Message-Id: <v0310060eae8c183bb2f5@resnick1.isdn.uiuc.edu>
In-Reply-To: <326665724f925a0@xs4all.nl>
References: <199610171347.JAA25040@ig.cs.utk.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Thu, 17 Oct 1996 12:31:22 -0500
To: Jos den Bekker <josdb@xs4all.nl>
From: Pete Resnick <presnick@qualcomm.com>
Subject: Re: 8bits in unstructured header fields
Cc: drums@cs.utk.edu

On 10/17/96 at 10:43 AM -0500, Jos den Bekker wrote:

>on Thu 17-10-96 09:47 Keith Moore wrote:
>
>>The DRUMS charter says 'New functionality is expressly inappropriate'
>>to keep the group from getting into such ratholes.
>
>The charter also says:
[...]
>So, really, I don't understand.

I think I can make this clearer:

Multiple private agreements are not acceptable to put in a standard,
especially if they conflict. Though most of Europe uses 8859-1 as the
interpretation of 8bit data in headers, that is not true of other places in
the world. You might want to write an informational document which says
"Most of the world uses 8859-1 as the interpretation of 8bit in headers",
but I don't think this is something that DRUMS needs to be involved in.

The only acceptable practice that we have come across with regard to
character sets is labeling. That's why MIME has charset labels.

It's possible to use 1522 labeling of headers to specify character set. One
solution to your problem would be to require 1522 labels for character set,
but introduce a new encoding for 8bit (as opposed to QP or Base64) with
1522 labels. Introducing a new encoding for 1522 labels is clearly out of
the scope of DRUMS, since the implications of such a mechanism are not well
understood.

Now, you may (like me) think that 1522 labels are ugly. Perhaps there is
another way to label the charset of headers. Perhaps we could define a
"Header-8bit-charset:" header which was required to appear before all 8bit
headers and indicated the default character set to interpret 8bit
characters in headers. Again, this would be a new mechanism and clearly out
of the scope of DRUMS.

I personally think (and others would likely disagree with me) that it would
be a fine idea to allow 8bit headers without 1522 labeling, though perhaps
I would require ESMTP 8BITMIME and some additional header to specify
character set. But I think this is far too much change to do under the
auspices of DRUMS.

pr

--
Pete Resnick <mailto:presnick@qualcomm.com>
QUALCOMM Incorporated
Work: (217)337-6377 / Fax: (217)337-1980