Re: Language tag education and negotiation

On 5/5/2008 7:39 AM, Frank Ellermann wrote:
> Asmus Freytag wrote:
>
>   
>>> What I really would like to express is "I prefer original text
>>> in french or english, then translated text in one of these two
>>> languages." But that's work for RFC 4646 ter :-)
>>>       
>    
>   
>> I have precisely the same problem! You express it very clearly.
>>     
>
> +1  It infuriates me when xyzzy.com asks me if I want to go to
> xyzzy.de, or worse just redirects to xyzzy.de, when I prefer
> xyzzy.com for some reason of my own.  But that is unrelated to
> language tags, it is some "Geo IP" stuff, unfortunately often
> "working as designed" (from their POV).
>
>   
We don't think that the issue as originally described is unrelated to 
language negotiation at all.

Language negotiation, as implemented seems to make the implicit 
assumption that where users provide more than one language, the second 
(and following) are to be treated as fallback, so *any* page in the 
first language, is preferable to the user, whether its a complete 
translation or just a partial service intended for those monolingual in 
the those languages.

What you are describing is the further assumption by some sites, that if 
your first choice of language matches some of their subsidiaries, that 
you would rather work in that 'locale'. Treating the language preference 
as what you call 'geo ip' appears to be a clear misuse of the language 
negotiation feature, whereas the other issue is one of a shortcut in 
implementation of a scheme, which, if correctly implemented and adopted 
(including the lack of assigning q values to the contents) might 
actually have led to the correct results.

But +1 to being infuriated about that misuse as well ;-)
A./

Received on Monday, 5 May 2008 15:31:52 UTC