VSCode/Language/Python/How To/Fix Stuck at Discovering Python Interpreters

• log 2024/Trimester/t3/w17/d3 - Wed, 2024/12/25 - debugging VSCode/Language/Python and VSCode/Extension/Python - VSCode/Proj/24t3/LearnVSCode/03 Basic Python
  • 13:12 - I debugged this for over two hours today while working through 11:02 - ((2ca9c3c8-0810-4adc-ab6e-2a45fba87e46))
  • I got stuck on configuring the interpreter after installing VSCode/Extension/Python. I tried everything. Here's what worked.

  • Problem Summary

    • When I upgraded from Macbook/My to Macbook/My2, going from Chips/amd/intel64 to Chips/arm/M2, I had a bunch of old virtualenvs that pointed to versions of python that were broken on my machine. In addition, I had many old projects that used outdated, broken versions of python on my machine.
    • When VSCode's python extensions start up, they try to introspect into the environment to see what virtualenvs are available. In order to support the widest market, the startup script searches for all the known and supported kinds of virtual environments. Unfortunately, it appears that if even one of the versions of python or one of the Py/Virtualenvs errors out during VS Code's startup, it breaks the whole environment discovery process.
    • This is not generally well-known in the VS Code community, as far as I can tell. The only reason I figured this out was because of a comment that LazarusX left on the vscode-python github issue here:

      • It's migrated from my last machine, and only used a few years ago. I have no idea why it's scanned as a potential environment. However, IMO, there could be enhancement for this situation, because a broken environment should not block the entire interpreter discovery process. I also believe there is sort of regression here, because the extension used to work fine in spite of the broken environment.
        

    • Solution Summary

  • Key Technique - trace logs in the output panel of VS Code

    • I saw that the VS Code team was very active about responding to bug reports, but they always needed a user to get the trace or debug python logs. This technique helped me figure out what my problem was.

      • #VSCode #Keyshort to open the output view in the View menu > Output #card

        • Shift Cmd U
        • image.png
        • Mnemonics:
          • OUTPUT
          • If something went wrong, you're going to need to do a U Turn and look at the output logs

      • Select the Python menu in the View > Output window to view VSCode/Language/Python logs

        • image.png{:height 507, :width 154}
        • AFAICT there isn't a keyboard shortcut for this step.

      • Select the Settings Cog logo in the right menu portion of the Output View and select trace

        • image.png

  • Steps Taken

    • Uninstall any unneeded VSCode/Extensions including all python and jupyter extensions such as VSCode/Extension/Python

      •   414  code --list-extensions --show-versions\n
          415  code --uninstall-extension vue.volar
          416  code --list-extensions --show-versions\n

    • Destroy Pyenv/virtualenv associated with the project I was working on; Person/codekiln/GitHub/gitpa/acolyte

      • I removed .python-version files and also removed the pyenv virtualenv

      •   424  pyenv virtualenv-delete gitp-acolyte
          425  ls -la
          426  rm -rf .python-version
          427  rm -rf gitp-acolyte/.python-version

    • Reset PyPoetry

      • Reinstall Poetry using the official installer

        • Poetry Installation

        •   432  curl -sSL https://install.python-poetry.org | python3 -

        • Note that this script installs it into a virtualenv located in ~/Library/Application Support/pypoetry (where the config also is; see below)

      • Install PyPoetry/Shell/Zsh/OhMyZsh integration with Zsh/OhMyZsh

        •   438  vim ~/.zshrc # I had to add
            439  . ~/.zshrc
            440  poetry --version
            441  mkdir $ZSH_CUSTOM/plugins/poetry
            442  poetry completions zsh > $ZSH_CUSTOM/plugins/poetry/_poetry\n
            443  vim ~/.zshrc
            444  . ~/.zsrhc

      • Reset Poetry Config

        • I had blindly tried many different PyPoetry/Configurations (poetry configurations) while trying to debug this issue:

          • (gitp-acolyte)  ✘ Me@Home  ~/dev/gitpa/gitp-acolyte  ↱ cc/04-host-n-dist/03-ai-ep-yml  history | grep "poetry config"
              112  poetry config virtualenvs.prefer-active-python true\n
              385  poetry config virtualenvs.in-project true
              395  poetry config virtualenvs.in-project false
              447  poetry config virtualenvs.prefer-active-python true
              456  poetry config --list

            • note that poetry config virtualenvs.in-project set to both true and false were advocated for here
              • python - VSCode doesn't show poetry virtualenvs in select interpreter option - Stack Overflow (asked 4 years ago, 188 votes up)
            • I tried virtualenvs.prefer-active-python true because of this confusing page in the Pyenv/Docs/Managing-Environments:
              • Managing environments | Documentation | Poetry - Python dependency management and packaging made easy

                • If you use a tool like pyenv to manage different Python versions, you can set the experimental virtualenvs.prefer-active-python option to true. Poetry will then try to find the current python of your shell.
                  

                • For instance, if your project requires a newer Python than is available with your system, a standard workflow would be:
                  

                  • pyenv install 3.9.8
                    

                  • pyenv local 3.9.8
                    

                  • poetry install
                    

              • Ok, that definitely makes it seem like that's a good way to go for pyenv. However, what this actually does is not intuitive. And I ended up seeing some people who actually removed the configuration to solve it (which would have removed the default value of false)
                • poetry env info does not respect virtualenvs.prefer-active-python = true setting · Issue #5947 · python-poetry/poetry GitHub/python-poetry/poetry/Issue
                  • Note, after all the steps below, my results were different from the OP here in that they were the same whether I ran poetry run poetry env info or if I ran poetry env info

                    • (gitp-acolyte)  Me@Home  ~/dev/gitpa/gitp-acolyte  ↱ cc/04-host-n-dist/03-ai-ep-yml  poetry run poetry env info                            
                      
                      Virtualenv
                      Python:         3.13.1
                      Implementation: CPython
                      Path:           /Users/Me/.pyenv/versions/3.13.1/envs/gitp-acolyte
                      Executable:     /Users/Me/.pyenv/versions/3.13.1/envs/gitp-acolyte/bin/python
                      Valid:          True
                      
                      Base
                      Platform:   darwin
                      OS:         posix
                      Python:     3.13.1
                      Path:       /Users/Me/.pyenv/versions/3.13.1
                      Executable: /Users/Me/.pyenv/versions/3.13.1/bin/python3.13
                      (gitp-acolyte)  Me@Home  ~/dev/gitpa/gitp-acolyte  ↱ cc/04-host-n-dist/03-ai-ep-yml  poetry env info                      
                      
                      Virtualenv
                      Python:         3.13.1
                      Implementation: CPython
                      Path:           /Users/Me/.pyenv/versions/3.13.1/envs/gitp-acolyte
                      Executable:     /Users/Me/.pyenv/versions/3.13.1/envs/gitp-acolyte/bin/python
                      Valid:          True
                      
                      Base
                      Platform:   darwin
                      OS:         posix
                      Python:     3.13.1
                      Path:       /Users/Me/.pyenv/versions/3.13.1
                      Executable: /Users/Me/.pyenv/versions/3.13.1/bin/python3.13

        • These configurations are stored in a toml file; see Configuration | Poetry

          • This file can typically be found in one of the following directories:

            • macOS: ~/Library/Application Support/pypoetry

        • I read many articles and forum posts about using poetry with pyenv, and they all seemed to give contradictory advice. But the thing that ended up working for me was removing the customizations to the config that I did by removing this dir. In particular, I removed these customizations because

          • (gitp-acolyte)  Me@Home  ~/dev/gitpa/gitp-acolyte  ↱ cc/04-host-n-dist/03-ai-ep-yml  history | grep "pypoetry"     
              176  emulate bash -c '. /Users/Me/Library/Caches/pypoetry/virtualenvs/gitp-acolyte-kfI6rJB7-py3.13/bin/activate'
              361  emulate bash -c '. /Users/Me/Library/Caches/pypoetry/virtualenvs/gitp-acolyte-kfI6rJB7-py3.13/bin/activate'
              458  ls ~/Library/Application Support/pypoetry
              459  ls "~/Library/Application Support/pypoetry"
              460  ls ~/Library/Application\ Support/pypoetry
              461  vim ~/Library/Application\ Support/pypoetry
              462  rm -rf ~/Library/Application\ Support/pypoetry/config.toml

        • These changes stuck (after the below)...

          • (gitp-acolyte)  Me@Home  ~/dev/gitpa/gitp-acolyte  ↱ cc/04-host-n-dist/03-ai-ep-yml  ls -la ~/Library/Application\ Support/pypoetry/
            total 8
            drwxr-xr-x@   4 Me  staff   128 Dec 25 12:48 .
            drwxr-xr-x  179 Me  staff  5728 Dec 13 07:24 ..
            -rw-r--r--@   1 Me  staff     5 Dec 25 12:29 VERSION
            drwxr-xr-x@   7 Me  staff   224 Dec 25 12:29 venv

    • Recreate the Person/codekiln/GitHub/gitpa/acolyte Pyenv/virtualenv and specify the pyenv global version to be 3.13.1

      •   449  pyenv virtualenv 3.13.1 gitp-acolyte
          452  pyenv local gitp-acolyte
          453  pyenv which python
          469  pyenv global 3.13.1

    • Uninstall all unneeded python versions with Pyenv and their Pyenv/virtualenvs

      • The first part of the solution was to uninstall all old virtual environments and old versions of python.
      • shell

        #!/bin/bash
        
        # Get list of virtualenvs in ~/.virtualenvs
        old_virtualenvs=$(ls ~/.virtualenvs)
        
        # Get list of pyenv-managed virtualenvs
        pyenv_virtualenvs=$(pyenv virtualenvs --bare)
        
        # Loop through each virtualenv in ~/.virtualenvs
        for env in $old_virtualenvs; do
            # Check if it's managed by pyenv-virtualenv
            if ! echo "$pyenv_virtualenvs" | grep -q "^$env$"; then
                echo "Deleting unused virtualenv: $env"
                rm -rf ~/.virtualenvs/"$env"
            fi
        done
        

      •   473  pyenv versions
          474  pyenv uninstall 3.5.6
          475  pyenv uninstall 3.6.7
          476  pyenv versions
          477  pyenv uninstall 3.7.1
          478  pyenv versions
          479  pyenv uninstall 3.9.6
          480  pyenv uninstall 3.12.0
          481  pyenv versions
          482  ls -la /Users/Me/.pyenv/versions/
          483  ls -la ~/.virtualenvs/
          485  touch scripts/clean_virtualenvs.sh
          486  chmod +x scripts/clean_virtualenvs.sh
          487  vscr scripts/clean_virtualenvs.sh
          488  ./scripts/clean_virtualenvs.sh
          490  ls -la /Users/Me/.virtualenvs/
          491  rm -rf ~/.virtualenvs/2023-11-26\ Sun\ Chiaroscurro\ Project
          493  pwd
          494  poetry install
          495  poetry env info --path
          496  poetry env info --executable | pbcopy

    • Reinitialize PyPoetry inside Person/codekiln/GitHub/gitpa/acolyte

      • 494 poetry install

    • Start VS Code and reinstall VSCode/Extension/Python; associate it with the pyenv virtualenv that poetry install installed to

      • 495 poetry env info --path

        496 poetry env info --executable | pbcopy
        

      • now I did Cmd Shift P in #VSCode and did Python: Select Interpreter and then entered in the path to the virtualenv. It started working.