Select interpreter can't list the old Pipenv virtual environments

[amanchaudhary-95]

Does this issue occur when all extensions are disabled?: Yes

• VS Code Version: 1.88
• OS Version: Windows 11 x64

Steps to Reproduce:

1. Create a virtual environment with pipenv.

2. Open vs code and execute select interpreter command, if in the same folder in which environment is created the, it shows the interpreter. Image attached:
   

3. After that close the VS code. Create a sub-folder or change the directory. Again, open the VS code and execute select interpreter command. Now, all the old virtual environments are not visible. Image attached:
   

The issue is resolved after downgrading this extension to v2024.0.1

### Tasks

[anthonykim1]

amanchaudhary-95 Thanks for filing the issue.

I believe this is as designed from: #23102 as we "filter out any pipenv or poetry envs which do not belong to the current workspace."

Should we not be doing the filtering pipenv per workspace? Karthik Nadig (@karthiknadig)

[karthiknadig]

This is expected behavior. Both pipenv and poetry expect that you create a different environment per project. They often generate a hash for environment key based on the project. This allows the tools to detect the right environment for a project. By reusing the environment in a different project can subtly break behavior.

[amanchaudhary-95]

Karthik Nadig (@karthiknadig) If the behaviour is as per the expectation then at least make it detect the parent environment inside the subfolders e.g. let's say I've below folder structure:

Development(Parent Folder)
|----Project1
|----Project2
|----Project3

Now, If I create a virtual environment in the "Development" folder then it should be visible in Project1, Project2 and Project3. Currently, even that is not possible. Since Pipenv doesn't allow the creation of a nested environment hence it should not be the issue.

The reason why I'm asking for this is because I've multiple ML models and while in development, these models use the same Python libraries. It is redundant and space-consuming to create a new virtual environment every time I want to develop a new model.