Re: 8-bit transmission in NNTP

[Mark Crispin <MRC@xxxxxxxxx>]

On Fri, 09 Sep 94 15:45:15 -0700, Erik E. Fair wrote:
> As one of the original authors of NNTP, and as a participant in the
> MIME effort, I can authoritatively say that NNTP was indeed explicitly
> 7-bit NETASCII, and anyone just sending 8 bits is in violation of
> protocol.

This is the statement I wanted.

But just to be sure that I understand, is it correct that this statement also
means that it is a violation of the NNTP protocol to send 8-bit data that is
character-set labelled with MIME?  In other words,
	Content-Type: TEXT/PLAIN;CHARSET=ISO-8859-1
	Content-Transfer-Encoding: 8BIT

If we were to send 8-bit data, this is what we would do.  This isn't ``just
send 8-bits''.  But, I think that this behavior is presently banned.  Correct?

> The whole idea behind MIME was to get the user interface
> software changed to properly support character sets other than ASCII
> *independent* of transport and it does no damn good if we don't label
> the silly things which is where the "just send 8 bits" idiocy leaves
> us. This goes for NetNews too.

Agreed.  Death to non-MIME UAs.

On Fri, 9 Sep 1994 19:03:24 -0500, Steve Dorner wrote:
> Not fair enough.  The "idiocy" is for Internet software to reserve a bit
> for parity.

Steve, this is a false and misleading statement and you know it.  There are
enough of us old farts who remember quite well why Internet text is 7 bits.
That reason has nothing whatsoever to do with parity.

> The claim of the 8-bit NNTP proponents is that the NNTP infrastructure is
> already 8-bit clean, and so there's no need to be so pessimistic in our
> assumptions.  We can assume 8-bit clean, unless told otherwise.

The entire issue is whether or not this claim is true.  Please stick to this
issue and not to strawmen.  The purpose of my message was to test the claim.
Erik's message, and certain NNTP documents, imply that this claim is false.

> If in fact the NNTP infrastructure is 8-bit clean (I am not qualified to
> comment on that), then doesn't it make good sense to declare the protocol
> 8-bit clean, too, and avoid universally hated stuff like quoted-printable?

Since I only use MIME UAs, I never see quoted-printable; so I don't hate it.
What I hate is receiving a message which is unreadable because some stupid MUA
decided to ``just send 8-bits'' over a non-8-bit clean path.