Notebooks - Create new notebook not honoring default kernel

[claudiaregio]

Environment data

• VS Code version: 1.61.0 - insider
• Jupyter Extension version (available under the Extensions sidebar): v2021.9.1001294413

Steps to reproduce:

1. Create new notebook
2. Select kernel is always available despite having the default kernel for jupyter notebooks selected
   

Expected behavior

Notebook to be created with the default kernel selected

Actual behavior

Select kernel always available

*On a related note, is this a regression? Before core supported notion of default kernel, Jupyter extension used to automatically assign the kernel that was last used in a notebook.

[DonJayamanne]

Before core supported notion of default kernel, Jupyter extension used to automatically assign the kernel that was last used in a notebook.

Not sure we did this.
The only concept of default kernel we had still applies and its the top of the list, when you get the kernel picker.

lect kernel is always available despite having the default kernel for jupyter notebooks selected

How do you select a default kernel.
This is how it has worked from day 1, in native notebooks, users will always get prompted to select a kernel for all notebooks.

[claudiaregio]

I remember we had added functionality in the extension such that if a user opened a notebook it would assign the kernel to the one that was last used.

The gear in the kernel drop down now says "Set as default for jupyter notebooks", whereas previously it used to say "Set default kernel provider". Based on the changed tooltip I thought the functionality has changed but even if this was not the case I remember the Jupyter extension used to consistently create a new notebook with the previously used kernel.

[greazer]

Automatically selecting a kernel on a new notebook is something that the jupyter extension can't do on it's own.

Related to #130946. We're also getting survey feedback asking for this.

[burkeholland]

Ran into this in the Golden Scenario. It seems odd to me to pick a different interpreter for Notebooks that are in the same "project" if you will. It also takes several seconds for VS Code to connect to the interpreter.

select-interpreter.mp4

[xkz0777]

Ran into this in the Golden Scenario. It seems odd to me to pick a different interpreter for Notebooks that are in the same "project" if you will. It also takes several seconds for VS Code to connect to the interpreter.

select-interpreter.mp4

I'm learning d2l now and I also feel very inconvenient about it. Want some configuration options in the workspace's settings.json to solve this problem.

[DonJayamanne]

Given this is a very old issue and predates most of the kernel work done
I believe the underlying issue has been resolve since, or we can close this in favour of #130946