referencing IDNA2008 (and IDNA2003?)

[Andrew Sullivan]

On Thu, Oct 21, 2010 at 03:05:32PM -0600, Peter Saint-Andre wrote:

> In which case, perhaps all we really need is text to the effect that
> "when you're comparing two domain names (e.g., comparing a
> domain-attribute against a request-host), you need to do so in a way
> that ensures that the two strings really are equivalent; either IDNA2003
> or IDNA2008 will serve this purpose, and which approach to use is a
> matter of implementation".

Why not just say that you should use whatever IDNA rules the
application implements, following all the rules for domain comparison
in that implementation?  (Note that IDNA2003 domain name comparison
turned out to be slightly troublesome, because people sometimes
compared what we came to call U-labels and other times compared what
we came to call A-labels.  But like someone else said up-thread, the
implementation is already relying on whatever it is relying on.)

A

-- 
Andrew Sullivan
ajs at shinkuro.com
Shinkuro, Inc.