Re: [idn] prefix/suffixes in internet-draft

["Martin J. Duerst" <duerst@w3.org>]

At 00/11/23 10:51 +1100, Frank Ernens wrote:
>Marc Blanchet wrote:
>
> > - idn ace prefix for me is like a protocol number: a registry should assign
> > it, not the protocol author. Currently, IANA is taking that role for
> > protocol numbers. I don't know which organisation should do the assignment,
> > but certainly not the wg, not the author.
>
>If it is done well enough the prefixes, and the competing encoding
>schemes, will become de facto permanent and all software will have
>to support all of them. And later some large outfit may invent
>its own prefix and widely deploy it without registering it, hindering
>interoperability.

I agree that having different schemes, and then different prefixes
to distinguish them, is a bad idea. No guarantee anywhere anymore
that a domain name will be found.


>For comparison, both these problems have occurred
>with the charset registry (the [accidentally, I believe] unregistered
>charset being "Windows-1252").

There have been a lot of problems with the charset registry, but at
least some of them have been cleaned up recently, and windows-1252
is now registered. But the consequences of such problems are much
less severe than they would be in the case of ace schemes.


Regards,   Martin.

>I thought the situation before, in which a prefix separated IDN names
>from others but collisions between competing IDN schemes could occur,
>was ideal. It ensured that at completion of testing all the test
>names had to be converted to a single scheme.
>
> > But, this would be for testbed purposes inside their own domain name space.
>
>Yes, genuine testing shouldn't need prefixes.