I agree most of the tests one writes before integration only capture the most trivial stuff that seldom breaks. Now I write tests after a bug or some weird cond

I agree most of the tests one writes before integration only capture the most trivial stuff that seldom breaks. Now I write tests after a bug or some weird condition arise as not to fall in that trap twice. Capturing bugs as test before solving them also ensure the fix is proper and doesn't regress other stuff. The real issue is to make sure if anything happens, tests can replicate that