PostgreSQL compatibility

Connect a normal PostgreSQL client to the PostgreSQL wire listener. The default is 127.0.0.1:5433; the database name and user are accepted for driver compatibility while tenant/auth behavior is controlled by the server.

Authenticated application and tenant owner/admin sessions cannot use the startup user value to become postgres. Global schema, role, and privilege statements require a platform superuser; data statements remain subject to the authenticated tenant and SQL/RLS policy.

psql 'postgresql://postgres@127.0.0.1:5433/postgres'
PGPASSWORD="$RYME_PG_PASSWORD" psql -h db.example.com -p 5433 -U app -d app

Covered protocol behavior

• Startup and authentication, BackendKeyData, and cancel requests.
• Simple and extended query flows (Parse, Bind, Describe, Execute, Sync) with prepared portal invalidation.
• Explicit transactions, SAVEPOINT, ROLLBACK TO, RELEASE, and aborted transaction state.
• COPY FROM STDIN and COPY TO STDOUT, including nulls and escaping.
• LISTEN, UNLISTEN, and post-commit NOTIFY notifications.
• Common driver probes such as version(), current_setting, SHOW ALL, SET NAMES, SET TIME ZONE, and application_name.
• statement_timeout, SQLSTATE errors, RLS/catalog introspection, views, sequences, roles, grants, and persisted migration metadata.

Compatibility boundary

rymeDB is not PostgreSQL's executor, extension ABI, or storage format. Review unsupported or partially supported application features before migration:

Area	Guidance
Extensions	Do not assume arbitrary extensions, FDWs, PostGIS, or custom types exist.
PL/pgSQL	Common function/trigger metadata and simple timestamp triggers are supported; arbitrary procedural code needs review.
Query planner	EXPLAIN is available, but plans and indexes are not PostgreSQL plans.
Isolation	READ COMMITTED/REPEATABLE READ map to snapshot semantics; serializable validation is engine-specific.
Large COPY jobs	Chunk requests and verify constraints; HTTP COPY is capped at 10,000 rows per request.
LISTEN/NOTIFY	Notifications are ephemeral and process-scoped; use durable realtime topics for replay.

Run your ORM's generated SQL against a disposable rymeDB data directory and keep the compatibility test in CI. The repository's PostgreSQL tests are a reference, not a guarantee for every driver.

Migration helpers

Use PostgreSQL COPY for data, then apply schema through the migration ledger:

pg_dump --schema-only source_db > schema.sql
rymedb-cli migrate apply 001-schema "$(cat schema.sql)" --author migration-bot

For a full plan, see migration.