raw.plot(block=True) freezes the jupyter kernel forever

Describe the bug

When plotting continuous “raw” data and specify block=True, everything works well until you try to close the figure. Control is never handed back to the command line. This does not occur when using command line ipython, but does occur when using Spyder or a Jupyter notebook.

Steps and/or code to reproduce

In Spyder or a notebook:

import mne
data_path = mne.datasets.sample.data_path()
raw = mne.io.read_raw_fif(data_path +'/MEG/sample/sample_audvis_raw.fif')
raw.plot(block=True)

Close the plotting window. The Jupyer kernel now hangs.

Expected results

After closing the window, the Jupyter kernel should become responsive again.

Actual results

Jupyter kernel never becomes responsive.

Additional information

Platform:      Linux-4.4.0-148-generic-x86_64-with-debian-stretch-sid
Python:        3.6.8 |Anaconda, Inc.| (default, Dec 30 2018, 01:22:34)  [GCC 7.3.0]
Executable:    /work/modules/Ubuntu/14.04/amd64/common/anaconda3/latest/envs/neuroimaging/bin/python
CPU:           x86_64: 8 cores
Memory:        15.6 GB

mne:           0.18.1 (also tested this on current master)
numpy:         1.16.4 {blas=mkl_rt, lapack=mkl_rt}
scipy:         1.2.1
matplotlib:    3.1.0 {backend=Qt5Agg}

sklearn:       0.21.2
nibabel:       2.4.1
mayavi:        4.6.2 {qt_api=pyqt5, PyQt5=5.9.2}
cupy:          Not found
pandas:        0.24.2
dipy:          Not found

[wmvanvliet]

This still breaks for me in Spyder, but not in plain ipython or in a notebook:

from matplotlib import pyplot as plt
plt.plot([1, 2, 3, 4])
plt.show(block=True)

However, @mvonflotow’s suggestion of disabling Spyder’s active support did the trick. I’m guessing there’s some weird interaction between the Qt thread spawned by the QtConsole component that Spyder uses and the block=True behavior. I’ll file the issue there.

[cbrnr]

I cannot reproduce the issue, the minimal example does not block execution in both a regular Jupyter notebook and inside VS Code. Maybe you do not have the latest versions of the involved packages (i.e. Matplotlib, Jupyter)?