Adam Martin wrote:From my personal experience, the error logging clause is usually less performatic than other approaches like using FORALL with save exceptions. I don't have the real data to try on, but I did make an experience using my test data above and the error logging clause was much more slower than my original approach. But of course, my test data is setup to trigger more errors than success, so it puts a very high overhead on error logging. I suppose that error logging is implemented with an autonomous transaction, so the difference in performance is probably consequence of the excess of commits in the log table.One simple way is define the primary key on the table and create an error logging table for the primary key column.Yes, but the issue was not one of fuss or muss, but of performance. Will this method outperform the others?
No UNIONs, no DISTINCT, no fuss, no muss.