Object representation

[thomaskilian]

Hi,
another dumb question. If I create a boundary using the icon from the analysis or communication toolbox it places an object on the class diagramm. Placing a class with <<boundary>> stereotype brings me exactly the same representation, but I do have a class now. Shouldn't the object be distinguished by an underlined name

[sargasso]

However, it is a class!

Further to that, if you change one of the objects to a class it quite properly still isn't underlined.

Finally, if you turn off Use Stereotype Icons in the diagram the underlining is OK.

In summary, the underlining of instances when Use Sterotype Icons is ON is not "doing the right thing".

B

[thomaskilian]

However, it is a class!

Do you mean the icons from the toolbox? I can't agree since they create objects (that's what the properties window tells me).

not "doing the right thing"

So I'll report a bug for that.

[sargasso]

Do you mean the icons from the toolbox?

No I meant the stereotyped class.  The toolbox boundary, controller, entity icons all create objects.  Which as you say, "should" be underlined.

On rebrowsing the Superstructure I find no mention of stereotype icons as alternate representations though - so I guess whatever the tool implementer provides is OK .... that's what I meant by "doing the right thing".

rgrds
Bruce

[thomaskilian]

I mailed with Sparx regarding this topic. In general they follow your opinion. Since robustness diagrams are used separately from class diagrams the context should make clear that these elements are stereotyped objects, not classes. So if nobody else asks for a change they leave it unchanged.

[sargasso]

Just realised I didn't post this before.  

If you REALLY need the underlines the element context menu for the object wil let you set the underline on in the Set Font form.

( because I use this myself so often I forgot it. - Quite often I end up converting "abstractised"  :-/ elements from the business object models to classes for use in the logical class model.)

cheers
Bruce

[thomaskilian]

Just browsed through the UML 2.0 spec:

Notation
An instance specification is depicted using the same notation as its classifier, but in place of the classifier name appears an
underlined concatenation of the instance name (if any), a colon (‘:’) and the classifier name or names. If there are multiple classifiers, the names are all shown separated by commas. Classifier names can be omitted from a diagram.

I could not find any word about not underlining stereotyped objects.
Still I think I can live with not underlining because one can distinguish by the diagram type. On the other hand EA does not seem to behave correctly.