How to define zorder when using 2 y-axis?

Question

I plot using two y-axis, on the left and the right of a matplotlib figure and use zorder to control the position of the plots. I need to define the zorder across axes in the same figure.

---

Problem

import numpy as np
import matplotlib.pyplot as plt
x = np.arange(-10,10,0.01)

fig, ax1 = plt.subplots( 1, 1, figsize=(9,3) )
ax1.plot( x, np.sin(x), color='red', linewidth=10, zorder=1 )
ax2 = ax1.twinx()
ax2.plot( x, x, color='blue', linewidth=10, zorder=-1)

In the previous diagram, I would expect the blue line to appear behind the red plot.

How do I control the zorder when using twin axes?

---

I am using:

python: 3.4.3 + numpy: 1.11.0 + matplotlib: 1.5.1

Answer 1

This should work

ax1.set_zorder(ax2.get_zorder()+1)
ax1.patch.set_visible(False)

Answer 2

the following codes works

import numpy as np
import matplotlib.pyplot as plt
from matplotlib import ticker as tick


x = np.arange(-10,10,0.01)

plt.figure(figsize=(10, 5))    
fig = plt.subplot(111)

"""be attention to here. it's fig.plot, not ax1.plot
if you write ax1.plot, then it does not work.
"""
fig.plot(x, x, color ='blue', linewidth =10) 

ax2 = fig.twinx()    
ax2.plot(x, np.sin(x), color='red', linewidth =10) 

"""
It looks like the two axes have separate z-stacks. 
The axes are z-ordered with the most recent axis on top
"""
fig.set_zorder(ax2.get_zorder()+1)
fig.patch.set_visible(False)
plt.show()