Ideally, there should be either only one aggregated so that we can apply the thresholds on the global situation.

Alternatively, we could see to define different threshold for acceptance and the "standard" run.

What is problematic right now is that if you add new code and cover it 100% using a services/test, acceptance coverage report will get smaller, and the threshold will kick in, fail the build and somehow get in the way. This just happened to me a few seconds ago.

I think the global goal for this is that we aim at coverall as well as possible our code, hence my thinking that reports ought to be aggregated into only one, and the threshold applied once only too.