Re: RDDL: new natures

Elliotte Harold wrote:

> Jonathan Borden wrote:
>> A class has a (i.e. one) set or group of members. If a namespace  
>> URI identifies a class then what is the set of members:
>> a) the set of names in the namespace (for example)
>> b) the set of documents that validate to a given schema (for example)
> I don't buy it. First of all I don't think a namespace URI  
> identifies the set of names in a namespace. That's simply not how  
> namespace URIs are used. I've heard people assert that principle,  
> but attempting to follow that idea leads them down a maze of twisty  
> little passages, all different.
> In an XML document, a namespace URI identifies a local name as  
> belonging to a particular namespace. That's a subtly different  
> thing, but the difference is important. There is no one unique,  
> fixed set of names in a namespace to be identified; and working  
> from the assumption that there is leads to brittle software.
> Secondly, I don't think it matters if we use the namespace URI in  
> clearly different contexts for different results.

That is fine. I don't mean to assert that namespace URIs identify the  
set of names in a namespace (but I don't want to assert that they  
don't either which is really the point).

The way that I have always envisioned a RDDL Nature is that it  
identifies the "nature" of a namespace related resource in such a  
fashion that software agents can make reasonable assumptions about  
how the resource is intended to be used.

A specific example:

When I say that the rddl:nature of is "XML  
Schema", this is intended to assert that it is reasonable to assume  
that ought comply with the "XML Schema"  
specification i.e. validate as an "XML Schema".

A more precise way of saying this is that the assertion that the  
<rddl:nature> of <> is "XML Schema" implies  
that <> is a member of the class of all  
"XML Schema"s.

With such an assertion a software agent might reasonably conclude  
that lack of XMLSchema validation on the part of < 
foo.xsd> would be an error.

What I *don't* want to say is that <> is a  
member of the XML Schema namespace. Using < 
XMLSchema> as the URI for the nature of "XML Schema" creates this  
ambiguity for ***software agents***. I understand that you, Elliotte,  
being an intelligent human being can distinguish this contextual  
difference, but the type of logic that you are using to do this is  
actually rather complicated. This is admittedly a technical issue,  
but as far as I can see a real one.


Received on Sunday, 10 December 2006 15:41:17 UTC