The only ways to prevent a race-condition on insert are either to obtain an exclusive lock on the table or by the creative use of a unique constraint.
Anyway, as I mentioned elsewhere in this thread, there *is* a way to deal with this issue in oracle, and it is the "returning ... into" clause (coupled with bidirectional binding). That is, to my knowledge, the best (and nearly the only) way of handling this.
If you *absolutely* needed to have a different way of doing it (not that I can think of why you would want another way... but hey, TIMTOWTDI, I guess), it would be possible to make use of Oracle's dbms_output channel, and have your on-insert trigger do a dbms_output.put(:new.id), and read from the dbms_output channel in your perl code. I speak from experience, though, and this method sucks (for some reason I searched and searched and didn't find the "returning id into" construct until after I had implemented it via dbms_output, and that was a mess. I was glad to go back and replace it with the right way).
------------ :Wq Not an editor command: Wq
|Replies are listed 'Best First'.|
Re^4: Sequences, last_insert_id, SQLite, and Oracle
by jplindstrom (Monsignor) on Jun 15, 2005 at 09:35 UTC
by etcshadow (Priest) on Jun 15, 2005 at 19:45 UTC