Date: Thu, 20 Aug 1998 03:19:24 +0200
Message-Id: <199808200119.DAA00593@quill>
From: Norbert Bollow <nb@thinkcoach.com>
Prefer-Language: de, en, fr
To: drums@cs.utk.edu
In-reply-to: <19980819231448.26937.qmail@cr.yp.to> (djb@cr.yp.to)
Subject: Re: VRFY discussion

D. J. Bernstein <djb@cr.yp.to> wrote:

> > When CGIs and MUAs start routinely using VRFY in order to alert users
> > of probable typos in e-mail addresses, then "real" VRFY support is
> > something which offers a real benefit
> 
> That approach is inherently unreliable,

No. The approach is reliable in the sense that it will never report a
probable typo when in fact the address is valid. That is the only kind
of reliability which is required for this kind of procedure.

> often unacceptably expensive,

I disagree. If implemented in a sensible way, it will not be
unacceptably expensive.

> and often in conflict with local policy.

I'm claiming that it is quite possible that this will change. (See my
previous messages for the details).

> None of this is in dispute.

As you have seen, all of your above statements are in dispute.

> There is a much less expensive approach that reliably produces the same
> benefits and more, without any policy problems: namely, automating the
> process of copying addresses, so that typos simply don't happen.

Sure... this should be done whereever possible. But it's not always
possible (there are situations where there is no alternative to typing
in an e-mail address) and it's not 100% reliable (an address copied from
somewhere could contain someone else's typos, or it could be outdated).
Wouldn't you hate it if you spent an hour or two on writing a letter to
someone only to find out afterwards that the only e-mail address which
you have for this person is invalid? Wouldn't it be much better if you
got the warning that the To: address is invalid soon after copying that
e-mail address from whatever source to the To: header field of your
e-mail message?

> Standard examples include ``cut and paste'' and ``Reply To'' and
> ``automatic From creation.'' Many users are so well coddled by the
> computer that they never have any opportunities to type bad addresses!
> 
> ---Dan
> Binary qmail distributions are allowed! http://pobox.com/~djb/qmail/dist.html
>