Re: [dnsext] URI RRTYPE review - Comments period end Aug 15th

[Patrik Fältström <[e-mail]>]

I am working on an update to this document given the comments on the mailing list.

For this issue, IRI, I have think the best thing is to allow one of:

- A URI
- An IRI but only using UTF-8 charset, and that the IRI is such that it can be converted to a URI and back again.

A new version that reflects this change will be released shortly.

   Patrik

On 26 jul 2010, at 02.43, Phillip Hallam-Baker wrote:

> Use the same RR for both.
> 
> As far as the infrastructure is concerned there is absolutely no
> difference. DNS does not need to know the semantics of the bits being
> shipped.
> 
> IRIs are going to have to fit into slots where URLs are used. In most
> cases the application will not know which is being used.
> 
> The justification for IRIs is that they are friendlier for human use.
> This is another technology meant to make things easier for human use.
> There is really no reason that the target should be another human
> friendly format. But there is no particular harm either. Provided that
> the target is something that the app can just use as a blind chunk of
> bits.
> 
> 
> 2010/7/25 Masataka Ohta <[e-mail]>:
>> Klaus Malorny wrote:
>> 
>>> dumb question: I don't consider myself an expert in that respect, but
>>> wouldn't it make more sense nowadays to use UTF-8 encoded IRIs (RFC
>>> 3987) instead of URIs? Or is there a need to keep everything ASCII-safe?
>> 
>> UTF-8 is causing a lot of problems, some of which was predicted
>> but some are new, with CJK unification and will never be stable.
>> 
>> Patrik;
>> 
>>> But I take on the task to check with the IRI people really fast,
>> 
>> They are the wrong people to ask, because only the positive
>> answers are expected.
>> 
>>                                                Masataka Ohta
>> 
>> 
> 
> 
> 
> -- 
> Website: http://hallambaker.com/
>