[Mondrian] Native vs. Regular Non Empty <Level>.MEMBERS Issue

Julian Hyde jhyde at pentaho.com
Fri Apr 11 03:10:24 EDT 2008


General principles:
 
1. Default behavior should be the same as SSAS.
2. Behavior should be the same for native and in-memory non empty.
 
If you don't like the default behavior, you can add a property, disabled by
default of course.
 
I would view differences between 'native non empty' and 'in-memory non
empty' behavior as serious bugs. It's extremely confusing for an end-user to
get different results, and it makes it more difficult to support the
product, since people don't generally post their settings when they have a
bug.
 
Based on these principles, what action should we take? Sounds like option
#1, except that the bug would be high priority. Correct me if I'm wrong.
 
Julian



  _____  

From: mondrian-bounces at pentaho.org [mailto:mondrian-bounces at pentaho.org] On
Behalf Of Ajit Vasudeo Joglekar
Sent: Thursday, April 10, 2008 3:13 AM
To: Mondrian developer mailing list
Subject: Re: [Mondrian] Native vs. Regular Non Empty <Level>.MEMBERS Issue



I propose the following 

We roll back the changes done in change list #10641. After doing this, bug:
1722959 will reopen and can be marked as a non issue and bug: 1909516 will
get fixed. As far as 1909516 is concerned it is possible to add check around
IgnoreUnrelatedDimension and get it working. But this makes ValidMeasure
dependent on IgnoreUnrelatedDimension which is not a good idea. Hence the
only way out is to roll back change #10641 

As far as the difference in behavior between native non empty and in memory
non empty (motivation behind bug 1722959) regarding the handling of empty
rows/ columns with literal "1" calc members is concerned we have couple of
options: 
1) can log a new low priority bug fixing of this will involve changing
native non empty behavior to not to eliminate the above said rows/columns.
Opposite of bug 1722959 
2) document it and live with the difference in behavior. whoever doesn't
want this native non empty behavior may turn of the the feature using
property 

Requesting comments 

Thanks 

-Ajit

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.pentaho.org/pipermail/mondrian/attachments/20080411/a29d8790/attachment.html 


More information about the Mondrian mailing list