Re: [boost] Regression testing unobservable behavior?