The difference between aggregations and compositions in the DATEX model seems obscure. There is no ecognisable benefit of using these two different UML constructs. As the model is not an object or computational model, but a model of tarnsfer syntax for data exchange, substructures are always fully contained and under control of the super structure. Hence I propose to use composition only and turn all aggregations in th emodel to compositions.
DATEX II v3.0 will no longer use aggregations, only compositions.