Notebook to agree with Poetry settings

[rolyp]

Running a notebook via poetry run jupyter notebook in the project folder (or starting poetry shell and then running jupyter notebook) should, I think:

• ensure the Python version in the notebook (when saved) agrees with the version chosen by poetry
• ensure all the Python dependencies specified in pyproject.toml are available

Unfortunately, neither of these things seems to be true: when @edwardchalstrey1 saves his notebook, it records the Python version as 3.7.4, and when I try import pytest, it fails with ModuleNotFoundError. The former in particular is annoying as we now get spurious differences when collaborating on the same notebook.

[edwardchalstrey1]

Fixed this by doing poetry add jupyter ipykernel - will PR

[rolyp]

@edwardchalstrey1 This all looks great. Annoyingly, I'm still getting a Python discrepancy (it seems I have 3.9.1 installed into my local environment, whereas the notebook is now saying 3.9.7, presumably the version you have installed?). I assumed that poetry.lock would somehow fix the Python version in a way that satisfies the python = "^3.9" constraint in pyproject.toml, but I can't see anywhere that the Python version is set in poetry.lock. (See this issue for discussion.)

I'm going to experiment with changing the Python version constraint in pyproject.toml from python = "^3.9" to python = "3.9.7".

[rolyp]

Another annoying Poetry feature seems to be that if you change the Python version (to one that isn't installed locally), then none of poetry update, poetry lock and poetry install will attempt to change the Python version (and thus discover that it isn't available). But if you then re-clone the repo and attempt a poetry install from scratch, it will fail.