Project Mentat: a store for evolving data in Rust
summary
Project Mentat presents a concrete implementation: a Rust-embedded store that layers a transaction log and strong schema on top of SQLite to get persistence, full-text search, and cross-platform portability. It targets encrypted sync with client-side conflict resolution, provides materialized views for fast queries, and exposes FFI-backed SDKs for Java, Swift (and soon Kotlin). Current work focuses on merging algorithms, deletion/forgetting from the append-only log, and production integration.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 0:48Mozilla User Data Team
- 1:57Problems with storing data
- 4:58Sharing good data
- 8:36CQRS
- 14:44Under the hood
- 16:17Demo
- 18:50Stable datum
- 19:47Attributes
- 20:22Query
- 20:46Compile
- 21:33Results
- 25:53Android component
related talks
19m"Embracing Monorepo and LLM Evolution"
20mRust as foundation in a polyglot development environment
35mPersisting Relations Across Time and Space
01h 00mMarten: Postgresql backed Document Db and Event Store for .NET Development
22mGraphite-Rust: A Fast Time Series Database Implementation
38mSimpler state management with Dart on the server and synced databases
47mMarten as a Document Store for .NET Applications
48mStrengthening the Base: Laying the Groundwork for a more robust Rust SDK