Jump Links

[uml.modeller]

Hi,

I'm in the process of creating a relatively complex component diagram in which several assembly connector cross over each other. To avoid any ambiguity I want the crossing connector to do so using a small semi-circle or similar notation.

In Rational Software Architect I can do this via a setting called Jump Links in the properties panel. However, I cannot find a similar setting in EA 7.5. I'm assuming this must possible as its a pretty basic feature

Thanks in advance

Jonathan

[Luis J. Lobo]

By now, this is not available.

Try using different line colors, or use the auto-layouts possibilities.

[uml.modeller]

Thanks for the reply.

The auto layout algorithms aren't intelligent enough to resolve the problem unfortunately. I'm also reluctant to start introducing arbitrary colour schemes.

If I comes down to it I may just end up editing the exported image in GIMP, but that seems like a pretty poor option. I'm just surprised there isn't an option to do this.

Thanks

Jonathan

[Geert Bellekens]

Jonathan,

You could also look at it another way: The crossing connectors rule:

If you can't avoid crossing connectors on a diagram, there's too much on it

In other words, if you have this issue then it might be a better idea to find a way to split your diagram in more digestible parts.

Geert

[uml.modeller]

I've divided my design into a number of white-box subsystem components. The problem is that many of these subsystems rely on a set of common services. For example, a basic CRUD service, and a DeadLetterHandler service for erroneous JMS messages. I can actually lay out the diagram in such a way as to avoid the crossovers but this layout breaks a couple of rules I've been trying to follow. Namely, that subsystems are grouped according to their layer, and that dependent components are to the left of the services they require.

Breaking up the diagram is an option. If I was to go down that route I'd probably provide separate white-box diagrams for each subsystem, with an additional black-box diagram containing all the subsystems. Modelling all the subsystems as black boxes would make the nodes smaller, so I could be more liberal with my spacing.

Any thoughts?

[Geert Bellekens]

Breaking up the diagram is an option. If I was to go down that route I'd probably provide separate white-box diagrams for each subsystem, with an additional black-box diagram containing all the subsystems. Modelling all the subsystems as black boxes would make the nodes smaller, so I could be more liberal with my spacing.

Any thoughts?

That seems like a pretty good idea to me.

Geert

[uml.modeller]

One additional thought I had was related to providing an abstract subsystem to model the dependencies on common services. The problem is I'm not sure how the semantics of component generalisation works with respect to ports and assembly connectors. For example, if my abstract subsystem provides a port with required interfaces for the common services, how would I delegate to the components that realise a concrete subsystem? Or is that in fact possible?

[Paolo F Cantoni]

One additional thought I had was related to providing an abstract subsystem to model the dependencies on common services. The problem is I'm not sure how the semantics of component generalisation works with respect to ports and assembly connectors. For example, if my abstract subsystem provides a port with required interfaces for the common services, how would I delegate to the components that realise a concrete subsystem? Or is that in fact possible?

Hi Jonathan,

It might be best to move this question as a separate topic in the UML processes category.  It sounds like a good topic for that category - and makes it easier to search for if someone else  is interested.

Paolo

[uml.modeller]

No worries. I've started a new thread for this.

Regards

Jonathan