Cedric's work on TestNG has touched off a series of debates about testing philosophy. The topic du jour is whether it's necessary and acceptable to define one test case that is dependent on another.

I'm really puzzled as to why there is even a debate here (the answer is "yes", BTW). The prevailing groupthink seems to be that since JUnit doesn't let you do it, you shouldn't do it.

I find this a bit frustrating, but Cedric is fully willing to entertain the discussion. I've come to really admire Cedric for his ability to be patient with folks who are a little short on clues. I guess that's how you get to be such serverside a rock star.