[Mondrian] Problem with aggregate tables..

Will Gorman wgorman at pentaho.org
Sat Dec 8 22:47:22 EST 2007


Hello Everyone,

I'm ready to check in the changes necessary to get Shared Dimensions
working again with Virtual Cubes.  In order to accomplish this goal, I
started refactoring out RolapStar's levelToColumnMap and
relationNamesToStarTableMap, which is the direction Mondrian is moving
towards with the new RolapCubeMember data structure.  This impacts the
various SqlConstraint API interfaces and classes, so a lot of code has
been touched.  For example the addLevelConstraint API call looks like
this now:

    public void addLevelConstraint(
        SqlQuery query,
        AggStar aggStar,
        RolapLevel level);

I was able to completely phase out relationNamesToStarTableMap, but
there is still work to be done to phase out levelToColumnMap completely.

Two simplifications in the code that are included in these changes
include the management of RolapCubeDimension's ordinal and
RolapHierarchy's addToFrom.  globalOrdinal has been completely removed
along with the lookups that existed in RolapCube, and the realiasing
that took place in addToFrom is no longer necessary because it is
computed already within RolapCubeHierarchy during the initialization
process.

I hope to check these changes in Monday night.  If there are any
questions I can answer before then, please feel free to post them here.

Will

On Mon, 2007-11-26 at 17:30 -0500, Will Gorman wrote:
> This weekend I started looking at 1833528 in more detail.  The majority
> of changes that will be necessary are in SqlTupleReader and the
> SqlConstraint API.  Now that Mondrian has RolapCubeHierarchy, which
> knows about its Cube and Star, the levelToColumnMap and
> relationNamesToStarTableMap maps are not necessary and will be phased
> out.  Today, these data structures limit the ability for shared
> dimensions within virtual cubes.
> 
> If folks would like to help out, the best area to contribute at the
> moment would be in additional tests that you would like to see work or
> worked before the RolapCubeMember check-in.  The more Virtual Cube and
> Shared Dimension unit tests we have, the more I'll be certain that we
> are going in the right direction with this new approach.
> 
> I'll be working on this over the next couple of weeks, and will
> hopefully have these test cases back on-line along with additional tests
> which will demonstrate more advanced shared dimension functionality
> with-in virtual cubes.
> 
> Thanks,
> 
> Will
> 
> On Sun, 2007-11-25 at 21:41 +0000, Julian Hyde wrote:
> > > So, what I'm proposing is this.  If someone has followed the 
> > > process and 
> > > submitted a regression test for something they expect to have working 
> > > forever in the future, then anyone else who wants to disable 
> > > that test 
> > > as part of some other change needs to raise the issue on the mailing 
> > > list first, and get it resolved before going ahead with checkin.
> > > 
> > > Would that be acceptable?
> > 
> > I'm OK with that. Will and I made the mistake of communicating back-channel
> > on this issue. I approved his check-in because it didn't conflict with the
> > December/January timeframe for mondrian 3.0, and I didn't know about your
> > release schedule.
> > 
> > Will, can you make it a priority to re-enable this particular test.
> > 
> > Julian
> > 
> > _______________________________________________
> > Mondrian mailing list
> > Mondrian at pentaho.org
> > http://lists.pentaho.org/mailman/listinfo/mondrian
> 
> _______________________________________________
> Mondrian mailing list
> Mondrian at pentaho.org
> http://lists.pentaho.org/mailman/listinfo/mondrian




More information about the Mondrian mailing list