Something I wanted to try out required python 3.13, and I'm currently on 3.12 on my dev machine. Time for an upgrade! `conda update python` just spent a dozen minutes contemplating it's navel, so I killed that. `uv python install` on the other hand? 4.35 seconds and done. - ThreadSky

shrydar.bsky.social • 90 days ago

Something I wanted to try out required python 3.13, and I'm currently on 3.12 on my dev machine. Time for an upgrade!

`conda update python` just spent a dozen minutes contemplating it's navel, so I killed that.

`uv python install` on the other hand? 4.35 seconds and done.

Comments

seanholm.bsky.social•90 days ago

Have you tried Mamba? It’s a drop-in replacement for conda and it’s pretty darn fast.

shrydar.bsky.social•90 days ago

I've not! I might have to do a compare and contrast now, thanks for the tip.

I started using conda about three years ago because I wanted to mess about with some ML things before starting work somewhere it was pertinent, and brew+vanilla pip were having issues getting pytorch etc installed.

shrydar.bsky.social•90 days ago

Definitely time for a change now, conda was also very sluggish when installing various radio-astronomy related things a few weeks ago.

shrydar.bsky.social•90 days ago

(argh! "its," of course)
(also, note that that was 4.35 seconds over wifi, just tethering my laptop to my phone's 4G internet connection)

shrydar.bsky.social•90 days ago

Ok ok, so that doesn't update my default conda environment - it just updates what new virtualenvs use by default (and ensures it's already ready to go - `uv venv` runs in less time that it takes to hit enter on the command), but I think I'll be trying out dropping anaconda altogether pretty soon.

johnfredcee.bsky.social•90 days ago

I manage my Python installs manually with batch scripts to switch environments. It’s nice and simple here in the Stone Age.

seanholm.bsky.social•90 days ago

Same here :)

shrydar.bsky.social•90 days ago

I mostly just use the one conda environment that activates by default, and rarely use much more than numpy + scipy + Pillow + matplotlib, and save virtualenvs for when I've got a thing that needs deploying, or for experimenting with a library I don't want to pollute my default env with..

shrydar.bsky.social•90 days ago

...but I just did a `pip freeze|wc` and I'm a bit horrified. Definitely time for some housekeeping.

shrydar.bsky.social•90 days ago

Haha nice. I was just looking at maybe using something like https://github.com/hyperupcall/autoenv for activating virtualenvs when I enter project directories..

johnfredcee.bsky.social•90 days ago

I don’t think I use enough Python for really fancy tooling!

Comments

Posting Rules

Reply