RE: namespaceDocument-8: possible interaction with Namespaces in XML 1.1

# Since *nobody* can *ever* commit to maintaining anything in
perpetuity, 
# given that in the short run we're all dead and in the long run there's

# the 2nd law of thermodynamics, do you thus feel that all URLs should 
# abandoned forthwith in favor of URNs?

A good reductio ad absurdum, but not what I'm advocating. I believe
that some people might judge that a URN meets their needs better
than an HTTP URL, when selecting a namespace name, and that the W3C
web architecture has no need to strongly prefer one over another.

Part of the review criteria for registering a URN namespace
is the evaluation of the registrar's claims of "permanence".
In that sense, there's at least some judgment about permanence
of URNs, and an attempt to insure it.

Some URN namespace owners might be willing to allocate URNs
for resources they don't control, didn't author, or make no
claims of authorship about, even while providing some assurance
of permanence, while some namespace authors might have difficulty
obtaining a similar promise for http URLs, purl.org notwithstanding.

Here's a different kind of argument:

At this level, "http" and "urn" are just URI schemes, and there
is no strong architectural reason to prefer one over the other.

Building "http" into the web architecture deeply is bad design;
it makes it unnecessarily fragile. And if you're going to allow
schemes other than "http", then you might as well allow "urn".

# The trouble is, we seem to lack consensus; in particular on the claim 
# that URNs have a superior quality of "permanence" in some sense.  Our 
# interchange right here is evidence of that.

I'm not claiming that "all URNs are more permanent than all URLs",
surely that is false. I'm not claiming that "some URNs are more
permanent than all URLs", since, at some level, a URN is a kind
of a URL. In fact, I don't even want to try to make a claim
about permanence, but rather one about "belief about permanence"
and the architectural principle of not disallowing workable
choices unnecessarily.

Personally, I think one advantage of "urn" over "http" might
be for namespace assigners would be that there's no need to
put up web space or get any web traffic.

Let me turn this around: why do you think it's important to stamp
out the practice of using "urn" URIs for namespace names?


Larry
-- 
http://larry.masinter.net

Received on Wednesday, 9 April 2003 14:36:52 UTC