Analytics

DuckDB does the querying and computing; Parquet does the packing and carrying. Put them together, then compile DuckDB to Wasm, and web, PWA, Tauri, and Electron apps can analyze chunky datasets locally. This post makes the division of labor concrete with a three-engine benchmark.

DuckDB 负责查和算，Parquet 负责装和带走。把两者放在一起，再编译成 DuckDB-Wasm，就能让网页、PWA 和 Tauri/Electron 应用在本地分析大块数据。本文用一个三引擎跑分 demo，把这套分工讲清楚。