[mdr-users] UML relationships and well-formed UML models

Hi all,

I have a question about the rules concerning where a UML relationship
may appear in a model, with respect to the elements participating in the
relationship.  It has come up in a ArgoUML discussion about the
possibility of supporting concurrent model development on top of MDR,
where a model would be partitioned into sub-models that could modified
in isolation.

If anyone is interested in the actual discussion about concurrent model
development in ArgoUML, you can visit the following link:

http://argouml.tigris.org/issues/show_bug.cgi?id=3497

The following is my conjecture: "A relationship must be in the same
namespace as the elements participating in the relationship, or in some
containing relationship".  For example, given a class "A" and another
class "B", a relationship between A and B must be in a namespace that
contains both A and B.

Is my conjecture true? 

If it is, is it because of a rule in the UML spec or the XMI spec, or
both?


And finally, out of mere curiosity, what would happen if I tried to
create a relationship that violated the rule in MDR (assuming that it is
a rule)?

One last note. If there is anybody on this list that would like a open
source UML tool to support concurrent model development, you should join
the ArgoUML project http://argouml.tigris.org/  and then vote for the
feature by visiting the link above.  If you have any questions on how to
do this, please contact me directly.


cheers, 

roy