Bravo! You got it! Rather than spend a lot of time trying to find flaws in JDO, which there are really few as a Java centric technology, I was trying to point out in my article the flaw in the approach of continuing to map attributes to relational databases when object-relational technology allows us to define our classes, that is attributes and behavior, in a persistence layer.
It's rather comical that I had to send my "correct" code to onjava.com three times and it still got mashed in the process of being posted. Either way, does one obvious defect in an article's listing make the code of poor quality? Only if you're so taleneted as to always write perfect code.