Re: httpRange-14 Adjunct: 302 is Valid for Non-Information Resources

On Tue, 2007-12-04 at 17:29 +0000, Richard Cyganiak wrote:
> Yes. But the vast majority of HTTP URIs are for traditional web pages,  
> and those don't return any RDF. We are left to guess what they  
> identify. Web pages? People? Things? Automated tools have no way to  
> tell, and hence it's not really possible to make RDF statements about  
> those URIs with any confidence.
> This sucks, because RDF was originally created to ... wait for it ...  
> express metadata about those traditional web pages.
> httpRange-14 axiomatically declares that for all those URIs, the  
> “naïve” interpretation is correct: They identify “the Google home  
> page”; “Richard's homepage”; “the TAG blog”; and so on. They do not  
> identify companies, people, and so on. 

Unfortunately not. It declares that you need to invoke a network
operation which may tell you whether the URI denotes a document or not.
Those types of URIs can still denote things other than documents

> Hence it becomes viable to use  
> RDF for saying things about web pages.

RDF is viable without http range.

> > Why do we need this weird implicit and useless classing
> > via HTTP responses?
> Because otherwise, we would have no clue what  
> identifies. Thanks to that 303 thing, we know that it identifies a web  
> document, and a snapshot of it currently sits in my browser window.

We still don't have a clue what it denotes, just narrowed the range of
things it could denote. You need some RDF to find out what it denotes,
but if you have that then knowing it's a document hardly seems relevant
- you can learn that from a single triple. Http range uses a lot of
network mechanics to convey a single triple. Perhaps it would have been
more useful to have a header that meant "this is a document" because we
could have used that to say other things too.

> Best,
> Richard


Received on Wednesday, 5 December 2007 00:09:31 UTC