RE: How to analyze a odd column distribution

Hi Alfonso

>I have a table where I have a date column with historic information in
>the same table, if date =3D '31-dec-2999' then is actual record if the
>date is different of 31-dec-2999 the is an historic record. Now there
>are about 5 million records and half of them are historic so there are
>2.5 million records with 31-dec-2999 and it is confusing the
>optimizer.

Exactly for this reason such thing should not be implemented.

>Can anybody give me any suggestion to tell the optimizer this odd
>distribution? I already try using histograms and didn't get a good
>execution plan and I would not like to use hints.

How did you create the histograms? Can you give us an example of a query = that doesn't work?

Chris

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