Hmm, what do you do with a dimension that is in more than one context? If you want to include that dimension in more than one family, it seems to push you into aliasing the table.
Doesn’t generally tend to happen. The nature of the reporting often means I need to create aliases anyway. But it’s only a guideline! I’ve not yet used IDT in anger …
Really? It does in the Universes that I’ve built.
Isn’t that the whole point of contexts - allowing users to create reports that combine both conformed Dimensions and Measures from numerous fact tables, joining all of the data in the resulting cube?
I get the impression that IDT families are a bit of a gimmick.
I use families all the time. They help me with development. I put each fact in it’s own family. I put conformed dimensions in a family. I put derived tables in a family. It helps me quickly distinguish between different types of tables. And, I can export the family from one universe, and import it into another, so I can use the same conventions across universes.
Of course, it’s a feature that has no impact on the end user, so it’s a completely optional feature. If you don’t find it useful, don’t use it.
Yeah - the nature of our business is a bit different! Very few of my universes even have contexts, and conformed dimension are even rarer. But yes, a bit gimmicky.
It’s interesting though - and I’m probably derailing the thread. Most of our data sources aren’t dimensional (for a variety of reasons). And while there are disadvantages, everything is easy to understand and track back to source. The one truly dimensional model is a nightmare as I genuinely haven’t got a clue where anything comes from or how it works and the guy that built it is long gone…