Java Bean Connectivity and Closing Connections (XI 3.0)

Question

Hi guys,
We have existing Java Beans that we would like to use with Crystal Reports. Our current Java beans return a disconnected GridModel and explicitly close the connection immediately after so that they can be returned to the connection pool (Oracle).
Once we modified these beans to return ResultSets we found that closing the connection also closes all the associated ResultSets and Statements.
What is the best practice here?
Modify the Java bean to close the connection in the finally block?
Will Crystal clean up after itself?
I also see there is something called a CachedRowSet which is disconnected:
http://java.sun.com/j2se/1.5.0/docs/api/javax/sql/rowset/CachedRowSet.html
Is this supported by Crystal?
Thanks,
Kevin D Lee =)

Answer 1

Hello Kevin,
Enterprise XI 3.0 CRConfig.xml points to Java JRE 1.5 provided with the install, so you should be ok with the CachedRowSet. 
It's used by other customers for disconnected operation.
Since you're going to Oracle, you'd likely encounter issues mapping Oracle fields to POJO property types.
Furthermore, POJO Factory libraries that were shipped with Crystal Reports XI Release 2 aren't shipped with Enterprise XI 3.0, so you'd not find that an out-of-the-box option. 
Another alternative to CachedRowSet is to define callback method in your JavaBeans class that Crystal will call when it's done with the data source. 
This has been implemented in XI 3.0 (track ADAPT00877915) - excerpt from the track note:
<JavaBeans>
    <CacheRowSetSize>100</CacheRowSetSize>
     <JavaBeansClassPath>c:\javabeans\</JavaBeansClassPath>
     <CallBackFunction>CrystalReportsLogoff</CallBackFunction>
</JavaBeans>
      Customer can configuration the <CallBackFunction/> to let CR know
      which function should be invoked when loging off.
Advantage of CachedRowSet is that you can immediately close it after you've read all the data, disadvantage would be the default implementation stores all data in memory.
Advantage of the callback method is that it would use the JavaBean you've supplied (and not load all data immediately in memory), but callback won't be called immediately, but only after the report is done with.
Sincerely,
Ted Ueda