Re: [idn] New protocol proposal: IDNRA

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

At 12:31 AM +0800 8/28/00, James Seng wrote:
>"As RACE could either be rendered in ASCII or I18N glyphs, rendering
>engine for RACE SHOULD have an option for user to select the preferred
>display."

Something like this would be useful. We'll add it.

>  > This can be discussed further, and the best API should be chosen. The
>>  important issue for _this_ draft is that the API used is a different
>>  than what is currently used.
>
>My take that if the API is in UTF-8 as compared to UCS-4, then
>
>Application takes native, convert to Unicode, convert it to UTF-8 to
>pass onto Resolver. Resolver then take UTF-8, convert it to Unicode, do
>all its funky nameprep, and convert it to RACE.
>
>The additional processing overhead of converting to and from UTF-8 could
>be removed if we were to use UCS-4 directly. And that seem to be the
>common API interface.
>
>But I may divert too much into the implementation here...

I don't think this is a diversion. It is quite appropriate for the WG 
to say to us "you must add an API to this" or "it's OK to not have 
the API here". And, if it is the former case, it would be very useful 
for real implementors to tell us what they would and would not like 
to see in the API. I chose UTF-8 for the suggested format because it 
is quite popular on this list. It may or may not be popular for 
implementors working in today's systems who would have to write 
resovlers. Comments are appreciated here.

>  > I don't feel that is needed as in the picture you see in 2.1, there
>>  is no difference between the different kind of DNS servers. I.e. as
>>  soon as you leave the resolver and use the DNS protocol, you use ONLY
>>  RACE encoded domainnames.
>
>Erm. That's true...but would be nice to complete the whole picture so
>there is a clear statement that caching server is involved in the
>process but it doesnt really matter...

I disagree. We are using the same picture that the IDN Requirements 
document uses, which makes all of that a single unit. I don't want to 
call out any part of the unit and say "it's like the rest of the 
unit."

--Paul Hoffman, Director
--Internet Mail Consortium