Hi,
Linking a dimension object is possible even if the list of values for the two queries is different.
Lets say query a retrieves clients A,B,D,F and query b retrieves clients B,C,F. Then the ‘linked’ version retrieves A,B,C,D and F.
It will be the superset, so consistent with a full outer join.
Now if you have measures from query A they can go in the table block.
The lines that have data (in this case for client A,B,D and F) will show the measure value for that dimension value.
Same applies for the measures from query B (in this case lines for clients B,C and F will show values).
Other (unlinked) dimension objects coming from query a OR from query B can go in the block.
If they result in more lines, the measure value from the ‘opposite’ query will show the ‘total’,
the calculation is only available on the linked dimension value level.
If you have other dimension objects to add you can either turn them to measure (by putting a max() around them)
or you can make them a detail of the linked dimension object.
Notice that in both cases it is only valid if you have just ONE value for each dimension value.
To clarify this another example;
if query a retrieved client (dim)/product (dim)/revenue (meas) and query b retrieved client (dim)/month(dim)/payment (meas),
putting client and both product and month in one table block would not make sense.
In this case, make the common dimension object(s) a section master and put the queries a and b in two separate table blocks.
They CAN be positioned next to each other so they look like one.
Hope this helps,
Marianne
marianne
(BOB member since 2002-08-20)