[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: New Internet Draft on registering IDNs
On Tue, Mar 25, 2003 at 09:31:09AM -0800,
Paul Hoffman / IMC <phoffman@xxxxxxx> wrote
a message of 16 lines which said:
> Greetings. I have just submitted a new Internet Draft that gives
> suggestions on how to register IDNs.
There is a big problem with the bundle approach described in this
draft, a problem I discovered while trying to implement it. I designed
a table, as specified in the draft, for the French language (for
multi-lingual countries or for multinational domains like '.eu', the
problem will be worse). My table may be too large, allowing all
Latin-1 (ISO-8859-1) characters, even those not used in French but it
is still too small for the entire European Union.
With this table, I experience a dramatic explosion. Many domains
generate a bundle of several thousands of names, sometimes more. If I
implement Option 1 of the draft ("Allocate all labels to the same
registrant, making the zone information identical to that of the input
label."), my zone file will explode :-)
The problem, as I see it, is that the draft uses variant tables which
work on a per-character basis, without any concern that not all
combinations mean something. Most of the words in a bundle have no
meaning in French and there is no real reason to keep them.
I understand that there is no easy option (using a dictionary will not
work since many domain names are not in any dictionary).
Did anyone try a bundle approach on his zone?
Here is the table, for those interested. It is simply
"accent-insensitive". I regard any composed character as a variant of
the plain character.
# Variant table for the French language
# See Internet-Draft draft-hoffman-idn-reg-00
# Designed at AFNIC
# Stephane Bortzmeyer <bortzmeyer@xxxxxx>
# - (hyphen)
# Ligature oe