[Bug]: mpl.rc_context within jupyterlab prevents plot to show up

[fmaussion]

Bug summary

This is arguably a very weird bug, but it is easy to reproduce. I don't know if the problem is in mpl or jupyterlab.

When using the bit of code below in a notebook, the plot shows once (and only once). Calling the function again prevents the plot to show again (see video).

I am at loss as to why this would happen 😅 . This didn't happen before, but I have no clue as to whether an update to mpl or jupyterlab created the issue.

Code for reproduction

import matplotlib as mpl
import matplotlib.pyplot as plt

figsize=(3, 3)

def plotter():
    with mpl.rc_context({"figure.figsize": figsize}):
        plt.plot([1, 2, 3])

And then:

plotter()

Actual outcome

mpl-2023-01-20_17.34.46.mp4

Expected outcome

The plot should show up on the second call.

Additional information

No response

Operating system

Linux

Matplotlib Version

3.6.3

Matplotlib Backend

module://matplotlib_inline.backend_inline

Python version

3.10

Jupyter version

3.5.2

Installation

None

[fmaussion]

Just to clarify: without the context manager the code works just fine:

mpl-2023-01-20_17.35.33.mp4

[fmaussion]

OK, I tried with a matplotlib version where I know the code definitely worked, and it didn't.

I am pretty confident the problem arose after a jupyterlab update. Should I report there?

[fmaussion]

Sorry for the noise, but I realize I don't even need the wrapper to create the bug - updated all the posts above.

[QuLogic]

It looks like a static plot, so possibly a change in https://github.com/ipython/matplotlib-inline

[fmaussion]

It's a pretty niche bug but if anyone encounters this, the straightforward workaround is to use a self-made context manager instead of mpl.rc_context:

class mpl_figsize:
    """Homegrown context manager to avoid a bug in mpl/jupyter:
    https://github.com/matplotlib/matplotlib/issues/25041
    """
    def __init__(self, figsize):
        self.figsize = plt.rcParams['figure.figsize']
        plt.rcParams['figure.figsize'] = figsize
    def __enter__(self):
        return None
    def __exit__(self, type, value, traceback):
        plt.rcParams['figure.figsize'] = self.figsize

[mwaskom]

Maybe dupe of #23298?

[fmaussion]

Maybe dupe of #23298?

Looks very similar indeed, but still happens with mpl 3.6.3

[tacaswell]

If you add %matplotlib widget does it work?

[fmaussion]

If you add %matplotlib widget does it work?

Mmh I cant get ipyml to work (installed via conda-forge), but the behavior of the disappearing error message seems to indicate that something very similar is at play:

mpl-2023-01-23_18.25.12.mp4

[github-actions]

This issue has been marked "inactive" because it has been 365 days since the last comment. If this issue is still present in recent Matplotlib releases, or the feature request is still wanted, please leave a comment and this label will be removed. If there are no updates in another 30 days, this issue will be automatically closed, but you are free to re-open or create a new issue if needed. We value issue reports, and this procedure is meant to help us resurface and prioritize issues that have not been addressed yet, not make them disappear. Thanks for your help!

[timhoffm]

Confirmed this bug still exists.

This is very weird. It has something to do with the figure being created inside the context manager, but the value of the context does not matter.

This slight modification (explicit plt.figure(), empty context) still exhibits the bug

def plotter():
    with mpl.rc_context({}):
        plt.figure()
        plt.plot([1, 2, 3])

but moving the figure creation out of the context manager does correctly show the figure

def plotter():
    plt.figure()
    with mpl.rc_context({}):
        plt.plot([1, 2, 3])

I've confirmed in both cases that no current figure exists and a new one is created - i.e. the inline backend correctly closes the figure after it has been shown.