Q1 is mapping on lookup permanent or transitional?

[John C Klensin]

--On Thursday, April 02, 2009 11:35 -0400 Andrew Sullivan
<ajs at shinkuro.com> wrote:

> On Tue, Mar 31, 2009 at 12:05:59PM -0400, Vint Cerf wrote:
>> Q1: Should the proposed mapping on lookup in a revised
>> IDNA2008   protocol specification be a permanent feature of
>> the protocol or   should it have a finite lifetime? Should it
>> be required or optional?
> 
> In my opinion, any plans for a finite lifetime are just
> fantasy.  If we have mapping, it will be a permanent part of
> lookup.

I think that depends quite a bit about what is said about it.
If, for example, the design is to follow the general principle
whose details I got wrong in Protocol-11, i.e., one tries an
IDNA2008 lookup without mapping and, only if that fails does one
try to interpret the label using a restricted variant of
IDNA2003, there will be lots of incentives to get rid of it,
including the potential performance hit associated with
sometimes needing several DNS lookups to determine that the
label does not exist, the code footprint (especially on
smaller/embedded devices) associated with keeping two sets of
tables around, etc.  I'd even expect to see some
performance-sensitive users (geeks in industrialized countries,
but fairly normal users and those who support them in places
where Internet links are slow and expensive, pressing
implementers to turn mappings off to avoid the delays associated
with doing them.

You may well be right and probably are, but I wouldn't be
certain about it.

    john