Notes on Notebooks: Is Jupyter the Bringer of Jollity?
summary
Jupyter is an interactive notebook ecosystem with a web client, server, language kernel and IPYNB JSON file store; GitHub hosts about 10 million notebooks and a Glasgow MOOC shows novices and scientists adopt them. Concrete problems: cluttered UI, nonsequential execution that breaks top-to-bottom assumptions, saved outputs without kernel state, no easy way to import notebooks as libraries, poor JSON diffs for versioning, and lack of in-notebook introspection — prototypes exist but robust toolchains are still missing.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:54Evolution of Developer Tooling
- 2:14notebooks are popular
- 2:38Jupyter server
- 3:35novices like notebooks
- 4:05scientists like notebooks
- 4:50origin of notebooks
- 6:51notebooks are user-friendly
- 8:40Channelling Zeitgeist
- 9:36Notebook structure linear sequence of cells
- 11:03Well-rehearsed problems with notebooks
- 11:13UI is overwhelming
- 12:20unintuitive evaluation orders are possible
- 13:01Partial persistence is confusing
- 14:00More interesting challenges for notebooks
- 14:12More interesting challenges no modularity
- 16:14Conclusion: Notebooks are here to stay
related talks
33mSignificant Roadblocks to Usefulness for Jupyter Notebooks and a Recipe to Fix them
01h 00mInteractive Computing with Jupyter: Past, Present, and Future
25mOvercoming 5 Hurdles to Using Jupyter Notebooks for Data Science, by the JetBrai...
24mJupyter Notebooks for Ops
49mBest tools for Jupyter notebooks
25mThe First Notebook War
28mUnlock the Full Potential of Jupyter Notebooks
21mThe Story in the Notebook: Exploratory Data Science using a Literate Programming Tool