Re: [idn] Suggested clarifications of the IDN requirements doc

[Harald Alvestrand <Harald@Alvestrand.no>]

At 11:12 09/08/2000 -0400, RJ Atkinson wrote:
>          The proposed new text above is far too limiting.

I was hoping to probe the consensus on that.

>
>
>          For example, it is also important to also be able to map an IDN
>to an MB record (for key management reasons; see RFC-2230; RFC-2230
>is supported in BIND and in more than 1 IPsec implementation, btw).
>
>          For IDN to be really useful, it needs to fully support all
>current DNS operations and mappings.  If we don't require this,
>the net result might be that IDN names are some sort of 2nd-class
>component of the overall DNS, whereas we need for IDN names to be normal,
>universally supported, 1st class components of the overall DNS.

I am not sure about that. An architecture that treated IDN as a new layer 
of abstraction on top of DNS names has some attractive features.

>          We should retain the original text, except that it would be
>prudent to add to the original text this one sentence from Harald:
>
>          "The protocol MUST NOT allow an IDN to be returned to a requestor
>          that requests the IP-to-(old)-domain-name mapping service."
>
>That one sentence helps clarify the requirement that we don't break
>existing deployed systems when IDN is deployed.

I wonder: do we really have the requirement to map from IP address to IDN name?

This requirement is particularly complex to meet, I think; if we require to 
map both to meaningful IDN names and meaningful hostnames, we invite 
increased complexity in this space.

               Harald

--
Harald Tveit Alvestrand, alvestrand@cisco.com
+47 41 44 29 94
Personal email: Harald@Alvestrand.no