Bundling vs Mapping

Mark Davis mark at macchiato.com
Wed Feb 25 19:12:15 CET 2009


Good questions. My take below.

Mark


On Tue, Feb 24, 2009 at 22:13, Patrik Fältström <patrik at frobbit.se> wrote:

> On 25 feb 2009, at 00.26, Mark Davis wrote:
>
>   1. That is, unless a local mapping is added.
>>  2. That means, effectively, that every reasonable implementation will
>>  need to add a local mapping so as to not have disgruntled users.
>>  3. But there is no standard mapping anymore as part of the protocol.
>>  4. That means that different implementations can use different mappings.
>>  5. Given no guidance, we know that they will.
>>  6. That means we will end up with a very significant interoperability and
>>  security problem.
>>
>
> This I agree with of course, but after separating the domain name from the
> mapping from the application specific implementation suggestions, the next
> questions arrive:
>
> 1. Does IDNA2008 documents have to be blocked until the rest of the "full
> picture" is ready as documents?


Because of the major interoperability issues in the transition from IDNA2003
to the draft IDNA2008, we have to have all of our ducks in a row before
release, so "yes".


>
> 2. Does the IDNA2008 documents include "enough" text about this? if not,
> what should be added where?


No. By allowing allowing arbitrary local mappings, it leaves
interoperability in a muddle. It's not a matter of just changing explanatory
text.


>
> 3. How much of the mapping can and should be the same regardless of the
> application(s)?


All. If everyone doesn't map the same way, we are going to have a horrible
mess.


>
> 4. Why has not work continued on the pre-i-d that Mark worked on, should
> that work continue?


The indications that we have gotten all along is that that the authors of
IDNA2008 were not interested in that. We have continued that effort in the
Unicode Consortium, so as to at least ameliorate the situation by having a
specification that people can apply uniformly. It would, of course, be
better if it were part of IDNA2008.


> 5. What pieces (ranging from all to none) of the UTC work in this area
> should be adopted by IETF?


I believe that all should be. That would meet the main goals of IDNA2008,
but also provide the necessary interoperability with IDNA2003.

>
> 6. What mappings (and similar) do applications already do today?


Today, I have not heard of any significant implementation of IDNA that
doesn't apply the mappings defined by IDNA2003, except for two variations
that are widely implemented:

   - Updating the normalization mappings to the latest version of Unicode
   (which, as you know, do not have any significant effect).
   - Applying the mapping to the entire domain name (only affects domain
   names with some very unusual characters, like
U+2488<http://unicode.org/cldr/utility/character.jsp?a=2488>( ⒈ )
DIGIT ONE FULL STOP).

These are documented in http://www.unicode.org/reports/tr46


>
>
> Etc...
>
>   Patrik
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.alvestrand.no/pipermail/idna-update/attachments/20090225/42c3c4a5/attachment.htm 


More information about the Idna-update mailing list