RE: URI's for processes, stocks, file hashes, UPC's, etc..

* Noah Mendelsohn
| Steve Pepper writes:
| > One important consideration (if this really *is* to be an open
| > standard) is that your users should be able to figure out which
| > "subject" (web page, stock, product, service, etc.) the URI is
| > intended to denote, without necessarily having to use your service.
| > In other words: given an URI used as an "identifier" for an
| > arbitrary subject, it should be easy to discover what that subject is.
| Well, I think you're being a bit quick in suggesting that as a design
| point.  In fact, there's a lot of emphasis on the Web on the opacity of
| URIs.  See [1] for an early exposition of this principle by Tim
| Berners-Lee, and "The Architecture of the World Wide Web" [2] for another
| explanation. Indeed, the TAG is has nearly wrapped up publication of a
| finding devoted entirely to what you should or should not infer about a
| resource from inspection of its URI.  The draft finding is The Use of
| Metadata in URIs [3].  I suggest that those interested in the suggestion
| above consult all of these resources, as this is a topic that's been
| carefully considered as the Web has evolved.

I agree with you 100%, so I guess my posting wasn't clear enough :-)

My point was *not* that it should be possible to inspect the URI and figure
out what the subject is, but rather that the URI should *resolve* to
something that can be inspected. This is entirely in accordance with the Web
architecture and it is the key principle behind the PRI proposal.

| Also, I strongly concur with Mark Baker's suggestion to use http-scheme
| URIs to solve the problem that is the subject of this thread.

This was also my main point (along with the recommendation that the URI
should resolve to something useful).

There is another TAG finding underway that is relevant in this connection
[1]. Its key message could (almost) be summarized as "New URI schemes
considered harmful: use HTTP URIs".


[1] URNs, Namespaces and Registries

Received on Tuesday, 19 September 2006 13:38:14 UTC