Hacker News Reader

petrizhangon Show HN: ParqDB – Vector search in the browser from Parquet over HTTP"

Thanks! SQLite with ranges is a pretty neat approach too. It’s great to see more people exploring browser-native databases.

petrizhangon Show HN: ParqDB – Vector search in the browser from Parquet over HTTP"

Not exactly. DuckDB itself does not have native vector index support; the vector search functionality comes from extensions such as VSS, which use their own index format. Those indexes are managed by DuckDB and are not designed to be shared as standalone artifacts or queried directly from a browser.

ParqDB takes a different approach: the vector index itself is stored in Parquet format. This makes the index a portable dataset that can be published, shared through object storage, and queried directly by clients (including browsers via WASM) without a database server.

More details and examples are available in our GitHub repo: https://github.com/parqdb-io/parqdb

petrizhangon A fast path for decoding fixed-length lists in Parquet

Thanks! I’ll take a closer look at parquet-rs and see if I can prototype a fast path for fixed-size, non-null lists.

petrizhangon A fast path for decoding fixed-length lists in Parquet

I ran into exactly this while profiling a Parquet-based vector index on DataFusion. The vectors all have a fixed dimension, but LIST decoding and rebuilding the Arrow arrays still showed up as a significant CPU cost. Have you looked at whether a similar fast path could be added to parquet-rs?