Repository navigation
[Bug]: Crash in Windows 10 if polar axis lim is lower than lowest data point. #26765
Description
Bug summary
This example causes matplotlib to silently crash on Windows, but not on Mac. The example below is a minimal example. Looks like important detail is that the data extends below the radial axis minimum limit.
I have another bug related to log scale polar axes, these might be related:
Code for reproduction
import sys
import matplotlib as mpl
import matplotlib.pyplot as plt
import numpy as np
def plot():
color_step = 3
fontsize_y = 8
# Create a polar plot
fig = plt.figure(figsize=(10, 10))
ax = fig.add_subplot(111, projection="polar")
ax.set_rscale("log")
# KEY LINE!!!!!!!!!!!1
# Crashes in windows if lower limit is higher than lower datapoint.
ax.set_rlim(101, 20000)
n_angles = 3
n_freqs = 10
_angles = np.linspace(0, 180, n_angles)
Angles = _angles[:, np.newaxis] * np.ones(n_freqs)
# KEY LINE
# Lower r_axis value is 100, which is lower than low axis limit.
freqs = np.linspace(100, 1000, n_freqs)
Freqs = np.ones((n_angles, 1)) * freqs[:, np.newaxis].T
print(freqs)
Mags = np.random.rand(*Freqs.shape) * 100
colormesh = ax.pcolormesh(
# Angle / 57.29,
Angles / 57.29,
Freqs,
Mags,
shading="gouraud",
# cmap=colormap,
# vmin=color_lo,
# vmax=color_hi,
)
plt.show()
if __name__ == "__main__":
plot()Actual outcome
Plot window briefly shows, then disappears. I tried a couple version of Python and three backends. Crash occurs on plt.show or plt.savefig
Expected outcome
Plot displays. (works on Mac)
Additional information
No response
Operating system
No response
Matplotlib Version
3.7.3
Matplotlib Backend
QtAgg
Python version
3.8
Jupyter version
No response
Installation
pip
Activity
QuLogic commentedon Sep 13, 2023
This appears to be an infinite loop in Agg;
QuLogic commentedon Sep 13, 2023
It looks like some NaN are getting through to Agg, which is producing some bad behaviour:
I think in most cases, we ignore segments of paths that have a NaN, so we should be doing the same for such triangles.