PDA

View Full Version : [Mondrian] [Fwd: Eigenbase perforce change 8900 for review]



John V. Sichi
03-20-2007, 09:16 PM
"The CrossJoin change was to make the optimizer less optimistic"

Sounds like we will end up with what Julian calls a pessimiser :)

JVS

-------- Original Message --------
Subject: Eigenbase perforce change 8900 for review
Date: Tue, 20 Mar 2007 15:30:01 -0800 (PST)
From: Richard Emberson <remberson (AT) edgedynamics (DOT) com>
To: Andreas Voss <tonbeller (AT) a-voss (DOT) de>, Bart Pappyn
<bppn (AT) users (DOT) sourceforge.net>, Julian Hyde <jhyde (AT) users (DOT) sourceforge.net>,
John V. Sichi <jsichi (AT) gmail (DOT) com>, Matt Campbell
<Matthew.Campbell (AT) thomson (DOT) com>, Sam Birney <sbirney (AT) jaspersoft (DOT) com>

http://p4web.eigenbase.org/@md=d&c=6PU@//8900?ac=10

Change 8900 by emberson (AT) bortei (DOT) head on 2007/03/20 15:28:08

MONDRIAN
Revised RolapResult evaluation algorithm and the CrossJoin non-empty
optimization algorithm.
The CrossJoin change was to make the optimizer less optimistic so that
it would include values that occur in corner cases. It also
cached information between invocations.
Added ability for Calc's to cache evaluation data in the Query object.
Made the ResultLimit check also work when an Iterable, rather than a
List, is used to generate the RolapAxis object.
Enable test RolapResultTest.java testD2
_______________________________________________
Mondrian mailing list
Mondrian (AT) pentaho (DOT) org
http://lists.pentaho.org/mailman/listinfo/mondrian

Richard Emberson
03-21-2007, 08:50 AM
Our main application developer created a calculated
measure that, basically, returned null if the
query was executed at the 'ALL' Customers level
but returned a value based upon the dataset for
Customer levels not at the 'ALL' level.
Well, this, of course, completely flies in the
face of the basic assumption of the original
crossjoin optimizer where, if there is no data
at the 'ALL' level, then there is no data below
that level and therefore no reason to include the
cell in the crossjoin.
With the checked in of the 'less optimistic' optimizer
this aberrant case, as an example, is also handled correctly.

Richard


John V. Sichi wrote:
> "The CrossJoin change was to make the optimizer less optimistic"
>
> Sounds like we will end up with what Julian calls a pessimiser :)
>
> JVS
>
> -------- Original Message --------
> Subject: Eigenbase perforce change 8900 for review
> Date: Tue, 20 Mar 2007 15:30:01 -0800 (PST)
> From: Richard Emberson <remberson (AT) edgedynamics (DOT) com>
> To: Andreas Voss <tonbeller (AT) a-voss (DOT) de>, Bart Pappyn
> <bppn (AT) users (DOT) sourceforge.net>, Julian Hyde <jhyde (AT) users (DOT) sourceforge.net>,
> John V. Sichi <jsichi (AT) gmail (DOT) com>, Matt Campbell
> <Matthew.Campbell (AT) thomson (DOT) com>, Sam Birney <sbirney (AT) jaspersoft (DOT) com>
>
> http://p4web.eigenbase.org/@md=d&c=6PU@//8900?ac=10
>
> Change 8900 by emberson (AT) bortei (DOT) head on 2007/03/20 15:28:08
>
> MONDRIAN
> Revised RolapResult evaluation algorithm and the CrossJoin non-empty
> optimization algorithm.
> The CrossJoin change was to make the optimizer less optimistic so
> that
> it would include values that occur in corner cases. It also
> cached information between invocations.
> Added ability for Calc's to cache evaluation data in the Query
> object.
> Made the ResultLimit check also work when an Iterable, rather than a
> List, is used to generate the RolapAxis object.
> Enable test RolapResultTest.java testD2
> _______________________________________________
> Mondrian mailing list
> Mondrian (AT) pentaho (DOT) org
> http://lists.pentaho.org/mailman/listinfo/mondrian
>


--
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.
_______________________________________________
Mondrian mailing list
Mondrian (AT) pentaho (DOT) org
http://lists.pentaho.org/mailman/listinfo/mondrian