aspect argument to FacetGrid uses the opposite convention from matplotlib

[shoyer]

This is somewhat unfortunate. I'm not entirely sure what we should do about it, though.

matplotlib

plt.imshow([[0]], aspect=2)

or

plt.plot()
plt.gca().set_aspect(2)

seaborn

sns.FacetGrid({}, aspect=2)

[mwaskom]

Eek. Is the matplotlib aspect in figure or data coordinates, by the way?

[shoyer]

It's an axis level setting so I think it has to be data coordinates. (Which I think is the intent with FacetGrid, but not exactly the case due to matplotlib's messiness.)

On Fri, Oct 30, 2015 at 5:19 PM, Michael Waskom notifications@github.com
wrote:

Eek. Is the matplotlib aspect in figure or data coordinates, by the way?

Reply to this email directly or view it on GitHub:
#746 (comment)

[mwaskom]

No, the FacetGrid aspect is just size of the thing in inches, it's invariant to the scale of the data being plotted. So at least they are doing different things?

[mwaskom]

FWIW I think the implementation in seaborn is "correct", assuming people correctly interpret aspect=a as giving an aspect ratio of a:1.