Dear Paimon Community, 

I hope this message finds you well. I'm writing to share a project I've been 
working on and to gauge community interest in it. 

I've developed a DuckDB extension for Apache Paimon that enables direct 
querying of Paimon tables using DuckDB. This extension allows users to quickly 
explore and analyze Paimon data locally without the need to set up and 
configure larger distributed engines like Trino or StarRocks. This makes it 
particularly useful for: 
- Quick data exploration and ad-hoc analysis 
- Prototyping and development workflows 
- Local testing and validation 
- Lowering the barrier to entry for Paimon adoption 

**Project Details:** 
- Repository: https://github.com/luoyuxia/duckdb-extension-paimon 
- Current Status: Experimental (0.0.1-beta release available for macOS, tested 
with DuckDB 1.4.2) 
- Supported Table Types: Log tables and primary key tables with Deletion Vector 
enabled 
- Technology Stack: Built on top of paimon-rust with a Rust FFI layer and C++ 
integration 

**Technical Approach:** 
The extension is built using paimon-rust as its foundation. I've been 
maintaining a fork of paimon-rust (https://github.com/luoyuxia/paimon-rust, poc 
branch) that adds read capabilities, which serves as the core for this DuckDB 
extension. This approach leverages Rust's performance and safety 
characteristics while providing a clean integration path with DuckDB through 
FFI. 

**Future Plans:** 
If the community finds this project valuable and interesting, I would be very 
happy to: 
1. Continue refining and improving the DuckDB extension for better integration 
with DuckDB's features and ecosystem 
2. Contribute the read capabilities back to the paimon-rust project, which 
could unlock more integrations between Paimon and the broader Rust ecosystem 

I'm eager to hear your thoughts and would welcome any feedback, suggestions, or 
collaboration opportunities. 

Thank you for your time and consideration. I look forward to hearing from the 
community. 

Best regards, 
Yuxia 

Reply via email to