UML14 — Aggregation is poorly defined

Summary: 1. Aggregation is poorly defined. It is not clear whether it is meant to be
configurational and/or invariant (reading different parts of the documents
give different signals). Instead it concentrates on lifetime dependency
and unique connections. I believe it is intended to be invariant but not
necessarily configurational. Firstly, that definition needs tightening and,
to me and Jim Odell, being configurational seems to be a prime requirement
for aggregation. Without it we really have a membership (yet still whole-part)
relationship.
Secondly, whatever the definition decided upon, its use in the metamodel must
be clear and clean. At present, I believe many of the "black diamonds" in the
metamodel really are not strong aggregations (whichever definition I try
to use).

Considered and declined.