Re: [idn] Why follow IDNA with UTF-8?

[Patrik Fältström <paf@cisco.com>]

--On 01-07-16 01.50 +0000 "Adam M. Costello" <amc@cs.berkeley.edu> wrote:

> Martin Duerst <duerst@w3.org> wrote:
> 
>> But the current IDNA draft explicitly specifies that zone files must
>> be in ACE, and therefore prohibits such implementations.
> 
> Well that's just silly.  Paul and Patrick, the IDNA spec should not say
> that master files must use ACE, it should say that DNS servers must
> accept master files that use ACE and may also accept master files that
> use other encodings, like UTF-8.

If that is in the text (I don't have it in front of me) yes, that is silly
and should of course be changed. What charset the user uses when entering
data into his favourite DNS server is up to the implementation.

What Paul and I want to say is that the zone data (in some meta-way) is ACE
encoded in the master server, and the master server operates on that data.
How the data is entered, converted or whatever is out of scope. The same
would be true for UTF-8 by the way given the ability for users to have
editors and operating systems which doesn't use _NAMEPREPPED_ UTF-8
natively in files which later are read by the DNS server.

And, regarding UTF-8 or ACE, the most important thing is the nameprep and
not what encoding the labels uses. I am sorry to say that I am extremely
worried that if we go with an encoding that is implemented (like UTF-8)
people will not do nameprep.

   paf