Cart
[Home] [By Thread] [By Date] [Recent Entries]
On Thu, 30 Jan 2003 14:23:22 -0500, Mark Baker <distobj@a...> wrote: > That's reasonable, except for one thing; the more general form is > already a success. If you've ever used a firewall, you used the more > general form, because the client didn't have to do a gethostbyname() > (i.e. treat it as a locator) on the authority component of the URI. > > The only place in *any* HTTP request-response chain that treats a URI as > a locator, is the next-to-last node, because it has to locate the last > node on the network. Every other node, including the origin server, > treats it as an identifier. That's a good point ... an HTTP URL used as a cache identifier is not really a "locator." Still, "URLs are useful as a way of generating unique identifers" is a nice trick even without all the stuff (that some of us seem to consider voodoo) about abstract resources and representations thereof. It seems to me that it's the *uniqueness* of URIs that leads to most of their power, both as locators and identifiers, and I'm not at all convinced that the uniqueness property depends on the resources/respresentation abstraction. Clearly there are some very powerful ideas underlying the success of the Web, and URLs/URIs are clearly one of the key principles. I simply think there's a lot of room for alternative theories of exactly why that is. The resource/representation paradigm is clearly one of them and must be taken seriously, but not IMHO treated as axiomatic. I'm sure we would best serve humanity by agreeing to disagree on this rather than abuse yet another mailing list with the debate :-)
|
