I worked on a demo image to have jupyterlab working inside docker and accessible from the host machine.

The documentation and usage is detailed here.

But I’d like to take a moment to summerise some pros of this solution:

1. It allows a standardised development environment.
2. It ensure any analysis performed in it is fully repeatable.
3. It runs on pretty much any machine that runs docker.
4. If you screw things up, you can delete the container and start from a brand new one. To keep data you just need to mount a volume (as explained in the documentation).
5. You can install libraries that only run on linux.