nbstripout or auto-formatting

[adrn]

In my own projects with tutorials and jupyter notebooks, I like to use pre-commit to run nbstripout to remove notebook metadata and cell output to clean up git diffs and the version control history. I also now use black (again with pre-commit) to apply auto-formatting to the notebook cells.

Would you be open to a PR that sets this up for the Session16 notebooks?

[adamamiller]

We also post solutions in the repository along with the notebooks. I'm not familiar with these tools (though they sound like they would remove lots of headaches for our program), but I want to be sure that in a notebook that has solutions - where it's essential to see the output - we don't lose anything.

[adrn]

Oh, I should have given a summary:

• nbstripout will remove metadata and output from notebooks, but it's possible to set directories and glob patterns to exclude.
• black auto-formats code in a consistent, pep8-compliant way. I don't love all of its choices, but it removes the need to think about code formatting. (As you might expect, it can be a divisive tool and has led to many debates!)
• pre-commit is a way of automatically running things at the git commit stage, so it provides a way of running nbstripout and black to clean up and reformat notebooks before they even get committed to a repo.

To start, we could try just enabling it for the non-solution notebooks in Session16?

I was also thinking it might make sense to make each session its own repository within the LSSTC-DSFP organization rather than a subfolder (cloning the repo takes a long time!), which would allow for more fine-grained control over testing and repository infrastructure (installation requirements, CI, etc.), but just a thought.