Re: Value of content negotiation? [was RE: content negotiation anti-principle]

On Friday 10 January 2003 04:15 am, wrote:
> In our systems, we use URNs that do not specify encodings,
> language, coverage, revision, etc. and dynamically resolve
> such information at run time based on system and user properties                                                                              
> in order to resolve a link to an actual resource. 

The "user properties" is fairly significant here. In most situations where I 
have seen a need for something like content negotiation, content negotiation 
itself has been insufficient.

> (If someone *wants* to specify the precise encoding, language,
> revision, etc. they can, but it makes the link fragile)

Right, but that's a tradeoff that the author should have the ability to 
make.... and for some things, only the author *can* make reliably.

> Putting the "required" MIME type in the link is simply making
> the links that much more fragile and defeating the purpose
> of content negotiation.

Precisely. In the cases where content negotiation is *not* desired, and a 
*specific* set of representations *must* be returned, this is a good thing.

Received on Friday, 10 January 2003 08:59:26 UTC