PR Summary

Adresses #11386, #5456. See also #11417 that properly fixes the logit issues...

Code like below caused anything that required the tightbbox of the axes to fail forlogit axes:

This PR does not fix logit axes, but instead ignores non-finite values of the text positions when calculating the bbox, since those are assumed to be invalid. I suppose someone could also figure out how to fix logit, but I'm not too interested in digging into it.

import matplotlib.pyplot as plt
import numpy as np
fig, ax = plt.subplots()

np.random.seed(1)
y = np.random.normal(loc=0.5, scale=0.4, size=1000)
y = y[(y > 0) & (y < 1)]
y.sort()
x = np.arange(len(y))
ax.plot(x, y)
ax.set_yscale('logit')
print(ax.get_tightbbox(fig.canvas.get_renderer()))

Before

/Users/jklymak/matplotlib/lib/matplotlib/transforms.py:402: RuntimeWarning: invalid value encountered in double_scalars
  return points[1, 0] - points[0, 0]
Bbox(x0=-inf, y0=nan, x1=1155.5, y1=nan)

After

Bbox(x0=-21.6925048828125, y0=58.15555555555555, x1=1155.5, y1=848.3)

PR Checklist

• Has Pytest style unit tests
• Code is PEP 8 compliant
• New features are documented, with examples if plot related
• Documentation is sphinx and numpydoc compliant
• Added an entry to doc/users/next_whats_new/ if major new feature (follow instructions in README.rst there)
• Documented in doc/api/api_changes.rst if API changed in a backward-incompatible way