Selected pipenv virtualenv is written to python.pythonPath

[Urkallinger]

Since some time (I can't say when exactly) the virutalenv is not recognized in my project. I open the project folder with vs code and as soon as the python extension is loaded, the system wide installation of python is recognized as interpreter, although there is a pipfile in my project folder.

Environment data

• VS Code version: 1.54.1
• Extension version (available under the Extensions sidebar): v2021.2.636928669
• OS and version: Windows 10, 20H2
• Python version (& distribution if applicable, e.g. Anaconda): 3.7.3
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): pipenv virutalenv
• Value of the python.languageServer setting: Pylance

Expected behaviour

the virutalenv is recognized and I don't have to set the "python.pythonPath" setting explicitly.

Actual behaviour

The system-wide Python installation is selected as interpreter.
When I change the interpreter to the pipenv virtualenv, the absolute path is entered into my settings.json.
This is problematic because the virutalenv is in my user directory and so the path is not correct for all developers involved.

Details

The corresponding virtualenv is marked as "(cached)" in the interpreter selection.

I already tried the following commands:

• Python: Clear Workspace Interpreter Settings
• Python: Reset Stored Info for Untrusted Interpreters

both had no effect.

i also reinstalled vs code and deleted the folders in %appdata%, but still the virutalenv was not detected.

[kimadeline]

Hi @Urkallinger, thank you for reaching out!

We made some changes to interpreter discovery (#12020), could you give them a try?

To do so, open your machine's setting.json (use the command "Preferences: Open Settings (JSON") and add the following:

"python.experiments.optInto": [        
        "pythonDiscoveryModule"
    ],

[Urkallinger]

Hi @kimadeline,
thanks for the quick response!

I have copied the setting into my settings. I also deleted the entry "python.pythonPath" from the workspace settings. After that I restarted vs code. Unfortunately the system wide Python was selected as interpreter again.
I executed the two commands "Python: Clear Workspace Interpreter Settings" and "Python: Reset Stored Info for Untrusted Interpreters" and restarted. Unfortunately also without success.

I have installed pipenv (version 2020.11.15) in my system-wide Python. When I call the following command with it, it gives me the correct path to virtualenv in my user directory:
pipenv --venv

[kimadeline]

Sorry for the response delay, and thank you for your patience!

A couple of things here:

• The 2 commands that you ran ("Python: Clear Workspace Interpreter Settings" and "Python: Reset Stored Info for Untrusted Interpreters") have an effect only if you are in the DeprecatePythonPath experiment.
• We don't perform pipenv discovery on extension activation anymore (Remove pipenv discovery from interpreter discovery on activation #11127), so your environment will not be auto detected until you select it at least once. Does your environment appear in the list of interpreters when you click on the interpreter in the status bar (bottom left)?

[Urkallinger]

My environment is correctly detected and offered in the dropdown:

As soon as I select my environment then, I get the entry

"python.pythonPath": "C:\\Users\\user\\.virtualenvs\\envname\\Scripts\\python.exe",

in my workspace settings.

If I delete this entry, the system-wide python installation (C:\Python37\python.exe) is selected again. This remains so after a reboot.

[kimadeline]

Could you copy here the output for Python in the Output panel (View→Output, change the drop-down the upper-right of the Output panel to Python)?

Also, just to confirm, how did you create your virtual environment? Does the issue happen with newly created environments as well?

[Urkallinger]

Output panel

User belongs to experiment group 'pythonDiscoveryModule'
User belongs to experiment group 'pythonaacf'
User belongs to experiment group 'pythonSendEntireLineToREPL'
User belongs to experiment group 'pythonNotDisplayLinterPrompt'
User belongs to experiment group 'pythonDiscoveryModulecf'
User belongs to experiment group 'pythonTensorboardExperiment'
User belongs to experiment group 'PythonPyTorchProfiler'
User belongs to experiment group 'ShowExtensionSurveyPrompt - control'
User belongs to experiment group 'CollectLSRequestTiming - control'
> python c:\Users\username\.vscode\extensions\ms-python.python-2021.3.680753044\pythonFiles\pyvsc-run-isolated.py -c "import sys;print(sys.executable)"
> python c:\Users\username\.vscode\extensions\ms-python.python-2021.3.680753044\pythonFiles\pyvsc-run-isolated.py -c "import sys;print(sys.executable)"
Python interpreter path: C:\Python37\python.exe
Starting Pylance language server.
> conda --version

how did you create your virtual environment?

The virtualenv is created and also updated as follows (these steps are automated by a python script):

• creation of the virtualenv by pipenv (C:\Python37\python.exe -m pipenv run python)
• install pip 20.2.4 (C:\Python37\Scripts\pipenv.exe run python -m pip install pip==20.2.4)
• clean (C:\Python37\Scripts\pipenv.exe run pipenv clean)
• sync (C:\Python37\Scripts\pipenv.exe run pipenv sync --dev --pypi-mirror <mirror1> --pypi-mirror <mirror2> --pypi-mirror <mirror3>)

Does the issue happen with newly created environments as well?

Yes, even then this behavior occurs.
I proceeded as follows:

• mkdir vscode_pipenv_test
• cd vscode_pipenv_test
• pipenv --python 3.7
• code .

The system-wide python is now selected as interpreter. If I select the just created virutalenv, the path is entered in the settings. If I delete the entry, the system-wide python is selected again.

The problem occurs on the notebook I have from my company. Can it be a permission problem?

[kimadeline]

I may have misunderstood your issue 🙈 Can you try setting the following in your global settings.json:

"python.experiments.optInto": [        
        "pythonDiscoveryModule"
]

and then reload VS Code and select your interpreter?