- Jupyterhub works well for large classes
- Students can install everything they need from Github to Jupyterhub even when not command line savvy with
git clone URL
- Jupyter notebooks work well as lab manuals, but notebooks would be much more useful if text formatting was better
- Google docs are better as lab manuals since I can edit and fix things live, thats always necessary
- The Jupyter notebook bash kernel is not perfect, and it’s export-as-PDF is mediocre
- Anything can go wrong, test everything, no really, everything, twice. Test in the student’s working environment, it will be very slightly different and surprise you
- Have students repeat new skills several times, in slightly different contexts, nobody gets this the first time, and it builds muscle memory
- Always have a story with a problem, and then introduce the skills to solve it, not vice versa
- Some students will do things you do not expect, have ways to catch them and help. For example, students submitting lab reports of >1000 (one thousand!) pages need catching early and advising on the correct use of
grep
- You may have hidden gems of students who do all the work set, then something extra and inventive. Pay attention to them also, struggling students are not the only ones who deserve your time.