> All I'm asking is that people not make false statements about it. I'm
> not even asking for apologies---I just want the record set straight.
> Mike Nolan claimed, incorrectly, that qmail's VERPs don't work reliably.
> He owes the readers a correction.
I still don't know that they do, all I know is that Dan's claiming they do.
However, I hereby freely admit that Dan Bernstein, the author of qmail,
states that qmail VERPs work reliably. (But I doubt my confession will end
This is sort of a negative proof, in the absence of evidence that qmail
cannot figure out SOME arcane or misguided e-mail bounce. (That is what
we're talking about, right? I hope I don't have that confused, too! This
whole thread is dealing more with semantics than with specifics.)
So, a claim that it works reliably is not proven and probably CANNOT BE
PROVEN, just not disproven. It's like saying that a piece of software is
bug-free because no bugs have been discovered yet. In statistics, they call
this Type II error, the probability of accepting as true a false hypothesis.
The Type II error on Dan's statement that qmail VERPs work reliably may well
be VERY LOW, and Dan obviously feels strongly that is the case, but it almost
certainly isn't zero.