RE: How could RDDL be distributed ?

[Leigh Dodds <ldodds@ingenta.com>]

> -----Original Message-----
> From: Eric van der Vlist [mailto:vdv@dyomedea.com]
> Sent: 16 January 2001 14:09
> To: xml-dev
> Subject: Re: How could RDDL be distributed ?

[...]

> However, probably still missing something, I wonder of xmlcat documents
> are attached to instance documents.

xmlcat documents (Catalogs) aren't associated directly with instance 
documents, except via the Pub/SysId.

It might be worth adding an additional lookup method so that even 
without a DOCTYPE, a NS-URI can be used to examine an xmlcat.

> Also wouldn't we be multiplying the layers by using both and wouldn't it
> be better to integrate xmlcat features or vocabulary within RDDL for
> RDDL own purposes ?

Well you could argue this the other way and say because Catalogs 
are well-known, and there's also an Oasis committee working on 
updating them, that its RDDL thats adding a layer!

I think that the two layers are distinct. xmcat is a location 
mechanism, and RDDL is the means to define a directory of resources.
They both have clearly defined roles in my opinion.

IIRC there was some resistance to conflating the two layers early 
on the RDDL discussions.

> Also, I think it would be nice to get the 3 options to define the
> association between the instance document and the RDDL (through the
> tools, through the instance documents and through the RDDL).

I'm not sure adding stuff into the instance document is really necessary.
You already have the NS-URI - what else do you really need?

Even with schemas the instance document details are merely 'hints'.

Cheers,

L.