Can that be installed through Conda? And will it work with jupyter-console?

[ubaldot]

I am wondering if it is possible to install it through conda?

And, once installed, if I can run the kernel with $ jupyter-console --kernel=matlab

[prabhakk-mw]

Hi @ubaldot

1. No, we have not published the jupyter-matlab-proxy package to conda.
   You will not be able to install it through the command conda install jupyter-matlab-proxy

One could however use PIP to install the package from within the conda environment. For example;

$ conda activate testEnv
(testEnv)$ python -m pip install jupyter-matlab-proxy

2. No, This package does not currently support jupyter-console.

$ jupyter-console --kernel=jupyter_matlab_kernel
Jupyter console 6.4.4

MATLAB
In [1]: ver
Kernel needs to be started by a Jupyter Server. Please use JupyterLab or Classic Notebook while using MATLAB Kernel for Jupyter.
In [2]: 

I'll create an internal request for this capability.

[ubaldot]

Thanks for the reply.
Could we keep this issue open until the kernel will be ready to be opened also by jupyter-console?

[ubaldot]

Hello! Is there any update for this issue?

[prabhakk-mw]

Hi @ubaldot,
Thanks for circling back, @krisctl is currently working on putting the infrastructure in place to enable this capability.
We expect this to be available for use in the coming weeks, and will update this issue when its ready.
Thanks!

[ubaldot]

Thanks! You mean both the features conda and jupyter-console?

[prabhakk-mw]

@ubaldot ,
No, I was only referring to using the kernel from jupyter-console.

We do not plan to publish to conda.
Is there a workflow that is not well served via our current release strategy to PyPI?

[ubaldot]

Ah ok, thanks! Looking forward to jupyter console integration!

There are many differences between PyPI and conda-forge for instance, with the second preferred especially when it comes into scientific work, plus several other things such as making sanity checks of the environment before installing packages and so on. You can also pack binaries with conda not just python code. Conda-forge is taking traction fairly a lot.
My suggestion is you guys to check conda-forge. It's not just hype. Once you have a package on PyPI it is rather straightforward to make a conda package as well. :)

[zhangxiaoxing]

I agree that familiarity with both Conda and pip is essential for anyone utilizing Conda in the first place, as it enables comprehensive package management. While leveraging both tools can enhance versatility, I'd like to highlight that deploying projects solely through Conda often projects a more polished and professional image. In contrast, relying on a mix of Conda and post-hoc adjustments via pip might inadvertently convey a less refined, 'rapid development' aesthetic – akin to a prototype rather than a finished product. Just my 2 cents.