Re: Minimal RDDL [NamespaceDocument-8]

Dare Obasanjo wrote:
> I love it. I only have 3 questions 
> 1.) Who has the authority to create new natures and purposes? 

Borden and me!  Buahahaha!  Under Reasonable And Non-Descriminatory 
licensing terms of course... sorry.

They're URIs, so anyone can.  One of the nice things, I thought, about 
the original RDDL was, which was a nice 
handy set of pre-cooked purposes like #runtime-validation and 
#authoring-validation and #normative-definition and 
#human-documentation.  I wonder if there's any organization the world 
that could/should/would buy into owning/maintaining this, so even though 
it's handy, it may not be practical.

> 2.) Using namespace URIs  for nature/purpose probably doesn't cut it.
> The W3C XML Schema working group may not rev the namespace for
> subsequent versions of the REC and the XSLT working group is definitely
> not going to. So should RDDL nature/purpose contain versioning
> information or should that be embedded in the targets of the href? 

I think we don't want to go there.  Anyone who's worked in publishing 
systems knows that schema versioning is a horrible, horrible rats'-nest; 
if we can find a reliable interoperable way to say "this is an XML 
Schema" or "This is a CSS stylesheet" or "This is a set of rendering 
classes in a .jar", those are huge 80/20 points and we should stop there.

> 3.) Is there a mechanism for dealing with multiple href targets that
> have the same nature and purpose besides UAs just picking the first one?
> The first that comes to mind is having multiple stylesheets for the same
> XML document. One could say that you should have different purposes for
> each stylesheet which then takes us back to question (1). 

It turns out to be impractical to rule out duplicates, which is what I 
wanted to do originally.  So I guess you just have to deal with them.


Received on Friday, 14 February 2003 22:47:28 UTC