[standards-jig] Jabber i18n proposal

Sebastiaan Deckers cbas at screaming3d.com
Sat Apr 26 19:16:57 UTC 2003

Iain Shigeoka wrote:

>In particular, you shouldn't just assert a lang because the other side of
>the conversation may not support it (and then what do you do?). Negotiating
>the 'best' lang seems pretty specific to users (let alone clients). For
>example, I may request english and if a service can't provide english, it's
>a failure because I'm a mono-language loser. However, another person, may
>request german, but be able to work with english as a second choice if
>german wasn't available.

Multi-step negotiating is not required.  In almost every case, the 
client simply puts xml:lang="[locale]" in the requesting message or iq, 
and the receiving side does what it can with that information.
If the sender requests xml:lang="de" (german) but the recipient only has 
a string table with "en" (english) and "fr" (french), then the recipient 
chooses the language it responds in.  That is a lot more convenient than 
having yet another negotiation process for a fallback language 
preference.  IMO.


More information about the Standards mailing list