Z-Order for graph edges

[Paolo F Cantoni]

There doesn't seem to be any way to apply Z-Order to GraphEdges (lines).

Why not?

Paolo

[thomaskilian]

Not needed so far? Or maybe it forces you to avoid crossings (which is not a bad intention IMHO).

[Eve]

I believe that the reason for this is because if connectors/GraphEdges/lines were to go under a diagram object it would appear as if they were going to that object, and since the whole point of diagraming is to convey a message, giving a wrong message like that would be much worse than not being able to set the Z-Order for them.

Simon

[Paolo F Cantoni]

Thanks Simon,

While I appreciate the point about lines going over/under shapes, there is also the problem of the tree style Generalisation.

Here, because of the way in which the current graphical design implements this feature, where you have any multiple inheritance, you get overlaying of lines so that the resultant diagram is hard to interpret.

You can go some of the way to alleviate this by using line appearance but there is still the problem of the relative Z-order of the lines with respect to other lines at the class connection points.

I have a pretty complex diagram in a particular area -since the concepts I'm modelling are, of their very nature, complex and have resisted previous analysis using narrative only.  

By using EA and UML I have been able to clarify matters for the users, but the diagrams are hard to follow because of this problem.  I haven't been watching the diagram long enough to note whether the implied (deduced) Z-order of the lines is "sticky", but I should still have control over them.

Paolo