On Fri, 2002-02-15 at 19:09, Tom Lane wrote:
> "Gordon A. Runkle" <gar@integrated-dynamics.com> writes:
> > [ tale of poor statistical results in a near-unique column ]
> [ new algorithm and patch ]
Hi Tom,
This works much better (I did SET STATISTICS back to 10). Not always
'-1', but almost always negative.
Is "-0.503824" the same as "503824 with a predicted increase in the
number of distinct values" (as opposed to using "-503824")?
I checked a couple of other tables for which my field is a bit less
nearly-unique, and they have better stats now also. I've attached a
file with distributions and stats, if you're interested.
And, getting to where the rubber meets the road, my queries are running
well again!
I'm running a batch of other queries (which were unaffected by this
problem) overnight to see if they still work well.
Are you planning to include this patch in v7.2.1, or would it require
too much testing by others?
Thanks for all your hard work, it is greatly appreciated,
Gordon.
--
"Far and away the best prize that life has to offer
is the chance to work hard at work worth doing."
-- Theodore Roosevelt
