Re: [idn] Re: An idn protocol for consideration in making the req uirements

[Mark.Andrews@nominum.com]

	The problem is not that the servers will die, not that many will
	anyway as it has always been a requirement that they cope with
	8 bits in labels.  Upgrading servers, if required, is the easy part.

	What is hard is fixing gethostbyaddr().  Gethostbyaddr() quite
	correctly validates its input today and rejects non RFC-592 /
	RFC-1123 names.

	Similarly gethostbyname().

	Changing the C library involves dealing with a lot of inertia.
	Schemes which augment the C library rather than modifying it
	are much likely to succeed.

	Mark
> 
> I agree with Martin on this.
> 
> Somebody mentioned cultural sensitivity in this (sub)matter.
> Well, for those of you  that have not suffered through about
> a decade of "Q-P" I can tell you that 1) it is culturally
> very offensive, and 2) it is still leaking through in raw
> even in the most advanced e-mail clients now and then.
> 
> No scheme that reencodes non-ASCII into ASCII will ever
> be culturally acceptable, since it will, I emphasise: WILL,
> leak trough to end users, for all future use, as ASCII text.
> 
> And still no-one has even made it plausible that some DNS
> servers will 'fall over and die' if presented with 8-bit
> data.  And if some DNS servers do, then they are so
> vulnarable to attack that they should be upgraded or
> decommissioned ASAP anyway.
> 
> 
> 		Kind regards
> 		/kent k
--
Mark Andrews, Nominum Inc. / Internet Software Consortium
1 Seymour St., Dundas Valley, NSW 2117, Australia
PHONE: +61 2 9871 4742                 INTERNET: Mark.Andrews@nominum.com