On Fri, Jul 26, 2002 at 09:27:58AM -0400,
Valdis(_dot_)Kletnieks(_at_)vt(_dot_)edu wrote:
On Thu, 25 Jul 2002 18:45:00 EDT, Keith Moore said:
All naming systems start with some context; otherwise they'd be trying to
impose a tree structure on the entire universe. restricting DNS URNs
to the real DNS is a reasonable design compromise.
Agreed.
The part I obviously am unable to get caffeinated enough to understand is
what purpose the timestamps serve - there's *no* way to know what the
value of turing-police.cc.vt.edu:01July2002 was (don't bother with 'dig',
that will get the the IP address of its docking station, which changed
recently - the laptop's been seen at lots of different IP addresses).
Or is the intent *NOT* to provide something that's resolvable/usable, but
to provide an audit trail of "the URN as it resolved at timestamp FOO" (a
la the Wayback Machine)?
Correct. They very explicitly don't want these things to be resolvable but
are mearly using then as unique tokens for disambiguating pieces of XML
namespaces. They could just as easily use UUIDs but those have issues
with uniqueness if you don't happen to have a MAC address handy. And they're
awfully cryptic and very prone to transcription errors...
-MM
--
--------------------------------------------------------------------------------
Michael Mealling | Vote Libertarian! | urn:pin:1
michael(_at_)neonym(_dot_)net | |
http://www.neonym.net