Again: two facts and one conformed dimension

Question

Dear all,
I know this has been asked quite often but I haven't found a solution for my problem here. Believe me I have done a lot of searching across the net ..
What I want to do is the following:
I have two fact tables with a number of common dimensions. A report should show the two facts across these common dimensions.
To keep it simple I start with only one common dimension.
Here's what get's to the db from OBIEE:
*SELECT Revenue."Date (YYYYMMDD)" saw_0, Sales."Day Day Code" saw_1, Revenue."TPV (Total Payment Volume in Millicent [€])" saw_2, Sales.DEBIT_AMOUNT saw_3 FROM "Company KPIs" ORDER BY saw_0, saw_1, saw_3*
Note that *Revenue."Date (YYYYMMDD)"* and *Sales."Day Day Code"* are the same field from the common time dimension. I just put them in different presentation tables for the end users.
TPV and DEBIT_AMOUNT are facts from the two different cubes.
Everything fine, right?
Now I look at the nqquery.log:
RqList
Dim Time.Day Day Code as c1 GB,
Dim Time.Day Day Code as c2 GB,
*Tpv:[DAggr(Cube Revenue.Tpv by [ Dim Time.Day Day Code, Dim Time.Day Id, CUBE_SALES.DEBIT_AMOUNT] )] as c3 GB,*
CUBE_SALES.DEBIT_AMOUNT as c4 GB,
Dim Time.Day Id as c5 GB
OrderBy: c1 asc, c2 asc, c4 asc
now the bold thing looks odd to me. Is he trying to aggregate one cube based on the other? Could be the problem, but where to influence this?
And finally this translates into this query:
WITH
SAWITH0 AS (select distinct T4110.DEBIT_AMOUNT as c1,
T281.DAY_DAY_CODE as c2,
T281.DAY_ID as c3
from
DIM_TIME T281,
CUBE_SALES T4110
where ( T281.DIMENSION_KEY = T4110.DIM_TIME ) ),
SAWITH1 AS (select distinct SAWITH0.c2 as c1,
SAWITH0.c2 as c2,
*cast(NULL as DOUBLE PRECISION ) as c3,*
SAWITH0.c1 as c4,
SAWITH0.c3 as c5
from
SAWITH0)
select SAWITH1.c1 as c1,
SAWITH1.c2 as c2,
SAWITH1.c3 as c3,
SAWITH1.c4 as c4
from
SAWITH1
order by c1, c2, c4
As you see, the tpy ist gone and a NULL appears. That's what I see in the report. TPV is always NULL.
So where did I miss a configuration ???
Thanks, Knut

Answer 1

You only need to set the content levels on the measures themselves when you want non-conformed dimensions in the report, as your working with fully conformed dims just make sure your levels are set on the content tab for each Logical Table source in the Logical fact table. Leave the non-conformed dim hierarchies as blank.
Remember - You should put a hierarchy on each logical dim , even if there is no real hierarchy, just have a total -> Detail.
You should always always always set your content levels for logical table sources, both in Dimensions and Logical fact LTS's.
If they are greyed out, make sure your not in read only, have checked the object out if on-line mode, failing that, perform the changes off-line then re-start your BI Server.