Get a copy of dbcpconnectionprovider.class and copy it into WEB-INF/classes/org/hibernate/connection/ (create the directories if they don’t exist)

Then add the following to hibernate.cfg.xml….

<property name=”hibernate.connection.provider_class”>

org.hibernate.connection.DBCPConnectionProvider

</property>
<property name=”hibernate.dbcp.initialSize”>8</property>
<property name=”hibernate.dbcp.validationQuery”>SELECT 1</property>
<property name=”hibernate.dbcp.testOnBorrow”>true</property>
<property name=”hibernate.dbcp.testOnReturn”>false</property>
<property name=”hibernate.dbcp.testWhileIdle”>true</property>
<property name=”hibernate.dbcp.minIdle”>4</property>
<property name=”hibernate.dbcp.timeBetweenEvictionRunsMillis”>120000</property>
<property name=”hibernate.dbcp.numTestsPerEvictionRun”>3</property>
<property name=”hibernate.dbcp.minEvictableIdleTimeMillis”>120000</property>

And it should keep your connections pooled so you no longer suffer from this issue.

Download the latest archive

Extract the pentaho-reporting/libs/ directory and dump its contents in

jboss/server/default/deploy/pentaho.war/WEB-INF/lib/

When you restart the system, if there is a load of error messages, remove the gnujaxp.jar file as it caused me a right load of problems.

So publish an analysis cube with the JPivot stuff, load it up and work out how you want your table displayed and setup. It may be an idea to clean up the code a bit but once that’s done copy it out, and paste it into the report wizard/designer and it should all work, just make sure you use the same cube

If you want to load the selections for a secure filter/prompt in a xaction dynamically, then in the editor select a Get Data -> relational

Input the query to retrieve the relevant information and then setup the result-set at the bottom of the page and assign the columns.

The in the Secure Filter/Prompt do the usual just selecting the dynamic result-set. As long as the details are correct you will see the query results next time you run the xaction.

For some reason Members and Calculated Members have different syntax to format calculated members add them in a similar way to the code here.

<CalculatedMember name="Rolling Budget Margin" dimension="Measures">
<formula>
((([Measures].[Current DAY_ID] / [Measures].[Month DAY_ID])) * [Measures].[Budget Margin])
</formula>
<CalculatedMemberProperty name=”FORMAT_STRING” value=”#,#00;;\Z\e\r\o”/>
</CalculatedMember>

MDX history

MDX stands for Multidimensional Expressions it is a query language for Online Analytical Processes(OLAP) Databases and cleverly it allows relatively easy calculations using formula’s similar to spreadsheets. To my surprise it was actually invented by a guy at Microsoft, they’ve come up with a useful specification and actually seemed to keep it pretty quiet. So unsurprisingly it runs on MSSQL, MYSQL, Oracle, DB2, the usual culprits.

How it works

Well MDX with Pentaho works by creating a Cube which references tables in a main database. In this Cube you will define ‘Dimensions’, ‘Hierarchies’, ‘Levels’ and ‘Measures’ these create the levels of information that the MDX server uses to reference. So for example you may have:

<Dimension name="TIME" foreignKey="DATE_ID">
<Hierarchy name="TIME" hasAll="true" allMemberName="All time" primaryKey="DATE_ID">
<Table name="TIME"/> <Level name="YEAR" table="TIME" column="YEAR_ID" uniqueMembers="false"/>
<Level name="QUARTER" table="TIME" column="QUARTER_ID" uniqueMembers="false"/>
<Level name="MONTH" table="TIME" column="MONTH_ID" uniqueMembers="false"/>
</Hierarchy>
</Dimension>

So as you can see first we define the dimension which in this case is called ‘Time’ which references the foreign key ‘Date_ID’.

Next we define a hierarchy which in this case is also called ‘Time’.

Then we define the levels which are reasonably self expanatory these decend in diminsihing size as it quite obvious with this date example.

<Measure name="TOTAL_SALES" column="TURNOVER_EUR" aggregator="sum" datatype="Numeric" formatString="# ##0,###"/>

Lastly we can create Measures which do pretty much do as the name suggests and we use them as the core data in our tables. The name is the name you want to see in the cube, the column if the column name from the database, there are several different aggregators your can set as with datatypes and formatStrings.

But don’t worry this doesn’t all have to be coded by hand Pentaho have developed a Cube Designer which you can download and use to create and upload the cubes to your server. This is much easier than creating by hand, the only problem being at the moment your can’t reopen a saved cube for editing, so once you’ve published your cube check it to make sure its as you expect before you close the designer, because you can always make adjustments and re-upload. Once the cube is published you can always edit the xml file directly, it will be called cubename.mondrian.xml. Take a look and check out its structure.

Anyway, I’ll be back later with the basics of starting query writing.

Also a big thanks to the folks at the forums(especially Taqua) for answering my rather stupid and plain obvious questions, and for bug fixing.

I’ll also blog about other stuff that I’ll try to keep separate so check back to see whats been going on, and I promise to make it look a bit more aesthetically pleasing over the next few weeks.

Tom