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

["Adam M. Costello" <amc@cs.berkeley.edu>]

Patrik Fältström <paf@cisco.com> wrote:

> 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.

I don't recommend trying to talk about how the names are represented
internally.  Perhaps the idea you want to express is this:  An ASCII
query Q matches a name N iff Q matches ace(nameprep(N)) using a
case-insensitive ASCII comparison.  One obvious (but not required)
implementation strategy is for the server to precompute ace(nameprep(N))
for each name N when it starts up, and then act just like an old
ASCII-only DNS server.  Another strategy is to use a separate
preprocessor to perform ace(nameprep(N)) on the master file, and
actually use an old ASCII-only DNS server.

Keith Moore <moore@cs.utk.edu> wrote:

> perhaps we should state things in terms of "exchange format" and
> "local format".
>
> "exchange format" is the format used to exchange zones in a flat file;
> DNS implementations should be able to dump and accept zones in that
> format. for the time being, it should use ACE, for compatibility with
> old DNS implementations.
>
> "local format" can be anything the DNS implementation wants (web form
> input, XML, S-expressions, Morse code, whatever) and the standard
> shouldn't place any restrictions on it.

I like this.

AMC