smithara
commented
5 years ago My plan for this situation is as follows:
-
to document a list of supported packages on the default kernel - see https://swarm-vre.readthedocs.io/en/latest/software.html (work in progress)
-
invite people to suggest new packages, which we can then consider to check compatibility and relevance (having too many packages could become difficult to manage if dependencies are problematic)
- I am starting to put together an appropriate list of popular packages which I will test myself and put forwards
-
write instructions for how people can add their own Jupyter kernel, but this might have to be more of an "advanced user" option with less support available
This issue to analyse the following use case.
A user may need to install additional Python libraries to run its own code or its own Python libraries. This could be handled "at system level", i.e. adding them to the Python distribution used by the VRE (i.e. the default Python kernel), or providing the user the possibility to create Python environments (and kernels) in its own Jupyter workspace.