Re: RFC 2119 violations and DRUMS mismanagement

["D. J. Bernstein" <djb at cr.yp.to>]

Let's try an example.

The smtpupd draft, like RFC 1123, defines a simple type of mailing list,
and requires that all SMTP servers let their users build such lists.

To the extent that this requirement matches user demands, it is useless.
To the extent that it doesn't, it is imposing extra implementation costs
for no benefit.

Why should LISTSERV, for example, be required to support naive mailing
lists without bounce protection? And why should firewall implementors
waste time and code to provide a feature that their users don't want?

I shouldn't have to ask these questions. There is no interoperability
excuse for the requirement. It's a barrier for new implementors. It
violates RFC 2119. The editor should have immediately rejected it.

Dave Crocker writes:
> Indeed, the real problem here is an unhappy working group member who does
> not wish to accept the very strong consensus judgements of the group.

This requirement is _not_ the DRUMS consensus. I recall objections from
Eric Thomas (LISTSERV), Jim Conklin (listproc), Philip Hazel (exim), and
of course myself (qmail).

The particular type of ad-hominem attack shown above, misrepresenting
the DRUMS consensus to avoid rational discussion, is standard practice
for a few people on the DRUMS mailing list.

> all of us have financial interests in these activities

I don't.

Frankly, I don't see why any MTA implementor, let alone an implementor
such as Newman with a financial interest in the results, should be
allowed to chair an MTA working group. How does the IESG expect to
protect itself from lawsuits?

---Dan