nb_conda_kernels has more features, for ex: it would allow us to run tools installed in those environments using %%bash from within the notebooks

we just need to install it within whatever environment has jupyterlab installed. Our Dockerfile seems to use this one as its primary base image, which seems to install jupyterlab in the base environment. So all we need to do is add this to the Dockerfile:

mamba install -y -c conda-forge nb_conda_kernels

PRs from forks do not have access to Github action secrets (source). This means that all PRs from a fork will fail to build.

It would be nice if there was a way to whitelist the TA fork of this repository so that secrets could be made available to it. This is particularly important because the TAs of the class change each year, and we would like to be able to allow the new TAs to commit to this repo without having to ask the DataHub team for push-access each year.

It would also be nice if there was a way to whitelist merges from our fork. I believe this is possible through a system similar to that of the conda-forge Github organization.

While reading through the Dockerfile, I noticed that one of the environment variables might be misspelled?

Shouldn't this be R_LIBS_SITE (with an S after the LIB), instead?

As far as I can tell, the setup on Datahub has been working so far, so I'm not sure it's a major concern to address this. But I thought I would mention it just in case. Feel free to close this issue if it isn't something to worry about.

Our conda environments are re-solved upon each build but, in theory, a re-solve is only necessary when a yml environment file changes.

• We should use conda-lock to create lock files for each yml file, and we should make sure to cache them
  • This might require creating a separate portion of our Dockerfile for this
• Whenever a yml file changes, we should detect it and go update the corresponding lock file in the cache
• Then, within our Dockerfile, we should copy the lock files from the cache into each build and then use them to install each environment

See also this description of how to use conda-lock with Docker.