osx backend not usable with ipython/jupyter from conda?

[tiagopereira]

Bug report

Bug summary

I am unable to use the osx backend from matplotlib with ipython or jupyter from conda (miniconda or anaconda, makes no difference). Since the alternative qt backend is suffering from the critical #13066 issue, this is quite important.

Code for reproduction

Python from conda does not appear to be installed as a framework, and this gives an error when trying to use the osx backend from ipython or jupyter. I followed the instructions on the error message (reinstall python.app), but I still get the same problem (see below). If I use pythonw and run things from there, the backend works fine. But there is no ipythonw to mimic the same behaviour. How are we supposed to use the osx backend from ipython? Or any modern backend? The Qt backend suffers from from the critical #13066 issue, which leads to hanging ipython sessions after a few minutes of use for me.

%matplotlib osx

Actual outcome

~/anaconda3/lib/python3.7/site-packages/matplotlib/backends/backend_macosx.py in <module>
     12
     13 import matplotlib
---> 14 from matplotlib.backends import _macosx
     15
     16 from .backend_agg import FigureCanvasAgg

ImportError: Python is not installed as a framework. The Mac OS X backend will not be able to function correctly if Python is not installed as a framework. See the Python documentation for more information on installing Python as a framework on Mac OS X. Please either reinstall Python as a framework, or try one of the other backends. If you are using (Ana)Conda please install python.app and replace the use of 'python' with 'pythonw'. See 'Working with Matplotlib on OSX' in the Matplotlib FAQ for more information.

Expected outcome

No error.

Matplotlib version

• Operating system: macOS 10.13.6
• Matplotlib version: 3.0.3
• Matplotlib backend (print(matplotlib.get_backend())): osx
• Python version: 3.7.3
• Jupyter version (if applicable): 1.0.0
• Other libraries: ipython 7.4.0

Matplotlib and python from anaconda or miniconda (defaults or conda-forge, makes no difference).

[dstansby]

I think this might have improved in 3.1, and the osx backend no longer requires a python framework build. Could you try the release candidate of Matplotlib 3.1 (pip install --pre matplotlib), and see if it fixes the problem?

[tiagopereira]

@dstansby I just tried with matplotlib 3.1.0rc1 and indeed the problem goes away!

[dstansby]

Great! There should be a full release of 3.1 in the near future, so I'm going to mark this as fixed, thanks for reporting.

[tacaswell]

3.1 went out today, its on pypi and conda-forge, may take a bit longer for defaults to pick it up.