FIX: compatibility with collab

[mmcky]

This PR looks to fix compatibility with google collab new julia option.

This updates the generated download notebooks with julia kernel name rather than julia-1.11 which is version specific.

see #330

[github-actions]

🚀 Deployed on https://6809beaba8c5c54bfc06f90e--vibrant-allen-f39ec9.netlify.app

[jlperla]

Hey @mmcky I wasn't sure how to test this correctly. Note that the julia kernel shows up in the _config.yml but also at the top of each of the myst files. They look like:

---
jupytext:
  text_representation:
    extension: .md
    format_name: myst
kernelspec:
  display_name: Julia
  language: julia
  name: julia-1.11
---

Things seem to execute here though. Maybe what ends up happening is that the embedded kernel is used for the myst execution, but the generated notebooks are left with the more general naming from the config?

I downloaded one of the notebooks from the netlify output here and the relevant stuff in the ipynb is

 "metadata": {
  "date": 1745469061.0018864,
  "filename": "julia_by_example.md",
  "kernelspec": {
   "display_name": "Julia",
   "language": "julia",
   "name": "julia"
  },
  "title": "Introductory Examples"
 },
 "nbformat": 4,
 "nbformat_minor": 5
}

So this seems to match the change to just julia. When I open this in VS Code's ipynb editor, or in jupyterlab, I think you need ot manually choose the kernel. That might be confusing.

But for all of that, it seems like there may be a tradeoff here. If you leave the kernelname as julia-1.11 then I think that the jupyter/jupyterhub/etc. will correctly open it when you run (without having to manually change the kernel type whenever you open the file).

With colab is it that when choosing julia it automatically opens without manual selection of the kernel? But if you put up the current notebooks with juila-1.11 does it just force you to select the runtimt type? If so, maybe that is preferrred since there are more users of jupyterlab/VS Code than there are of colab?

[mmcky]

thanks @jlperla

So this seems to match the change to just julia. When I open this in VS Code's ipynb editor, or in jupyterlab, I think you need ot manually choose the kernel. That might be confusing.

Oh that is a downside. I thought using Julia would select the latest kernel version. I think that's how the python side works. Thanks for letting me know @jlperla.

Happy for you to make the call on this tradeoff. Please feel free to close when you're ready.