[Mondrian] Course grained measure groups

Julian Hyde julianhyde at gmail.com
Mon Mar 3 15:06:51 EST 2014


On Feb 28, 2014, at 1:52 PM, Matt Campbell <mcampbell at pentaho.com> wrote:

>> ...I think the general approach for granularity should be the same as for dimensionality — if the measure doesn’t exist at the current attribute’s granularity, force the member to one where the measure exists. I think the effect would be very similar to ValidMeasure.
> Yes, that does sound similar to ValidMeasure.  SSAS and IBM Dynamic
> Cubes took the opposite approach with multi-granularity and set
> attributes at lower granularity to NULL.  I guess ideally it should be
> up to the schema designer whether to inherit higher level attribute
> values or not.

I think SSAS & IBM have the right approach. If you want to inherit, you can create a calc member that inherits. But by default you don’t inherit.

Julian


More information about the Mondrian mailing list