Legacy Issue Number: 1390
Summary: 2. In the same vein, it is clear that the "inheritance" connection used
in all parts of the metamodel is in
fact a generalization relationship. I believe this to be the best choice.
However, I would ask whether ALL of the generalization relationships in the
metamodel really do fulfil the criterion of generalization i.e. we can
say yes to the question "is the subclass A KIND OF the superclass?".
My biggest query here, which I have asked many times, is "Is a Generalizable
Element a kind of Namespace?" as shown in Figure 6 of semantics document.
Grady"s answer to me was "the simple reason is that superclasses form
a namespace". Forming a namespace sounds to me like aggregation or
membership and not generalization.
Reported: UML 1.1 — Tue, 19 May 1998 04:00 GMT
Disposition: Resolved — UML 1.2
Considered and declined.
Updated: Fri, 6 Mar 2015 21:35 GMT