>> Then maybe the number of bugs for an architecture can bematched to >> the power-on-time for the machines for that architecture. > >So your solution is to replace requiring financial donations to >requiring more hardware donations? Cold boots are by far the biggest >cause of hardware failure, this risk is far too high for some of the >older machines that consequently have more expensive and harder to find >parts.
To clarify the situation: the machines are on all the time. And as much as possible, they are always building something, whether it be ports or builds, hoping that some of the address space randomization or such will spot bugs. Also, if a new change goes into the source tree which creates a problem, we want to spot it as soon as possible, before the developers involved have become distracted in other directions. > Besides, how are you going to find bugs on powered-off machines? No kidding.
