[Mondrian] RE: High Cardinality check in

[Julian Hyde]

First there's some cleanup to do. As I said in a previous email:


When you have a perforce account, there are some cleanup items:

*	javadoc for new IterCalc methods 

*	document new property HighCardChunkSize. MondrianProperties.java
describes how to document a property. Documentation should make sense to a
DBA; even I don't understand the current documentation in
MondrianProperties.java. 

*	document schema changes (see my TODO: in schema.html) 

*	review tuning guidelines in view of new high-cardinality support

Please do all of these before committing any other changes.

Then the next step is to check in a minimal fix to allow the regression test
suite to pass when a dimension is defined as high-cardinality. I think
that's what you're suggesting here. To test it, don't add a high cardinality
dimension to FoodMart.xml. Use a combination of the
mondrian.test.highCardDimensions property (which allows you to make a
particular dimension or dimensions high-cardinality for the entire suite)
and temporary schemas (as used in many examples in SchemaTest). Make the
absolute minimum code changes - I think there's only a ClassCastException to
fix.

Before you check in, run 'mondrian/bin/megatest' - this runs test suite
against various configurations of jdk, database, and property settings to
enable grouping sets, aggregate tables, etc.

When you've completed those, and only then, you can check in the remainder
of the high cardinality change. 

Please Cc mondrian at pentaho.org on your emails - I want the other developers
to know what's going on.
 
Julian


> -----Original Message-----
> From: Luis F. Canals [mailto:luisf.canals at gmail.com] On
> Behalf Of Luis F. Canals
> Sent: Friday, May 30, 2008 2:20 PM
> To: julian.hyde at sqlstream.com
> Cc: Jorge López Mateo; Javier Giménez Aznar
> Subject: High Cardinality check in
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi Julian,
>
> firstable, sorry for the delay. I've promissed you some
> modifications 
> to have high cardinality up and running on thursday and we're on 
> friday....
> But, good news: we've made some (but not too many) changes to move 
> nearest the high cardinality.
>
> The only thing we need to finish high dimensions management is to 
> modify an attribute dimension (particullary "highDimension") at 
> runtime in the test, to avoid changing FoodMart.xml.
> With this modification, high cardinality dimensions are ready to use 
> at least for simple cases.
>
> In this point we have two options:
>       a) submit changes and add test to certify high
> cardinality later (on 
> two or three days)
>       b) wait for having test completed and submit everything
> in two or 
> three days
>
> What do you think?
>
> Thanks a lot and best regards.
>
> - - javier, jorge & luis
>
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.8 (Darwin)
>
> iEYEARECAAYFAkhAb5EACgkQh9NqBdPYVw2kagCgqwGhIWFjl0gqpDDY1Q89Fm+q
> JWAAoJotyuK9pFXr2mJft8GPOlrDRuKh
> =b/9P
> -----END PGP SIGNATURE-----
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.pentaho.org/pipermail/mondrian/attachments/20080601/09550ac4/attachment.html