Stephan Szabo wrote:
> > If don't understand. We already have a unique index on the SERIAL
> > column, so why bother rejecting an insert/update that supplies the
> > value? We need the column to be unique, and that is forced, but why
> > prevent _any_ unique value from being used.
>
> One reason is that the sequence won't respect those inserted values and
> you'll get uniqueness errors on statements that don't give a value for the
> column where you'd expect to be getting a working autogenerated value.
Oh, it causes problems later on. Interesting. However, preventing
INSERT/UPDATE seems quite extreme.
-- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610)
359-1001+ If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square,
Pennsylvania19073