Errors-to: owner-drums@CS.UTK.EDU
Date: Thu, 1 Jun 1995 12:17:32 -0700 (PDT)
From: Mark Crispin <MRC@Panda.COM>
Sender: Mark Crispin <mrc@Ikkoku-Kan.Panda.COM>
Subject: re: address syntax
To: John Gardiner Myers <jgm+@CMU.EDU>
Cc: drums@CS.UTK.EDU
In-Reply-To: <4jnUhHK00WBw09xulD@andrew.cmu.edu>
Message-Id: <MailManager.802034252.19114.mrc@Ikkoku-Kan.Panda.COM>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII

On Thu,  1 Jun 1995 14:47:47 -0400 (EDT), John Gardiner Myers wrote:
> Things which are at the top of my hit-list are:
>
> 1) The ``# number'' construct in the element nonterminal of 821

Good idea.  This relic from 8-bit NCP addresses is mostly because nobody
really knew what addresses would look like back then.  Due to endian concerns,
it's ambiguous how this translates to an IP address anyway.

> 2) The use of domain-literal in the sub-domain nonterminal, as opposed
> to the domain nonterminal (e.g. "foo.[128.2.10.102].com") in 822 and
> corresponding constructs in 821.
> 3) The ability in 822 to use quoted-pairs inside of domain-literals

These too, will not be missed.  I don't think we'll ever use these
capabilities in real life.

Another great idea!  Why not go further and get rid of non-numerics inside of
domain-literals?

So why not something like:
     domain            ::= domain-name / domain-literal
     domain-name       ::= atom *("." atom)
     domain-literal    ::= "[" domain-octet *("." domain-octet) "]"
     domain-octet      ::= 1*DIGIT      ;; 0 <= value <= 255

Or just remove ".", "[", and "]" from specials, and then we could have:
     domain            ::= atom

Let some other (DNS?) specification decide what a domain-literal looks like.

> 4) The ability in 822 to use quoted-strings for individual words of a
> multiple word local-part, as opposed to a single quoted-string as a
> local-part.

Yeah.  Very few people handle this right.

> 5) The ability in 821 to use \-quoting inside a local-part

Uh, there's a problem with this.  \-quoting is the only way to get <">, "\",
and CR (but not LF!) inside a quoted-string.  Same thing for comments.  I
don't think it's possible to do this.

> I'm also wondering whether we should slap a "free insertion of
> whitespace and comments not allowed inside this nonterminal"
> requirement on domain.  Your suggestion for redefining domain happens
> to have this effect.

This is already a rule; see C.2.1 in RFC-822.