Re: Optimzer Oddities

Thanks for the response.... you are correct - I just wanted a worded explanation of the mechanics behind the join method....

I can interpret the statement much easier now, thanks for your time..

Matt

"Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote in message news:<c6rls6$i89$1_at_hercules.btinternet.com>...
> If you want someone to predict what the
> execution path will be, then you need to
> supply a lot more information. If you just
> want a clue about how the statement could
> be "legal" or "manageable" -
>
> Join A to B on emplid, recording the values
> for a.name, b.location, a.effdt, a.emplid
> b.start_time, a.sal where a.sal > 1000
>
> For each row in the join result, query C
> for:
> max(effdt)
> where c.emplid = {constant 1}
> and c.start_time > {constant 2}
>
> except that constant 1 is the a.emplid
> from the join row and constant 2 is
> b.startime from the join.
>
> If the result of the subquery matches the
> a.effdt from the join row, report the join
> row.
>
>
> Of course, Oracle may rewrite the query to look
> completely different, and the mechanics may
> therefore be nothing like my description - but
> logically, what I've described is what has to happen.
>
>
> --
> Regards
>
> Jonathan Lewis
> http://www.jlcomp.demon.co.uk
Received on Fri Apr 30 2004 - 02:33:22 CDT