AW: Sharp-S and Final Sigma Consensus Call Results

John C Klensin klensin at
Wed Dec 9 22:12:52 CET 2009

--On Wednesday, December 09, 2009 16:02 -0500 Andrew Sullivan
<ajs at> wrote:

> On Wed, Dec 09, 2009 at 03:55:48PM -0500, Vint Cerf wrote:
>> wouldn't mapping of pvalid chars would disrupt the
>> A-Label/U-label   equivalence?
> Well, what do you mean by "mapping"?
> Suppose ß were PVALID.  If an application always took "ß" and
> converted it to "ss" before handing it to the IDNA subsystem,
> how is the U-label spelled?  With ß or ss?  (I can make an
> argument either way, I think.)

The argument against remapping PVALID characters goes something
like this.

Suppose that, in an application, I decided to always map the
ASCII "k" to Greek "κ" (I vaguely remember, from some years
ago, a proposal to do just that).  Now, because your
organization has registered the ASCII string "" but
not the IDN "shinκ", no user of that application is
going to be able to and resolve your domain as intended.   I
don't know how to pass a rule against that which would be
somehow enforceable, but I think it should be clear that it
isn't a really good idea.

More generally, if ISPs or regulations in Upper Slobbovia can
succeed in blocking DNS access to all domain names based the
Lower Slobbovian language by requiring that applications map key
characters in the latter language to nonsense, I suggest that is
a problem for the concept of a global Internet.  Nothing they
couldn't do in other ways, of course, but not something we
should encourage, either.


More information about the Idna-update mailing list