Monorepo

[valentinsulzer]

Starting a summary issue to collect thoughts / to do about setting up monorepo infrastructure. The idea would be to keep everything in a single (this) GitHub repository, but have clearer splits between different "projects" within the repo. The main benefit of this is that it makes it easy to keep dependencies up to date, and requires less maintenance of infrastructure. A downside is that it requires smarter testing and infrastructure to keep CI running time down. A good example is scipy.

To do:

• [ ] Set up tests to only test files that have been changed and downstream ones
• [ ] Split file structure into separate "projects" (move tests and docs to pre-file?)
• [ ] Add tags to issues to identify projects

Possible project split:

• Core (expression tree, discretisations, solvers etc)
• Models
• Parameters
• Pack (future; separate pip install?)
• Optimization (future; separate pip install?)

[brosaplanella]

If we split the core, might be good to make it self-contained and pip-installable as well.

[valentinsulzer]

flake8 is one of the things that gets much slower with more files, but it can take a path, so we can easily run it on individual folders

[valentinsulzer]

Possible useful information here https://stackoverflow.com/questions/63129163/how-can-i-correctly-include-a-path-dependency-in-pyproject-toml

We would need to move the contents of pybamm into pybamm/src and then move setup.py, etc into pybamm