Databases/PostgreSQL: pg_mentat 1.3.0, pg_infer 1.0.0, pg_tre 1.1.1, Textpattern issues with MariaDB 11.8

posted by Roy Schestowitz on May 23, 2026

• PostgreSQL ☛ pg_mentat 1.3.0 released -- Datomic-compatible Datalog inside PostgreSQL

  I am pleased to announce the first public release of [pg_mentat] (https://github.com/gburd/pg_mentat), a PostgreSQL extension that implements Datomic's data model -- immutable facts (datoms), schema-first attributes, a full Datalog query compiler, the pull Hey Hi (AI) time travel, and Hey Hi (AI) transactions -- entirely inside PostgreSQL.

  pg_mentat is built with pgrx 0.17 in Rust and supports PostgreSQL 13 through 18. The current release is 1.3.0, the "Postgres Extension Family" release, which adds Datalog where-fns that bridge into rum, pg_trgm, fuzzystrmatch, pgvector, pg_infer, PostGIS, and several other extensions as soft dependencies (nothing pg_mentat ships requires any of them, but where-fns light up automatically when they are present).

  An optional companion daemon, mentatd, speaks the Datomic client wire protocol (EDN, Transit+JSON, Transit+MsgPack) over HTTP for applications that already expect it.

• PostgreSQL ☛ pg_infer 1.0.0 released -- transformer model knowledge as SQL relations

  I am pleased to announce the first public release of pg_infer, a PostgreSQL 18+ extension that exposes the internals of small transformer language models -- gate activations, feature labels, learned associations, embeddings -- as SQL-queryable relations and a custom index access method.

  pg_infer is not "natural language to SQL." It is not "SQL to natural language." There is no chat interface, no agent loop, no prompt template generating queries. pg_infer brings model inference into the query plan as an operator the planner can cost, schedule, parallelize, and combine with ordinary predicates and joins. The model becomes a first-class data source -- a set of relations the planner can scan, filter, and join -- not an external service the database talks to.

• PostgreSQL ☛ pg_tre 1.1.1 released -- an approximate-REGEX index AM for PostgreSQL 18+

  I am pleased to announce the first public release of [pg_tre] (https://codeberg.org/gregburd/pg_tre), a native PostgreSQL 18+ index access method for approximate-regex matching.

  pg_tre indexes text columns through a three-tier filter funnel (BRIN-style range bloom -> sparsemap trigram postings -> per-tuple bloom) backed by Ville Laurikari's TRE library for the heap recheck. The result is genuine Levenshtein-distance regex matching ("find text within k edits of this pattern") driven through a real IndexAmRoutine, with WAL coverage, VACUUM awareness, and REINDEX CONCURRENTLY support.

• Ruben Schade ☛ Textpattern issues with MariaDB 11.8

  I recently updated a MariaDB server to 11.8 LTS, their latest long term support release. Immediately I started seeing issues on Textpattern 4.9.1 sites I sysadmin for people. Creating new articles works, but editing existing articles results in the following error shown across the bottom of the screen: [...]