Re: Transactions: good or bad?

Well, it certainly would be a tall challenge, that's for sure.

>
> Your pie in the sky is provably impossible to cook :)
>
> To quote the great logician Jean-Yves Girard "a cause des ces fichues
> idees" (sorry for the lack of accents on my keyboard). In English it's
> "because of those bloody ideas" (that they totally do not have). For the
> whole delicious paper google on Jean-Yves Girard Les fondements des
> Mathematiques.

Thank you.

> Test can only prove the presence of bugs, not their absence.

But what makes a test a test and a proof a proof? And, aren't proofs themselves often rather wordy? Isn't the proof for Fermats Last Theorem that someone came up with a few years back some 400 pages long?

> Yes, there are. But it has nothing to do with your current claim (that
> you can exhaustively test computer programs).

My current claim was that the you could use the information in the program to generate test cases sufficient to do a test, coupled with an expert system for weaning out already proved things. Theoretically, I think you could do this, and then have a wired up world where everything was a gigantic peer to peer bus of formation knowledge expressed through correctly working software.

> Yep, and related to logical assertions, there are like tons of problems
> that are provably undecidable, impossible, NP-complete or otherwise
> infeasible.

If they are provably undecidable, then, how do people decide them? I mean, travelling salesman is NP complete but we still have algorithms that express a "best shot". Received on Thu Jun 12 2003 - 13:56:39 CEST