W3C home > Mailing lists > Public > www-zig@w3.org > March 2003

Re: Betr.: RE: requesting XML records

From: Mike Taylor <mike@indexdata.com>
Date: Fri, 28 Mar 2003 10:55:28 GMT
Message-Id: <200303281055.h2SAtSA22099@badger.miketaylor.org.uk>
To: matthew.dovey@las.ox.ac.uk
CC: Theo.vanVeen@kb.nl, a.powell@UKOLN.AC.UK, www-zig@w3.org

> Date: Fri, 28 Mar 2003 10:38:38 -0000
> From: "Matthew Dovey" <matthew.dovey@las.ox.ac.uk>
> > If there is not a way to express what we want in XML schema, 
> > why can't we express it in plain English and give it a URI.   
> We already have such a URI - the namespace URI.

Kind of, yeah.  As Bob the Angry Flower observes
(http://angryflower.com/status.gif), "Your idea
both terrifies and intrigues me".

> As far as I can see we have people who want to request a record
> which conforms to a given structure i.e. request by schema, others
> who want a record containing certain elements i.e. which uses
> (non-exclusively) a particular namespace in the way DC works.

Yeah.  It does seem that what Theo wants is really just to say "I
prefer my records to contain elements that smell like _this_."  Which
really is more like -- or at least less unlike -- requesting a
particular namespace than a particular schema.

> The distinction I think is comparable to GRS.1, as to whether you
> want a record which uses particular tag-sets or a record which
> comforms to a particular profile such as GILS.

Yes, this is a strong analogy.  Thing is, we never felt it necessary
in GRS-1 days to provide a means for clients to say "Please give me
records containing tagSet-foo elements", and for what I think are good
reasons -- it would lead only to the vaguest and least useful kinds of
interoperability.  Do we really want to change this decision in the
XML world?

I honestly don't know whether this is A Good Idea or not.  I can see
strong arguments both ways.

> A possible problem arises if the same URI is being used for both -
> which is why I feel some way of saying whether you are requesting by
> schema or by namespace is needed (and ComSpec seems to have this in
> Z39.50 v3).

Yes -- if we decide to support both of these, they _must_ be specified
in separate places, or be otherwise disambiguated.

 _/|_	 _______________________________________________________________
/o ) \/  Mike Taylor  <mike@indexdata.com>  http://www.miketaylor.org.uk
)_v__/\  "Unencumbered with facts as I am, I will comment" -- Drew
	 Larson in alt.folklore.computers, now the official Usenet

Listen to my wife's new CD of kids' music, _Child's Play_, at
Received on Friday, 28 March 2003 05:56:06 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:26:05 UTC