RE: "Hash URIs" and content negotiation

-- Alan,

> On Nov 8, 2006, at 11:05 AM, Xiaoshu Wang wrote:
> > Why not the same?
> 
> I don't think that there should be any controversy that 
> different language translations are different things. 
> Different people worked on them, there are different 
> typographical errors, each might be slightly wrong or 
> non-idiomatic in a different way. I think that if you asked a 
> prominent translator whether their work product was the 
> "same" as the original they might be (quite reasonably) offended.

Fundamentally, we all get a representation of the resource.  Even if it is a
document, waht you retrieved is a "copy" of the document on the server but
not THE copy of the server.  The Accept header distinguishes the
representation by MIME type.  The Accept language further distinguish the
representation by the type of natural language employed, if any.  They are
all part of the mechnisms that prevent the proliferation of alias URI.  

> The choice is whether you want the naming of things to be 
> driven by the nature of the things, or the nature of the user 
> interface. I think it's safer to name them by the former.

Right. It is a choice.  Then when I serve a document (in either different
langauge or format), I think the provided representation would adequately
represent the content given the situiation.  The content served is a result
of negotiation for the purpose of communication. For instance, if a client
want a low resolution image in jpeg, and you insist giving them a high
resolution png image, who would you think be very upset? I think you
preceding argument misunderstand the problem, content negotiation has
nothing to do with the content-equality.    
 
> > Let's assume "the machine" to be a special kind of human being but 
> > only speaks RDF. Since human won't (at least not willingly) 
> understand 
> > RDF and machine-humans won't understand HTML (at least not 
> > consistently), how do you supply descriptions to both 
> human-humans and 
> > machine-human? Wouldn't the solution come down to content- 
> negotiation 
> > in some form?
> 
> It will come down to content understanding and decision 
> making in some form. I just don't think it should be in the 
> network protocol.  
> For HTML, for example, there is a mechanism to embed rdf. 
> That there are alternate translations could be embedded in 
> that rdf, and a browser, on a user's behest, could certainly, 
> upon reading about these translations, automatically redirect 
> to the translation in the user's preferred language.

But where would it be if not in the transportation layer?  How much wasted
bandwidth will be there if we use the embedding technique.  It is as if to
serve an HTML document with all possible human language translation.
Besides, what if a software agent only knows RDF? The embeding solution
would have to force every agent to implement all kinds of parser (HTML, PDF,
etc...), you think that is a good solution? Honestly, I don't see it.

Xiaoshu

Received on Wednesday, 8 November 2006 17:07:35 UTC