Can't install matplotlib in Conda env w/ Python 3.8

5
votes

I'm replicating my Python 3.7 Conda environment for Python 3.8 on an Ubuntu 18.04 system. I started with a plain jane Conda env with python=3.8 as the sole requirement. Then I started installing packages from my 3.7 environment.

Most of the process went swimmingly, but matplotlib failed. Conda search shows a 3.8 version of matplotlib, but the conda install command is unable to resolve a nonconflicting set of packages. Numpy, ipython, etc all installed fine. After much spinning of the little spinny thing it complains:

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Package python conflicts for:
python=3.8
Package pip conflicts for:
python=3.8 -> pip
matplotlib -> python[version='>=3.7,<3.8.0a0'] -> pip
Package certifi conflicts for:
python=3.8 -> pip -> setuptools -> certifi[version='>=2016.09|>=2016.9.26']
Package wheel conflicts for:
python=3.8 -> pip -> wheel
matplotlib -> python[version='>=3.7,<3.8.0a0'] -> pip -> wheel
Package python-dateutil conflicts for:
matplotlib -> python-dateutil
Package ca-certificates conflicts for:
matplotlib -> setuptools -> ca-certificates
python=3.8 -> openssl[version='>=1.1.1d,<1.1.2a'] -> ca-certificates
Package setuptools conflicts for:
python=3.8 -> pip -> setuptools
matplotlib -> setuptools

I can't quite tell what it means. Maybe the version dependency for matplotlib itself is incorrect, or there's no official matplotlib Conda package for Python >= 3.8.0a0? Conda search suggests there should be something:

% conda search matplotlib
...
matplotlib                     3.1.1  py38h5429711_0  pkgs/main

Here's the output of conda info:

     active environment : python38
    active env location : /home/skip/miniconda3/envs/python38
            shell level : 2
       user config file : /home/skip/.condarc
 populated config files : /home/skip/.condarc
          conda version : 4.8.1
    conda-build version : not installed
         python version : 3.7.3.final.0
       virtual packages : __glibc=2.27
       base environment : /home/skip/miniconda3  (writable)
           channel URLs : https://repo.anaconda.com/pkgs/main/linux-64
                          https://repo.anaconda.com/pkgs/main/noarch
                          https://repo.anaconda.com/pkgs/r/linux-64
                          https://repo.anaconda.com/pkgs/r/noarch
          package cache : /home/skip/miniconda3/pkgs
                          /home/skip/.conda/pkgs
       envs directories : /home/skip/miniconda3/envs
                          /home/skip/.conda/envs
               platform : linux-64
             user-agent : conda/4.8.1 requests/2.22.0 CPython/3.7.3 Linux/4.15.0-74-generic ubuntu/18.04.3 glibc/2.27
                UID:GID : 1000:1000
             netrc file : /home/skip/.netrc
           offline mode : False

Not sure why it reports 3.7.3 as the Python version. Perhaps that's in the root environment? I clearly have 3.8.1 installed:

(python38) polly% python
Python 3.8.1 (default, Jan  8 2020, 22:29:32) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>>
2
pythonmatplotlibanaconda
Possibly related: stackoverflow.com/questions/58557593/… – Gabriel
Does this answer your question? Current Matplotlib version compatible with Python 3.8? – Gabriel
Thanks for the pointers. I have freetype headers installed, so that's unlikely to be an issue. My install attempts never get to the point of compiling anything (besides, aren't Conda packages architecture-dependent binaries?) Also, note that my environment with Python 3.7.6 installed does have the latest Conda package version of matplotlib installed (3.1.1). It sure seems like a Conda dependency nit. I will keep digging. – smontanaro
Simplest demonstration of the problem... This succeeds:<br/> conda create --name test python=3.7 matplotlib while this fails:<br/> conda create --name test python=3.8 matplotlib – smontanaro

2 Answers

7
votes
conda install -c conda-forge matplotlib

It worked for me. My environment is python 3.8 and Ubuntu 18.04.

5
votes

As of now, March 2020, you must unfortunately downgrade your conda executable (in your base environment) to install matplotlib. Here's the github discussion.

Try this:

conda activate
conda config --set allow_conda_downgrades true
conda install conda==4.6.14

conda create --name test_env
conda activate test_env
conda install matplotlib

This should work.