RE: Different execution plans for same query with same cost

We ran into bugs in 9.2.0.3 where FORCE would cause certain queries to reparse every time they executed...that may explain why it's reparsing the query. But it still doesn't explain why the plan would change.

By the way- a point of interest- we used to use CURSOR_SHARING=3DSIMILAR with histograms on ALL INDEXED columns. I see you're doing the same with histograms. If you have CURSOR_SHARING=3DSIMILAR and a histogram'd column is in the where clause Oracle will reparse a statement EVERY time it executes, regardless of the value used...this is a "feature".

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of jo_holvoet_at_amis.com Sent: Thursday, March 03, 2005 11:39 AM
To: oracle-l_at_freelists.org
Subject: RE: Different execution plans for same query with same cost

Actually CURSOR_SHARING=3DFORCE. Long story, crappy third-party app (no=20 binds, very little normalization, ...).

mvg/regards

Jo

--
http://www.freelists.org/webpage/oracle-l