Is it a viable strategy to use micromamba with conda-forge for our data science tasks? We’re considering this approach to bypass the licensing conditions associated with conda’s default channel.
Please see the conda-forge blog post for a more thorough discussion.
A workflow starting with a miniforge or mambaforge baseline will use conda-forge as the default environment, and be free of any references to ToS-encumbered channels:
anaconda.comanaconda.org/m2anaconda.org/anaconda
To be absolutely sure, one wouldn’t use any anaconda.org channels. I only know of the the quetz mirror as a non-shady alternative, but it says beta right on the tin:
https://beta.mamba.pm/get/conda-forge/