gerv at mozilla.org
Thu Aug 22 10:41:18 CEST 2013
On 21/08/13 22:52, Vaggelis Segredakis wrote:
> I have personally raised many concerns about idna2008 on these
> discussions. Even today I find it worrying that we expect that the
> client software will transform capital letters to small, if at all they
> are accepted, and then it will proceed with the idna2008 translation.
> Will this ever really behave canonical? Will there be a standard way?
I think the possibility of inconsistency here is one of the things TR46
is trying to address. I certainly agree that consistency between
browsers on this mapping is highly desirable.
> We, as a registry, faced some hard choices at that time. For sometime
> the final S letter was proposed to become an invalid character, leaving
> all male names in the Greek language misspelled in the url fields. We
> accepted the better scenario, maybe not the best possible solution but
> no other choice was offered. The idna2008 is a compromise between many
> expert's opinions, not something the registries commissioned or signed upon.
What solution would you have preferred?
> We agreed that to have the final sigma was a better way forward than not
> having it, although this fact would present compatibility issues that we
> did not ask for. We will try to mitigate all risks, educate the users
> etc. However, since the incompatibility was inevitable, if the question
> became "should we ditch idna2003 for something completely new instead of
> evolving it to something partially compatible", some interesting
> proposals appeared in the mailing list and could become a basis for this
> discussion but were rejected by most experts, in favor of maintaining
> *some* compatibility.
You would have preferred that we start again?
> We would be happy to work with you or any other interested parties, to
> help mitigate any risk of using Greek character domain names, especially
> those containing the final sigma or typed in and translated from capital
> letters. Please feel free to contact me.
Here I am :-) If we use TR46 mapping, with Nontransitional Processing,
does that work for you? Or would you prefer something different?
More information about the Idna-update