@dstansby the eval is just to prevent redundant definition of the string 'AutoDateLocator(maxticks=8)' and the locator itself, i.e. it's meant as a shortcut of writing e.g.

locators = [
    ('AutoDateLocator(maxticks=8)', AutoDateLocator(maxticks=8), '2003-02-01', '%Y-%m'),
    ...

Maybe using eval is not really best practice but it reduces the possibility of mistakes/inconsistencies between the code and the text in this case.

I guess I but the argument that the eval is useful here...

Maybe using eval is not really best practice but it reduces the possibility of mistakes/inconsistencies between the code and the text in this case.

Would a repr on the objects (assuming it exists) print anything different from the eval?

@story645 repr(AutoDateLocator(maxticks=8)) gives '<matplotlib.dates.AutoDateLocator object at 0x000001CE4F687CA0>'.

Defining nice reprs wouldn't be enough, however, as we also need the arguments of the function calls.

Defining nice reprs wouldn't be enough, however, as we also need the arguments of the function calls.

Why couldn't those be in the repr?

@story645
I only have very basic knowledge of python in this respect, maybe I'm completely wrong:

When I try AutoDateLocator.__repr__ = lambda self: f'AutodateLocator(maxticks={self.maxticks})' then repr(AutoDateLocator(maxticks=8)) returns 'AutodateLocator(maxticks={0: 8, 1: 8, 3: 8, 4: 8, 5: 8, 6: 8, 7: 8})', i.e. the class variable maxticks. I don't know how to get the value passed in during the function call. But even if this is possible, I find defining all the repr rather verbose and a bit hacky as compared to a simple eval.

In an ideal world we would define sensible __repr__s on each of the locators/formatters so they could be used - I think that's a stretch goal, and this PR is worth considering regardless of that.

Thanks @StefRe for your very helpful contributions!

@jklymak Thanks!
What about the rrulewrapper issue I mentioned at the very beginning? Open a new issue for this? Should I (for the time being) remove the reference to rrulewrapper as there's no link:

Sure, open an issue. Seems the solution is to add a docstring to rrulewrapper if it is public API...