Re: [SQL] Bug: Sequence generator insert

2003-11-30 Thread Iain
I just re-read your message, and it occurred to me that for the second insert you may have provided an explicit value for the 'id', is that right? You may need to give more detail about the inserts. I'd be mildly surprised, but not concerned if the insert that failed used an explicit value for the

Re: [SQL] Bug: Sequence generator insert

2003-11-30 Thread Iain
Hi Colin, You have done a nice job of demonstrating the documented behaviour of sequences. Sequences are designed for speed and high concurrency. They do not guarantee contiguous numbers and are not included in any transaction as I understand it. regards Iain - Original Message - From:

Re: [SQL] Bug: Sequence generator insert

2003-11-28 Thread Bruno Wolff III
On Tue, Nov 25, 2003 at 18:37:41 -0500, "Burr, Colin" <[EMAIL PROTECTED]> wrote: > > However, even though the new record failed to be inserted into the table, > the sequence generator was still updated. > > The sequence generator should only be updated if the record is successfully > inserted i