Elsewhere (
http://www.sparxsystems.com/cgi-bin/yabb/YaBB.cgi?num=1119859101/7#7) I have put forward the proposition that the Association "lozenge" should be regarded as a specific
rendering of the Association Class (where the current binary AssociationClass is merely a
specific form of the n-ary Association element).
Now, I'm not sure when this functionality came into EA, but as of at least v7.1, when you convert an ordinary class into an AssociationClass, the attached Association also becomes an "AssociationClass" when you open it's properties.
Now this causes confusion for shape scripting - since if I create a shape script, am I attempting to apply it to the vertex or the edge? However, that's not what I'm discussing here.
It seems to me that when you convert an "ordinary" Class into an AssociationClass - there should be some indication that this has occurred. Similarly with the linked edge (which appears to have transitioned from Association to AssociationClass). Now some might argue that this latter transition is a bug, but I'm not so sure.
If the edge and the vertex are
both on the same diagram, then EA follows the UML requirement to link them with a Notelink. However, when both the edge AND the vertex are NOT on the same diagram, it would be useful to have a visual signal that the other (linked) element exists.
Trying to be consistent with my previous pronouncements, it seems to me that the addition of a small lozenge glyph in the top right of the Class and the addition of a small lozenge glyph in the centre of the Association would be sufficient to provide such an indication. These indicators would mean: "There is another part of the construct that is not shown on this diagram".
What do others think?
