Re: [idn] Re: Just send UTF-8 with nameprep

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

> > Using these IDNs to answer legacy queries will absolutely
> > definitely cause some kind of unintended consequence, and we
> > will get a lot of heat for the failure.
> 
> The chance that when a legacy resolver passes through a local encoding
> and this matches with a UTF-8 entry is extremely low. The reason for
> this is that UTF-8 has a very peculiar byte pattern.

I agree that it is low probability. However, for charsets like MacRoman
and Windows-1252 that reuse the C1 codepoint space, there is a measurable
probability of it happening.

IMO, if we are to take a risk, we should simply say that any label which
contains eight-bit characters must be processed through the IDN path, and
that any eight-bit queries which arrive via the legacy path are to be
rejected (we may also want a new RCODE to signify the exact failure).
Otherwise we end up in a situation where freak labels have to be compared
to the UTF-8 profile and then processed or rejected based on match level.

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