pyenv virtualenv not work for matplotlib

[shixudongleo]

I installed a framework enable python 2.7.8. And matlablib is working fine under this python 2.7.8.
After that, I created a virtualenv, matplotlib is not working inside the virtualenv.

A simple test:

$ env PYTHON_CONFIGURE_OPTS="--enable-framework --enable-ipv6 --enable-unicode --with-threads" pyenv install 2.7.8
$ pyenv shell 2.7.8
$ pip install numpy scipy matplotlib
$ python -c "import matplotlib.pyplot as plt" #this works 

$ pyenv virtualenv 2.7.8 nipype_env
$ pyenv activate nipype_env
$ pip install numpy scipy matplotlib
$ python -c "import matplotlib.pyplot as plt" #this doesnot work
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "/Users/shixudongleo/.pyenv/versions/nipype_env/lib/python2.7/site-packages/matplotlib/pyplot.py", line 114, in <module>
    _backend_mod, new_figure_manager, draw_if_interactive, _show = pylab_setup()
  File "/Users/shixudongleo/.pyenv/versions/nipype_env/lib/python2.7/site-packages/matplotlib/backends/__init__.py", line 32, in pylab_setup
    globals(),locals(),[backend_name],0)
  File "/Users/shixudongleo/.pyenv/versions/nipype_env/lib/python2.7/site-packages/matplotlib/backends/backend_macosx.py", line 24, in <module>
    from matplotlib.backends import _macosx
RuntimeError: 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 Working with Matplotlib in a virtual enviroment see 'Working with Matplotlib in Virtual environments' in the Matplotlib FAQ

I am new in using pyenv and pyenv virtualenv. In previous experience with virtualenv, I have not encounter such a problem.

I have found differences in directory structure between 2.7.8 and nipype_env.

$ ls ~/.pyenv/versions/2.7.8
Python.framework envs             lib
bin              include          share
$ ls ~/.pyenv/versions/nipype_env
bin                lib                share
include            pip-selfcheck.json

I have found a fix in this link. But I am still wondering whether there is a more elegant way support by pyenv virtualenv to fix my problem.

[twoframesperminute]

Matplotlib expects a Framework install on OS X in order to manipulate GUI elements. You can install Framework versions with a command like this:

env PYTHON_CONFIGURE_OPTS="--enable-framework CC=clang" pyenv virtualenv <python_version> <env_name>

The fact that it worked in your first example is because you used --enable-framework.

[yyuu]

RuntimeError: 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 Working with Matplotlib in a virtual enviroment see 'Working with Matplotlib in Virtual environments' in the Matplotlib FAQ

Read error messages.

[brifordwylie]

I had to do something a bit different in order to get my pyenv virtualenv to work.

1. Run @psomhorst suggested config opt on the ^base^ python (in my case 3.6.6)

   $ env PYTHON_CONFIGURE_OPTS="--enable-framework CC=clang" pyenv install 3.6.6
   

2. Uninstall my virtualenv (if you already have one that you want to reuse)

   $ pyenv uninstall py3
   

3. Reinstall my virtualenv

   $ pyenv virtualenv 3.6.6 py3
   

And then of course activate my virtualenv like normal.. and then everything worked great.

[keisuke-nakata]

a simple solution (and seem the most elegant for me) is putting matplotlibrc at your config dirctory (config dir depends on how matplotlib prioritizes its rcfile. see https://matplotlib.org/users/customizing.html) and set backend: Agg.

[jwuttke]

For reference: After installing Tk to the computer it is necessary to rerun pyenv install <version>. It is not necessary to uninstall and reinstall pyenv.