ipympl: Saving png figures matplotlib 3.3.0 results in "AttributeError: delete"

Describe the issue

When trying to save png figures with ipympl, interactive widgets, and the latest matplotlib (3.3.0), the saving fails with the following error message:

import matplotlib
matplotlib.use('module://ipympl.backend_nbagg')

from matplotlib import pyplot as plt
fig = plt.figure()
fig.savefig('test.png')

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-3-c4b3ac84131c> in <module>
      6 from matplotlib import pyplot as plt
      7 fig = plt.figure()
----> 8 fig.savefig('test.png')

/usr/local/lib/python3.8/dist-packages/matplotlib/figure.py in savefig(self, fname, transparent, **kwargs)
   2309                 patch.set_edgecolor('none')
   2310 
-> 2311         self.canvas.print_figure(fname, **kwargs)
   2312 
   2313         if transparent:

/usr/local/lib/python3.8/dist-packages/matplotlib/backend_bases.py in print_figure(self, filename, dpi, facecolor, edgecolor, orientation, format, bbox_inches, pad_inches, bbox_extra_artists, backend, **kwargs)
   2223                 self.figure.set_edgecolor(origedgecolor)
   2224                 self.figure.set_canvas(self)
-> 2225             return result
   2226 
   2227     @classmethod

/usr/lib/python3.8/contextlib.py in __exit__(self, type, value, traceback)
    118         if type is None:
    119             try:
--> 120                 next(self.gen)
    121             except StopIteration:
    122                 return False

/usr/local/lib/python3.8/dist-packages/matplotlib/cbook/__init__.py in _setattr_cm(obj, **kwargs)
   2064         for attr, orig in origs.items():
   2065             if orig is sentinel:
-> 2066                 delattr(obj, attr)
   2067             else:
   2068                 setattr(obj, attr, orig)

AttributeError: __delete__

Saving figures as pdf or pngfigures with an older Matplotlib version (3.2.0) works well.

Versions

Matplotlib version: 3.3.0

python -c "import sys; print('\n',sys.version); import ipympl; print('ipympl version:', ipympl.__version__)" && jupyter --version && jupyter nbextension list && jupyter labextension list

–>

3.8.5 (default, Jul 20 2020, 18:32:44)
[GCC 9.3.0]
ipympl version: 0.5.7
jupyter core     : 4.6.3
jupyter-notebook : 6.0.3
qtconsole        : 4.7.4
ipython          : 7.16.1
ipykernel        : 5.3.4
jupyter client   : 6.1.6
jupyter lab      : 2.2.2
nbconvert        : 5.6.1
ipywidgets       : 7.5.1
nbformat         : 5.0.7
traitlets        : 4.3.3
Known nbextensions:
  config dir: /usr/etc/jupyter/nbconfig
    notebook section
      jupyter-js-widgets/extension disabled
  config dir: /usr/local/etc/jupyter/nbconfig
    notebook section
      bqplot/extension  enabled
      - Validating: OK
      itkwidgets/extension  enabled
      - Validating: OK
      jupyter-datawidgets/extension  enabled
      - Validating: OK
      jupyter-leaflet/extension  enabled
      - Validating: OK
      jupyter-matplotlib/extension  enabled
      - Validating: OK
      jupyter-js-widgets/extension  enabled
      - Validating: OK
  config dir: /etc/jupyter/nbconfig
    notebook section
      jupyter-js-widgets/extension  enabled
      - Validating: OK
JupyterLab v2.2.2
Known labextensions:
   app dir: /usr/local/share/jupyter/lab
        @jupyter-widgets/jupyterlab-manager v2.0.0  enabled  OK
        @jupyterlab/hub-extension v2.0.2  enabled  OK
        jupyter-matplotlib v0.7.3  enabled  OK
        jupyterlab-datawidgets v6.3.0  enabled  OK

About this issue

Original URL
State: closed
Created 4 years ago
Reactions: 3
Comments: 21 (9 by maintainers)

Most upvoted comments

For me it was fixed with the new release 0.5.8 after the fix from @ianhi . But I have to admit I moved on to the next matplotlib release already. Are you using ipympl 0.5.8?

Edit: I just saw on the readme page that ipympl 0.5.8 requires matplotlib 3.3.1. Couldn’t you move to this one? Or is 3.3.0 necessary?

scimax on Oct 20, 2020

and if this fixes the problem, then I would say this is the optimal solution as it does not involve core Matplotlib picking up any more complexity and reduces the exposed public surface on the ipympl side.

tacaswell on Sep 8, 2020

@tacaswell sorry I keep diverting away from the final solution of implementing better handling in matplotlib, unfortunately I don’t think I can help there. But! I do have a workaround that can be implemented in ipympl that seems to give all the correct behaviors and I don’t think will messing with the context manager:

It turns out that _is_saving only shows up in one place in the codebase: https://github.com/matplotlib/ipympl/search?q=_is_saving&unscoped_q=_is_saving Most importantly, it is not used at all on the javascript side so I don’t think it actually needs to be defined using traitlets. So if you delete the line defining it using traitlets then everything works as expected because it inherits _is_saving from its super class:

diff --git a/ipympl/backend_nbagg.py b/ipympl/backend_nbagg.py
index de37bab..8b88609 100644
--- a/ipympl/backend_nbagg.py
+++ b/ipympl/backend_nbagg.py
@@ -199,7 +199,6 @@ class Canvas(DOMWidget, FigureCanvasWebAggCore):
     _current_image_mode = Unicode()
     _dpi_ratio = Float(1.0)
     _is_idle_drawing = Bool()
-    _is_saving = Bool()
     _button = Any()
     _key = Any()
     _lastx = Any()

Which allows

%matplotlib widget
import matplotlib
from matplotlib import pyplot as plt
with plt.ioff():
    fig = plt.figure()
plt.plot([0,1],[0,1])
fig.savefig('test.png')
display(fig)
print(f'Matplotlib version: {matplotlib.__version__}')
print(fig.canvas._is_saving) # no attribute error!!

to give the expected output:

I know this isn’t the optimal solution but how would you feel about using to fix the immediate problem?

ianhi on Sep 7, 2020

this is also a problem if you display(fig).

I think is an important bug. That error is a bit opaque to me though. @tacaswell @martinRenou @thomasaarholt do any of you have thoughts on where to start here?

ianhi on Aug 13, 2020

ipympl: Saving png figures matplotlib 3.3.0 results in "AttributeError: __delete__"

Describe the issue

Versions

About this issue

Most upvoted comments

ipympl: Saving png figures matplotlib 3.3.0 results in "AttributeError: delete"