Re: I am he and you are me and we can all ID together

Brian McBride <bwm@hplb.hpl.hp.com> wrote:

>>> 3.    r2 is the resource named by the resource attribute if present or a new
>>> resource. If the ID attribute is given it is the identifier of this new
>>> resource.
>> As Dave Beckett kindly pointed out, this contradicts the previous definition
>> of the ID attribute, which was to provide an identifier for the reified
>> statement. Thus, the M&S is unclear on the meaning of this attribute, since
>> it defines it twice.
> 
> I assume Dave is referring to:

[ snip ]

Yes. Sorry for not including that in my original message.

>> Since this section specifies treatment of the attribute syntax, some
>> interpretations of the spec may believe that these rules are only to be
>> followed when the abbreviated syntax is used.
> Personally, I have interpretted the spec to mean that ID on a property element
> is treated differently when attribute syntax is used and when element syntax
> is used.  It seemed the only interpretation that worked.  I wonder what the
> different parsers do.

This would be possible, but seems difficult since the syntax for referencing
resources is identical in both element and attribute syntax. Because of
this, a Description that only had resources as objects would be exactly the
same whether using attribute syntax or not. Example:

<rdf:Description>
  <a:b rdf:resource="#a">
</rdf:Description>

> Having ID denote the reified statement in attribute syntax doesn't work in
> general as there may be more than one attribute, thus more than one statement.

I may be mistaken, but it was always my belief that an ID used in attribute
syntax would fall on the Description element and thus would be interpreted
as the ID of the subject, not the reified statement.

> So its my belief that the intent of the spec was to specify different
> interpretations for ID in these different circumstances, so I've added a
> reference to this thread under:

If this is true, then it should be clarified how implementers are to
determine which format is being used, since, as I have shown above, in some
cases they are indistinguishable.

> It may be that someone wants to resubmit it as an issue that its wrong for ID
> to be interpreted differently in to very nearly similar circumstances.

This is close to what I am showing here.

>> This may partially be the cause of confusion in
>> #rdfms-not-id-and-resource-attr and so may be related.
> Can you explain more clearly how these might be related.

I simply meant that the interpretation that the ID was naming the object
resource may have been the reason that the above issue was not allowed in
the grammar. If one was to use this interpretation, the ID would be naming a
resource which already had a URI, which would seem to be a mistake.

-- 
Aaron Swartz <me@aaronsw.com>| ...schoolyard subversion...
  <http://www.aaronsw.com>   |  because school harms kids
AIM: JediOfPi | ICQ: 33158237|  http://aaronsw.com/school/

Received on Monday, 16 April 2001 14:11:24 UTC