Activate conda environments using the path instead of the conda env name

A project-local conda environment will be seen by VS Code, but it will fail to activate automatically. I would guess this is because it looks at the end of the path for the environment name and always activates by name, rather than by path. However, for environments created with the -p pathname flag, this will fail (as you can’t even set environments name for them with -n).

Environment data

• VS Code version: 1.30.1
• Extension version (available under the Extensions sidebar): 2018.12.1
• OS and version: Windows 10 V 1803
• Python version: 3.7.0, conda 4.5.12
• Type of virtual environment used: conda

Steps to reproduce:

1. Create empty directory (mine was D:/TST) for project and add a random .py file (for ext to load), open in VS Code.
2. Run conda create -p ./localenv python=3.7 in that directory.
3. Select this environment as the Python interpreter.
4. Make sure "python.terminal.activateEnvironment": true (or unset) in settings.
5. Open new terminal (EDIT: as in interactive terminal, e.g. a REPL; for me, this was the cmd, but running code in the REPL has the same effect: it fails to activate the env, and thus runs in the default environment).

Expected behaviour

Automatic environment activation by path, rather than name:

D:\TST>activate ./localenv

(D:\TST\localenv) D:\TST>

OR

D:\TST>activate "D:\TST\localenv"

(D:\TST\localenv) D:\TST>

Actual behaviour

When opening a terminal:

D:\TST>activate localenv
Could not find conda environment: localenv
You can list all discoverable environments with `conda info --envs`.
D:\TST>

[lpravda]

I think I have found the solution eventually in some other thread here. However, don’t really know which one, so I can’t give a proper credit to the person who came up with this. However, putting:

    "terminal.integrated.env.osx": {
        "PATH": ""
    }

into settings.json file did the trick, until the bug is fixed.

[EBjerrum]

I have this issue too. Can’t use various features when I have conda environments created with -p. In my case it’s necessary to use prefixes in conda environments due to integration with our compute cluster. Looking forward to seeing a solution.