[sc34wg3] TMCL 6.7 Overlap Declaration, Global Validation Rule

Lars Marius Garshol larsga at garshol.priv.no
Sat Jun 20 04:03:16 EDT 2009

* Patrick Durusau
> Err, but isn't the "overlap" of a supertype/subtype relationship  
> between
> two topics different from say the overlap of a topic being of type
> father and also of type employee?

It's different, but the two are related.

> That is to say that shouldn't I be able to distinguish between saying
> that t2 may be a subtype of t1 versus t2 may overlap with t1?

You are able to distinguish, of course, because they are different  
association types. So there's no confusion of the two.

> As I read the declaration and global validation rule, those two cases
> are being treated as the same?

No, what the rule is saying is that the following (assuming it's a  
complete topic map) is not OK:

   lmg isa person; isa creature.

Because unless explicitly declared to be OK, two topic types are not  
allowed to overlap.

However, the following *is* allowed:

   person ako creature.
   lmg isa person.

Even though by the semantics of supertype-subtype (ako) lmg is now  
both a person *and* a creature.

The reason we have that extra check in the global validation rule is  
that supertype-subtype states not just that the sets of instances  
overlap, but that in fact one is a subset of the other. (Every person  
is a creature, is what the subtyping is saying.)

So you could say that supertype-subtype implies an overlap, but has  
stronger semantics.

--Lars M.

More information about the sc34wg3 mailing list