Specifying file doesn't seem to work correctly?

[emacs-jupyter]

Specifying file doesn't seem to work correctly?

[non-Jedi]

Probably a dup of #16.

[nnicandro]

Is your Org version >= 9.2? The semantics of :file has changed in Org >= 9.2, you also need to specify :results link for it to work. Specifying :file by itself now means to write whatever the source block returns to file.

So it should be

#+BEGIN_SRC jupyter-python :session py :file ./test.png :results link
import matplotlib.pyplot as plt
plt.plot([1,2,3,4])
#+END_SRC

However if you specify :async yes everything should work without :results link. You may also need to add %matplotlib inline before the plot call, see https://ipython.readthedocs.io/en/stable/interactive/plotting.html#id1.

Also note, there is an issue about consistent use and meaning of header arguments #40.

[AoifeHughes]

I'm using org 9.1.9 - Updating to use async works. Thanks!

Any idea why the async is required?

[nnicandro]

For :async yes we have a custom result insertion method that doesn't go through org-mode (the :async no case does) and therefore doesn't follow the semantics of org-modes header arguments in the same way. The main reason this is done is so that when :async yes is specified, results are inserted in the buffer as they come in, much like a Jupyter notebook. In the :async no case, the results are collected first and then inserted (through org-modes result insertion) after the code block finishes.

I opened #40 to address the consistency of how header arguments are handled in the :async yes vs :async no case.

[nnicandro]

Closing this as #40 is more relevant.