Parameterized Kernels

[kevin-bates]

Hello all. I recently closed the draft Parameterized Kernel Launch due to lack of traction and (primarily) time to advocate for it. I also believe, now that I better understand the JEP process, that it was not necessarily in the spirit of a JEP and should be more of a community effort - both in terms of advocacy as well as design/implementation.

As we discussed in today's server meeting, this issue exists to have a discussion on this topic and get a pulse for parameterized kernel's need and participation. Should we find an interest, then perhaps we can gather some folks together as an ongoing working group where we can meet regularly or use a small portion of the server dev meeting to discuss and report progress.

Please post any comments you might have regarding this topic, its utility, and in what capacity you'd like to participate (if applicable).

[welcome]

Thank you for opening your first issue in this project! Engagement like this is essential for open source projects! 🤗

If you haven't done so already, check out Jupyter's Code of Conduct. Also, please try to follow the issue template as it helps other other community members to contribute more effectively.

You can meet the other Jovyans by joining our Discourse forum. There is also an intro thread there where you can stop by and say Hi! 👋

Welcome to the Jupyter community! 🎉

[kevin-bates]

To this day, I am asked how to accomplish something relative to (primarily remote) kernels and, more often than not, the answer is "this should be handled by parameterization" - so I firmly believe this would be a useful feature. I also believe parameterization is primarily about the environment in which the kernel is running (i.e., resources) rather than kernel-specific functionality (although any solution needs to account for these). So, in that sense, I suppose I'd be +1 for a reboot and would be happy to participate in its design and implementation.

[echarles]

Thx for opening this @kevin-bates

As said during today meeting, I plan to demonstrate the work done in https://github.com/deshaw/ksmm that is a way for JupyterLab to create custom kernelspecs based on templates. The parameters are baked in the templated based on jsonschema.

The demo will be given during next JupyterLab meeting and is IMHO a good teaser to get inputs from users (the ones behind the screens and using JupyterLab) and to generate interest, contributions as part of an advocacy initiative.

Please add me to the list of participants to the working group.

cc/ @mlucool