when I insert/update many rows at once using INSERT ... SELECT into a
table which has plenty of CHECK constraints, the error message that
Postgres returns has no indication of which row failed the constraint
check. The attached patch tries to provide information in a similar way
to how duplicate items in a UNIQUE constraint are handled.
Originally, I tried to simply check the new row's t_ctid, but it was
always (0,0) -- I guess that's expected, maybe it's still in memory at
that time and maybe such nodes don't have a ctid assigned yet.
Please let me know if this patch is suitable for inclusion. It's based
on REL9_0_STABLE, because that's the version I'm running.
I'd like to thank intgr on IRC for his feedback when I was wondering
about the t_ctid.
With kind regards,
Trojita, a fast e-mail client -- http://trojita.flaska.net/
pgsql-hackers by date
|Next:||From: Peter Eisentraut||Date: 2011-11-07 05:13:05|
|Subject: -Wcast-qual cleanup, part 1|
|Previous:||From: Mark Kirkwood||Date: 2011-11-07 04:55:48|
|Subject: Re: Measuring relation free space|