Combining Embedded Total and Unsolved Dimensions In A Single Cube

There's an interesting thread on the OTN OLAP Forum at the moment. 

"I am trying to define a cube using the cwm2_olap packages, where not all dimensions have embedded totals.

Dimension A is Time with a hierarchy of Day>Month>Year and I want embedded totals for that. Dimension B only has a very small tree (three levels, 10 lowest level nodes, so I guess I can do without ETs.

Is it possible to specify this cube using the cwm2_olap routines? After all cwm2_olap_table_map.map_facttbl_levelkey only has one parameter STORETYPE to say 'ET' or 'LOWESTLEVEL'."

Heiko Becker comes up with a good answer which it's worth posting here;

"If you are in the ROLAP world ... You need just to define unsolved, since you are able to use the materialized views to resolve your problem of a partial aggregation.

What I mean is, you can use the DBMS_ODM to create materialized views and in this creation process you can select which level combinations should be in the materialized view and which should aggregated on the fly.

If you are in the MOLAP world, just define your cube a fully solved, since in the anayltic workspace you can define aggregation on the fly by using an aggmap.

So in my opinion no need to define different metadata strategies for the cube."