[Mondrian] Aggregate Tables and Supremely Dodgy SQL

Tom Barber tom at wamonline.org.uk
Thu Feb 26 09:13:17 EST 2009


Thanks for the replies guys,

Will I tried a simple copy in of 3.1 to replace 3.0 and it throws and error when I try and run a jpivot....

java.lang.NoSuchMethodError: mondrian.resource.MondrianResource.getThreadOrDefaultLocale()Ljava/util/Locale;

Tom
----- Original Message -----
From: "Will Gorman" <wgorman at pentaho.com>
To: "Mondrian developer mailing list" <mondrian at pentaho.org>
Sent: Thursday, 26 February, 2009 5:46:07 AM GMT +00:00 GMT Britain, Ireland, Portugal
Subject: RE: [Mondrian] Aggregate Tables and Supremely Dodgy SQL

I've fixed a lot of aggregate table bugs in the trunk, which will be coming out soon in 3.1.   This issue may be fixed already.  Tom, could you try the latest snapshot of Mondrian and see if the problem goes away?
 
http://ci.pentaho.com/job/mondrian/
 
Will

________________________________

From: mondrian-bounces at pentaho.org on behalf of John V. Sichi
Sent: Thu 2/26/2009 12:34 AM
To: Mondrian developer mailing list
Subject: Re: [Mondrian] Aggregate Tables and Supremely Dodgy SQL



Tom Barber wrote:
> Is this a bug, my crappy schema design(its not real its just my tests), or something else?

Hmmm...the agg designer did not generate this:

<AggForeignKey factColumn="cust_code"
aggColumn="d_customers_Customer_Code" />

Instead it generated this:

<AggLevel column="d_customers_Customer_Code"
name="[Customer.Customers].[Customer Code]">
</AggLevel>

So the bug would appear to be that the agg rewrite at MDX query time is
deciding to join in the dimension table, but it has no foreign key
mapping available, so you end up with the big slow cartesian product,
which will also yield incorrect results--whereas the SQL didn't actually
need any join at all (it could just group by d_customers_Customer_Code).

JVS

_______________________________________________
Mondrian mailing list
Mondrian at pentaho.org
http://lists.pentaho.org/mailman/listinfo/mondrian




_______________________________________________
Mondrian mailing list
Mondrian at pentaho.org
http://lists.pentaho.org/mailman/listinfo/mondrian





More information about the Mondrian mailing list