New Features 🎉
- Add support for indexing the Postgres
ltreecolumn type.
Stability Improvements 💪
- Significantly improved test coverage via concurrent property testing.
pg_searchwill now write WAL entries for its buffers, allowing for crash recovery.- Many fixes for planning of joins to allow the custom scan to trigger in more cases.
- Significant test coverage improvements for joins.
- Fix for
ORDER BYon duplicated column names atop joins. - Fix for race condition in detoasting during VACUUM.
- Added an error when mixing incompatible proximity operators.
- Remove minimum CPU requirements, as they did not show a performance benefit.
Performance Improvements 🚀
- Improved the performance of large term set queries, and some types of hash join.
- Introduce a first version of a MPP plan partitioning mode, to improve parallelism and remove redundant work during joins.
- Improved performance of Top-K queries on
pdb.score.
New Contributors 👋
- Daniel Oliveira
- Joshua Lockerman
- Jun Jie
- Nihal Pandey
- Ritik Kumar
- S Bala Vignesh
- Saad Tajwar
- Yerkebulan Tulibergenov