You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
However, a short while later (and with no indications in the log where this would happen), it seems a large part of the environment gets reinstalled and consequently overwritten by pip/PyPI:
It's definitely micromamba related but I don't think it's a micromamba bug, rather a conda list bug. It's identifying the Python packages as coming from pypi but they are not because we are not running pip at all in the CI. We could run micromamba list there and compare too. So, in short, yes it's a bug but it's only cosmetic and doesn't hinder functionality afaict.
Noticed something weird in CI recently... Provisioning begins normally
which is populated 100% from conda-forge.
However, a short while later (and with no indications in the log where this would happen), it seems a large part of the environment gets reinstalled and consequently overwritten by pip/PyPI:
From looking at https://github.com/conda-forge/conda-forge-ci-setup-feedstock/blob/main/recipe/run_conda_forge_build_setup_win.bat, I cannot tell what would be causing this (also hasn't changed recently). It also doesn't seem to happen on unix.
Full details:
The text was updated successfully, but these errors were encountered: