Fix error that occurs when minorticks are on multi-Axes Figure with more than one boxplot #27647
Conversation
There was a problem hiding this comment.
The test case could be further simplified
- remove random (the values from the boxplot don't matter, may as well just do
np.arange(10)or something - reduce number of subplots (2 will do, there is likely a way of getting it with just one plot, but may be slightly contrived, as the
sharexis part of the cause for the specific presentation)- plotting 2 boxplots at the same
positionsshould probably do it)
- plotting 2 boxplots at the same
import numpy as np
import matplotlib.pyplot as plt
fig, ax = plt.subplots()
ax.boxplot(np.arange(10), positions=[0])
ax.boxplot(np.arange(10), positions=[0])
ax.boxplot(np.arange(10), positions=[1])
ax.minorticks_on()
fig.draw_without_rendering()has the same behavior (I included one at another position such that np.unique has at least 2 values (and thus doesn't go into the return [] case just under the changed line of code)
Assertions then don't need the all, as there is only one Axes.
|
Owee, I'm MrMeeseeks, Look at me. There seem to be a conflict, please backport manually. Here are approximate instructions:
And apply the correct labels and milestones. Congratulations — you did some good work! Hopefully your backport PR will be tested by the continuous integration and merged soon! Remember to remove the If these instructions are inaccurate, feel free to suggest an improvement. |
Fix error that occurs when minorticks are on multi-Axes Figure with more than one boxplot (cherry picked from commit b03407b)
PR summary
Closes #26484. Having
minorticks_on()in a multi-Axes Figure that contains more than one boxplot and shares the x-axis was raising an OverflowError.As @ksunden pointed out in #26484 (comment), the boxplots set the Locator and Formatter of one axis to be FixedLocator/FixedFormatter, without handling repeated values. This leads to multiple repeated values for the Major ticks, which then the
AutoMinorLocatordoes major[1] - major[0] to get the spacing of the major, which returns zero and then divides by 0, leading to the error.I've implemented the option on the first bullet point of the above comment to call
np.uniqueon themajorticklocsinAutoMinorLocator. This fix remove repeated values frommajorticklocsto avoid the zero division that was causing the error.PR checklist