On 4/26/06, Tantek Çelik <[EMAIL PROTECTED]> wrote: > > The naming 'uri' vs 'uid' aside, would it be reasonable to RECOMMEND > > that a URI is used (thus including URLs) and leaving the door open to > > less useful ids should people want to use them? > > Yes, and I have just added similar details to uid-brainstorming, preferring > URLs first, then URNs.
I'm glad there's some progress in this discussion, but you're still trying to come up with a general rule for disparate things. At the risk of throwing major confusion into this discussion, but with the thought that it might help clarify things further, I'd like to whip out FRBR again. The FRBR model says that when talking about stuff, we can think in terms of four levels of abstraction; from top to bottom: 1) work - an abstract creation 2) expression - some realization of a work (say, an english language text) 3) manifestation (a physical production, like a book) 4) item - the specific thing you have in your hand or on your screen URLs are in essence focused on 4; they are about locating items. With web documents, that URL is in essence equivelant to the manifestation too. If one wants to get article x, one goes to one particular url. No problem. But when we're talking about stuff that exists independently of the web, this breaks down, which is why the value of more abstract uris for identification. If one uses a urn to indicate a book isbn, we are at the manifestation level, and using that makes it possible to then locate any one of thousands of individual items, If one uses an asin to indicate a movie, say, one is identifying the work level, which can then be used to locate multiple expressions (theatrical releases, special editions, etc.), whcih in turn can have multiple manifestations (DVD vs. VHS), etc. So I'd say that URLs should only be preferred where one is referring to a particular item whose canonical location is in fact on the web. E.g. when you have a web resource, use a URL. Otherwise, prefer a urn, and then perhaps other similar options such as info. Can we perhaps agree on that? Bruce _______________________________________________ microformats-discuss mailing list [email protected] http://microformats.org/mailman/listinfo/microformats-discuss
