Re: inherited composition

Responding to Wettin...

Given

 [X]<#>----[A]<|----[B]----<#>[Y]

does above imply that [B] is associated to [X] XOR [Y]?

I think UML is a tad vague on this issue. It depends on how one interprets...


Compare to:

[X]<#>----[B]----<#>[Y]

....this. One can argue this is an XOR situation where a B is part of an X or a Y and it may be swapped between them. That is based on the premise that a Part can only be part of one Whole at a time. The UML description seems to support this view, albeit in a quite vague way.

OTOH, one can envision situations were a Part might legitimately be a part of different Wholes at the same time. For example,

[WheelAssembly]
     <*>
      |
      |
  [BrakeShoe]
      |
      |
     <*>
[BrakingSystem]

for an automobile where [WheelAssembly] is the thing on the end of the axle rather than the thing with the tire on it. One would have a hard time rationalizing some sort of nesting where a [WheelAssembly] was a part of a [BrakingSystem], so the Wholes are unrelated. Yet the BrakeShoe is a valid part of both. I would hope the UML definition was not intended preclude this sort of thing. But that would eliminate the XOR view.


*************
There is nothing wrong with me that could
not be cured by a capful of Drano.

H. S. Lahman
hsl@xxxxxxxxxxxxxxxxx
Pathfinder Solutions  -- Put MDA to Work
http://www.pathfindermda.com
blog: http://pathfinderpeople.blogs.com/hslahman
(888)OOA-PATH



.