lwasser
commented
2 years ago I think we can close this easily with a few fixes but i want input from the group
- jinja has been removed as we removed distribute.py so it's not called anywhere except as a requirement in the setup.py file. ok to remove?
- nbclean has been removed already (via which PR i think karen's cleanup pr maybe removed some of the old stuff??)
- distribute.py has been removed via some pr - probably also karen's cleanup pr.
- papermill is called here in the circleci.yml file. given we no longer use this approach (autograding) can we remove it. i suggest we do. it was a super cool build when we had it but way beyond the current scope of this package.
we should also remove this folder as it's 3 years old and no longer relevant. i think this was an old notebook where we were trying to create a demo for autograding
Currently our dependencies reflect an older version of abc-classroom where we w ere building notebooks via circle ci using papermill. The dependencies including:
notebook.pycould also be removeddistribute.pyuses this - we could potentially deprecate this module as we are not using it. this is how we built the feedback template hosted on circle ci for easy viewing.notebook.pyuses this: this is how we ran the notebook on circle ci. to make sure the notebook ran, we needed to normalize the kernel to ensure it found the right kernel. All of this can be done with github actions so we could also deprecate this module as it is now out of scopenotebook.pycalls this as wellThe docs are failing (see #414 ) now because of the version of jinja installed and i think it's a good time to revisit what dependencies we actually do use in abc-classroom and remove unneeded ones to make the package more light weight.
I think it would be best to clean notebooks using nbconvert as i'm not sure if nbclean is actually supported anymore.
Chris hasn't worked on it since 2018. the one thing nbclean does is allow you to scrub PARTS of a cell. so i t was developed to implement the nbgrader functionality that strips hidden tests from the notebook. i believe nbconvert will force us to remove entire cells. With that said, i don't think that is a terrible thing to do just that - tell the user if they want to remove tests they will tag the cell and it will be stripped when feedback is created. you could then have hidden tests in cells and visible tests in other cells if you wanted to keep both in the notebook.
here is the code for removing cells - https://github.com/jupyter/nbconvert/issues/1194 it looks like the docs didn't update when the pr was merged to update the code. i think i know why...