Clemens Brunner

Re comparisons with `plt.plot()`, seaborn has functions that plot more complex figures (with multiple axes) that still do not return a figure (e.g. https://seaborn.pydata.org/examples/many_facets.html), but do it through some container...

@agramfort yes, I agree that it is too big of a change RN (and I did hear some people wonder about `_ = ` and `;` so I'd avoid these...

@agramfort it is broken to me, I don't want to explain to students why there are two identical outputs. This is very confusing and a bad UX. If it's easy...

> I agree, IMHO it's broken and it's confused me and my students/colleagues. Adding the semicolon is weird, I never have to do this anywhere but in MNE. It's not...

Thanks @drammock, that's already a lot of information to chew on. But the workaround is not just a single keystroke. I'm talking about using a notebook to write course material,...

If I'm calling a plot function I'm expecting that exactly one plot appears. Everything else is a workaround. If you really need to keep the figure for later, then yes,...

I think I found a nice solution. What if we didn't return a `Figure`, but a small custom object which contains the `Figure` as an attribute? This would solve the...

I think our CSP implementation has a problem with rank deficiencies. This problem keeps popping up (e.g. https://mne.discourse.group/t/csp-returning-linalgerror/3567), and I just stumbled into this error myself with a completely different...

I've wrapped the lines black-style. I'm OK with the proposed description style, the only thing I'd add is the default at the end.

I'm also wondering why almost all keys are prefixed with `MNE_` – why? It would be easier if we just removed this prefix.