@anntzer Sorry I'm still not too familiar with this Git system. Have I now created 2 commits or will the second superseed the first automatically?

You have two commits. You can squash them into one yourself (see squashing at https://git-scm.com/book/en/v2/Git-Tools-Rewriting-History, probably a good thing to learn) but you can also leave it as it is and we can squash it (or not) at merge time; either way is fine.

You are still running afoul of PEP8 though (https://travis-ci.org/matplotlib/matplotlib/jobs/283269976#L2977).

Well, I didn't use Git on my computer but the online editor (simply clicking the pen symbol). So I guess squashing is not possible here?
I thought this was a small, quick fix, now it turns out several people have to work on it for hours. :-( Sorry for that.

Sorry for the meta comment, but why not use "Squash and Merge"?

We do use squash and merge (on a case by case basis) but we may as well try to teach the right way (tm)...

In any case this looks good now.

@jklymak Meta comments are good, we should adopt how we do things as it makes sense to.

@ImportanceOfBeingErnest I took the liberty of fixing the trailing white space issue.

Both of the issues have been style related. We run pycodestyle (with only a sub-set of the warnings / errors enabled) on travis and make sure that it passes. This seems trivial and can be a bit frustrating, but it does help to keep the code style consistent across the code base.

Looks like there's still some whitespace somewhere that travis doesn't like.