Bo Lorentsen <bl@netgroup.dk> writes:
> I use normal tabel ID (SERIAL and BIGSERIAL) all over the place for FK
> constaints, but I use OID in one special situation. When I insert a
> single row into a table, I like my low level code to be kompatible with
> mysql ( mysql_insert_id ), and fetch the row that I just inserted. This
> I do by using the PGoidValue function, and then select the row by the
> oid. This works quite nice .... but when a table get large, it become a
> big search (seq scan) so I have added an index on oid's on the table
> where I use this trick, and this have helper :-)
The thing you have to worry about is the possibility of duplicate OIDs
once your DB has been running long enough for the OID counter to wrap
around (2^32 OIDs). You should make sure that index is specifically
declared as UNIQUE, so that any attempt to insert a duplicate OID will
fail. That might be enough for you, or you might want to add logic to
your application to retry automatically after such a failure.
regards, tom lane
