Conda env defined $PYTHONPATH is ignored - breaks pylint, refactoring,...

[TotallyFred]

Issue Type: Bug

I set up Code Studio to use a conda environment called LMCrypto for my project.

Here is the content from settings.json that was generated by selecting the environment from the bottom left blue status line:

(LMCrypto)# cat ../.vscode/settings.json
{
"python.pythonPath": "/Users/fdetienn/anaconda3/envs/LMCrypto/bin/python"
}

This environment contains a PYTHONPATH:

(LMCrypto)# echo $PYTHONPATH
/Users/fdetienn/Documents/Programming/LazyMaestro/standalone/lmcrypto

When CodeStudio runs pylint (upon saving the file), all the import for my own modules are underlined in red with a pylint-import error .

When I run pylint from a terminal (including in code studio's terminal), pylint properly detects the import path from PYTHONPATH.

I.e. the problem only appears when Code Studio runs pylint on its own.

Seemingly, code refactoring is also broken.

This was working nicely before my vacation and seems to have appeared with the latest upgrade.

Extension version: 2020.8.109390
VS Code version: Code 1.49.0 (e790b931385d72cf5669fcefc51cdf65990efa5d, 2020-09-10T17:39:53.251Z)
OS version: Darwin x64 19.6.0

System Info

Item	Value
CPUs	Intel(R) Core(TM) i7-7820HQ CPU @ 2.90GHz (8 x 2900)
GPU Status	2d_canvas: enabled flash_3d: enabled flash_stage3d: enabled flash_stage3d_baseline: enabled gpu_compositing: enabled metal: disabled_off multiple_raster_threads: enabled_on oop_rasterization: enabled opengl: enabled_on protected_video_decode: unavailable_off rasterization: enabled skia_renderer: disabled_off_ok video_decode: enabled webgl: enabled webgl2: enabled
Load (avg)	3, 3, 3
Memory (System)	16.00GB (0.10GB free)
Process Argv
Screen Reader	no
VM	0%

[karthiknadig]

Try these steps when working with conda. We have some know issues with conda due to limitation in conda itself and VS Code:

1. Turn off auto activation. To do this, disable these two settings, ideally only for the project you are working on:
   
2. Exit VS Code.
3. Open Anaconda Prompt, navigate to your project.
4. Activate the environment that you want.
5. Run code . or code project.code-workspace to launch VS Code.

This will give you an activate environment, and the extension will not have to try and activate every time a terminal is created.

[TotallyFred]

Thanks! I had completely overlooked these options all the way at the bottom.

In my case:

• Activate on load was disabled.
• Activate in terminal created by the extension was enabled.

I tried your proposal and it works.

I also tried the reverse: I activated both options and it works without having to fire up a terminal. Ie everything works as expected (pylint works again, the first terminal jumps into the conda environment, and all) which is more comfortable.

I have no idea how "activate on load" got disabled. I'm pretty sure I didn't change it (but only 99%). Or maybe they're new options?

Now I read you when you say there are some known issues with conda. Maybe I'm getting lucky but I'll keep my eyes open. Can you point me to the existing bugs to be aware of the symptoms ?

[karthiknadig]

We have not changed those settings or added new ones. So i am not sure what went wring there.
You can see the issues with conda and terminal here: #5344 #4568