0.15.0 - Paradedb — ContextQMD

<Note> `0.15.0` contains breaking changes. After upgrading, all BM25 indexes must be recreated with `REINDEX`. </Note>

New Features 🎉

Match Query

Introducing a new query builder function: paradedb.match. Similar to Elasticsearch's match query, paradedb.match is meant to be the standard query function for full text search, allowing for custom tokenizers and fuzzy matching. It supersedes the paradedb.fuzzy_phrase query.

JSON Term Set

paradedb.term_set now supports JSON fields.

Performance Improvements 🚀

Modified the block storage layout for faster reads. This breaking change necessitates existing indexes to be recreated.
Significantly improved VACUUM times when the table contains updated or deleted rows - VACUUMing one million updated rows went from 20 minutes to 0.8 seconds.

Stability Improvements 💪

Fixed a bug where datetime values with sub-second precision were truncated to second precision.
Fixed an issue where VACUUMs could not receive SIGINTs.
Fixed concurrency issues around concurrent UPDATE, VACUUM, and parallel custom scans.

New Contributors 👋

AdamGS made their first contribution in 2146.

Full Changelog

The full changelog is available on the GitHub Release.