Re: Translations

M.T. Carrasco Benitez:
>

>> It has to use *some* q values anyway, doesn't it?
>
>Some value that has this effect as the q value is a parameter for some 
>calculations.  I always want the Spanish first, even if the English is of 
>much better quality.

Do you mean that you want the language preferences to override all other
preferences?  You can do that with

  Accept-Language: es, en;q=0.02; fr;q=0.01

just specify a huge gap between the appreciation of Spanisch and everything
else, then you will always get Spanish first. [Note: I'm not sure if `es'
specifies Spanisch]

Anyway, following comments at the December IETF, I'm now revising the
transparent content negotiation draft to make the existing server-side
quality calculation algorithm fully optional.  If you want to (and it seems
you do), you will be able to specify another algorithm which does not use q
factors and multiplication at all.  Of course, it will be up to server
authors to decide whether they want to support your algorithm, the current
one with multiplications, or both.

However, I believe that the current algorithm is the best one anyone can
come up with for now.  We will need a extensive experience with real content
negotiation before we will be able to invent something sufficiently better.

>
>> I don't know what you mean by "best doc only from friendly servers",
>
>servers to which the language preference is transmitted first.  Mechanism 
>to be defined: a simple list maintained by the user in his client or 
>probably something more elaborated.
>
>> why that would be needed.
>
>Privacy purpose: I trust the friendly severs. 

There are a lot of privacy-enhancing technologies you can deploy on top of
HTTP/1.1 or TCN, but I do not think we need to define standards for such
mechanisms.  The security sections of the drafts contain many helpful hints,
and the rest can be left to the marketplace.

>Regards
>Tomas

Koen.

Received on Monday, 20 January 1997 03:51:51 UTC