the massive amount of work it would require)

So is there just no way that Python package managers can retroactively declare upper bounds on their support for the libraries they depend on?

But our goal is to evolve uv into a "Cargo for Python": a comprehensive project and package manager that's fast, reliable, and easy to use, wrapped in an opinionated workflow. Think: a single static binary that bootstraps everything you need to be productive with Python.

Announcing uv: an extremely fast Python package installer and resolver, written in Rust. uv is designed as a drop-in replacement for pip, pip-tools, and virtualenv. astral.sh/blog/uv

@pganssle cool! sent you an email

@konstinx Hey, I am interested in meeting up, do you want to set up a time to meet? Shoot me an e-mail (ganssle.io/contact) to coordinate?

i'll also be at #PyConUS if you're interested in chatting about maturin, pyo3, evolving packaging standards, python tooling in rust or just python packaging in general

@domenkozar there's also some (very different looking) data in #PythonPackaging" target="_blank" rel="nofollow noopener">lp.jetbrains.com/python-develop… and #GeneralPythonUsage" target="_blank" rel="nofollow noopener">lp.jetbrains.com/python-develop…, but they didn't explicitly ask for other package managers

@domenkozar i copied it from the pdf linked from discuss.python.org/t/packaging-vi… : drive.google.com/file/d/1U5d5Si…

i've reimplemented PEP 440 (the python version standard) and wrote about the depths of PEP 440 and how to evolve it for the future cohost.org/konstin/post/5…

@samuelcolvin they now also showed up in a new repo of mine just to complain about maturin 🫠

WOW being a rust hater looks like really hard work... github.com/pydantic/pydan…

@SDuninHorkawicz For the cosine search, have you seen github.com/facebookresear…? That should solve the maximal size of the target database, without much optimization we got e.g. 313k pLM embeddings searched all-against-all in 77s, and their benchmarks go up to 2 billion embeddings

@SDuninHorkawicz I benchmarked length-dependence rather extensively in frontiersin.org/articles/10.33… . Funnily enough for single domains, performance generally increases with length. (knnProtT5: cosine similarity on flattened embeddings of Pfam ful-length proteins, 2nd picture is knn on CATH <20% PIDE)