Re: The lost meaning of the HTTP protocol in URIs

sorry for the delay in the reply and thanks for your answer.

On Sunday, Sep 14, 2003, at 22:15 Europe/Rome, Sandro Hawke wrote:

>
>> My humble proposal is to come up with a best practice that, basically,
>> stops using HTTP for URI that are not meant to be dereferenced. 
>> TimBL's
>> Car, for example. I proposed the simple "uri:" scheme, but I'm happy
>> with anything, rdf: res: abs: or even urn:
>
> I went through the same thought process (and some of the same
> emotions) when I first encountered RDF.  So I proposed a URI scheme
> like you suggest, one which is defined to have no dereference
> mechanism, and merely serves to help people share the space of
> possible names [1].  It's been hard to justify to the IESG, though, so
> it is not yet published as an RFC.   I originally called it "tann",
> then with my co-author, "tag", ... and we're vaguely in the market for
> a new name, since this Technical Architecture Group came along.  :-)
>
> Anyway, I stopped pushing very hard, because I found an answer to your
> question:

ok

>> Now, since this is obviously too simple for me being the first one to
>> propose this, what am I missing?
>
> What you're missing is that derefencing is very useful.  URIs should
> work in browsers, and network-aware RDF-based systems should be able
> to use derefencing to do their jobs better.  For instance, some of the
> more advanced RDF validators (eg [2]) follow the links and look for
> logical inconsistencies or other signs of errors in your document IN
> THE CONTEXT OF THE WEB, as found through the URIs used in your 
> document.

This makes perfect sense.

> I'm also hopeful that someday pasting a URI found in RDF into a
> browser will give you readable documentation, but that hasn't
> materialized very well yet.
>
> This is not covered in the current RDF specifications.  There is still
> a lot of work to be done.  The TAG recently recognized this as an
> issue [3], and with the help of the Semantic Web Coordination Group is
> starting up a group to try to move this along. [4]

I think it would be very helpful to explain the rationale around this 
use of the HTTP: scheme in "URIs that aren't URLs but might well be 
used like URLs to retrieve metadata about the URI"

Now, I have another question: should all "URI which are not supposed to 
be URL" considered like "potential URLs to retrieve metadata about 
themselves", or this is just a subset of them?

--
Stefano.

Received on Friday, 19 September 2003 17:49:59 UTC