Performance Improvements 🚀

Incredibly Fast Top N by Field

#1820 introduced efficient handling of ORDER BY <field> LIMIT queries, where <field> is indexed as a fast field. This optimization is automatically applied by the query planner when these queries are executed.

CREATE INDEX Speedup

Introduced settings to tune the resource usage of CREATE INDEX, which can significantly improve index creation times by allowing Postgres to use all of the hardware on the host machine.

Parallel Index Scans

#1824 introduced parallel index scans, which improves certain COUNT(*) queries by 23x in local testing.

Stability Improvements 💪

Index Cleanup

0.11.0 introduced a regression where dead index files were not getting cleaned up, which caused index sizes to balloon. This has been fixed.

BM25 indexes created on 0.11.0 will need to be recreated for this fix to take effect.

Prepared Statements

Fixed a bug where prepared statements in the form of field @@@ $1::text were getting incorrectly parameterized.

Full Changelog

The full changelog is available here.