HTTP has had an "Accept-Language" option since HTTP/1.1, almost 14 years old now:
http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.4
It seems to be supported by all of the popular browsers, and presumably is straightforward to use on the web server side.
Apparently this feature is not used?
Yes, in my experience it is mostly ignored.
I guess my question is that if companies are going to ignore browser settings - which have the actual user language information - does it really make sense to add language information to IP databases?
In my opinion, the correct place to store user language information is at the client. It is foolish to assume every user of an IP address block wants to see content in the same language, or to assume they even speak the same language. Statistically speaking, the language-attribute might be right most of the time, but certainly not all of the time, unlike settings made by the user of the browser. It seems to me that relying on some arbitrary location data to guess the desired language for content is pretty strong in the software community. I've even seen Windows software installing itself automatically in Finnish just because my date and time format was set in Finnish, ignoring system locale and location. I would rephrase the question above: If companies are already ignoring settings made by users, should we encourage this behaviour by providing more data for guesswork? I don't think we should. I'd also like to hear comments on how this works in countries like Belgium and Switzerland, where location based language assumptions are even more likely to go wrong. ____________________________________ Tero Toikkanen Nebula Oy, Finland