Re: [HACKERS] Refactor handling of database attributes between pg_dump and pg_dumpall

Bruce Momjian <bruce@momjian.us> writes:
> Oh, I see what you mean.  I was just worried that some code might expect
> template1 to always have an oid of 1, but we can just call that code
> broken.

Ever since we invented template0, it's been possible to drop and recreate
template1, so I'd say any expectation that it must have OID 1 has been
wrong for a long time.  This change will just make the situation more
common.

            regards, tom lane