Using Targit 2k14 build 6624 I am facing this problem.

I have a cube using two measure groups from two separate DWH fact tables. First measure group contains invoiced lines and second measure group contains ordered lines. On all lines in both groups there is a field called "TRANSSTATUS" which can contain a 0 for invoiced or a 1 for ordered. This field serves as a foreign key for a dimension "Sales transaction status".

This works well when using the dimension in the grid (crosstab, table, chart), however, when I put the dimension as a criterion and open it for selection, it shows only one member "Invoiced".

I have unticked the "Filter criteria" from both options in the user settings, but the problem persists.

If I connect to the cube from Excel and create a pivot table and use the dimension as a filter, it works without incident.

What do I do to make sure that I can select "Ordered" from the dimension when used as a criterion?

[img=http://imgur.com/PGi7D03]Missing dimension members[/img]

Edited by user Monday, June 13, 2016 11:36:41 AM(UTC)  | Reason: Not specified

Hi Páll,

Thanks for responding. I just tried it but with the same result as before.

/Raasmus

Thanks for the help. I set the "Non empty criteria" (my own translation from Danish) to "No" and now it returns the total dimension member list.

However, I still see this as a bug as the cube clearly has both members in it from the get-go.

I don't think zero/NULL have anything to do with it. When trying several times I saw that sometimes it was the dimension member with the zero/0 key that was shown and other times it was the dimension member with one/1 showing. I think it has more to do with TargIT querying only one measure group to find the dimension members available for the list.

Again, thanks for helping.

This is the MDX for the Query that returns all members


This is the MDX for the Query that returns only members from one measure Group from the [CustomerSales] cube.


I'm not an MDX Expert, but if I remove the "NON EMPTY" clause from the second MDX statement, it returns only the data from one fact table of the cube with values, and the other shows a [null] value - even though plenty of transactions reside in both fact tables.

Colour me confused :)

I still suspect SSAS to over-optimize 0 (cero).
If you are fiddling try to change your MDX calculation to re-define the calculated measure as
1+"measure1"+"measure2".
Thus you never get 0 (cero)