Message-Id: <v0311070bb03b45df21b9@[130.237.158.12]>
In-Reply-To: <6dU$GEUUcsB@khms.westfalen.de>
References: <199709072145.RAA01085@spot.cs.utk.edu>
 <199709072145.RAA01085@spot.cs.utk.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Wed, 10 Sep 1997 03:53:08 +0200
To: kai@khms.westfalen.de (Kai Henningsen), drums@cs.utk.edu
From: Jacob Palme <jpalme@dsv.su.se>
Subject: Re: Incorrect use of the word SHOULD in relation to Message-ID

At 09.12 +0200 97-09-08, Kai Henningsen wrote:
> The *only* reason I can see to make this SHOULD instead of MUST is
> compatibility with 822. OTOH, IMHO, this could just as well be done by
> saying MUST generate, but also MUST cope if missing - that's the way we
> handle all the other differences to 822.

This is true. Even if it is possible to construct other arguments
for SHOULD instead of MUST, the real reason is probbably why
it is a SHOULD is to keep compatibility with old software, coded
according to RFC822 which did not have this MUST.

The standard hast to say that messages without Message-ID
are allowed on certain exceptional circumstances.

What I was discussing was whether the use of the word SHOULD is correct,
at least if we use definition of SHOULD in RFC2119. That definition
does not allow the word SHOULD for things which we only allow to
keep compatibility with old software.

I suggest we use words like OBSOLETE or DEPRECATED when we talk about
messages without Message-ID. In that way, we more clearly say that
a conservative generator always generates them, but a liberal recipient
is willing to accept messages without them.

------------------------------------------------------------------------
Jacob Palme <jpalme@dsv.su.se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/~jpalme