Yet another python packager
...............insane that such a popular language still doesn’t have this basic problem solved.I use poetry and it works really well. I would consider it solved but that doesn’t mean there isn’t the possibility of a better solution.
It’s a really bold claim. Every time a new package manager and/or dependency resolver comes around, we have the exact same headline
There are 14 competing standards…
It is a bold claim, but based on their success with ruff, I’m optimistic that it might pan out.
I do enjoy ruff a lot, but only time will tell
have there been a lot of them?
pipx, poetry, pipsi, fades, pae, pactivate, pyenv, virtualenv, pipenv
Let’s hope this next one will be the true standard.
pyenv, virtualenv, pipenv, aren’t package managers… they are virtual environment managers / creators and use pip for package management.
We’re using poetry and it solves our problems. I’ll have to look into uv, but I don’t feel in any rush to switch away from poetry.
I’ve been mostly a poetry guy but have tested out uv a bit lately. Two main advantages I see are being able to install Python (I relied on pyenv before) and it’s waaay faster at solving/installing dependencies.
Yeah, it certainly looks nice, but my problems are:
- everything runs in a docker container locally, so I don’t think the caching is going to be a huge win
- we have a half-dozen teams and a dozen repositories or so, across three time zones, so big changes require a fair amount of effort
- we just got through porting to poetry to split into dependency groups, and going back to not having that is a tough sell
So for me, it needs to at least have feature parity w/ poetry to seriously consider.
You should be using dockers cache mounts
https://docs.docker.com/build/cache/optimize/#use-cache-mounts
uv is still faster with a cold cache
and uv does have dep groups
Just tell me how uv is financed …
In my field we rely on conda and I hate it every day.
Why, out of curiosity?
We do geodata science and rely on some pretty specific C++ libraries that are only distributed via conda. While on unix-based systems it’s possible to get some of them from other channels or even building them from source, we mostly have Windows machines in production where we are not that flexible. Docker is unfortunately no solution due to security concerns.
If you are asking why I hate it: It’s bloated, uses more space than needed and it’s rare I can reproduce an environment from the environment file without running into errors. Using it feels unintuitive, I still google command after years. It was very slow until recently, when the libmamba solver was finally integrated. Last but not least licensing is a pain in the ass.
I’ve been using micromamba/mamba and not had solving issues like I did with conda. Im glad conda integrated libmamba.
Question: why were docker containers deemed security risks?
I share the same frustration trying to replicate an environment. I’m glad I can avoid it these days, the community needs a way out of the conda lock-in.
Interesting. We use conda via micromamba for my own project, as it makes the install for end-users much easier when they can just run a shell script, to install python, cuda, and all the dependencies needed.
Got toml file support yet? Then I’m happy to talk :)
they do, just use project management commands like
uv+ {add,remove,sync,lock,run}Looks like it has basic support:
required-python = "..."dependencies = [ ... ]
Once it gets dependency groups, I’ll try it out. I’m currently using
poetry, which works, but I’m always interested in better perf.it already has dep groups; e.g.
uv add --optional staging pytestthen
uv sync --extra stagingto install / uninstall packages accordingly.
They have a
--devshorthand for dev dependencies, but it seems the dependency group PEP is not final, so there isn’t a standardized way of doing this yet.Private PyPI too?
We’re coming from poetry but it’s slow and needs its own .venv, so a UV binary would be very nice.
yeah, it works with private pypi
I don’t find it in the docs…how to set my PAT?
