Re: Fwd: Eeek - .NU Domains using

["Martin J. Duerst" <duerst@w3.org>]

At 08:09 00/01/10 +0100, Harald Tveit Alvestrand wrote:
> At 17:29 09.01.00 -0800, Bill Manning wrote:
> 
> >I'd be strongly in favor of not mixing glyphs between sets.
> 
> but we can't tell whether we are doing this without defining what a "set" 
> is, which gets us started down the implementation slippery slope.
> 
> We can say that "there is no requirement for mixing A with B inside a 
> single label", where A and B are different subsets of the set of all 
> characters.
> Giving a few examples is probably Good Enough for the requirements.

Yes, I guess. Giving examples where we know that mixing in the
same label is a requirement is also very helpful.

> A requirement that it should be illegal to do such mixing can be read as 2 
> things:
> - A technical requirement on the solution that it should be easy to identify
>    the subset a character comes from, and either prohibit the mixing or say
>    that there is no requirement on implementations that they support this
> - An administrative requirement that no such label should be registered.
> 
> The first one belongs in this group, IMHO. The last one doesn't.

Specifying the administrative requirement as such is not the job
of this group. However, I think it is possible that we discover
in the solution stage that it is better to deal with certain issues
on the client side or on the registration side, rather than in the
protocol itself. In such cases, I would expect that we:
- Document an example recommendation for administrative requirements.
- Make sure that there are some implementations that can do the
  checks we propose, even if only as a proof of concept.

That would still leave the registration guys free to decide whether
they want to adopt this or not, but we couldn't be blamed for dumping
part of our requirements onto other parts of the infrastructure
without having done our homework.


Regards,   Martin.





#-#-#  Martin J. Du"rst, World Wide Web Consortium
#-#-#  mailto:duerst@w3.org   http://www.w3.org