Any JUnit fans here

Question

Just started dabbling with it here now and it seems to be the Biz, any opinion?

Answer 1

When I find bugs I write a JUnit test to mimic the
behavior right away. Then I fix the bug and make the
test pass. This brings up a good but often overlooked point: As a general rule, it's good to write the tests before you implement the code you're testing, but write them as if the code being tested were already complete. You'll get failures at first, but when you see the failures switch to successes, you can be more confident that it's because your code works, and not just because your test missed something. No guarantees of course, just a somewhat higher confidence level. Plus, as was pointed out, it forces thinking like a client of your code, which in turn can illuminate design shortcomings.
On the flipside, if the tests start succeeding too soon, then you know you've got a problem with your test code. So this technique has the benefit of providing some level of sanity checking on the tests themselves.
Fowler and the whole JUnit and XP camp are pretty vocal about writing the tests first, IIRC. I don't follow that guideline rigidly myself, but I do generally write the tests quite early--usually after the code being tested has some skeletal structure, but before it's complete.