Hey! Sure :wave: My name's Tom Solberg, and I work...
# welcome
g
Hey! Sure đź‘‹ My name's Tom Solberg, and I work as a machine learning engineer in the games industry. I've been using Bazel and various Python tools (Poetry, Conda, venv) professionally for a couple of years, but they don't work well together in a monorepo. I'm privately investigating Pants as an easier-to-use bazel that works better than the classical Python tools for large repositories. To learn more about how Pants works I'm also building a plugin for kustomize/kubernetes.
đź‘‹ 8
h
“easier-to-use bazel with a Python focus” is a solid description of Pants, TBH!
Interesting plugin idea. What would the functionality be?
g
Managing my cluster(s) and integrating the whole flow from "source on disk" to putting that in an image, pushing it, putting the sha in a deployment and deploying to a cluster. Same as e.g. https://github.com/bazelbuild/rules_k8s.
I'm not sure if easier-to-use yet; but definitely better at Python. So far I'm struggling a lot with "thinking like pants"... Bazel is a lot more explicit, so I'm fighting the "magic" with how I define targets etc. Like this: https://github.com/pantsbuild/pants/issues/17042
h
Oh, sorry we missed your question on Slack, I’ll respond on the ticket.
[When I think of it being “easier to use” I usually think of two things: dep inference saving me from writing explicit deps in BUILD files, and the ability to act on files rather than targets (e.g.,
./pants test path/to/my_test.py
)]