Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Message-Id: <v04101003b1ec021569a6@resnick2.qualcomm.com>
Date: Mon, 3 Aug 1998 20:54:02 -0500
To: drums@cs.utk.edu
From: Pete Resnick <presnick@qualcomm.com>
Subject: draft-ietf-drums-msg-fmt-05.txt and CR/LF

I've sent in draft-ietf-drums-msg-fmt-05.txt to Internet-Drafts. Hopefully
it should be posted soon. It is also available for those who want it early
at:

<http://www.eudora.com/~presnick/draft-ietf-drums-msg-fmt-05.txt>

or if you prefer FTP:

<ftp://resnick1.qualcomm.com/Public/drums/draft-ietf-drums-msg-fmt-05.txt>

There is one major issue that I did not address in the draft that was in
the minutes of the last meeting, but it was simply because I was having a
hard time reading consensus (probably because I think just about everyone
besides kre and djb may have tuned out): Bare CRs and LFs.

Here's my take of the two positions. In both cases, bare CRs and LFs are
considered "bad things" to be sending. Correct me if you think I
misrepresent here:

1. Keep bare CR and bare LF in the obs- (accept) syntax. That means that
you MUST accept bare CR and bare LF as regular old control characters when
they appear in a message. How you deal with these, or any other control
characters, is up to the receiving implementation, but there is an
expectation of "reasonable" handling.

2. Remove bare CR and bare LF from the obs- (accept) syntax. That means
that you don't have to accept bare CR or bare LF in a message. You can
effectively "do whatever you want" with a message that contains bare CR or
bare LF. Of course, as per the admonishment in -05, programs crashing or
messages being silently deleted is never on the list of "whatever you
want". However, the expection of "reasonable" handling is gone; the message
may end up bounced or transformed in some way, but is in any event outside
of the scope of a legal message.

The syntax in -05 should be correct so as to handle bare CRs and LFs in a
message; the syntax being wrong in -04 was what sparked the issue. If you
want them left in, you needn't say anything, but if you are opposed to bare
CRs and LFs in the accept syntax, please restate those objections so I can
figure out whether consensus is against the document now.

pr
--
Pete Resnick <mailto:presnick@qualcomm.com>
QUALCOMM Incorporated
Work: (217)337-6377 or (619)651-4478
Fax: (217)337-1980 or (619)651-1102