Hardwood 1.0.0.CR1, the first release candidate for the Java Apache Parquet library, is now available. Key highlights include a reworked ColumnReader API that introduces a Validity type (inspired by Apache Arrow) for handling nullability in flat and nested/repeatable columns, and initial geospatial support for Parquet's GEOMETRY and GEOGRAPHY column types with predicate push-down. The release also adds FLOAT16 support, improved RowReader accessors, more efficient multi-column filter evaluation, split-aware reading for distributed engines like Apache Flink, and a documentation overhaul following the Diátaxis framework. 50 issues were resolved in this release, with Hardwood 1.0 Final expected within a week.
526 Impressions