Re: A character is in the eye of the beholder
Keld Simonsen wrote:
>We discussed the architecture of character handling, i18n etc
>in the IAB, and noticed that sorting is not only dependent on
>language, but on the users preferences, where language and culture
>and country is prominent ingredients. So we said that this layer
>in our model should be 'locale' oriented in the sense of C and
>POSIX locales, where we also would see it implemented in many
>cases. So I would rather recommend that we do not mix language
>and the locale, but that we introduce a locale facility in HTTP.
Well, language tags already allow to specify more 'locale'-oriented
stuff. You could for example say fr-ch or fr-ca. Of course, this
may not be perfect, and you may have other reasons for disentangling
language and locale.
But I think that if possible, locale aspects should be handled
locally, i.e. the server should send a list, with the indication
that it should be sorted. We tried to go into such a direction
when we proposed markup for dates and mesures in the first
i18n HTML draft; that was premature at that time.
Also, it should be considered that language and locale interact
in various ways. I could immagine that even to a Danish viewer,
having Danish sorting sequence inside a German text might not