Great news for #python developers! #dependabot now supports #uv #astraluv . Nothing stands between your codebase and almost-instant #dependencymanagement!
Hey everyone thanks so much for all your replies! Wanted to circle wagons since I found a VERY satisfying conclusion!
They JUST added an experimental flag that makes the uv installed python into your current default by adding it to $HOME/.local/bin/python and ensuring that's in PATH. The command is:
uv python install --preview --default
Kudos to the #astraluv folks for all their amazing work!
Update: I am now embracing lock files in CI in most packages I control.
I’m very used to CI being a gamble: every time I push with confidence and see CI going red, my first reflex is to suspect something changed in the environment. I'm just coming to terms with the idea that *it doesn't need to be this way*; especially in projects that already a special CI job whose sole purpose is to catch incoming incompatibilities, it makes more sense that *other* jobs just be stable.
#astraluv #Python
The thing with #astraluv is I don't think I understand lock files, and how they’re supposed to be used/maintained. Maybe it's because I mainly develop and maintain #Python libraries and it doesn't make as much sense in that space: I need to be able to test a range of versions of my dependencies (I have little control over my users’ environment), and lockfiles tend to add artificial restrictions to these ranges, to the benefit of (maybe ?) my immediate confort but to the detriment of my users.
New #webdev related blog post here, introducing my take on what a #Django project template could be for an advanced usage. Obviously #astraluv is there, but also #justsystems , #esbuild , and... #SOPS 
It's very early stage so please tell me what do you think about it 
https://david.guillot.me/en/posts/tech/proposal-for-a-django-project-template/
pipx run uv install uv 
uv is a tool written in Rust that replaces as a 95% drop-in replacement pip, pipx, pip-tools and a few other things.
So far it's looking pretty cool. It's faster, I guess, and the output is a bit neater and more "interactive" or "live".
One great thing is that you can have a globally installed uv and it will operate correctly in whatever your current Python environment, whereas with pip-tools you always need to take care to run the one installed in the environment you're compiling the pinned dependencies for, or it will calculate python version wrong for the dependencies that are conditional on that.
I just converted my open source projects to use uv for installation, with the new setup-uv action ( https://github.com/astral-sh/setup-uv ) and tox-uv ( https://pypi.org/project/tox-uv/ ). (They already used uv’s 'pip compile' command to pin test dependencies.)
For the smaller projects with fast test suites, I saw up to a 3x improvement, thanks to all the setup overhead shrinking (e.g. 1m40s -> 40s).
#astraluv 0.2.15 is out and includes a new tool that I’ve wanted for so long: `uv pip tree` displays the whole dependency tree in your current virtualenv !
https://pypi.org/project/deptree/ Already implemented this but it's much more convenient to have a version of this tool that doesn’t actively “pollute" the environment it’s meant to display.
kudos to @charliermarsh and team 
