Re: [Idna-update] IDNA and combining sequences

[John C Klensin <[e-mail]>]

--On Saturday, 10 March, 2018 21:57 -0800 "Asmus Freytag (c)"
<[e-mail]> wrote:

>...
> However, independent of the details of what may have been
> exchanged a decade ago, I strongly  suspect that going over
> that ground is going to do little in terms of helping to make
> progress forward.
> 
> Instead, the sole question we should be asking is what can we
> do to effectively improve the experience involving IDNs.

I concur.   The history is not, however, completely irrelevant.
It can be important to avoid repeating prior mistakes and,
perhaps, to calibrate some proposals going forward.  

In particular, there is some non-technical history that may
constrain what we can consider today.  In addition to a change
of model, IDNA2008 included a few changes that invalidated some
labels that were possibly in use ("possibly" because I'm talking
about the whole DNS now, not just the root zone or second-level
names) and changed the interpretation of others.  Those changes
were controversial and arguably remain so, but there was broad
consensus in the registry community that they were worth it if
(and perhaps only if) that type of change never occurred again.
If we take that as a constraint today, a guideline that says
"don't do so-and-so unless you have a very good reason" would
still allow registries to "grandfather" existing names or make
other rules about them as needed.  A protocol change, especially
one with lookup-time implications, would not have that property.
That is one of several reasons why the three-level distinction I
tried to make in my earlier note is important.

best,
   john