Contexts and join types in IDT: Best Practices?

[Former Member]

I am working on my first BI4 universe, and asking myself some existential questions about context creation.

Coming from the pre-BI4 UNV world, I am familiar with contexts, and IDT seems to have a much more user-friendly interface for managing them. However, I was wondering if someone with some more hands-on experience with IDT could give me some advice concerning the new "neutral" joins. Is it better to always explicitly include ALL joins that are compatible with a context, or to rely on the "neutral" designation to include them? I can see where neutral joins are quite handy for universe maintenance -- no longer have to remember to add joins manually to multiple contexts when they're added to the universe! -- but I'm not sure how to use them from a best practices perspective.

Do you only include/exclude explicitly for joins that cause a problem between two tables?  For example, I have tables A and B, they can be reached through two different paths, I create two contexts to separate things.  Context 1 explicitly excludes the joins in one path and explicitly includes the joins in the other; context 2 does the opposite.  All other joins which are compatible with both contexts are designated "neutral". Is this the way to proceed?

[Former Member]

Hello Emily ,

You may refer here

I hope it helps.

Cheers,

Fadoua

[Former Member]

Hi Emily!

I have always adopted one rule of thumb:

Where I have more than one Fact table sharing Dimension tables, I only explicitly INCLUDE Joins from my context relevant Fact table to the immediate dimension tables (and explicitly EXCLUDE all joins from the other Fact tables to the dimension tables directly joined to those).

All other joins (i.e. everything else outside the loop) are 'neutral' as they would normally belong to all contexts if in Universe Designer .unv world...

I find however, that using the 'Detect Context' utility most times marks several of the joins I would deem 'neutral' as explicitly included or excluded. Not sure why, perhaps something to do with perceived cardinality. I accepted some of these once and ended up with horrendous incompatibility issues...

However, the above approach has never failed me - yet 🙂

Lawrence Onunekwu

Tricord BI

[Former Member]

Hey there,

Does anyone have an answer to this questions? I am also having a little trouble in understanding how the new logic of context definition works. I've read the document suggested but it didn't help much