On Fri, Jul 06, 2007 at 02:42:33PM -0400, Christopher Browne wrote:
> That would be a case where there could be some "win" from joining
> deletes together at least somewhat...  So I think this feature is one
> that can't be so quickly ruled out...

I also wonder, though, whether it mightn't be exposed by EXECUTE or
something instead, presumably with a lowish lock level.  I'm just
worried about the potential for this to turn into a big expensive
operation for ordinary cases because we've optimised the admittedly
awful mass-delete behaviour (made even worse because PostgreSQL
itself is sort of brutal with mass deletes, due to all the dead
rows).  Something that was special that way would also reduce the
exposure from hairy, complicated bits in the main line code, which as
others have already noted is somewhat mysterious to many developers.

A

-- 
Andrew Sullivan  | ajs at crankycanuck.ca
The fact that technology doesn't work is no bar to success in the marketplace.
		--Philip Greenspun