Simon Riggs wrote:
> I've been looking at ways of doing INSERT then UPDATE, but it doesn't
> seem very easy to avoid unique index violations in that case. So doing
> the UPDATE first then INSERTs later seems like the way to go.
INSERT has to be first to avoid a race condition (see my previous
emails). We just need a way to get a unique index violation and
continue with the UPDATE. We have savepoints now so it certainly seems
possible.
-- 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