[URN] Re: URI documents

David G. Durand writes:
 > While several proposed URL spaces have no notion of hierarchy, some
 > do, and of those, _some_ but not all, may sensibly be used with
 > "relative addresses" of the "relative URI" sort. So the "hierarchy
 > properties" may not apply globally to all forms of URI. On the other
 > hand, _where_ hierarchy can be applied, it should be done in a
 > uniform way, so that knowledge of naming scheme is not required in
 > order to parse and properly resolve relative URIs.

I agree completely.  One addition I would make: In cases where hierarchy
doesn't apply for a scheme, but the scheme still uses '/', if relative
URIs are never used relative to URIs in that scheme, then there should
never be a problem with the lack of support for hierarchy.  But if
hierarchy were later introduced for some reason, that would be a problem.

 > I agree with that relative URNs may well be a bad idea, nd they are
 > certainly not well understood (what is the "base URI" in a
 > protocol-independent context?)  

We discussed relative URNs at some length last spring or so, and I was
under the impression that the problems with relative URNs were more of
an uncertainty about the nature of the beast rather than a clear danger.
Concerning the base URI problem, there is a clear answer: the client
should either use what it is told to use, or it should use the last URI
(not the first) that it used to resolve to the resource.

By the way, I am of the belief that hierarchical URIs (including URNs)
are necessary for scalability, just as DNS itself uses hierarchy for
scalability.  The use of hierarchy to support relative URIs is
secondary, but also useful for some of the same reasons they are useful
relative to the run-of-the-mill http URLs.  Multiple URIs (including
multiple URNs) for a single document are allowed, and will occur.

--
Daniel LaLiberte
 dlaliberte@gte.com  (was: liberte@ncsa.uiuc.edu)
 liberte@hypernews.org

Received on Tuesday, 6 January 1998 15:59:52 UTC