Simon Riggs wrote: > > The real question is why anyone would actually perform that kind of > UPDATE. It doesn't really make much sense to increment a PK value. > > PostgreSQL is good at supporting things people want and need, so > differences do exist in places that are fairly low priority. > I agree it makes less sense to modify PK that way and that's not what we were doing. The case we went through is that we have a unique index on a table that contains a date field. While we rolled the dates forward it happens to "collide" with the existing data in the transient state and failed the update. I don't think this is that weird. There are different ways to get around the way PostgreSQL behaves, just a little surprise about that since that doesn't seem right from a purist's point of view. -- View this message in context: http://postgresql.1045698.n5.nabble.com/Another-unexpected-behaviour-tp4610242p4616541.html Sent from the PostgreSQL - general mailing list archive at Nabble.com. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
