[iratewizard]

The former. Business logic changes regularly. New phases of projects frequently require changes that break the unnecessary business logic test code.

Test code should fail when an important underlying mechanism regresses. If it always fails on the next phase of the project because some hapless idiot needed to cover every line, the signal to noise ratio gets weaker and the purpose of the test code is lost.

[nescioquid]

Thanks for the clarification -- that's what I suspected, but thought I must be misreading or misunderstanding, but I also don't know your situation and trust you're making the right set of trade-offs here. If your system is today a cuttlefish, yesterday a cow, and tomorrow a crab, cow tests don't do you much good. That sounds like a crazy environment, and I'd maybe have a care that those hapless idiots are sane people in an insane place.

[iratewizard]

You sound like you're in that category if you can't differentiate and think that it's surprising for software to have a flexible business logic layer.