Re: [EAI] UTF-8 in Message-IDs

["Charles Lindsey" <[e-mail]>]

On Tue, 23 Aug 2011 15:42:40 +0100, John C Klensin <[e-mail]> wrote:

> To strengthen the distinction, one would need actually need a
> lot more than "truly pure UTF-8 environment".  Identifier
> comparison in a non-ASCII environment is actually quite
> difficult in the general case, requiring a lot of additional
> comparison rules and/or restrictions (see RFC 6055 for a partial
> discussion).  While we take it for granted these days, that is
> even true in the ASCII environment because we have to specify
> choices between case-dependent and case-independent comparison.
> But, in the DNS case, even if we substituted "pure UTF-8" for
> the IETF-invented Punycode encoding, we would still need fairly
> extensive rules about permitted characters, normalization, etc.,
> to make things work -- and even more extensive rules if any sort
> of mapping (or non-exact comparison) was permitted.

All of which is exactly why I oppose allowing UTF-8 in Message-IDs (which  
get treated like identifiers in many situations). Fine if the job is done  
properly by specifying the normalizations etc, but not otherwise.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: [phone]                       
   Web: http://www.cs.man.ac.uk/~chl
Email: [e-mail]      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5