>2) Given that it is impossible to test all aspects of a software in all contexts and
>with all users, it is impossible to prove that code is defect free. If that's the case,
>then how would you ever know when to stop? Add to this the fact that code runs on
>equipment in contexts for users not envisioned during implementation, and you have an
>impossible ideal. Does this matter? I do. One can put 90% of one's budget to protecting
>one's house against, say, a falling satellite. And miss the dry rot in the foundation.

Agree totally. In this vein, would we consider the FPW2.6 "problem" running with faster CPUs a bug? Those CPU speeds did not exist at the time the software was in production, so it could not be tested.

Could they have had forethought and prevent this from happening?
Yes.

Should they have spent the time and resources to think of that and come up with a solution?
That's debatable.

When the problem appeared a couple of patches soon followed and I'm Ok with that.