coderefinery / jupyter-retired Goto Github PK

the difference between regular git diff output and nbdime-diff output is not apparent to new users

before going into the basics_of_jupyter notebook, spend a few minutes showing either the Gravitational Wave or the Activity Inequality papers and their supplementary info which is in jupyter notebooks. This also creates a connection to the Git lessons since the material is on GitHub.

Optional: clone the respective repos and try to reproduce them (one notebook from Activity Inequality material may fail due to python environment issue)

since the code snippet is too long and it creates 5 figures

leave out some magic commands, perhaps list them at the bottom for reference

showcase services like binder.org

We are not sure why. Import statement gave no error but no widget appeared on the website. Across several browsers.

since it gives away the solution

after the presentation, open a new blank notebook to demonstrate the cells, buttons, keyboard shortcuts, etc. Then open introduction notebook and go through it

e.g. impractical beyond a certain size of the code

like sharing notebooks, jupyterlab, jupyterhub, further reading

leaving it inside the python part results in some people missing it

the answer (involving how gitbash is set up) is in this thread:
jupyter/help#181

See good list here: https://twitter.com/lexnederbragt/status/940275053484281856

remove a few minor things like %capture, %pastebin which aren't that important.
shorten %debug demo a little

IMO the time spent to explain the "git diff" exercise is disproportionate to the benefit. On the other hand last time we ran out of time with the nice interact examples which I think should come first. IMO the "git diff" part can be left out if time is short since I find it more important that people see the power of notebooks rather that everybody understands the pitfalls of JSON diffing.

• RTD supports notebooks
  • some nbextension is required
• https://mybinder.org/
• https://notebooks.azure.com/
• https://colab.research.google.com

We demonstrate how to mix R and Python, but don't discuss when that might be useful. We should mention possible use cases (e.g. python person who wants to use R code from colleagues, or prefers the plotting capabilities of R, etc)

it's not sufficiently apparent or clear on https://coderefinery.github.io/installation/python/#jupyter