I don't recommend any attempt to unify check/handle and panic/defer/recover. They don't behave the same because they aren't doing the same: one is controlled ha

I don't recommend any attempt to unify check/handle and panic/defer/recover. They don't behave the same because they aren't doing the same: one is controlled handling of normal errors with reduced boilerplate, the other is "crashing with cleanup".