I always have a lot of problems setting up conda (other than hating the idea of "polluting" my home folder/system with other packages...), and I'm quite used to devcontainers as a way to have a common and stable development environment across computers or even developers (if you are interested :-)
This adds a simple configuration for a devcontainer environment, which inherits the micromamba devcontainer image (https://github.com/mamba-org/micromamba-devcontainer), and installs all the packages specified in the environment-dev.yml file, along with a few sane vscode defaults. This image uses micromamba and not conda, which I find to be much faster and much lighter on my disk space.
If you want to try it you need to run .binder/postBuild before running jupyter to use xvfb
Yes, micromamba is much nicer to use. I use it on my computer, it's easy to set up, has no base environment.
I set CONDA_PKGS_DIRS and CONDA_ENVS_PATH to control where the data is stored.
I always have a lot of problems setting up conda (other than hating the idea of "polluting" my home folder/system with other packages...), and I'm quite used to devcontainers as a way to have a common and stable development environment across computers or even developers (if you are interested :-)
This adds a simple configuration for a devcontainer environment, which inherits the micromamba devcontainer image (https://github.com/mamba-org/micromamba-devcontainer), and installs all the packages specified in the
environment-dev.yml
file, along with a few sane vscode defaults. This image uses micromamba and not conda, which I find to be much faster and much lighter on my disk space.If you want to try it you need to run
.binder/postBuild
before running jupyter to use xvfb