Date: 8 Mar 1996 17:09:43 GMT
Message-ID: <19960308170943.22256.qmail@koobera.math.uic.edu>
From: djb@koobera.math.uic.edu (D. J. Bernstein)
To: drums@cs.utk.edu
Subject: Re: Message format document outline

> I don't believe we need to go this far.  If something is obsolete,
> we need not be afraid to say so.

I suspect that practically everything you refer to as ``obsolete'' is
generated by some of today's MUAs, in conformance with today's
standards. Users will be very surprised if some of tomorrow's 822bis
parsers can't handle today's valid 822 messages.

Of course, that's just an argument against _sudden_ change. You can
safely change most protocols by first saying must-not-send, then waiting
(usually a few years is enough) for the practice to die out, and finally
relieving parsers of any responsibility.

The difficulty with mail is that mail doesn't die out. There are
billions of stored RFC 822 mail messages. They will still be around when
our current mail software is a distant memory. I believe that DRUMS has
a far greater responsibility to the owners of those messages than it has
to the authors of current mail software.

The 822bis stamp of approval should mean that an MUA can read all 822
messages _and_ the extensions that 822bis allows _and_ that it doesn't
generate certain discouraged 822 syntaxes. It shouldn't be an excuse for
breaking compatibility.

---Dan