Thanks for your information, Nuno. We upgraded Peoplesoft databases from 9.2.0 to Oracle 10.2.0.2 about a year ago. During the testing phase, I found that I had to set the following hidden parameter to make the application running smoothly. After the 10.2.0.3 patchset was applied to those databases in , I had not rechecked if any of the following parameters need to be adjusted. *._complex_view_merging=false *._gby_hash_aggregation_enabled=FALSE *._optimizer_cost_based_transformation='off' *._unnest_subquery=false Limin Guo. On 11/4/07, Nuno Souto <dbvision@xxxxxxxxxxxx> wrote: > > Alex Gorbachev wrote,on my timestamp of 5/11/2007 4:28 AM: > > > Thanks Wolfgang. Sounds encouraging. > > > > On Nov 4, 2007 11:03 AM, Wolfgang Breitling <breitliw@xxxxxxxxxxxxx> > wrote: > > > >>I have worked with several Peoplesoft databases running the gamut > >>from HR, SA (Student Admin), Financials/Supply Chain and EPM on > >>Oracle 9i and 10g and none of them has optimizer_features_enable > >>"backdated". We set specific _ parameters though: > >>_unnest_subquery=false > >>_complex_view_merging=false > >>_gby_hash_aggregation_enabled = false ( nothing to do with > >>Peoplesoft but the default true can produce wrong results ). > > > Nearly missed this one. > We're now on 10.2.0.3, after a bad experience with 10.2.0.2 > and .48.10 Peopletools. > we still use _unnest_subquery for the darn date range subqueries > in Peoplesoft but > _gby_hash_aggregation_enabled is not > needed in 10.2.0.3: the bug that caused that problem > was fixed in this release. I have confirmed this > with a query that caused the problem in 10.2.0.2 and > it is indeed gone. > > -- > Cheers > Nuno Souto > in showery Sydney, Australia > dbvision@xxxxxxxxxxxx > -- > http://www.freelists.org/webpage/oracle-l > > > -- Regards, Limin Guo.