Parquet reader refactor: move to a Streamly-based streaming pipeline#188

Open

sharmrj wants to merge 29 commits intoDataHaskell:mainfrom

sharmrj:streaming-parquet

Commits on Mar 20, 2026

Commits on Apr 4, 2026

Fixed an issue where the parquet parser was using ~2x the amount of memory it should have been
sharmrj
committed
Changed Parquet Zstd decompression to no longer stream
sharmrj
committed
Use `FileBufferedOrSeekable` for the `RandomAccess` instance for `LocalFile`
sharmrj
committed

Commits on Apr 19, 2026