Making progress on the mapping question

Mark Davis mark at
Wed Apr 1 03:09:28 CEST 2009

We need to separate some very different cases. Take a string S, and look at
the results under each implementation. By A-Label or XN-Label below, I
mean*the IDNA2008 definitions.
1.* Identical. *IDNA2008 produces an A-Label, IDNA2003 produces an A-Label,
and they are the same. This is the best case.

Example: http://ö <>.

2. *Different. *IDNA2008 produces an A-Label, IDNA2003 produces an A-Label,
but they are different. This is the really bad case.

Example: क्‌ख.com <>
= क\u094D\u200Cख.com <>, with a ZWNJ

3. *Mappable. *IDNA2008 does not (old draft) produce an A-Label, IDNA2003
produces an A-Label. This is the mappable case, aka M-Label. People will
continue to expect these to work.

Example: Ö <>

4. *Forbidden.* IDNA2008 does not produce an A-Label, IDNA2003 produces an
XN-Label that isn't an A-Label. This is the "symbol" case.

Example: I♥ <>

I think implications of the rough consensus from the meeting is that:

   - Forbidden and Different are not part of protocol, but that we should
   explicitly describe a transition strategy for dealing with those cases. Note
   that only Different requires a double DNS lookup.
   - Mappable is a required part of lookup protocol, but forbidden in
   registration. That way we have a single, uniform, compatible mapping on the
   user side, and avoid misunderstandings in registration.
   - We have a SHOULD on only storing A-Labels or U-Labels.

That is a compromise position that appears to meet everyone's core
requirements, even where it is not the preferred position.


On Tue, Mar 31, 2009 at 08:15, Paul Hoffman <phoffman at> wrote:

> One of the things that a few people said at the mics last week was that we
> need to be much more specific about what we mean when each of us says
> "mapping".
> At 7:41 AM -0400 3/30/09, Vint Cerf wrote:
> >1. first look up under IDNA2008 rules
> >2. If a domain name is found, return the corresponding results
> >3. If a domain name is not fund, apply IDNA2003 mapping
> >4. If a domain name is found, return the results
> >5. If a domain name is not found, report that no such domain name exists
> Take i<heart> A fairly reasonable definition for
> "mapping" in step 3 is "follow the steps in RFC 3490 for ToASCII". If so, in
> step 4 that domain name will return positive.
> At 9:17 AM -0400 3/31/09, Vint Cerf wrote:
> >We are not going to revisit this for the Nth time. The WG long ago
> >concluded to drop these symbols from IDNA2008 and nothing has changed.
> That sounds like you emphatically do not want step 3 to return positive for
> i<heart> Thus, you (and each of us) need to give a
> more complete definition of "mapping".
> On a related note, Mark's proposal from yesterday defines mapping quite
> precisely, but does not follow the steps above because the mapping is done
> before what you have as step 1. So, the WG needs to decide both what it
> means by "mapping", and when that will be applied.
> _______________________________________________
> Idna-update mailing list
> Idna-update at
-------------- next part --------------
An HTML attachment was scrubbed...

More information about the Idna-update mailing list