Dependency management and Anaconda ToS

Ray Libraries (Data, Train, Tune, Serve) Ray Serve
1

Hello everyone, first timer here.

I was reading the deployment guide, specifically on the Dependency Management section, and it states that conflicting package versions between services are handled by conda envs. I’d just like to double-check whether this management does not violate the Anaconda ToS (most importantly, does not use the default channel and does not use any anaconda tools).

Thanks!

2

I think it should be fine, the feature calls into the user’s existing conda installation, using the public conda command line API to programmatically activate and create conda environments on the user’s machines.

3

Indeed, I gave it some more thought and I ended up ditching my old Anaconda environment and switched to a miniforge-based one. Thanks!