multiple Matplotlib libraries

[ernestmartinez]

While trying basic plots in jupyter I get
AttributeError: module 'matplotlib.cbook' has no attribute 'popall

google search suggested multiple installed matplotlib libraries to be culprit.

jovyan@jupyter-admin:~$ conda list matplotlib

packages in environment at /opt/conda:

matplotlib 2.1.2
matplotlib 2.0.2 py36_2 conda-forge

jovyan@jupyter-admin:~$ pip list |grep matplotlib
matplotlib (2.1.2)

Trying to uninstall conda libs caused:

jovyan@jupyter-admin:~$ conda uninstall matplotlib
Fetching package metadata
Solving package specifications: .

The following packages will be REMOVED:

matplotlib: 2.0.2-py36_2 conda-forge

Proceed ([y]/n)? y

ERROR conda.core.link:_execute_actions(339): An error occurred while uninstalling package 'conda-forge::matplotlib-2.0.2-py36_2'.
PermissionError(13, 'Permission denied')
Attempting to roll back.

PermissionError(13, 'Permission denied')
Current conda install:

           platform : linux-64
      conda version : 4.3.34
   conda is private : False
  conda-env version : 4.3.34
conda-build version : not installed
     python version : 3.6.3.final.0
   requests version : 2.18.4
   root environment : /opt/conda  (writable)
default environment : /opt/conda
   envs directories : /opt/conda/envs
                      /home/jovyan/.conda/envs
      package cache : /opt/conda/pkgs
                      /home/jovyan/.conda/pkgs
       channel URLs : https://conda.anaconda.org/conda-forge/linux-64
                      https://conda.anaconda.org/conda-forge/noarch
                      https://repo.continuum.io/pkgs/main/linux-64
                      https://repo.continuum.io/pkgs/main/noarch
                      https://repo.continuum.io/pkgs/free/linux-64
                      https://repo.continuum.io/pkgs/free/noarch
                      https://repo.continuum.io/pkgs/r/linux-64
                      https://repo.continuum.io/pkgs/r/noarch
                      https://repo.continuum.io/pkgs/pro/linux-64
                      https://repo.continuum.io/pkgs/pro/noarch
        config file : /opt/conda/.condarc
         netrc file : None
       offline mode : False
         user-agent : conda/4.3.34 requests/2.18.4 CPython/3.6.3 Linux/4.4.0-116-generic debian/stretch/sid glibc/2.23
            UID:GID : 1000:100

[jlewi]

/cc @ojarjur

[willingc]

Looks like you have a Python version mismatch with conda package. Delete the entire conda/Anaconda directory and reinstall.

[ernestmartinez]

The issue is having conda and pip installed libs for matplotlib, the fix being to pip uninstall matplotlib. Unfortunately I cannot sudo nor become root in order to delete anything as a Jupyterhub user.

[willingc]

@ernestmartinez Are you able to do the following within a notebook cell?

%%bash
python3 -m pip uninstall matplotlib

Or at the command line:

python3 -m pip uninstall matplotlib

[jlewi]

@willingc I believe we are using a combination of conda and non-conda packages; I'm guessing because not everything we want is in conda (e.g. TFX and beam). Is this the right thing to do? Is there a better way to handle this.

[jlewi]

If I'm understanding comment 1 it looks like conda is installing matplotlib 2.0.2 but there is a newer version 2.1.2 installed as well.

It looks like matplotlib 2.2 is the latest release.

Can we fix this just by uninstalling the conda installed version in our docker images using the command @willingc provided above.

[willingc]

The best way would be to create an environment.yml file to use with conda where you can conda install and pip install from one file. Example from repo2docker Example from JupyterHub docs - this is the more straightforward example

In general, I would use conda-forge for the channel.

The yaml file will list conda packages and then pip packages. Here's an example from the conda docs: https://conda.io/docs/user-guide/tasks/manage-environments.html#creating-an-environment-file-manually

[willingc]

@jlewi The conda cheatsheet is helpful too. https://conda.io/docs/user-guide/cheatsheet.html

[pdmack]

Is this still an issue? All I see in the current 1.8 notebook is:

matplotlib                2.2.2                     <pip>

[jlewi]

Marking this fixed based on @pdmack's comment.