Michael Waskom

This is easy enough to accomplish by stacking context managers: ```python with sns.axes_style("whitegrid"), sns.color_palette("Set2"): sns.boxplot(tips, x="total_bill", y="time", hue="sex") ``` Not convinced we need an additional function here.

Closing as matplotlib now raises when both singular and plural forms of these parameters are passed, so there's an explicit exception for the reprex here. It's not the most clear...

I'm going to close as there are no specific plans to expand the variables that can be shown by `catplot` and this is a general enough design decision / limitation...

This is now obviated (without explicit effort to enable it) by #3294: ```python sns.boxplot(data=tips, x="day", y="total_bill", widths=[.2, .4, .6, .8]) ``` It doesn't work with `hue`; enabling that would be...

In v0.13+, `barplot` has an `err_kws` that can be used here: ```python ax = sns.barplot(tips, x="day", y="total_bill", errorbar="sd", err_kws=dict(clip_on=False)) ax.set(ylim=(0, 25)) ``` 

This appears to now generally work with the categorical refactor.

After a few PRs touching the scale machinery but mostly https://github.com/mwaskom/seaborn/pull/3488, the given example here no longer raises.

Note also that the `Path` mark in the objects interface also represents nulls with a gap.

What is that logging call supposed to achieve that is different from “warning” then?

To circle back to the beginning of the thread, Python logging is “opt in” in a very crude sense and a line of code somewhere in your dependency tree can...