emacs-jupyter and inline images in org-mode

emacs-jupyter is an emacs interface for jupyter kernels. It can be installed using

(use-package jupyter
  :straight t)

If ob-async is installed, then jupyter-python needs to be added to the list of languages that are not run asychronously. (Refer this issue). This can be done by

(setq ob-async-no-async-languages-alist '("jupyter-python"))

An example python block:

print("ab")

ab

1. emacs-jupyter works doesn’t work in server

When I first installed emacs-jupyter I found a weird situation where I could not run jupyter code blocks in org-mode if emacs was started as a daemon. On the other hand, if it was started separately as a window, everything ran fine. The issue was that I was adding /home/boticelli/.local/bin/ (path to jupyter executable) to exec-path at the end of the config files. When opening emacs from a terminal as an independent session, it was considering this added path somehow, maybe by inheriting it from the shell itself, but it was. I moved that line to near the beginning of the init.el file itself and this seemed to solve the issue. This issue was very similar to the Pipenv in emacs issue that I had recently.

2. Why emacs-jupyter

Running python code in normal org-mode source block is nice and easy. But using them to plot images was a bit painful/difficult. Since emacs-jupyter uses ipython kernel, it allows some magic commands like %matplotlib inline etc to make this easier. It also gives %timeit which is really useful.

Now, with emacs-jupyter, I can do imports like this and make plotting easier for myself.

%matplotlib inline
import numpy as np
import matplotlib.pyplot as plt

fig, ax = plt.subplots()
ax.hist(np.random.default_rng().normal(0, 1, 50000), bins=100);