tables document [Re: IDNA comments]

John C Klensin klensin at
Tue Jul 15 17:05:56 CEST 2008

--On Tuesday, 15 July, 2008 15:51 +0200 JFC Morfin
<jefsey at> wrote:

> At 04:11 15/07/2008, Kenneth Whistler wrote:
>> Registration of *domain names* is under the purview of the
>> Registry Managers.
> The decisions concerning the ccTLD tables are under the
> purview of  the Registry Manager.

Just to be absolutely clear before we descend into another rat
hole (I don't believe that the rest of the note, including the
stated copyright concerns, are relevant right now)...

Throughout the IDNA2008 documents, when terms like "Registry" or
"Registry Operator" are used, they are used interchangeably with
"Zone administrator".  Those terms apply to _any_ zone in the
DNS, at any level. 

That nomenclature goes very far back in the history of the DNS,
far enough back that, if anyone had been able to conceive of
ICANN, the conception probably would have been considered as a
particularly severe nightmare.

Because of the way the DNS is organized, every zone
administrator has to make decisions about what sorts of strings
to permit in labels and what strings to not permit.  I make
those decisions for JCK.COM, you presumably make them for
JEFSEY.COM, Harald makes them for ALVESTRAND.NO, and someone
makes them for SYBASE.COM (I suspect it isn't Ken, but neither
know nor care).   

Sometimes those decisions are constrained externally: by ICANN
with respect to a handful of TLDs, by governments for some TLDs
and elsewhere (anti-obscenity laws in some countries may prevent
the use of some strings even fairly deep in the DNS tree, or at
least make them illegal), by corporate policies, or by the
implications of infringing other people's trademarks.  And
sometimes they are constrained by protocol considerations, not
of the DNS (or IDNA) protocols but by the fact that if, for
example, one wants to use a particular string with email and
another for service location, those protocols impose some limits
on what can be in those strings.  In some sense, IDNA can be
seen as a way to remove the external, non-DNS-protocol,
constraints that prevented the use of non-ASCII strings that
were intended to be interpreted as parts of domain names.

We have also always understood how a zone administrator at level
N of the DNS can impose its will about external constraints on
zone administrators at level N+1 and below.  It simply makes
registration conditional on obeying whatever rules it lays out
and declines to create the delegation (or withdraws it) if those
rules are not agreed to or followed.   In practice, there are
often a lot of problems with that theory, but they are not
protocol problems, or problems for this WG or the IETF more

None of those external constraints, or how they are defined and
administered, are part of the agenda of this WG.  As soon as you
start talking about ccTLDs as compared to other types of TLDs or
domains much lower in the tree, or about how policies are
propagated, you are off-topic or simply not understanding what
is going on here at best.

As I have said many times before, I'm in favor of
voluntary-compliance "best practices" recommendations that
describe what ideas are good, what are bad, why, and how to tell
the difference.  But none of those are part of this WG's charter


More information about the Idna-update mailing list