Re: UML 15 association and CompositeViolation

Thanks Stefan,

I missed the [0..1]  for namespaces:(  The JMI impl seems to accept two 
composite relations to the same container.

/anders

Stefan Ocke wrote:

>----- Original Message -----
>From: "Anders W. Tell" <[email protected]>
>To: <[email protected]>
>Sent: Tuesday, August 19, 2003 12:50 PM
>Subject: Re: [mdr-dev] UML 15 association and CompositeViolation
>
>
>  
>
>>Thansk Stefan,
>>
>>Yes I noticed that, setting the namespace to the association work
>>though. Do you know where in the UML spec I can find that Ends dont have
>>a namespace ?
>>
>>    
>>
>
>On page 2-14, you can see that the relation between Association and
>AssociationEnd is a composition. The semantics of a composition is, that
>there can be at most ONE composit.
>The multiplicity at Association (1 , NOT 0..1) states that every
>AssociationEnd must have an Association as composites. So, it is implicit,
>that there can be no other composites than Association, especially not a
>Namespace. (page 2-13 shows, that the relationship between Namespace and
>ModelElement is a Composition as well, but with multiplicity 0..1, which
>means that a modelelement may have a namespace.)
>
>Stefan.
>
>
>
>  
>