RDF namespace conventions

-----Original Message-----
From: David Carlisle <davidc@nag.co.uk>
To: timbl@w3.org <timbl@w3.org>
Cc: xml-uri@w3.org <xml-uri@w3.org>
Date: Saturday, May 20, 2000 8:01 PM
Subject: Re: SIgh [was: Irony heaped on irony]


>
>> I would like to say that the statement that dereferencing the namespace
URI
>> was "not a goal" I took on review as stating that it was not a goal of
the
>> Namespace spec itself.  I assumed that it allowed other specs to do it.
>
>Even I agree with the above statement:-)
>
>> I had no idea that the XML community would decide that in their
>> wisdom they would _prohibit_ other groups from doing it.
>
>You are not prohibited from doing it, but you are `prohibited' from
>using dereferencing to schema as use cases for changing the spec.
>That is, you should firstly consider the use cases that _are_ a goal.


I'm sorry, but if the spec (and attendant expert interpretation)
disallows dereferncing of the URI, then prohibting chnaging the spec
on those grounds effectively prohibits dereferencing URIs.
(was this an accident or a slick process play by an anti-URI group?)

What do we do with the use cases which are prohibited, I wonder.

>> The RDF group was that which was asked to use XML for consistency and
>> interoperability. They required every RDF property, hence every XML
element,
>> to have a well-define identity in the URI space,
>
>which is why it is so odd that RDF use of namespace URI is so bizarre.
>Why doesn't it arrange to insert a / (or anything) if the namespace URI
>doesn't end with a /, rather than just tacking on an element name on the
>end and hoping for the best. I've even seen it suggested somewhere that
>_everyone_ should distort their namespaec URI to account for this broken
>behaviour.


It is bizarre, isn't it?   I understand (I wan't there), that some people
wanted
the RDF property "red" to be http://example.org/colors/red  so that they
could dereference an individual
resource coresponding to the property itself. Others wanted someone seeking
information on "red"  to ask about http://example.org/colors#red . The
result
was - guess what? A compromise - whether you wanted "/" or "#" you had
to put it onto the end of the namespace name.  It works, as far as I can
see, but if it gets in
the way of (say) xml-schema making consistent URIs for elements, then I
could
see a transition strategy where we check few if any people actually use "/"
then
 transition RDF to (a) ignore a trailing "#" on the namespace URI and (b)
automatically insert one between the namespace and localname when generating
the URI-refeernce for an RDF property.

>David


Tim

Received on Sunday, 21 May 2000 09:14:48 UTC