> > If we rolled back sequence numbers, we would have to lock the table
> > until the transaction commited or was rolled back. That is too much
> > locking so was not worth it.
> >
>
> That seems like a pretty big thing to sacrifice. Did sequence locking work before
> 6.5? It was my impression that it did.
No, sequence numbers of aborted transactions never were saved.
> Do you know if there is a workaround? In my particular situation, it is critical
> that all instances of the sequence actually get used. If a transaction is
> aborted, I lose an instance and everything gets messed up.
I recommend hand-rolling a sequence number, similar to how the sequence
mechanism works, but explicitly lock your sequence table, retrieve the
value, and the commit/abort will unlock the table.
-- Bruce Momjian | http://www.op.net/~candle maillist@candle.pha.pa.us | (610)
853-3000+ If your life is a hard drive, | 830 Blythe Avenue + Christ can be your backup. | Drexel Hill,
Pennsylvania19026