Thanks, Nigelyou have no idea how often and long I looked at that and never saw it. That, of course explains it. Now I have to find the flaw in the process that builds f00_tmp.
At 02:21 AM 11/10/2008, Nigel Thomas wrote:
WolfgangAll your rows have the same column values except for the amtN columns, the PD column and the two (currency?) columns which appear to be C14 and C15, and C13 (with values B and U). By the way, the PD and FY columns appear to be reversed (surely FY would be 2008 and PD would be 1 to 4).In foo_tmp you have one set of 12 rows for C13='B' (4 values for PD * 3 currency combinations) but you have two sets of 12 rows for C13 = 'U' - making 24 rows compared to the 12 rows for C13='PD'.So naturally given 2 rows in the input for each distinct tuple in foo_tmp you get 2 rows in the output... or am I missing somethingAnd they are sorted (what was the plan - a sort merge?) so like rows are shown together (rather than being in the order shown for foo_tmp)Regards Nigel2008/11/10 Wolfgang Breitling <<mailto:breitliw@xxxxxxxxxxxxx>breitliw@xxxxxxxxxxxxx>Can someone show me the errors of my way, please.
Regards Wolfgang Breitling Centrex Consulting Corporation http://www.centrexcc.com