Changes in the Behavior of Functions Between ReportNet 1.1 and IBM Cognos BI

In ReportNet 1.1, double counting could occur when applying count or count distinct to a query item. This occurred primarily when querying multiple query subjects that were joined 1-to-N in the model if the counted item existed in the query subject on the 1 side of the join. The explanation is that count or count distinct was applied after the join operation.

Count distinct worked only when the item to which count distinct was applied was the column used in the join. In this case, the column to be counted could have identical values for the different values used in the join condition.

In IBM® Cognos® Business Intelligence, improvements to the count functionality cannot be handled through an automatic upgrade. The new approach avoids double counting on the 1 side of a 1-to-N join. The explanation is that count or count distinct is now applied before the join operation.

There is now a lesser need to use count distinct and, when used, it will be successful more often. Count distinct is no longer required to overcome double counting. Instead, it can be used as intended to select distinct values that exist in a query subject.

When you want to count repeated occurrences of a value, we recommend that you do one of the following:

Apply a count operation on a column based on a query in which an explicit join occurs.
This applies the count after the join.
Count rows in a report by using a layout calculation object or by counting the literal value 1.