[Mondrian] Re: Explicit aggregate tables

Richard Emberson remberson at edgedynamics.com
Tue May 22 17:43:05 EDT 2007

If one uses the AggName element, then one MUST also
specify ALL of the mappings using: AggFactCount,
AggIgnoreColumn, AggForeignKey, AggMeasure and
This in-schema aggregate table recognizer is
"explicit", ALL information must be specified.
On the other hand, the default recognizer does
pattern matching and nothing has to be specified
in the schema file.
Here, we do not use the explicit aggregate table
recognizer simply because there is so much that
has to be specified. We built our own version of
a default recognizer that matches what we want.

In an earlier version of the documentation, the
AggName element had no child elements but thats
now been corrected - it always requires child
elements (or it will not be very useful having
identified no columns, levels, etc.).


Julian Hyde wrote:
> Richard,
> Is there a way to explicitly choose particular aggregate tables without 
> specifying the roles of all of the columns?
> I have read 
> http://mondrian.pentaho.org/documentation/aggregate_tables.php#Recognizing_explicit and 
> the example therein. But I would like to just write
>  <Cube name="Sales">
>   <Table name="sales_fact_1997">     
>     <AggName name="agg_c10_sales_fact_1997"/>
>   </Table>
>   ...
> without any embedded AggFactCount, AggForeignKey, AggMeasure, AggLevel 
> elements, and have it figure out the purposes of the columns. I tried 
> AggPattern instead of AggName, still no luck.
> I also found that when I made a mistake, it printed
>     Too many errors, '1', while loading/reloadin aggregates.
> but did not tell me what the errors were!
> Julian

Quis custodiet ipsos custodes:
This email message is for the sole use of the intended recipient(s) and
may contain confidential information.  Any unauthorized review, use,
disclosure or distribution is prohibited.  If you are not the intended
recipient, please contact the sender by reply email and destroy all
copies of the original message.

More information about the Mondrian mailing list