To: drums@cs.utk.edu
From: josdb@xs4all.nl (Jos den Bekker)
Message-ID: <3265602a7a6960@xs4all.nl>
Subject: Re: 8bits in unstructured header fields
References: <19961016152739.1a1aae79.in@lacroix.wildbear.on.ca>
Reply-To: josdb@xs4all.nl
Organization: Seni Translation Services, Amsterdam
Date: Wed, 16 Oct 1996 23:22:26 +0200 (MET DST)
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

on Wed 16-10-96 15:29 Jack De Winter wrote:

>Doesn't the argument's in this argument center around how to make
>a current practice that is not covered acceptable?

I think not allowing 8bits in unstructured header fields while
allowing them in the body is a clear lapse, an omission, and not in
keeping with the spirit of RFC 822. The current practice, moreover,
has evolved because it is practical, logical and economical, not
because of some whim.

>I have learned
>from some conversations in the past that there are certain people
>who will always think they are right, no matter what everyone else
>says.

?

>However, assuming that such a thing is implemented, my concern would
>be making ti backwards compatible with older implementations?  I.e
>how could we include 8bit headers without breaking current systems
>that are not expecting ANY 8bit information in the headers.

Which systems would that be? Do you mean there are systems that do not
expect 8bits in the header, but do expect them in the body? I have
never heard of such systems. As far as I know, there are systems that
cannot handle 8bits and will therefore strip the 8th bit, but they do
so indiscriminately, of course, in the header AND in the body. Tough
luck for both sender and receiver, but nothing untoward happens. Then
there are systems that can handle 8bits, and those systems will just
pass them on where they occur in message parts that have no structural
meaning, that are not parsed by the "lexical analyzer", i.e. the body
and unstructured header fields. What's the problem?

The only problem would be if transporting agents started parsing
unstructured header fields, like "Subject:" and "Keywords:" and
comments in general. But that is a violation of RFC 822. And anyway,
who on earth would want to do that? And why?

If 8bits are allowed in the body of messages, they should be allowed
in unstructured header fields as well. If they are forbidden in
unstructured header fields, it is a ludicrous inconsistency to allow
them in the body.

I know of no versions of sendmail and inews that cannot process
messages with 8bits in unstructured header fields properly. But maybe
someone else does.

-- 
 Jos den Bekker
<josdb@xs4all.nl>     http://www.xs4all.nl/~josdb
                              -*-
  "He who can, does. He who cannot, teaches." - George Bernard Shaw