SiPy Kernel for Jupyter - mauriceling/mauriceling.github.io GitHub Wiki

Citation: Ambel, W, Ling, MHT. 2026. SiPy Kernel for Jupyter. Medical - Clinical - Research 2(3):59-65.

Link to [PDF].

Here is the permanent [PDF] link to my archive.

Jupyter Notebook and JupyterLab have become widely adopted platforms for interactive data analysis, teaching, and reproducible research. While many polyglot kernels enable multi-language execution through in-notebook language switching and in-memory data exchange; such approaches often blur language boundaries, resulting in increased coupling to kernel-specific abstractions, and complicate reproducibility outside the notebook environment. We present the SiPy kernel, a lightweight Jupyter kernel designed not as a multi-language interpreter but as a coordination interface for script-based scientific workflows. SiPy treats R, Julia, Python, and shell scripts as first-class external artefacts, executing them as standalone scripts rather than embedded notebook fragments. This design preserves explicit execution boundaries, enhances portability, and ensures that analytical components remain runnable outside Jupyter. The kernel integrates the SiPy interactive shell into Jupyter while maintaining minimal dependence on internal notebook APIs and avoiding interpreter embedding. Features include controlled execution with configurable timeouts, structured output capture, and support for HTML rendering within cells. By positioning Jupyter as a narrative and documentation layer rather than a computational substrate, the SiPy kernel offers an alternative architectural model for polyglot scientific computing that prioritizes transparency, reproducibility, and long-term maintainability over convenience-driven abstraction.