Pijul: Version-Control Post-Git
summary
Pijul claims a patch‑first VCS with associative and commutative merges that avoid Git’s silent reshuffles by representing files as directed byte graphs and using category‑theory pushouts to detect conflicts. It stores history in a transactional Rust key‑value engine Santa Claria with B trees, copy‑on‑write forks, crash safety and benchmarks that beat lmdb on single‑core workloads. A wasm/cloudflare workers hosting prototype is live at nest.pijul.org enabling partial clones and large file deltas.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 2:37Why VersionControl
- 5:28Requirements
- 6:34Associative merges
- 9:39Why would we want this
- 11:27Change perspective
- 12:45Operational transforms
- 15:10CRDTs
- 16:46Conflicts
- 18:10Category Theory
- 19:50Minimal Extensions
- 23:15Conflict
- 24:52Conclusions
- 28:21Challenges
- 30:07SantaClaria
- 32:21BTree
- 37:04Key Value Stores
- 40:00Pijul Hosting
- 46:26Conclusion
related talks
22mA Bird's-Eye View of Version Control with Pijul
29mBridging the worlds: pixi reimplements pip and conda in Rust
18mExtending Git without breaking it
52mWill Git Be Around Forever? A List of Possible Successors
53mThe Power of Modern Version Control
25mKnowledge Transfer by Source Code
22mShifting Left, Shifting Right
27mRevision Control for Structured Data
