🚀 go-pugleaf

In article <89RsVB3w165w@blues.kk.sub.org> kosta@blues.kk.sub.org (Kosta Kostis) writes:

>>   I *KNOW* of at least 3 systems (combinations of OS + hardware) that
>> will flat out crash if offered certain fairly common 8-bit
>> ISO-8859-like characters.  I'm not going to say which ones in public
>> because some lurker would go and use this to crash someone else's
>> machine maliciously.

>I have never said one can do that *without* changing some software
>at some places. So, what's the "nonsense" in here?

  The systems are mostly orphans.  In some cases, the vendor doesn't
even exist anymore.  The software effectively cannot be changed.  Such
systems are a large part of the reason that the DDN has had an awful
time trying to convert to the Domain Name System.  Heck some of the
older systems from still existing vendors used 6-bit characters rather
than 7-bit ASCII.  Ever programmed on a Cyber ?  Mayhaps not.  However
many such systems won't go away or be replaced soon for financial and
other reasons.

>If a system is not capable of storing and forwarding 8-bit octets,
>why is it connected to the Internet? Maybe we should simply not route
>anything over those systems. Old hardware got to be replaced someday
>and should it be only because of the energy it consumes...  ;-)

  The Internet does not require that any machine use any particular
internal word size or character size.  No RFC requires the use of
8-bit characters and no standards-track RFC is likely to require that
soon, if ever.

  Because of dynamic routing, the vagaries of MX handling, and other
factors, it isn't really possibly to prevent mail traffic from being
misrouted via any particular host.  This is why the ESMTP draft on
8-bit extensions REQUIRES that the sending host do a handshake to
verify that the receiving host says that it can handle 8-bit BEFORE
any 8-bit gets sent.  News uses a flooding algorithm for article
propogation and so there isn't really much user control over routing.

>>   The second is to have a standard for marking the contents of
>> messages in standard headers using standard syntax and standard
>> tokens.  If I understand Henry correctly, he is putting together a
>> draft, based on MIME, that addresses this second issue.

>Didn't MIME specify some encodings already? And wasn't ISO 8859-1
>the default for 8bit encoded text? (hm...)

MIME only talked about Internet email, not news.  Henry's draft will
apparently talk about news issues and so forth.