Philip Warner <pjw@rhyme.com.au> writes:
> At 12:37 PM 20/08/2004, Bruce Momjian wrote:
>> But that doesn't fix ascii dumps loaded via psql.
> It does; the ascii dump file is generated by exactly the same technique as
> pg_restore.
Right. Philip's suggestion would essentially use the same technique
that we previously adopted for portability of WITH/WITHOUT OIDS --- if
the "SET" fails, it won't stop the table from being created. (Note we
have to be careful that the semantics of the SET actually cause the
error to occur on the SET and not later on the CREATE. But that's
doable.)
It seemed like a reasonable idea to me...
regards, tom lane