Procedures/Options for resolving package conflicts among courses.

[felder]

Summary

Issue #2882 highlights a known problem of multiple courses sometimes requiring different versions of the same package. We currently do not have a lot of really good solutions. Currently we can:

1. Have one of the classes switch hubs.
2. Have one of the classes manually install a different version of the conflicting package.
3. Setup alternative environments (when possible) that have differing versions of the same package.
4. See if the courses are ok with changing the version of the package they are using.

User Stories

As an instructor my course needs a different version of a package that is currently required by another course.

Acceptance criteria

Datahub operations has better options for allowing multiple versions of the same package to co-exist on the same hub until course specific hubs are a thing.

Important information

#2882 #2782 No doubt there are other issues as well.

Tasks to complete

Options and documentation for implementing those options. No doubt different software systems will require different solutions. For example python vs R.

[felder]

In slack @ryanlovett writes:

"one way of addressing package conflicts like this is, other than course-specific images, is to have course-specific library paths. we could do it with an environment variable the same way that we bump RAM. i.e. "if user in course, append /path/to/stat-20/libs to R's library path". then in the class-libs function we can have the courses specify their own library path."

[balajialg]

Just to understand the complexity of different course requirements, Sharing the hub vs course mapping here