Re: [idn] UTC feedback

[RJ Atkinson <rja@inet.org>]

At 16:55 20/08/00, Marc Blanchet wrote:
>At/À 10:44 2000-08-19 -0700, Mark Davis you wrote/vous écriviez:
>>You are right: the list is quite short: simply the Turkish i. French and
>>a few other languages may strip accents from uppercase,
>
>depends on where you are (back on localisation, ...). Some french speaking countries use different rules for casing.

Interesting and useful data.

>> but it is
>>acceptable to retain them.
>
>for french, we should, imho.

Agree; I think that for compatibility with other languages having
intersecting character sets (e.g. Vietnamese), we need to retain
the accents, not lose them.  If lower-casing will ensure that accents
are preserved and upper-casing does not ensure this, we might consider
explicitly making the casing transformation to lower case.  It isn't
clear to me whether lower-casing really solves this issue for ALL
languages.

>>Moreover, there is a workaround for them:
>>register 2 names (with and without accents).
>
>well, then if you have 10 accented characters, you have to register every combination... a lot of $$$ for the registries, but not very useful for the user (both the registrant and the internet user that access that domain name)... also, since the registrant wants all those  combinations to be equivalent, then he needs to synchronise all this data together, not forgetting one.  for example, if he wants to change its dns servers for those domain names, then all should be changed.

I agree with this rationale for why the workaround isn't reasonable.

That aside, the proposed workaround is unreasonable in languages where 
a character without its accent is really different from a character 
with its accent.  We need to preserve accent marks and recognise
that a single character includes all of its accent marks.

Ran
rja@inet.org