Nathan Zook writes:
And is there any way to build trusted system out of small, verifiable pieces? Since the way they're connected could also be questioned, I suspect that when you put enough of them together it's just as bad as the case of a single, monolithic program. But this isn't my area, so I don't know.
No. This was essentially proved during the first third of this century.
Well, I haven't gotten a reply from Nathan Zook on this assertion, so can anyone else back it up with some references? Perhaps we're discussing different contexts, but proving correct systems composed of correct components is still a subject of active research.
nathan
I suspect that he's referring to Godels' Theorem, which shows that in any complete logic system it's possible to make undecidable statements. I'm not at all sure if this can be extended to computer programs. Even if it can, I suspect it only applies to contrived cases. Peter Trei Senior Software Engineer Purveyor Development Team Process Software Corporation trei@process.com