executablebooks
Provide a programatic way to add dependencies for jupyterlite
Context
Code that we want to execute may require additional dependencies to be loaded into the kernel. This can be done programmatically.
Currently, we can execute a cell on the kernel using piplite and install additional dependencies in python, however is there a javascript based API for doing this?
Either way thebe should provide a convenience APi for allowing this to be set from a configuration script, rather than being visible in the notebook or UI.
Also related, is it possible to pre-bundle additional dependencies in/with the kernel?
Proposal
No response
Tasks and updates
No response
You can install additional packages to the jupyterlite xeus-python kernel, and a recent jupyterlite PR ( https://github.com/jupyterlite/jupyterlite/pull/698 ) seems to add the abiliity to set a custorm default kernel, which may also be handy.
I've not had a chance to try it yet, but @yuvipanda's https://github.com/yuvipanda/repo2jupyterlite, as well as @martinrenou's https://github.com/jupyterlite/jupyterlite-sphinx are also in the hinterland of the "build an environment' requirement; so there may be approaches or Github Actions that might be shareable or common issues to resolve?
