matplotlib.ticker.FormatStrFormatter clashes with ax2.set_yticklabels when dual y-axis is used #7161
Description
- Matplotlib version 1.5.1-1ubuntu1, Python version2.7.12, Platform Ubuntu 16.04 x64
- I installed Matplotlib and Python from apt repository
- Example:
#!/usr/bin/python3
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
fig = matplotlib.pyplot.figure(figsize=(7,4), dpi=96, facecolor='#eeeeee', tight_layout=1)
ax = fig.add_subplot(111)
x = np.arange(0., 10., 0.1)
ax.plot(x, 12+x**1.5+0.1*x**2.5)
## Simple axes
ax.set_xlabel('expected maximum load (kg)')
ax.set_ylabel('needed bar diameter (cm)')
ax2 = ax.twinx()
ax2.set_ylabel('bar cross-section (cm$^{2}$)')
## left and right y-axes apply to the same data, but the right one has a given functional dependency on the left one
## (in this example, right ticks share the position with the left ticks)
ylim2 = ax.get_ylim()
print(ax.get_ylim())
ax2.set_ylim(ax.get_ylim())
def tick_function(r): return r**2 + 0.123 # 10.46*(p**1.68)
ax2.set_yticks(ax.get_yticks()[:-1])
ax2.set_yticklabels(tick_function(ax.get_yticks()[:-1]))
## ERROR? When uncommented, the values on the secondary y-axis are not processed by tick_function!
ax2.yaxis.set_major_formatter(matplotlib.ticker.FormatStrFormatter('%.2g$\\pm \\beta$'))
## Finish the plot + save
ax.set_title("with custom matplotlib.ticker.FormatStrFormatter")
ax.grid()
ax.legend(prop={'size':10}, loc='upper right')
fig.savefig('with_formatter_error.png')
In short, I define dual y-axes of a plot, and need the right y-axis to show some given function of the left one. But when I try to format the right axis using the FormatStrFormatter class, it takes the values which are not processed by my function.
I consider this to be a bug, preventing me from defining the values as 100.123+β, 400.123+β, 900.123+β ... on the right axis. None of the attached example figures is therefore acceptable.
Exchanging the order of the ax2.set_yticklabels and ax2.yaxis.set_major_formatter does not help. Switching to python2 does not help either. I did not test the latest git version yet.
