From: Jukka K. Korpela (jkorpela@cs.tut.fi)
Date: Wed Sep 21 2005 - 06:39:35 CDT
Antoine Leca wrote:
>>Would you like to have French dropped out just because
>>the repertoire of the device does not include "æ"?
>>
>>
>
>Not a relevant point: Æ is required for some Scandinavians orthographies
>which outnumbers French use.
>
>
I think it is still relevant as regards to French. Given a context where
the character repertoire is limited to a given collection of characters,
for one reason or other, it is relevant to ask which languages can be
acceptably written using that repertoire. It is not a matter of the
adequacy of the repertoire for pan-European use, for example. It is
simply about the set of languages for which the product or service can
be sold (or "sold").
As a producer of a hypothetical product that can display all the
characters usually listed as needed for French except for "æ", would you
be happy with the conclusion that the large potential user base in the
francophone world is excluded just because of this letter, which most of
the customers would hardly ever use?
>It depends on which point of view you are taking: marketers just do not
>care, since they know that such a minor defect will not prevent them to sell in France (or Spain);
>
Perhaps. But if the CLDR database says otherwise, their customers and
competitors might have a case against them. Perhaps even a legal case.
> users neither care, as it should be clear now;
>however, when you speak with officials of any class, politicians, people
>which are in representation or delegation of their country, they can be
>_very_ picky on such issues;
>
How very true. This is why politics and pragmatics should meet once in a
while. There is an inherent cost in adding characters to the repertoire
_required_ in a language (for different values of "require"). At the
level of principles, it is easy to throw in anything you prefer for
typographic or other reasons, if you have never met and never considered
potential implementation problems and costs.
> for example, I believe French was retracted for
>the list of languages supported by parts 1-3 of ISO/IEC 8859 in the 1998
>revision, because of the Œ (and Ÿ) issue.
>
>
That is my impression too. Similarly, ISO/IEC 8859-1 is nowadays not
characterized as supporting Finnish, due to the lack of s and z with
caron (š, ž), even though most Finns probably think that the language
can be acceptably (though perhaps not perfectly) be written without
them. (They only appear in the orthography of some words of foreign
origin, and in foreign names of course.)
Unicode does not make such considerations unnecessary. Character
collections will always be with us, even though we may hope that in
_some_ situations we can use the full Unicode repertoire.
In my opinion, CLDR should describe character requirements of languages
in a more layered way (and include punctuation characters), starting
from the rock bottom repertoire that is absolutely needed, then adding
different collections of characters that should be available (for
different values of "should"), or would be nice to have.
This archive was generated by hypermail 2.1.5 : Wed Sep 21 2005 - 06:40:29 CDT