> However my application code is assuming that duplicate rows will
> simply be ignored (this is the case in Ingres, and I believe Oracle's
> bulkloader too). I propose modifying _bt_check_unique() in
> /backend/access/nbtree/nbtinsert.c to emit a NOTICE (rather than
> ERROR) elog() and return NULL (or appropriate) to the calling function
> if a duplicate key is detected and a 'COPY FROM' is in progress (add
> new parameter to flag this).
If you have a UNIQUE index on the table, just throwing away duplicates
seems really bad to me. I know Ingres had that heapsort structure that
would remove duplicates. That may be an interesting feature to add as
an operation that can be performed.
-- Bruce Momjian | http://candle.pha.pa.us pgman@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