Hydrogen: Feature request: Toggle inline plots

[florian-wagner]

Dear hydrogen devs,

thank you for this awesome plugin. After a long journey from spyder over vim-ipython to the jupyter notebook with vim bindings, I found the perfect environment for me: atom + vim-mode + hydrogen.

Icing on the cake would be the opportunity to toggle inline plots. I find typing %matplotlib inline quite unusual in normal Python scripts and most of the time I want to use a mixture of both, i.e. inline mode to keep the overview and see if everything is working properly and interactive mode from time to time to inspect a figure and its data in more detail.

Any ideas on how to accomplish that?

[n-riesco]

We introduced the setting Startup Code precisely for this purpose. Something like this should work:

{"python2": "%matplotlib inline", "python3": "%matplotlib inline"}

[florian-wagner]

Yes that helps to set up inline plots by default, but is there a way to toggle them, i.e. send something like plt.ion() per atom keybinding to the python kernel used by hydrogen? Thanks a lot.

[n-riesco]

As far a I know, Atom doesn't provide a macro recording feature.

Perhaps atom-macros may be of help. I will experiment with atom-macros when I find some time, and post back the results here.

[lgeiger]

I have a (quite strange 😉 ) solution to your problem.

You can create a separate python kernel which uses the inline option. To do this:

• locate the python kernelspecs folder with jupyter kernelspec list
• in there should be a kernel.json file (if not run ipython kernel install)
• duplicate the whole folder
• and change change the display_name in kernel.json to something else (e.g. Python 3 inline).
• add {"Python 3 inline": "%matplotlib inline"} to yourstartup code`

Now you can switch in the menu between the original and the "Python 3 inline" kernels.

[n-riesco]

@lgeiger The problem with switching kernels is that toggling the display mode would require to re-run all the previous code.

[lgeiger]

Yes thats true, but I think changing the matplotlib display mode requires a kernel restart anyway.

[theotheo]

Where to put startup code?

[n-riesco]

@theotheo Edit > Preferences > Packages > Hydrogen > Settings > Startup Code

[wadethestealth]

https://stackoverflow.com/questions/43545050/using-matplotlib-notebook-after-matplotlib-inline-in-jupyter-notebook-doesnt view this thread for solutions/discussion. But it doesn't seem like toggling backends multiple times is recommended.