CRDT Research Project

A comprehensive research project comparing Conflict-free Replicated Data Type (CRDT) implementations for collaborative applications. This project evaluates YJS and Automerge through practical implementations in both TypeScript and Rust.

Project Structure

├── automerge-crdt-rs/     # Rust implementation using Automerge
├── automerge-crdt-ts/     # TypeScript implementation using Automerge
├── y-crdt-rs/             # Rust implementation using YJS/y-crdt
├── yjs-crdt-ts/           # TypeScript implementation using YJS
├── comparison.md          # Detailed performance and feature comparison
└── crdt.md               # Technical analysis of CRDT approaches

Research Findings

This project provides practical benchmarks and analysis comparing YJS and Automerge across multiple dimensions:

• Performance: YJS shows ~9% faster execution (0.646s vs 0.707s)
• Memory Efficiency: YJS produces smaller binary updates (92 bytes vs 158 bytes)
• API Design: Automerge offers explicit change tracking vs YJS's event-driven approach
• Use Cases: Different strengths for real-time collaboration vs audit-heavy applications

Getting Started

TypeScript Implementations

YJS Implementation:

cd yjs-crdt-ts
npm install
npm start

Automerge Implementation:

cd automerge-crdt-ts
npm install
npm start

Rust Implementations

Y-CRDT (YJS) Implementation:

cd y-crdt-rs
cargo run

Automerge Implementation:

cd automerge-crdt-rs
cargo run

Documentation

• comparison.md: Executive summary and detailed comparison of YJS vs Automerge
• crdt.md: Technical deep-dive into CRDT approaches for decentralized systems

Key Takeaways

• Choose YJS for performance-critical real-time collaboration
• Choose Automerge for applications requiring rich change attribution and audit trails
• Both libraries provide robust conflict-free synchronization for collaborative applications

Research Context

This research was conducted to evaluate CRDT solutions for decentralized knowledge graph applications, particularly those using IPFS and blockchain technologies for data storage and consensus.