Stability Improvements 💪

  • Fixed an issue where our custom scan pushdown code would discard some predicates if the query also contained a subselect
  • Fixed an issue where having too many segments would consume too much memory
  • In ParadeDB Enterprise, fixed an error that could cause a parallel CREATE INDEX to terminate early

New Features 🎉

  • The inet type can now be BM25 indexed

Performance Improvements 🚀

  • Optimized query performance for scenarios where there are many (i.e. several hundred plus) segments in an index
  • Operations over scalar arrays like IN are now pushed down to the BM25 index, if the field is fast

The full changelog is available here.