Hi, I just ran into the problem MONDRIAN-791, and applied the patch..
seems to work fine thankfully, but this was just a side problem.
The main problem is with the values returned when the last filter
element (when more than one member) have more than 1 child. E.g.
When I select a filter such as:

where Crossjoin({[Measures].[amount_inv_EUR]}, {[ProfitCenter].[BU2],

Where BU2 has 2 children and BU3 has 12, then the resulting cells are 12x value
Reversing the order {[ProfitCenter].[BU3], [ProfitCenter].[BU2]}
gives 2x values
Testing with another (BU4) with 3 Children gives me values 3x expected.
If I select just the children , even if many, the results are correct
(I'm assuming since each is 1 element and not a group of elements).

This seems to indicate to me something wrong with either the creation
of the set in the filter? or application of the filter ....

Any ideas, or even on where this might be happening?

Mondrian mailing list
Mondrian (AT) pentaho (DOT) org