David Johnson wrote an interesting item yesterday on persistence frameworks in Java. As I’ve mentioned before, I’ve always been a write your own SQL kind of guy, but the more I read about persistence layers, the more interested I become. I’m particularly intrigued because on our last two internal projects, we’ve wound up creating our own simple persistence layers from scratch. If we can use an existing package to provide more robustness and take some of the gruntwork out of the project, so much the better. This snippet seems to hold much promise:

Roller uses Castor, but not the Castor Source Generator. Instead, Roller uses XDoclet to generate the persistent data objects, Struts form beans, and the Castor Mapping file all from the same source meta-data.