Re: Another Outer Join Question

David Pattinson wrote in message <36647E0A.E70BA8FD_at_addease.com.au>...
>Andrew,
>
>I think you will be quite safe with your outer join here.
>It doesn't make sense for a restriction "y.y_key is null" based on a column
of
>the 'right hand' table to be performed 'before' a left outer join "x.y_key
=
>y.y_key (+)", since the values of many of the columns in the result set
could
>only be determined after the join has been performed. If any SQL optimiser
were
>to use such a plan it could only be described as a bug in my opinion.
>
>Regards, David.
>

Thanks for the reassurance David. I'm going to be uncharacteristically bold and use this method. But the nagging doubt remains. Trouble is, there's quite a lot riding on in and I just don't trust Oracle. In my last big Oracle project I had so much trouble with exploding partition-eliminated views, beserk parallel query slaves and so forth that I'm beginning to suspect that Oracle is out to get me. Maybe it already has.

I still have horrible visions of raising a TAR in 6 months time only to be told "no, that's not a problem, that's expected behaviour", just as my client is oiling the guillotine. Received on Wed Dec 02 1998 - 00:59:35 CST