Hardwood, a new open-source JVM library for reading Apache Parquet files, has reached version 1.0 just five months after inception. Created by Gunnar Morling, it offers a faster alternative to the official Apache Parquet Java implementation by using multi-threaded page decoding across all available CPU cores and maintaining a zero-mandatory-dependency profile. It provides two APIs — a row reader for general-purpose access and a column reader for high-throughput analytical workloads — and benchmarks show 16.5 million rows per second on 8 vCPUs. The library also includes a CLI with a TUI for inspecting Parquet file schemas. Write support is planned for future releases.
3 Impressions