FW: Preferred language in OpenID registration extension

[Recordon, David]

Moving this discussion on the specs list with Stephane's permission. :)

--David

-----Original Message-----
From: Stephane Bortzmeyer [mailto:bortzmeyer at nic.fr] 
Sent: Monday, January 08, 2007 7:32 AM
To: Josh Hoyt; Jonathan Daugherty; Recordon, David
Cc: bortzmeyer at nic.fr
Subject: Preferred language in OpenID registration extension

Hello,

I noticed in "OpenID Simple Registration Extension 1.0"
(http://openid.net/specs/openid-simple-registration-extension-1_0.html):

openid.sreg.language:
    End User's preferred language as specified by ISO639

ISO 639 is far from covering all the options people want with the
language. It does not allow to express regional preferences (such as
British orthograph vs. US orthograph) or scripts (which is not useful
for French or English but is very important for languages like Chinese
or Serbo-croatian).

I suggest to use instead the "language tags" defined in RFC 4646
(successor of RFC 3066). They are already the reference for protocols
like HTTP (Accept-language header) and XML (xml:lang attribute).

With language tags, an user can say things like az-Arab-IR (Azeri
language, in the Arabic script - Azeri is often written in Latin or in
Cyrillic - as spoken in Iran) which is more richer than what ISO 639 can
offer.

Funny, but the reference you give for ISO 639 is a W3C page which says
the same thing than me :-) BTW, you did not specify which ISO 639...
(639-1, 639-2 or the almost-ready 639-3).

My personal page on the language tags: http://ltru.generic-nic.net/