Re: [idn] Re: hostname history hell

["Eric A. Hall" <ehall@ehsco.com>]

Eric Brunner-Williams in Portland Maine wrote:

> >Does the example override the rule?
> 
> Of course not! All dns implementations that resolve a.isi.edu are
> broken!!!

Well, resolving it is something else. It is a valid domain name and is
certainly valid for lookups.

Being conservative in what gets generated: it is not valid for assignment
even if applications allow it to be used as a host identifier. Being
liberal in what gets accepted: it is not valid but it resolves so we will
use it. Applied globally: it is not valid and should not be used, and apps
that choose to ignore the rules do so at some risk (maybe not in this
particular case but certainly when it comes to IDNs there will be a large
amount of risk from ignoring the rules).

We do not have to stick with the two-char rule for i18n hostnames. We do
not have to stick with the no-leading/trailing-hyphen rule either.

-- 
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/