>>>One of my problems with database development is how to construct
analogously strong test cases in order to prove the code correct. <<<
I have found the best method is to be as random as possible. I think
coders subconsciously only test with data they think will work so they
don't have worry about working late. They're biased. We're all so
programmed to think in deterministic ways that we don't recognize the
power of pure random guessing.