Chuck wrote:
> peter wrote:
>
>>Nothing comes to mind right now, but as mentioned in entry#11 I do know
>>that Jonathan's book Cost Based Fundamentals does mention it in Chapter
>>7 and a few other places. Bind peeking was something introducted in 9i.
>>The performance guide does mention it.
>>
>>If you search asktom.oracle.com for "sqlplus and bind peeking" there
>>are some things on it.
>>--peter
>>
>
>
> I think the mystery is solved. It seems that the _unnest_subquery
> parameter was set to false. That's probably causing the optimizer to
> ignore the unnest hint. I'm still not sure why it's using it in the
> explain plan though. One would think that if the parameter keeps the
> optimizer from using a hint a run time, it would do the same when you
> run explain plan.
>
> Any thoughts on this?
Jonathan Lewis had some fascinating comments on the differences in
optimizer default behavior between 8i, 9i and 10g at UKOUG: Likely he
has published them elsewhere. 9i does fascinating things. And beyond
that I will leave it to him to speak on the matter if he chooses.
--
Daniel A. Morgan
http://www.psoug.org
damorgan_at_x.washington.edu
(replace x with u to respond)
Received on Tue Dec 06 2005 - 17:50:58 CST
