Semi-additive measures accross multiple dimensions
Currently you can only select one dimension in the cube. We have a requirement to average all dimensions on all hierarchy levels. That is on the very highest level, you want to see the average of everything in the cube and on all levels down.
Ben Pfrommer commented
I think you can do it in a rule, but if you had semi-additive measures able to do this across multiple dimensions that would be great.
Same problem when you have a "Daily Stock" cube (check stock levels on a daily basis) and your stock levels are of course not cumulative over time - when you have 3 time related dimensions for this (Year, Week, Day of Week) then using the semi-additive measures doesn't help you. Using 3 time-related dimensions makes sense because you can put them on rows and columns, whereas if you have only one (Date dimension) you can only put that on rows OR columns, i.e. you can't do an easy comparison across "same day last week" or "same week last year" etc.
So I think having semi-additive measures across multiple dimensions would be great. (alternatively having the ability to have only one time-dimension but stick it's attributes against rows and columns at the same time).
Sjoerd Kuipers commented
Isn't that a matter of defining a cube rule?