DuckLake v1.0: The Lakehouse Format Built on SQL Reaches Production-Readiness

DuckLake v1.0 is now production-ready — a lakehouse format that stores all metadata in a SQL database (SQLite, PostgreSQL, or DuckDB) rather than scattered files in object storage. Key new features include data inlining for small DML operations (avoiding the small file problem), sorted tables for query performance, bucket partitioning for high-cardinality columns, GEOMETRY and VARIANT type support, and experimental deletion vectors compatible with Iceberg v3. The release ships 108 merged PRs with 68 focused on reliability and correctness. The ducklake DuckDB extension is available in DuckDB v1.5.2. Community adoption includes clients for Apache DataFusion, Apache Spark, Trino, and Pandas, plus a hosted service from MotherDuck. Future plans include git-like branching, permission-based roles, and incremental materialized views.