Re: Talked to the xml.gov people

Michael Mealling wrote:

> Yes. Persistence is a social/management/policy thing. But you can make
> that a function of the scheme by specifying those policies for that
> scheme, which is what the 'urn:' scheme has done. Here's the key
> difference: if I detect some non-persistent behavior from
> 'urn:isbn:123456-87' then I know based on the scheme that it wasn't my
> error to have assumed the name was persistent. With 'http://foo.com' I
> have no way of knowing whether or not I get to make that assumption.

I've heard this argument 10^6 times and I still don't get it.  Let's 
see, 'urn:isbn:123456-87' and 'http://foo.com' are two characters 
strings, and treated as character strings, are equally persistent.  If a 
spec says "This {namespace|bank-account|book|you-name-it} is identified 
by URI 'xyz', then, in the frame of reference of that spec, it is so 
identified, and that identification is *not* a function of what xyz's 
URI scheme is.

In fact, the only meaningful difference that I can see is that if you 
use something in the http space, there's a well-documented way for you 
to use it to make representations available, and ubiquitous technology 
that everyone already has and anyone can use to retreive them.  Neither 
of these is true for URNs.

Put another way, I do not agree with your second sentence, that 
social/management/policy issues can be made a function of URI schemes.

What am I missing?

For the application in question here, which is XML namespace names, I 
just can't imagine why you'd ever invent a vocabulary and give it a name 
that can't easily be dereferenced to find out about the namespace.

> There is no consensus that 'http' is the one true scheme for everything
> so please stop claiming that it is...

I informed the xml.gov people that my position didn't represent consensus.

-- 
Cheers, Tim Bray
         (ongoing fragmented essay: http://www.tbray.org/ongoing/)

Received on Wednesday, 21 May 2003 16:22:43 UTC