Date: 7 Mar 1996 20:01:06 GMT
Message-ID: <19960307200106.18212.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

> Invisible continuation lines must not be generated.  There then
> becomes no requirement on how they are recognized or interpreted.

This is incompatible with existing practice. Yes, it's unfortunate that
such messages exist, but they do, and an 822bis parser shouldn't choke
on a valid 822 message. Mail messages stick around forever.

I suggest must-not-generate-but-must-interpret-properly.

> I am clarifying in the BNF which nonterminals are lexical

``Clarifying''? RFC 822, section 3.3, makes perfectly clear what its
lexical tokens are.

You are _changing_ the rules. For example, disallowing whitespace in
domains would break some RFC 822 examples. Has this been discussed?

  [ programs that understand only one timezone spec: ``GMT'' ]
> aren't such programs hopelessly broken?

Yes, they are, but there are enough of them that they're worth
supporting. RFC 822 lets me support them. RFC 1123 says I shouldn't.
You're saying that I can't. That's going too far.

> > John has suggested that, for something to be ``reasonably considered a
> > modern Internet MTA,'' it must store all mail messages in CRLF form.
> I don't recall making this statement.

You said that qmail ``has quite a ways to go to be reasonably considered
a modern Internet MTA.'' You then gave a list of seven items, of which
items 1 and 2 were that I ``should store messages in canonical RFC 822
format,'' by which you meant ``lines being CRLF-separated.''

> This is the definition that already exists in both RFC 822 and MIME.

RFC 822 doesn't adequately distinguish between content and encoding.
Anyway, even if that's the view taken in a few new RFCs, it's _wrong_.

A mail message is a sequence of lines. Users know this. Most commercial
UNIX mail software, including practically every OS vendor's built-in
mail software, uses this model. I object to any attempt to impose a
different definition.

---Dan