Hi all!

I'm one of the maintainers of conda and my colleagues in the packaging team at Anaconda have investigated this bug and relased fixed version of the Python package to solve this once and for all.

Note: This only applies if you have Python 3.8 or 3.9 installed. It also seems specific to Windows. Please let us know if you've also seen this on other platforms.

Problem

It turns out it isn't an issue in conda itself (even though it was triggered by a change in conda) and needs to be fixed by updating the build of Python that comes with Anaconda Distribution or miniconda installations.

The necessary patches to the Python package have been applied to 3.8 and 3.9.

Mitigation

To mitigate the problem for your installation, you'll need to update Python in your conda base environment (make sure to activate first!).

On Windows

  1. Open a new "Anaconda Prompt" from the start menu
  2. If enabled, deactivate the base environment with: conda deactivate
  3. Reactivate the conda base environment again with either of the following steps:
    • Run conda activate base or
    • Call the file C:\path\to\your\miniconda3\Scripts\activate.bat
      (please update your path accordingly, e.g. mine was C:\Users\jezdez\miniconda3\Scripts\activate.bat)
  4. Check which version of Python you have installed in your conda base environment:
    • Run conda list -n base python
    • Output should be something like, remember the Python version listed:
    • ​​​​​# packages in environment at C:\Users\jezdez\miniconda3:
      ​​​​​#
      ​​​​​# Name                    Version                   Build  Channel
      ​​​​​python                    3.9.15               h6244533_2  defaults
      
  5. Update the Python package depending of whether you have Python 3.8 or 3.9 installed (note the ):
    • For Python 3.8 run: conda install -n base python=3.8
    • For Python 3.9 run: conda install -n base python=3.9

Confirmation

Once updated, confirm that the update has fixed your environment:

  • Open a new "Anaconda Prompt"
  • Run conda create -n scipy-env scipy to create a new environment and install scipy into it
  • If all goes well, you can remove the test environment again with conda env remove -n scipy-env afterwards

Or try path/to/conda's/python -c "import ssl" in a deactivated enviroment.

Other environments

In case you've installed Python in any of your other conda environments, please follow the conda install python=3.8 or conda install python=3.9 step above while the environments are activated.

Select a repo