|
From: | Benja Fallenstein |
Subject: | Re: [Gzz] urn-5 "documents" |
Date: | Tue, 27 Aug 2002 19:39:45 +0200 |
User-agent: | Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.0) Gecko/20020615 Debian/1.0.0-3 |
Tuomas Lukka wrote:
On Tue, Aug 27, 2002 at 01:39:10PM +0200, Benja Fallenstein wrote:Re Tuomas' comment in urn5.tex, in the "Potential Problems" section.You say, "Anyone may spoof an id easily. Therefore, you can't really refer to an id outside of the current context: too easy to substitute another document for it."IMHO that's the wrong way to think about this. urn-5 ids *never* refer to specific data, there is no GET-type operation defined on them.Exactly. That's the point that we REALLY REALLY REALLY REALLY have to drive through. If *I* lost sight of it for a moment, think about Joe Random Reader.
Yes, I agree. I think we should make use of the fact that this is not specific to us: It is a new way of using URIs, introduced by W3C (I think). The two really prominent examples are XML namespace names and RDF "terms" (not the correct word, I don't remember the correct one). There is even at least one other URN namespace being registered that solely exists for this kind of id: the urn:tag: namespace (see taguri.org; e.g., urn:urn-tag:address@hidden,2002:a-readable-identifier). BTW, there is a term for making a new such URI, giving it a meaning: you "mint" it.
Our contribution really is a) that you don't even have to enter a domain name or email address to mint identifiers (which comes at the cost of non-readability), and b) the *idea* that this can be used in far more contexts than most people currently expect, namely that the computer can internally assign URIs to a great many objects, instead of the process common today in which all URIs are explicitly minted by humans.
- Benja
[Prev in Thread] | Current Thread | [Next in Thread] |