Bug summary

I’m trying to use a constrained layout in a visualization that contains an artist that is an instance of AnnotationBbox, and matplotlib raises a warning saying constrained layout is not applied. The visual effect is not evident in this simple example, but it becomes very clear once we have multiple panels.

Code for reproduction

import matplotlib
import matplotlib.pyplot as plt
from matplotlib.offsetbox import AnnotationBbox, TextArea

fig, ax = plt.subplots(layout="constrained")

ab = AnnotationBbox(
    TextArea("Some text", textprops={"size": 42}),
    (0.5, 0.5),
    xycoords="axes fraction",
    box_alignment=(0.5, 0.5),
    pad=0
)

ax.add_artist(ab)
fig.set_facecolor("w")
fig.savefig("annotation_box.png", dpi=300)

Actual outcome

UserWarning: constrained_layout not applied because axes sizes collapsed to zero. Try making figure larger or axes decorations smaller.

Expected outcome

No warning should appear

Additional information

The following works without any warning

fig, ax = plt.subplots(layout="constrained")
ax.text(0.5, 0.5, "Some text", size=42, ha="center")
fig.set_facecolor("w")
fig.savefig("ax_text.png", dpi=300)

The problem with the constrained layout is more evident if I have two or more panels.
One way of fixing it (i.e. getting rid of the warning and bad functionality) is to do ab.set_in_layout(False) before doing ax.add_artist(ab).

This was first posted on Discourse https://discourse.matplotlib.org/t/constrained-layout-does-not-work-well-with-annotationbbox/23301

Operating system

Ubuntu 22

Matplotlib Version

3.6.2

Matplotlib Backend

No response

Python version

No response

Jupyter version

No response

Installation

pip