Collaborative editing in Jupyter
summary
Shows that Jupyter now uses CRDTs (yjs in practice) to make notebooks locally editable and automatically converge, with both centralized Jupyter Collaboration over WebSockets and a peer-to-peer mode via Jupyter Share Drive / WebRTC in JupyterLite. Demonstrations include a suggestion system with merge/discard, a chat tied to an LLM agent, a document timeline/undo manager, and a server-side execution model (jerse) that preserves outputs after disconnects; widgets are being reworked toward collaborative, restorable state.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:52CRDTs in Jupyter
- 4:19Centralized vs distributed
- 7:12Local-first with CRDTs
- 10:31CRDT features in Jupyter
- 13:02Suggesting system demo
- 16:57Cross-platform demo
- 19:35Side execution in Jupyter
- 23:10Collaborative widgets
- 24:55Future of CRDTs
- 26:02Q&A
related talks
29mJupyterLab+Real Time Collaboration
28mCollaborative GIS editing in JupyterLab
37mBuilding a Collaborative Text Editor
40mNew algorithms for collaborative text editing
31mBrowser-based AI workflows in Jupyter
13mHow to make a real-time collaborative text editor in 5 easy steps!
29mCRDT: Conflict-free Data Types for Collaborative Editing and Offline-First
28mReal-time collaboration with Rails