[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[idn] I fear I cannot use IDN in the next 10 years
As the IDN list have been only focused on ACE and are now removing the
proposals that support IDN nativly in the DNS protocol, I fear that
I will not be able to use non-ascii domain names for the next 10 years.
To understand why, you have to look at how domain names are used on your
computer systems.
When I look at mine, I enter or handle domain/host names in many places,
for example: host-files, logs, web pages, text documents and databases.
In all these places the names are using my systems local native character set.
All programs handle that.
To use non-ascii names, the only acceptable way to do it is to enter them
natively in all places. Natively does not mean ACEencoded names, it means
using the native character set. My web pages will never containg ACE in
URLs, they will containg ISO 8859-1 characters just like the rest of
my web pages. My host files may not contain ACE, the must use ISO 8859-1.
To use ACE-encoded host names can never be an application specific modification.
The only viable way to do it is to convert all names when they leav the
DNS protocol into local character set. Under Unix this could be done
in the resolver libraries so that applications never need to know the
protocol used over DNS.
To allow mixed usage of ACE-encoded and native names will result in a
terrible mess. The only usable way is for all names to be nativly encoded
within a system.
The current groups wanting ACE-encoded names do not take this into
consideration. At least I have seen no discussions about it.
As all ACE-encoded names need to be decoded into native names when
entering a system, why not use UTF-8 over the wire to support simpler
handling of names?
There are two places where ACE-encoded names are needed:
1) When talking to ancient systems.
2) When not being able to represent an UCS name using the local native
character set.
But in all other cases UTF-8 will do and will result for systems
using UTF-8 natively that they can use dns names without any
convertion. This will never happen with ACE, no system will ever
use ACE-encoded text as native character set.
So forget about quick use of non-ASCII names because of ACE. As long as my
applications cannot accept a native name or an UTF-8 name, non-ASCII
names cannot be used. I am not going to expose my users to the
mess and confusion that ACE-encoded names will result in.
When I can register a domain name using non-ASCII characters and can
query and get it returned in UTF-8 or ISO 8859-1, and my system stop
rejecting names because they conatin non-ASCII characters, then I can
start using IDNs.
Dan