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
Recently, conda decided to switch per default to the mamba solver in their new version, to avoid the known speed issue of the classical solver when solving big environments :
Currently, pySDC testing is based on mamba because of that, but it seems that on a short term conda will use mamba under the hood, so one could use the latest development on conda (e.g parallel installation of packages) while keeping the mamba speed when solving environments.
Going back to conda as default installer would then maybe simplify the installation process (for tests, users, ...) while avoiding problems when mamba version is (temporarily) de-synchronized with conda (e.g this recent problem : mamba-org/mamba#2882).
I'd be happy to roll back to conda as soon as this works reliably. So far we did not have any problems with micromamba, being blazing fast and quite reliable.
Recently, conda decided to switch per default to the mamba solver in their new version, to avoid the known speed issue of the classical solver when solving big environments :
https://conda.org/blog/2023-07-05-conda-libmamba-solver-rollout/
Currently, pySDC testing is based on mamba because of that, but it seems that on a short term conda will use mamba under the hood, so one could use the latest development on conda (e.g parallel installation of packages) while keeping the mamba speed when solving environments.
Going back to conda as default installer would then maybe simplify the installation process (for tests, users, ...) while avoiding problems when mamba version is (temporarily) de-synchronized with conda (e.g this recent problem : mamba-org/mamba#2882).
@pancetta what do you think ?
The text was updated successfully, but these errors were encountered: