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

[Paul Hoffman / IMC <phoffman@imc.org>]

At 2:49 PM -0500 7/15/01, Eric A. Hall wrote:
>What severe problems?

If any DNS server in the search path for a name has not been updated, 
or if the user's resolver has not been updated, the query will fail. 
That seems pretty bad, yes?

>Is the need to upgrade ACE clients (and servers, unless you consider
>manually editing ACE gobbledygook in zone files to be a wonderful feature)
>also a severe problem?

No. Applications will need to be upgraded no matter what. Further, 
DNS servers do *not* need to be upgraded: the editing interface to 
the host file does. That is pretty trivial: a single script that 
converts the native-charset file (which does not have to be UTF-8: it 
can be any native charset) to the host file. It took me about 5 
minutes to write mine in Perl (using an existing ACE<=>UTF-8 routine).

>  > Have you now shifted
>>  from "ACE then UTF-8" to "never ACE"?
>
>For new applications, there is absolutely the possibility that they will
>be able to only specify the use of UTF8.

One more time: we are talking about protocols, not applications. And, 
even then "able to" is completely different than "needs to". Where is 
the need here? How will balkanizing the DNS help end users?

>  This is a good thing, as new
>applications can be written for international markets from the beginning,
>and UTF8 is becoming very common. UTF8-in-DNS is an enabler. Furthermore I
>would argue that it is absolutely critical to the next twenty years of
>protocol work.

Because...?

You still haven't given any examples of why it is needed, only one 
where it would make writing an application only slightly easier, but 
only if we already had UTF-8 names. Please reconsider your arguments 
about an upgrade path.

--Paul Hoffman, Director
--Internet Mail Consortium