TM1 Forum

Hi,

I have a Display Flag Cube. The intention of this cube is to feed other cubes (Budget, Costs etc) with a flag "1" so that when zero's are suppressed any row with a flag will still be visible. Then at the start of the budget process when the user enters his budget via an active form he can see the valid elements against which he can enter his budget. Without this flag he wouldn't be able to see any elements with zero suppressed because nothing has a budget yet. Without zero's suppressed he would see too many elements.

I have two questions:

1. From a design perspective do you think that this is a sensible solution? I have considered using subsets of dims to show fewer elements (only the ones relevant to the budget holder) and that way I wouldn't have to suppress zero's. However, it would involve creating a lot of subsets and I think this would create too much of a maintenance overhead.

2. Feeding the display Flag cube. The cube has 3 dims (Project No, Cost Centre, Measures (measures = original budget, actuals, committed). The rule for Actuals and Committed are based on external sources. The rule for Original budget is internal to the display flag cube:

['Original Budget']=N:
IF(!Project No@=subst(!Cost Centre,1,8),
1,STET);

I am having trouble feeding this rule. So far I have come up with:

['Original Budget']=>
DB(IF(DB('Display Flag',!PAAR_AFE,!T3,'Original Budget')=0,'','!PAAR_AFE'),
!T3,'Original Budget');

However, when I trace the feeders it tells me that the values are still not fed.

As usual, any advice would be welcomed.

Cheers

Hi
It is possible to get this working but my question would be how flexible the 'flagging of other cubes' needs to be and when it needs to happen.
Obviously there are models where the flagging needs to happen instantly when you change the 'Display Cube' - say user sets up a new (future) asset and needs to budget for capital right away.
But even in those situations you could run a TI process that takes care of the flagging.

If you pre populate your 'Display Cube' with actual data (lots of 1s) in the known combinations rather than using a rule it is easier to adjust later - just adding ones where needed or parameterize the TI that populates it - and might also be easier for you to get around the feeding problem.
Another process can take care of the flagging in the other cubes based on the 'Display cube'.
Sure it will need to run and lock and all but it is simple and if the valid combinations are pretty much fixed it is the easier solution.

Gregor,

many thanks for your input. The rule to populate the flag was killing the performance of the cube. I have deleted the rule and followed your advice with a TI as the flag does not need to be real time. The TI works and performance is unaffected. I have also scheduled the TI to run nightly.

Thanks again