Standardizing on IDNA 2003 in the URL Standard
gerv at mozilla.org
Thu Jan 16 11:53:23 CET 2014
On 16/01/14 10:39, Anne van Kesteren wrote:
> The sentiment in this thread that we can keep changing the rules also
> strikes me as bad.
I think that's unfair. IDNA2008 was created because IDNA2003 was
inadequate, in a number of documented ways. No-one is saying that we
should just keep changing the rules again and again. UTS46 is, among
other things, the mapping layer which IDNA2008 says should be
implemented, so they are not "two competing standards". Yes, there are a
couple of levels within UTS46 regarding how you deal with the four
exception characters, but that's hardly creating a whole new standard.
And more about those below.
> The whole reason we set standards
> is stability. So you can build on top of a foundation you know will
> not change. We take this pretty seriously in most places. Not taking
> it seriously for something as fundamental as domain names strikes me
> as wrong.
Fixing on IDNA2003 would permanently block all those scripts which have
been added to Unicode since 3.2 (is that right?) from ever being used in
domain names, and so block users of those languages from having IDN
names. If you decide to "fix" that, then you aren't using IDNA2003 any
more, and you are "changing the rules" in a way to which you have
indicated opposition - and worse, in a non-standard way.
It has always been my understanding, and I've had confirmation certainly
from the Germans, that the backwardly-incompatible changes in IDNA2008
relating to the four exception chars - Greek sigma, Eszett, ZWJ and ZWNJ
- are endorsed by the registries of the languages most affected. In
other words, as people closest to the problem, they still think changing
is less bad than sticking with IDNA2003. That should count for a lot.
More information about the Idna-update