[Mondrian] Degenerate dimensions and Aggregate tables

Pedro Salgueiro pedro.salgueiro at inductiva.pt
Tue Aug 28 10:30:13 EDT 2012


We are now able to use degenerate dimensions together with

We were debugging mondrian, trying to find the problem, and suddenly we 
decided to change the definition of our cube. Before, in the degenerate 
dimension we had as foreignKey, the key of the fact table (which in some 
point in time made sense to us).

After reading more carefully the Mondrian 
documentation(http://mondrian.pentaho.com/documentation/schema.php#Degenerate_dimensions), 
we saw that there is no need to include the foreignKey in such 
dimension. So, we removed the foreignKey, and the aggregate table 
started to be picked.

Could it be that this bug have been corrected without anyone notice 
about it?
At least in our situation it seems to be working ok.

Pedro Salgueiro


On 08/24/2012 09:55 AM, Pedro Salgueiro wrote:
> You are right, haven't thought about it those terms, if we are able to 
> maintain the aggregate table, we can also manage this small dimension.
>
> We will try switching to a real dimension, probably based on view.
>
> BTW, any idea on where to start digging to try to debug and solve this 
> issue?
> Pedro Salgueiro
>
> On 08/23/2012 06:35 PM, Julian Hyde wrote:
>>> On Thu 23 Aug 2012 05:47:22 PM WEST, Pedro Salgueiro wrote:
>>>> The cube we are using contains a very simple degenerate dimension
>>>> which only have a single level. We are using this degenerate dimension
>>>> in all our queries and it's not practical to create a fact table for
>>>> this dimension since it would be hard to maintain.
>> Does not compute. You can maintain an agg table but not a dimension 
>> table? Why not make the dimension table a view on the agg table?
>>
>>>> http://jira.pentaho.com/browse/MONDRIAN-563
>> On Aug 23, 2012, at 9:50 AM, Pedro Alves <pmgalves at gmail.com> wrote:
>>
>>
>>> Afaik, that issue still holds
>> I believe so, too.
>>
>> Vote for the jira case.
>>
>> Workaround to try: Create a real dimension table (which will have a 
>> small number of rows). Consider using a junk dimension -- I haven't 
>> tried these, though.
>>
>> Julian
>>
>> _______________________________________________
>> Mondrian mailing list
>> Mondrian at pentaho.org
>> http://lists.pentaho.org/mailman/listinfo/mondrian
>>
>



More information about the Mondrian mailing list