The application probably still treats id as unique, but nothing in the schema guarantees it. And you can’t recover the guarantee with a separate UNIQUE (id) con

The application probably still treats id as unique, but nothing in the schema guarantees it. And you can’t recover the guarantee with a separate UNIQUE (id) constraint: both MySQL and PostgreSQL require every unique constraint on a partitioned table to include the partition key columns. The uniqueness property has effectively been traded away. Not really? MySQL has AUTO_INCREMENT [1] PostgreSQL has SERIAL [2] and CREATE SEQUENCE [3] What am I missing? [1] https://dev.mysql.com/doc/refman/8.4/en/example-auto-increment.html https://dev.mysql.com/doc/refman/8.4/en/example-auto-increme... [2] https://www.postgresql.org/docs/18/datatype-numeric.html#DATATYPE-SERIAL https://www.postgresql.org/docs/18/datatype-numeric.html#DAT... [3] https://www.postgresql.org/docs/18/sql-createsequence.html https://www.postgresql.org/docs/18/sql-createsequence.html