I read through the current IDN draft on the plane coming to Japan, and
while I've no problem with most of the technical content insofar as I could
follow it, I feel rather uneasy about some aspects of its presentation.
I have one technical issue: having multiple label separator characters
(i.e. the different representations of '.') seems misplaced. It seems to
me like a confusion between user interface issues and protocol elements.
In reading this, I couldn't feel confident that I was properly
understanding what was being specified. I think a distinction should be
drawn between what is being specified, and a process for calculating that
value. If one were to use a different process, it can be very difficult to
prove that it gives the same answer as some other process.
Something else that makes me feel very uneasy about the current draft is
the way it defines what constitutes a valid IDN in terms of algorithms
described in two separate documents. This feels to me like putting the
cart before the horse: I think this (i.e. what constitutes a valid IDN) is
a fundamental idea which needs a crisp, easily understood description so
that (for example) any future developments to embed IDNs directly into DNS
don't get lumbered with legacy ACE code simply to determine what is a valid
IDN.
#g
-------------------
Graham Klyne
<GK(_at_)NineByNine(_dot_)org>