Explore enabling Non-Python kernel auto complete

[rebornix]

We might already have issues open for this request but I couldn't find that yet.

Notes:

• Feature will be behind an experiment
• Experiment will be rolled out to all insiders
• We will need to ensure completion requests to the kernel complete quickly, so as to prevent slowing down the kernel
• Python kernel completion cannot be disabled (it is on by default)
• Python kernel completion does not support displaying documentation (resolveCompletionItem not implemeneted)
• Kernel completions will be enabled by default
• Configurable non-python kernel completion trigger characters
• Setting to run off non-python kernel completions
• Merge Python and non-python kernel completions into the same settings (turn off and trigger characters)
  This will need to be done after the experiment, once were comfortable enabling this new feature (non-python kernels) for all
• Leave experiment for at least 2 months.

[Philonoist]

Yes please! As a concrete example, I would like to use Rust jupyter kernel (https://github.com/evcxr/evcxr/blob/main/evcxr_jupyter/README.md)
with completions. (rust-analyzer lsp integration exists both in plain jupyter lab, and in vscode).
If I can help here, let me know.

[maxandersen]

I see the option Enable Kernel Completions but not seeing any requests for it to the kernel.

how does one trigger it?

[DonJayamanne]

t not seeing any requests for it to the kernel.

@maxandersen What version of VS Code and Jupyter extension are you using?
This should be enabled by default in VS Code Insiders and the Pre-releaase version of the Jupyter extension (I think i mentioned that in a previous comment)

how does one trigger it?

This gets triggered via the characters . @ # in Java.
You can customize these trigger characters via the setting jupyter.completionTriggerCharacters

[maxandersen]

I'm running latest vscode insiders (Version: 1.85.0-insider (Universal)
Commit: 055458feab40fc308fc6d70b185068138c8cf8eb
Date: 2023-11-27T05:36:15.984Z (1 day ago) and the pre-release of jupyter extension v2023.11.1003331000.

I'll try it out some more and see whats going on.