Jupyter environment variables load only once ever per kernel.

[tbenthompson]

Applies To

• Notebooks (.ipynb files)
• Interactive Window and/or Cell Scripts (.py files with #%% markers)

What happened?

Once I run a notebook, the settings in the .env file become "stuck" across:

• restarting the kernel
• restarting VSCode entirely
• restarting my laptop entirely
• opening an entirely different notebook
• deleting and re-creating the conda environment with the same name

The only solution is to create a conda env/kernel with a new name at which point the .env file is reloaded. But, again, only once.

It seems like there must be some kind of cache inside the extension?

Below, you can see two interactive windows that I loaded from two different kernels:

1. On the left, this is the output from a kernel that I loaded before changing the .env file. This output is incorrect. The output here is after a full restart of VSCode.
2. On the right, this is the output from a kernel that I loaded after changing the .env file. This output is correct.

VS Code Version

Version: 1.69.0 (Universal) Commit: 92d25e35d9bf1a6b16f7d0758f25d48ace11e5b9 Date: 2022-07-07T05:28:41.511Z (3 days ago) Electron: 18.3.5 Chromium: 100.0.4896.160 Node.js: 16.13.2 V8: 10.0.139.17-electron.0 OS: Darwin arm64 21.5.0

Jupyter Extension Version

v2022.6.1001902341

Jupyter logs

No response

Coding Language and Runtime Version

Python 3.10.5

Language Extension Version (if applicable)

v2022.10.0

Anaconda Version (if applicable)

4.13.0

Running Jupyter locally or remotely?

No response

[DonJayamanne]

Thanks for filing this extension, i'll have a look at this soon.
Please note, we've identified a few issues related to the environment variables being loaded in kernels & plan on addressing them for the next release
See here #10359

[tbenthompson]

Oh, haha, it's great to hear that there are known .env problems. I thought I was going crazy with this strange bug.

[DonJayamanne]

@tbenthompson Please could you :

• Try this again
• Provide all of the output from the Jupyter output panel (you can get to that using the command Jupyter: Show Output)

[tbenthompson]

Sure, here: https://gist.github.com/tbenthompson/63b3697b8be29d3ef18a5d24771b5441
Thanks!

[DonJayamanne]

Should be fixed with this PR #10843
Please could you install the latet pre-release version of the Jupyter extension and try this again (in vscode insiders) & let us know.
If it still persists, please feel free to reopen this issue, or I can do that.