Re: [urn] URNs are not URIs (another look at RFC 3986)

[Barry Leiba <[e-mail]>]

I'm replying to one of John's messages for convenience, but this isn't
directly in response to that (for those who are following in a
threaded mail client).  I want to untangle one thing that's come up a
number of times: persistence.

We seem to have a lot of confusion because we have things called
"URNs", and we have a URI scheme "urn:".  Note that we do not have URI
schemes "uri:" or "url:".  That means that it's easy to talk about
URIs and URLs without confusion, but talking about URNs gets us
wrapped around an axle.

There's a type of identifier that we call a "name", which has certain
characteristics that distinguish it from identifiers that we'll call
"locators".  In particular, the name is just a name, like "Barry
Leiba".  That name identifies me (and perhaps other people, but that
doesn't matter for now).  It doesn't tell you where to find me.  It
doesn't tell you anything else about me.  "[e-mail]"
also identifies me, but it *does* tell you some of those other things
(it tells you where to send email to me; in URI terms, we'd say
"mailto:[e-mail]").

The name "Barry Leiba" can be more or less "persistent".  Reg Dwight's
parents might have thought his name would persist for his entire
lifetime, at least, but now he's Elton John.  I dare say that name
will persist long after his life is over.

So with URNs: their persistence varies.  RFC 2141 defines a type of
URN, a type of name, that begins with "urn:".  It tells us that these
names that begin with "urn:" "are intended to serve as persistent,
location-independent, resource identifiers," and it tells us something
of what that means.

But there are other URNs, other URIs that are names, which do not
(necessarily) have that characteristic, and which are not covered
under RFC 2141.  "ni:", for instance [RFC6920], defines names -- they
are most clearly NOT locators.  There are others.

Now, the notion of persistence varies from one type of URN, one type
of name, to another.  For "urn:", RFC 2141 describes it.  But it's
clear that "ni:" is meant to be used to name transient resources, as
well as long-lived ones.

Let's not get too stuck on this, and let's be clear what the scope is.
 I think we did ourselves a disservice by creating a generic name
("URN") and a scheme ("urn:") that can be so confused.

Barry