Bug summary

There is a strange change on the behavior of the line style specifiers of the hist method in matplotlib when moving from version 3.9.0 to 3.10.1. It seems like the linestyle and ls arguments are behaving differently. In addition, it is no longer possible to specify a customized line style for histograms via a tuple ls=(0, (1, 10)). This is not the case for the plot method.

In matplotlib 3.9.0:

In matplotlib 3.10.1:

Using the linestyle argument with (0, (1, 10)) as a value leads to the following error

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2,) + inhomogeneous part.

Note that when using ls there is no error, but the result is not what we would expect.

Code for reproduction

import matplotlib.pyplot as plt
import numpy as np

x = np.random.normal(0, 1, 10_000)

fig, ((ax0, ax1), (ax2, ax3)) = plt.subplots(2, 2, figsize=(8, 8))
ax0.plot(x, x, ls=(0, (1, 10)), lw=2)
ax0.set_title("scatter plot (ls)")
ax1.plot(x, x, linestyle=(0, (1, 10)), lw=2)
ax1.set_title("scatter plot (linestyle)")
# this does not display any error, but the histogram is incorrect
ax2.hist(x, bins=10, range=(-5, +5), ls=(0, (1, 10)), histtype="step", lw=2)
ax2.set_title("hist (ls)")
# this gives an error in 3.10.1
ax3.hist(x, bins=10, range=(-5, +5), linestyle=(0, (1, 10)), histtype="step", lw=2)
ax3.set_title("hist (linestyle)")
plt.show()

Actual outcome

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2,) + inhomogeneous part.

Expected outcome

Additional information

No response

Operating system

No response

Matplotlib Version

3.10.1

Matplotlib Backend

No response

Python version

No response

Jupyter version

No response

Installation

None