Re: [idn] Re: [JET-member 464] Re: Fw: Re: new membersinvitation

[John C Klensin <klensin@jck.com>]

--On Monday, 29 October, 2001 10:30 -0500 Eric Brunner-Williams
in Portland Maine <brunner@nic-naa.net> wrote:

>> However, if one says "we need to do this because otherwise we
>> have to do double (or more) registrations and those are too
> 
> We're discussing exponential behavior. Scaling is the problem
> we are trying to solve. Does an out-of-protocol SC/TC
> equivalance mechanism both exist and scale, for the existing
> problem constraints?

Beats me.  I'm still not sure I understand the stated problem
constraints (e.g., I'm not sure whether we really have to
"translate" an arbitrary combination of mixed TC and SC
characters to some other arbitrary combination of them or
whether the problem can sensibly be constrained to translating
entirely-TC strings to entirely-SC strings and/or back).  The
two, obviously, have widely different scaling properties if one
tries to go the multiple registration route as most of us would
not spend a lot of effort on a factor of two.

Conversely, there are some potentially serious problem
constraints that I'm not sure I see being addressed.  And I
prefer to not try to guess at how (or whether) some proposal
will deal with them until I see the proposal.   Among those
constraints is that we don't language-label and we don't insist
that the labels of an FQDN be homogeneous (or even predictable
from each other) wrt language or even script embedded in Unicode.

    john