I'm new to PosgtreSQL, a veteran with Informix. We are considering a migration. In Informix, it is recommended to create explicit named indexes on columns for primary and foreign keys prior to creating the constraints. Otherwise, the server create the indexes for you with meaningless names. This is not generally a problem, except when you dump the schema, you get all the constraint indexes in the DDL, exported as if they were explicitly created, but with the server generated names. It's a mess to sort through. What's the recommended procedure in PG? At first glance it appears that PG hides the implicit indexes from you at all times, including pg_dump. So it appears that explicit index creation can be skipped without leaving you with a mess later. Is this just a non-issue in PG? Jeff ---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings