Re: ** statistics for zero row table - Yes or No?
From: Kerry Osborne <kerry.osborne_at_enkitec.com>
Date: Sun, 23 Aug 2009 20:11:02 -0500
Message-Id: <67180759-ACD8-4923-AE22-991F564E1117_at_enkitec.com>
Well put Ric. Of course on the other hand, why would you ever join to a table with no rows? Unless the zero row table is sometimes not a zero row table. Like if it's being used as a temp table that sometimes has rows and sometimes doesn't. In which case, dynamic sampling would actually be better than stats that are sometimes dead wrong.
Date: Sun, 23 Aug 2009 20:11:02 -0500
Message-Id: <67180759-ACD8-4923-AE22-991F564E1117_at_enkitec.com>
Well put Ric. Of course on the other hand, why would you ever join to a table with no rows? Unless the zero row table is sometimes not a zero row table. Like if it's being used as a temp table that sometimes has rows and sometimes doesn't. In which case, dynamic sampling would actually be better than stats that are sometimes dead wrong.
Kerry Osborne
Enkitec
blog: kerryosborne.oracle-guy.com
-- http://www.freelists.org/webpage/oracle-lReceived on Sun Aug 23 2009 - 20:11:02 CDT