Doltgres 1.0 launches August 6th. Here's what has landed in the codebase over the past few weeks: replication support, automatic garbage collection, data-correctness fixes, and more.

DoltHub Blog

Doltgres 1.0 is launching August 6th, one week away. Over the past three weeks, 50 pull requests and 136 commits from 9 contributors have landed. Key features now complete include: remotes support (push/pull between Doltgres databases including PostgreSQL-specific entities like sequences and custom types), three replication modes (remote-based, cluster, and PostgreSQL replication) now production-ready, and automatic garbage collection enabled by default. On correctness, Doltgres hit 99% SQL logic test compliance and fixed a notable bind variable type inference bug that could cause silent data corruption with ORMs. Performance has improved past the original 3x Postgres multiplier goal, now targeting 2.6x to match MySQL's relative performance.

One Week Out: Doltgres 1.0 Update

1.0 Progress: Remotes, Replication, and Garbage Collection #

<p>The bind-variable fix may be as important as the headline features. Silent ORM corruption is the kind of edge case that makes teams hesitate on a new database; pairing 99% logic-test compliance with production-ready replication makes the 1.0 milestone feel much more credible.</p>


<p>The remote-based replication flow is elegant, but making a replica check and pull from the remote before every read turns freshness into a latency and availability dependency on that intermediary. For production users, metrics around pull lag, remote failures, and read behavior during an outage would be as important as the happy-path setup—especially because the three replication modes have different recovery semantics.</p>
