Drops of Jupyter: An Exploration of the Notebook Ecosystem
summary
Jupyter remains useful but fragmented: Google Colab provides a ready execution environment while notebooks stay in Git repos; reviewnb and nbdime make PRs readable; nbdev converts notebooks into literal codebases; Marimo stores source as Python functions and builds a dependency graph to avoid execution-order bugs and runs as scripts; Molab provides a Colab-like runner for Marimo; Quarto’s QMD is markdown with inline code and QMD runs code top-to-bottom during rendering enabling testable, publishable docs.
AI-written summary · about
chapters (22)
- 0:00Jupyter notebooks intro
- 1:03Jupyter notebooks flaws
- 3:33No perfect notebook format
- 4:56Jupyter origins and uses
- 6:22Machine learning and Jupyter’s rise
- 7:26TensorFlow and notebooks
- 8:31Docs as code & literate programming
- 9:32Notebook evolution pressures
- 10:12Git collaboration in notebooks
- 12:44TensorFlow notebook docs
- 14:28Notebooks for software engineering
- 15:53Marimo Jupyter alternative
- 18:42IPython Flow reactive notebooks
- 19:02Notebooks for docs and publishing
- 20:07Quarto docs publishing system
- 21:54Quarto documentation focus
- 22:14Takeaways and frustrations
- 22:58Closing remarks & Q&A
- 23:39Q&A beginner notebook use cases
- 24:42Q&A shared execution environments
- 25:47Q&A interactive Quarto files
- 27:11Q&A adding Quarto to docs
related talks
28mUnlock the Full Potential of Jupyter Notebooks
38mJumping into Jupyter Notebooks
01h 23mImproving Your Data Modeling Work Through Open-Source Software
32mSpin up Jupyter notebooks at scale and enhance your productivity
01h 18mProductionizing Jupyter Notebooks
26mTransforming a Jupyter Notebook into a reproducible pipeline for ML experiments
09mAccelerating ML Prototyping The Pythonic Way
58mNotebooks & .NET: A Marriage Made In Jupyter