[Mondrian] [Fwd: Eigenbase perforce change 9529 for review]

Rushan Chen rchen at lucidera.com
Fri Jun 29 19:35:47 EDT 2007


Ah...forgot to update that table. The lookup key for RolapNativeSet includes the key for the SQL constraint and the key for the inputs. The change added referenced base cubes to the key for the SQL constraint. So I think a more accurate description for RolapNativeSet key could be "key for SQL constraint(including referenced base cubes) + key for inputs".

Updating the doc.

Rushan


-----Original Message-----
From: mondrian-bounces at pentaho.org on behalf of John V. Sichi
Sent: Fri 6/29/2007 4:03 PM
To: mondrian at pentaho.org
Subject: [Mondrian] [Fwd: Eigenbase perforce change 9529 for review]
 
In this change, Rushan fixed an interesting bug resulting from the 
combination of virtual cubes with caching for the tuple set results of 
native NonEmptyCrossJoin.  Previously, the caching wasn't taking into 
account which fact table the referenced measures were coming from, so an 
MDX query which was identical to a previous one in all aspects except 
the referenced measure could get an incorrect hit in the cache lookup 
(and return the wrong results).  Her fix adds more discriminating 
information into the cache key.

Corresponding doc update is here:

http://docs.eigenbase.org/MondrianCacheOverview#Non-Empty_Tuple_Set_Cache

Rushan, I updated the table at the top of the page to state that the 
lookup key in this case is "SQL constraint + referenced base cubes"; is 
that accurate?

JVS

-------- Original Message --------
Subject: Eigenbase perforce change 9529 for review
Date: Thu, 28 Jun 2007 10:00:01 -0700 (PDT)
From: Rushan Chen <rchen at lucidera.com>
To: Andreas Voss <tonbeller at a-voss.de>, Bart Pappyn 
<bppn at users.sourceforge.net>, Julian Hyde <jhyde at users.sourceforge.net>, 
John V. Sichi <jsichi at gmail.com>, Matt Campbell 
<Matthew.Campbell at thomson.com>, Sam Birney <sbirney at jaspersoft.com>, 
Zelaine Fong <zfong at lucidera.com>

http://p4web.eigenbase.org/@md=d&c=6PU@//9529?ac=10

Change 9529 by rchen at rchen.fao.eigenbase on 2007/06/28 09:58:02

	mondrian: fix cache lookup for RolapNativeSet when two native set 
operations(e.g. non empty cross join) based on the same virtual cube 
reference the same dimensions, yet they are used in queries that 
reference different base cubes.

Affected files ...

... //open/mondrian/src/main/mondrian/rolap/RolapCube.java#102 edit
... //open/mondrian/src/main/mondrian/rolap/RolapResult.java#106 edit
... //open/mondrian/src/main/mondrian/rolap/SqlContextConstraint.java#16 
edit
... //open/mondrian/testsrc/main/mondrian/rolap/BatchTestCase.java#2 edit
... //open/mondrian/testsrc/main/mondrian/rolap/VirtualCubeTest.java#28 edit
_______________________________________________
Mondrian mailing list
Mondrian at pentaho.org
http://lists.pentaho.org/mailman/listinfo/mondrian





More information about the Mondrian mailing list