Re: [URN] Re: Relative URLs and URNs
Daniel LaLiberte <[email protected]>
| Newsgroups | gmane.ietf.url |
|---|---|
| Message-ID | <[email protected]> |
> Martin J. Duerst writes: > > > >First, relative operations are purely syntactical, so the client will > >just chop off on the URN and either construct something (or get > >stuck). This is actually in contrast with what Daniel suggests > >or supposes in some of his mails, namely that the browser is using > >some knowledge about the hierarchy for retrieval: It doesn't > >presently for URLs. Scott Seligman writes: > An excellent insight: we must be careful to distinguish between > syntactic transformations and semantic "knowledge". The rules for > processing relative URIs are an example of the former. Well, whether it is a syntactic transformation or semantic knowledge is ... just semantics (i.e. it's an arbitrary distinction), especially in this case. A syntactic transformation based on rules about a '/' separated list of components is essentially implicit, innate knowledge about the corresponding hierarchy. What kind of semantic knowledge are you thinking about that is knowledge about a hierarchy but distinguishable (not derivable) from the transformation rules? One thing that *might* be different (if we agree on the following) is that you cannot assume anything more about the hierarchy than you operate on given a set of relative URIs that climb around in the hierarchy. For example, if none of the relative URIs you operate on start with '.' or '..', they never climb up the hierarchy, so you don't know if you are allowed to, even if there are '/'s in the base URI. But what you do know about is at least partial knowledge of the hierarchy. But I don't think that I even agree with the previous paragraph. If URIs can only use '/' to imply hierarchy so that relative URIs can be used, there are some other things that the knowledge of the existence of a hierachy can be used for. Some web crawlers explore by looking at all prefixes of a known URL to find other URLs. These new URLs may return a directory index or an index.html file, for example. All these generated URIs can be requested, but any particular URI, even one that looks like an intermediate node in a hierarchy, might correspond to nothing. Something that exists for one user might not (apparently) exist for another user, for security reasons. But if an opaque URI scheme can use '/' for whatever it want besides hierarchy, you have to know at least whether the scheme is opaque before you can infer any hierarchy. -- Daniel LaLiberte ([email protected]) National Center for Supercomputing Applications http://union.ncsa.uiuc.edu/~liberte/