uv is spectacular
But I’m utterly shocked that UV doesn’t support “system dependencies”. It’s not a whole conda replacement. Which is a shame because I bloody hate Conda.
Dependencies like Cuda and random C++ libraries really really ought to be handled by UV. I want a true genuine one stop shop for running Python programs. UV is like 80% of the way there. But the last 20% is still painful.
Ideally UV would obsolete the need for docker. Docker shouldn’t be a requirement to reliable run a program.
> Dependencies like Cuda and random C++ libraries really really ought to be handled by UV.
Currently, there isn't a way for the packages to specify these dependencies.
In part because of the complexity of explaining exactly what is needed (and where to look for it, if you're expecting the system to provide it outside of what Python manages itself).
See https://peps.python.org/pep-0725/ and https://pypackaging-native.github.io/ for details.
Does uv use a sandbox or do process isolation?
I've switched to running any and all python projects in Docker as a way to ensure that low effort supply chain attacks doesn't easily get everything in my home dir. So even if I use uv, I'd only do that in a Docker image for now
No idea. Probably not?
Docker images are a productivity killer. I don’t want to waste even 1 second building an image. And all the hoops you have to jump through to enable rapid iteration aren’t worth it.
Docker Images are fine - I guess - for deployment. But for development I absolutely hate them.
Ideally uv pip install mypackage should work 100% of the time, but that's far from the case.