Michael Waskom

Another problem: `pointplot` already accepts a float for `dodge`, but applies it differently. Therefore, the following example doesn't work: ``` sns.boxplot(x="sex", y="total_bill", hue="smoker", data=tips, palette="pastel", dodge=1.1) sns.pointplot(x="sex", y="total_bill", hue="smoker", data=tips,...

Let's leave the PR open. The idea has merit but it does require a bit more thorough thinking about the various implications. I think that implementing a continuous `dodge` the...

With the new [objects interface](https://seaborn.pydata.org/tutorial/objects_interface.html), the `Dodge` transform has a `gap` option: ```python import seaborn.objects as so from seaborn import load_dataset tips = load_dataset("tips") so.Plot(tips, "day", color="sex").add(so.Bar(), so.Hist(), so.Dodge(gap=.1)) ```...

You can post them in this issue.

thanks @IsNotMyIP. on my view, I see a table of contents off to the right:  does it appear differently for you?

Closing as this has now been moved to the main docs: https://seaborn.pydata.org/faq

> Apart from the additional computational cost How much is this, in practice?

I think it's a good idea, but handling the fact that it doesn't exist on our minimal supported version of scipy or in fastcluster will be annoying.

> @mwaskom scipy/scipy#5238 is part of SciPy 1.0, which is your minimum version says setup.py. Yep those have been bumped recently so we would be good to go on that...

This is a weird one! Let's debug. First thought, this has something to do with using `FacetGrid` directly (which is discouraged). But no, we can reproduce using `relplot`: ```python g...