displaying attribute tags & constraints

[steve.capell]

Does anyone know how to get a class diagram to show tagged values and constrainsts at attribute level?

[Geert Bellekens]

Time to revive this one.
I would very much like to do now.

Anyone?

Geert

[Geert Bellekens]

I guess I'll need to send in a feature request.

The problem I'm facing (and I'm sure I'm not alone) is that certain users want to see certain characteristics of attributes on a diagram.
In this instance they want to know which attribute are part of the "core" data of an element.
So in the meta model we added a property to our attribute stereotype that indicates being "core". Unfortunately there is no visual indication on a diagram which attributes are "core" data and which are not.

As a result of this limitation we see our users pushing for all kinds of dirty workaround:
- Why can't we use a steretoype "core"?
- Why can't we use the isID field?
- Why can't we sue the Collection type field?

All hideous workarounds that pollute our clean model, just because these other things happens to be visible on a diagram.

I though about writing a shapescript, but that seems an awful lot of work to reproduce the whole class shape just because of some indicators we need.

Geert

[Glassboy]

I think the danger is you start to overload things.  How many of these concerns are adequately addressed with the use of notes on the diagram?

[Eve]

Shape scripts don't have the capability of customizing the attribute text, so that's not an option.

I would extend your existing stereotype with a core stereotype instead of adding a new tagged value to your stereotype.

It sounds like these "core" attributes are a different type of attribute, so it would match the intention of stereotyping. In addition, in my opinion that's much cleaner than having every other attribute having a tag saying that "core" is false.

[Paolo F Cantoni]

Hi Geert,

Do you have a formal definition of what "core" means?

That should lead you to the type of workaround may be best.

Notwithstanding that, it's time that element features (including attributes) stopped being lower class (pun intended) citizens of the model - and allow us to modify the graphical aspects of the feature on the diagram.

Paolo

[Geert Bellekens]

"Core" (as far as I understood it) means that this is part of the minimum set of data to be communicated in messages.
Bu they have others like this too like an "authentic indicator". An NO these are not stereotypes, but properties of stereotypes.

We are only prone to solve it with multiple stereotypes because these are visualized on a diagram, not because its the right thing to do.
It feels like multiple stereotypes are somewhat similar to multiple inheritance. Using multiple inheritance has proven to be only for the happy few with an IQ of 130+. Everyone else just gets fricking confused

So please, give me a way to somehow visualize some tagged values on the diagrams.

Thanks

Geert

Geert

[Glassboy]

So please, give me a way to somehow visualize some tagged values on the diagrams.

Well

• Notes
• text elements
• Note linked to Tagged value

[Geert Bellekens]

So please, give me a way to somehow visualize some tagged values on the diagrams.

Well

• Notes
• text elements
• Note linked to Tagged value

The idea is that I only need to change the tagged value of the attributes, and that this becomes apparent in all diagrams where this attribute is used. (similar to things like "isID" or using multiple stereotypes, which are the things that are now often abused because of this)

I'm afraid none of the options above give me that.

Geert

[Glassboy]

Could you potentially use primitives?

[Geert Bellekens]

Could you potentially use primitives?

Can you explain what you mean by that?

Geert

[Glassboy]

Could you potentially use primitives?

Can you explain what you mean by that?

Geert

Create a primitive and place it on the diagram and use it as the Datatype for your attribute.  Then you can display whatever feature of the attribute you like with existing functionality.

[Geert Bellekens]

Could you potentially use primitives?

Can you explain what you mean by that?

Geert

Create a primitive and place it on the diagram and use it as the Datatype for your attribute.  Then you can display whatever feature of the attribute you like with existing functionality.

Nah, wouldn't work. The datatype is already used as well, but thanks for the suggestion.

Geert

[Paolo F Cantoni]

In a similar situation, I've "hijacked" Scope.  Not overloaded, hijacked.  Scope had no effective meaning in the context I was using for that type of feature - so I grabbed that.  That allowed me to suppress features on the diagram using Feature visibility.

HTH,
Paolo

[Geert Bellekens]

I see. In most situations I wouldn't worry about it too much and just hijack/abuse/overload any of the available properties to get the job done.

The problem with this client is that they are a government standards organisation. And it be can be hard to enforce standards when you are not following standards yourself 

Geert