W3C home > Mailing lists > Public > public-rdf-in-xhtml-tf@w3.org > November 2005

Re: CURIEs, xmlns and bandwidth

From: Ben Adida <ben@mit.edu>
Date: Thu, 3 Nov 2005 18:19:15 -0500
Message-Id: <8A3840A2-10FD-4168-892E-C84EC7205596@mit.edu>
Cc: public-rdf-in-xhtml-tf@w3.org, iptc-metadata@yahoogroups.com
To: Norman Walsh <Norman.Walsh@Sun.COM>


Norman,

> You're (if I understand, which maybe I don't) proposing that instead
> of getting the prefix binding from the in-scope namespaces, you look
> somewhere else.

To be very clear: that's not true in the current definition of CURIEs  
[1]. The resolution of a CURIE is exactly the same as that of a  
QName. Now, IPTC is looking into ways of further compressing the  
current CURIE syntax so as to reduce the number of namespace  
declarations. It's important, however, not to confuse those issues.

In particular, take your example:

>   <myvocab pointer="x:foo" xmlns:x="http://example.org/">
>     <h:div someAttribute="x:foo"/>
>   </myvocab>
>
> the x:foo string in someAttribute might expand to a (uri,local-name)
> pair *other than* (http://example.org/,foo)?

the answer, according to current CURIE specs, is no. And I'm pretty  
sure that the IPTC and Misha would answer the same way. The question  
at hand is whether "x:foo" can be resolved without explicitly  
defining the "x" namespace. But if it's defined, I can't imagine a  
use case that would call for "x" to resolve to anything other than  
its in-scope definition.

> I think it's risky at best to introduce this conceptually alternate
> mechanism for associating URIs with short strings. I think it's just
> plain wrong to overload the existing prefix:local-name lexical form
> for these alternate short strings if they can occur in contexts where
> a QName might appear.

I want to stress again that we should not confuse issues here.  
Conceptually, CURIEs are meant to be very similar to QNames. The only  
goal is to allow for the expression of *any* URI in a compact way,  
something QNames are unable to do in a number of cases.

Let me put it differently: a QName is a valid CURIE. Moreover, a  
QName, resolved according to CURIE rules, resolves to exactly the  
same thing it would resolve to under QName rules. It's just that  
there are some CURIEs that are not valid QNames.

-Ben

[1] http://www.w3.org/2001/sw/BestPractices/HTML/2005-10-27-CURIE
Received on Thursday, 3 November 2005 23:19:30 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:15:00 GMT