[pacharanero]

Overview

Add a roadmap item to replace the current Jupyter notebooks with reproducible Marimo notebooks and build a small baseline of researcher-friendly workbooks for the digital growth charts platform.

Scope recorded

• Keep notebook dependencies optional.
• Make workbook execution reproducible from a clean environment and add headless CI execution.
• Preserve clinical calculations, reference data, and golden outputs during the format migration; review any example or expected-value changes separately.

Validation

• s/lint
• git diff --check

[eatyourpeas]

Are we sure about this (meaning I think that I am not sure)? I have never used Marimo but from what I have read you can read jupyter inside marimo but cannot read marimo inside jupyter? Surely supporting jupyter is the better option since people who prefer Marimo can still use this?

[pacharanero]

I'm still exploring Marimo, but I do prefer the way they work. Marimo works much better with Git. Jupyter makes changes to .ipynb data files every time the notebook is used, resulting in spam commits.

I'd ask you to read this https://marimo.io/features/vs-jupyter-alternative

[eatyourpeas]

I can see lots of benefits. But Jupyter is still the standard. It feels like dropping support for it would be a shame. Can we support both?

[mbarton]

I'd like us to have one set of examples and avoid duplication. Marimo git support is nice and their support for dynamic examples with UI controls is very useful. But I'm not sure we have enough of a reason to switch right now.

Either way I don't think to a consumer it matters hugely which notebook platform we use as they'll just be copy pasting things out into the one they are using themselves.