[sc34wg3] TMDM / XTM 2.0 vs. XTM 1.0 "reification" procedure

[Lars Marius Garshol]

* Lars Heuer
>
> a) Topic with subject identifier "foo", name with item identifier
>    "foo" => no problem, no special meaning

Originally, this was how TMDM did reification. Later, reification was  
made explicit, but this remained the mechanism for expressing it in  
XTM. The post-Atlanta XTM version does not even use this mechanism  
any more.

> b) Topic with subject identifier "bar", topic with item identifier
>    "bar" => no problem, no special meaning

As you yourself noted, the equality rule for topics implies that  
these two topics must merge.

I think you've raised a very good point here, and one that bears  
considering. I think it's probably worth going all the way back to  
first principles on this, to make sure it still makes sense.

If you are in situation a) or b) you've got a topic whose subject  
indicator is another topic map construct. Since the subject indicator  
is supposed to be a resource indicating what the subject of the topic  
is, clearly this means something beyond there simply, coincidentally  
being two equal URIs in the topic map in different properties.

There is also the issue that XTM 1.0 used, and will forever continue  
to use, this particular mechanism for indicating reification, which  
isn't entirely inconsistent with the semantics of these constructs.

To me it seems that the correct thing to do here would be to:

   1) keep the equality rule for topics as it is, and

   2) add a constraint stating that your case a) is not allowed.

The arguments against this seem mostly to be that:

   - TMDM has gone to FDIS ballot, and if it passes we really  
shouldn't be
     changing things any more (my counter-argument is that I think  
this is
     a bug fix, and therefore acceptable), and

   - that this means one more uniqueness rule for the implementations to
     maintain, which is far from free (to which my counter-argument  
is that
     I don't think we have any choice, given past history, and the  
semantics
     of these constructs).

Any thoughts on this?

--
Lars Marius Garshol, Ontopian               http://www.ontopia.net
+47 98 21 55 50                             http://www.garshol.priv.no