On Tue, May 10, 2011 at 10:29 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> That will be true only if you intentionally ignore the points Greg
> raised. If the table isn't entirely ALL_VISIBLE, then the choice of
> index will determine the ordering of the actual table probes that occur.
> There could be more or fewer page reads, in a more or less optimal
> order, depending on the index used.
However, note that this wasn't one of the cases I said I was going to
try to optimize in the first go-around anyway.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
