[Mondrian] Re: Possible Solution to MetricFilterTest bug

Richard Emberson remberson at edgedynamics.com
Thu Apr 19 13:03:30 EDT 2007


It would be great if you could check in a bug that demonstrates
the problem and then also check in the fix (assuming all other
tests still pass with your fix).
I can not remember the origin of the line of code you wish
to comment out. It is possible that it was added because
it was required for a private dataset and schema file.
If that was the case and no junit was created demonstrating
the need for that particular line of code, then, I guess,
too bad for me.
If you do decide to check it in the fix, could you only
comment out the line of code with a reference to the bug
and/or junit that fails if it is not commented out?
That way, if some private test fails, I will know where to
look. Thanks.

Richard


Will Gorman wrote:
> Richard,
> 
> I've got the MetricFilterTest.testMetricFiltersWithNoSubtotals
> non-native test working with a minor change to the CrossJoinFunDef, and
> wanted to make sure the solution sounds reasonable.
> 
> The function "nonEmptyListNEW" gathers up the measures for the query,
> and checks for any non-nulls in "checkData()".  The MeasureVisitorNEW
> class traverses the query looking for all measures.  All I changed was
> commenting out the "queryMeasureSet.add(member);" line if the measure is
> calculated.  So now only non-calculated measures are checked vs.
> calculated measures.
> 
> The problem in the specific query seemed to be that the tuple
> [Measures].[*Store Sales_SEL~SUM] was returning a non null value when
> checkData() was called.
> 
> Here is a link to the bug:
> http://sourceforge.net/tracker/index.php?func=detail&aid=1696772&group_id=35302&atid=414613
> 
> Thanks for your help!
> 
> Will
> 


-- 
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