Add PyJulia package in julia-notebook and datascience-notebook

[i0ntempest]

What docker image(s) is this feature applicable to?

julia-notebook

What change(s) are you proposing?

Add PyJulia package to julia-notebook and datascience-notebook

How does this affect the user?

PyJulia allows calling julia from python, which is very useful when using this notebook with Open WebUI where models can only call Python.

Anything else?

No response

[mathbunnyru]

PyJulia doesn't seem to be supported anymore: https://github.com/JuliaPy/pyjulia#pyjulia:

Ongoing development of the Python/Julia interface has transitioned to PythonCall.jl/juliacall, please consider using that package instead.

So, I don't think we will be adding PyJulia.

But we might add https://github.com/JuliaPy/PythonCall.jl, could you please create a PR where you add this package?
It would be great, if you added a test that it works as well

[i0ntempest]

PyJulia doesn't seem to be supported anymore: JuliaPy/pyjulia#pyjulia:

Ongoing development of the Python/Julia interface has transitioned to PythonCall.jl/juliacall, please consider using that package instead.

So, I don't think we will be adding PyJulia.

But we might add JuliaPy/PythonCall.jl, could you please create a PR where you add this package? It would be great, if you added a test that it works as well

Just saw that - that package would be even better. I'm not familiar with composing docker images, but I'll try adding packages to a running image and report back.

[mathbunnyru]

@i0ntempest I would be glad if you created a PR for your suggestion

To do that you need to:

1. Add a package here: https://github.com/jupyter/docker-stacks/blob/main/images/minimal-notebook/setup-scripts/setup-julia-packages.bash#L34
2. A a simple test here: https://github.com/jupyter/docker-stacks/tree/main/tests/image_specific_tests/julia-notebook
3. Take a look at our policy when we add new packages: https://jupyter-docker-stacks.readthedocs.io/en/latest/maintaining/new-images-and-packages-policy.html. It's mostly for maintainers, but still worth taking a look at.
4. Update: https://github.com/jupyter/docker-stacks/blob/main/docs/using/selecting.md

[mathbunnyru]

@i0ntempest would you like to work on this?

I'm happy to help you, if you have any questions.

[i0ntempest]

@i0ntempest would you like to work on this?

I'm happy to help you, if you have any questions.

Planned, just pretty busy with uni recently. Will report back when there's progress.

[benz0li]

@i0ntempest You may be interested in b-data's/my JupyterLab Julia docker stack or Data Science dev containers.

These container images allow to persist¹ the user's home directory. And because packages (both Julia and Python) are installed within the home directory, those are persisted, too.

Footnotes

1. using either a volume or bind mount ↩

[mathbunnyru]

@i0ntempest You may be interested in b-data's/my JupyterLab Julia docker stack or Data Science dev containers.

These container images allow to persist1 the user's home directory. And because packages (both Julia and Python) are installed within the home directory, those are persisted, too.

1. using either a volume or bind mount ↩

This issue is not about persisting home directories though.
As far as I know, your images also have no interaction between Julia and python, so they won't help here.

[benz0li]

As far as I know, your images also have no interaction between Julia and python, so they won't help here.

No, such a package is not installed by default.

But @i0ntempest can install anything at runtime that serves the purpose, e.g. PythonCall.

This issue is not about persisting home directories though.

Because one can persist the whole home directory (packages, configurations, etc.), there is no need to pre-install such a package.

[benz0li]

@i0ntempest You may test at https://demo.jupyter.b-data.ch. Select image Julia 1.11 + Python 3.12 (jupyterlab/julia/pubtools:latest).