Tim Pierce wrote,
| I dream about the benefits that we could get from VERP delivery --
| reduced CPU utilization, increased server efficiency, less listowner
| confusion, better word-of-mouth, and so on ...
OK, I'll agree about less listowner confusion, and Russ Allbery has already
explained the reduced CPU utilization and the increased server efficiency in
his response when Bernie Cosell asked, but how do VERPs improve the word-of-
If anything, I've seen that as a small downside to VERPs: unsophisticated
members see their own addresses in the return path and panic that they're
being forged. Or sometimes the headers will be corrupted above the RFC822
From: line, so the list member's MUA will see the From: line as part of the
body, ignore it, and try to construct a sender name from the return path,
which will include the recipient's address in the VERP portion, placing it
where even a brand newbie who doesn't know about displaying full headers will
see it; the reaction then is something to witness. (And of course, if it
happens on an unmoderated list, everyone argues "My address was on it!" "No,
you're wrong, mine was!") One public listhost that uses a modification of
ezmlm has tried to reduce the panic factor by using the word "sentto" earlier
in the VERP instead of "errors" as it used to or "return" as some others do.
[Another workaround is to assign a member ID to each subscription and to use
that in the VERP instead of the subscriber's actual address.]
While that's not enough to outweigh the advantages of VERPs, it certainly
does make me wonder why Tim says they improve the word-of-mouth.
Oh, speaking of VERPs, a couple months after my drawing the problem to their
attention, BigMailBox modified its webmail to allow equal signs in outgoing
addresses. Since sites that run it actually direct users to BMB's machines
rather than running copies of the software, the fix took care of all BMB
sites. Before then, as I was saying on this list last fall, users of BMB
sites could not send confirmations to addresses that included equal signs,
and BMB sites could not send NDNs to VERP addresses because of the equal
signs in them.