I've always accepted that all the tasks we want our digital machines to do are going to vary in the methodologies best used to implement them. I'm also hopefull that SQL folks at some point in their career come to the same conclusion as it would seem to be obvious to me, but I'm far from certain that this will be the case.