Re: Group by of 400 million rows to deal with dupes is it best way?

On Mon, 7 May 2001, jonathan_at_jlcomp.demon.co.uk wrote:

> One idea:
> Create a unique or pk constraint on the target
> columns - make the constraint deferrable,
> so that it is built with a non-unique index
> (8.1 feature), and create it disabled, non-
> validated.

Now that's excellent. I kept reading this guys question trying to figure a way to get at just the duplicated rows without having to do a count(*) ... group by ... having. This seems like great way.

Never tried it, but have stored it away to try in near future.

-- 
Galen
I don't want to be the rock.  Yeah, okay, what do you want to be?
I want to be the piece of glass.