To: "Randall C. Gellens" <RANDY@MPA15AB.MV.UNISYS.COM>
Cc: drums@cs.utk.edu
Subject: Re: smtpupd-02 
In-Reply-To: Your message of "Tue, 06 Aug 1996 14:30:04 MST."
             <EJHQ3003F6A6FA@MPA15AB> 
Date: Wed, 07 Aug 1996 08:47:48 +1000
Message-Id: <9585.839371668@munnari.OZ.AU>
From: Robert Elz <kre@munnari.oz.au>

    Date:        Tue, 6 Aug 1996 14:30:04 -0700 (PDT) (21:30 UT)
    From:        "Randall C. Gellens"  <RANDY@MPA15AB.MV.UNISYS.COM>
    Message-ID:  <EJHQ3003F6A6FA@MPA15AB>

    I can see restricting the characters that can make up a domain name,
    but I don't see the point in restricting the domain name syntax past
    what is needed to keep parsing reasonable.

This is reasonable.

    and I think the domain name should be the business of the DNS.
    If DNS can return MX records for something, why should we care
    what that something is?

But this isn't - the DNS can return MX records for anything, if we were
to allow anything the DNS would allow, then 822 would need to be able
to parse ...

	| Column 1 is under the 'v'
	v

	Date: (whatever it is)
	From: me@wherever.i.am
	To: user@some.domain

	ends.uni.edu
	Message-Id: <something@somewhere.or.other>


Note that the destination address is (encoded a little in C style)

	<user@some.domain\r\n\r\nends.uni.edu>

Perfectly valid DNS name, a bitch to parse in 822 (it would need to
be quoted, and would never be compatible with anything that now exists)

    By the way, was there ever any agreement on the syntax for IPv6 domain
    literals?

I don't recall any.

kre