Re: Mississippi Hißes

[Vint Cerf]

Alex,

I think your point is where many of us, including me, have ended up.

I believe the WG has gone to great lengths to explore various ways in  
which
our two problem characters can be introduced as PVALID. By allowing them
to be PVALID in protocol, we do not demand that every registry introduce
them (every registry is free to set its registration policies and some  
may
reject registrations including them). If they do not allow them to be  
registered,
and if some applications that use DNS are applying IDNA2003, then
some look ups will make it appear that sharp-S HAS been registered  
because
of the conversion to "ss" (assuming a target using the "ss" form  
actually
exists).

It is that kind of peculiarity that led the WG towards the conclusion  
that
A-labels and U-labels should be symmetric and that no mapping should
be permitted during the registration process.

I believe that the WG has provided a reasonable basis for moving
forward, both through the MAPPING advisory document and the
informal description of a transition tactic that allows registries to
proceed independently (which they will in practical terms in any case).

A transition plan that requires flag days (must be converted by X)
seems unrealistic. We cannot force or even check whether the flag
day has been met at all levels of registration (ie zone administration).

I may be in the minority but I believe that the "least harm" transition
example is reasonable but I also don't believe that we are in a position
to dictate in practical terms.

The offer made by Cary Karp on behalf of ICANN to draw the
question of IDNA2008 introduction into its IDN Guidelines work
strikes me as a very reasonable way to move forward. A much
broader set of parties affected by IDNs will participate in that
work and I believe our outputs are in satisfactory condition to
be taken up in that process.

Vint


On Dec 18, 2009, at 5:30 AM, Alexander Mayrhofer wrote:

>
>
>> these are good operational points. Ultimately, I think your line of
>> reasoning leads us to recognize that if we make previously mapped
>> characters PVALID, we must think of their use in registrations as
>> functionally independent. If registrant wants them to exhibit the
>> earlier mapped effect, the registrant will need to make their
>> independent resolution achieve this equivalence for him. This
>> suggests
>> to me that the registration process and the resolution
>> process should
>> be seen as very distinct. While we might adopt rules that operate
>> during registration to give the effect of "bundling" (essentially
>> assuring the registrant of getting all bundled forms), once this
>> registration occurs, it is up to the registrant to manage the set of
>> associated domain names to achieve equivalence, if that is desired.
>
> Right. Plus, the decision on how to handle the transition phase in  
> the registration phase must be left to the respective registry.  
> Registries are the organizations who have the best contact to their  
> local customer base, constituency and registrars - and i don't think  
> that a "one size fits all" policy is possible here, not even for the  
> singular case of the "ss/ß" transition.
>
> Alex