I always hared the environment lol. My stubbornness justified.
Academia
Conda is needlessly convoluted
If I need anaconda, I will install everything it includes one-by-one from their respective project repositories out of spite
This could potentionally be a really big deal for a lot of research labs... I know quite a few who have whole workflows that revolve around conda
.
What's even more troubling is that according to their pricing page, there are only "per user per month" subscriptions... That's is going to be really expensive.
That said, the article does mention that it will remain free for coursework.
I've switched to poetry and I'm very happy
My colleague sent me this Is conda Free?.
It appears that if you just use conda
and the packages from conda-forge
(ie. non-commercial packages), then that will remain free:
conda, the software program for package and environment management, is free to use by anyone. > It’s open source, too. The conda-compatible packages in conda-forge, Bioconda, and almost all other publicly accessible channels are free to use by any one.
That said, this is dated Aug 29, 2023, so who knows if that still applies?
hah, my installation is nearly broken, I suppose if they demand payment they'll start to offer support services to those institutions.
Who among us hasn’t tanked at least a dozen conda environments?
Good news, everyone: gave up trying to fix it, so I finally installed mamba from scractch (miniforge) and it's fast again! 😍
Can anybody explain what conda provides over python + pip or python + poetry? I've never needed conda in all my years writing python for various purposes - except data science, but pip install numpy
and pip install pandas
had me up and running in <1 minute for some test projects.
Do you use venv? Conda gives you that capability (which is absolutely critical with the state of interconnected and complex package relationships) plus a package manager. I’m not sure what conveniences conda may provide over pip/virtualenvs but I imagine there is something that conda gives as an advantage for package publishers. I’d be curious about that myself having not pushed any packages to pypi or a conda repo.