[
Lists Home |
Date Index |
Thread Index
]
Tim Bray scripsit:
> Hmm.... the more I learn about URNs, the fewer use-cases I can imagine. -Tim
The basic use case for URNs is having a collection of unique standardized
labels for things (computerized or not) and making them interoperate
with URLs even though retrieving representations is not practical.
Consider the news: URL scheme, which is functionally an URN. It doesn't tell
you how to retrieve the article (unlike the nntp: scheme); relative news:
URLs make no sense; it is a unique label for something (a Usenet article)
that exists in multiple copies around the world.
Similarly, the urn:newsml: scheme provides a unique label for Real
World news articles, with four subparts: domain name of originator,
date of ownership of the domain, local-part, version number. Again,
relative URIs make little sense here.
The ISBN and ISSN numbers for books and periodicals (the whole periodical,
not a particular number) also say nothing about retrieval and are mapped
by the urn:isbn: and urn:issn: schemes.
The urn:oid: scheme represents ASN.1 Object Identifiers, which are
hierarchical sequences of numbers used to label various computer-related
things. For example, urn:oid:1.3.6.1 is the Internet itself.
The urn:ietf scheme labels RFCs and their friends:
ftp://ftp.isi.edu/in-notes/rfc2648.txt is an URL for that particular
instance of RFC 2648, but urn:ietf:rfc:2648 is RFC 2648 itself.
Finally, there is an URN scheme for the persons in the Internet Whois
database, people who are technical or administrative contacts for
particular DNS domains.
--
John Cowan
jcowan@reutershealth.com
I am a member of a civilization. --David Brin
|
