At 04:08 PM 10/03/2004, Tom Lane wrote:
>None at the moment, but I'm open to suggestions. It seems like we might
>need different stats for equality probes than range probes.
What about my suggestion from August 2000:
"There might be a way to side-step the issue here. I assume that the index nodes contain a pointer to a record
ina file, which has some kind of file position. By comparing the file positions on one leaf node, and then
averagingthe node cluster values, you might be able to get a pretty good idea of the *real* clustering."
I don't use the CLUSTER command, but I have clustered data and would like
to be able to take advantage of the fact if possible. *If* the record
pointers can be used to indicate closeness, then the same approach of
randomly sampling index nodes would seem to work. Then again, maybe I don't
know enough about the storage techniques...
----------------------------------------------------------------
Philip Warner | __---_____
Albatross Consulting Pty. Ltd. |----/ - \
(A.B.N. 75 008 659 498) | /(@) ______---_
Tel: (+61) 0500 83 82 81 | _________ \
Fax: (+61) 03 5330 3172 | ___________ |
Http://www.rhyme.com.au | / \| | --________--
PGP key available upon request, | /
and from pgp.mit.edu:11371 |/