Re: New 8.1.6 bug of the day

Well I've done the relational algebra on this one and I guess this is going to force me to pospone an imminent migration to 8.1.5.

Not a trivial thing at all.

At least the datawarehouse instance has got backup code and SQL that allows it to run on a 8.0.5 instance.
(i.e. isn't solely dependant on add-ons that come with 8i).

No small inconvenience.

Never mind user-generated SQL, what about all the recursive SQL going on in the background!

Hmmmmmmm.

Proposed temporary workaround (basically coding conservatively, forcing operator precedence using brackets).

Anyone able to test on an 8.1.5 or 8.1.6 box that was giving the incorrect results before?

SELECT TABLE1.F1, TABLE1.F2, TABLE2.F1, TABLE2.F2, TABLE2.F3 FROM TABLE3, TABLE2, TABLE1
WHERE (TABLE1.F1='test1'

        AND TABLE1.F1=TABLE2.F3)
        AND TABLE1.F3=TABLE3.F1;

or

SELECT TABLE1.F1, TABLE1.F2, TABLE2.F1, TABLE2.F2, TABLE2.F3 FROM TABLE3, TABLE2, TABLE1
WHERE TABLE1.F1='test1'

        AND (TABLE1.F1=TABLE2.F3
        AND TABLE1.F3=TABLE3.F1);

David P.

Oracle Certified DBA.

(who's just about to phone Oracle Silver support)