Type-check on axes-instances created with subplots() fails

[Khris777]

Bug report

Bug summary

Creating an axes object using the subplots()-method creates an object of type matplotlib.axes._subplots.AxesSubplot, but the object can't be tested on this type, while an axes-object of type matplotlib.axes._axes.Axes can be tested successfully.

Code for reproduction

import matplotlib.pyplot as mp
import matplotlib.pyplot as mpp
fig,axes = mpp.subplots()
print(isinstance(axes,mp.axes._subplots.AxesSubplot))

Out: AttributeError: module 'matplotlib.axes._subplots' has no attribute 'AxesSubplot'

Compared to:

fig = mpp.Figure()
axes = fig.add_axes([0,0,1,1])
print(isinstance(axes,mp.axes._axes.Axes))

Out: True

Matplotlib version
Windows 10
Matplotlib 2.0.2
Python 3.6.1
Anaconda Distribution

[tacaswell]

Why do you want to do this? It is better to do duck typing.

The reason it is failing is the Axes classes returned by subplots are dynamically (ish) created and are not in the matplotlib.axes._subplots namespace (which is why you are getting an AttributeError).

If you must do type checking, I would use maxes.Axes or maxes.SubplotBase (check type(ax).mro() to see what classes are in the inheritance tree).