Sparx Systems Forum
Enterprise Architect => Bugs and Issues => Topic started by: MagnusH on December 18, 2009, 05:34:56 pm
-
Hi,
When using an actor in a diagram and put a stereotype on that actor the stereotype name is not shown in the diagram. Is there a reason for this? (I have not disabled it for the diagram, other stereotypes for packages are shown as expected)
// Magnus
-
Magnus,
Whether or not EA displays the stereotype(s) depends on the notation you use.
If you use the stickman notation then no stereotypes will show. Use the rectangular notation and you'll see the stereotype(s).
I think the idea behind this behavior is that the stickman representation already is an indication of the stereotype (<<actor>>) so there is no more need to show the stereotype.
If this is the reasoning, it is obviously flawed.
Geert
-
Actually Geert, I think this is a UML flaw, not particularly an EA fault.
Actor was "always" a stereotyped classifier. It may be different nowadays. But, it also had a specific notation i.e. the stickman and specific notations were "always" stereotype-specific.
However, to get back to the OP, I am wondering why you would need to stereotype a classifier which is by definition external to your model? Your system, "in toto", knows nothing about this beast apart from the fact that it is a something that can perturb its' well-being, if not its' equilibrium. So why is there a need to differentiate?
Putting this another way, "you" (the system) need to know nothing about an actor that invokes a use case, apart from how to interact with "it" in terms of a stimulus/response pair.
bruce
-
[size=18]...[/size]I think the idea behind this behavior is that the stick man representation already is an indication of the stereotype (<<actor>>) so there is no more need to show the stereotype.
If this is the reasoning, it is obviously flawed.
Actually Geert, I think <<actor>> is a keyword, not a stereotype (Superstructure 16.3.1 Actor (from UseCases)). Unfortunately UML uses the same notation for keywords as stereotypes. Then it compounds the problem by allowing stereotypes and keywords to be (effectively) intermingled - talk about conflation...
Obviously, I agree that any stereotypes added to an actor should be visible.
HTH,
Paolo
-
Bah! Go back and read the 1.1 spec! 'Twas and AFAIC 'tis a classifier that "you know nothing about".
(Mod, I wasn't being personal)
-
Thanks for the replies!
What I try todo is not in any sense "UMLish" but rather draw a diagram showing something non-tech. The sticky man is nice to use in that case but I want to have different "roles" and that's the reason I want to use stereotypes to show the different roles.
I suppose I have to use the shape editor to achieve what I want. No big deal.
// magnus
-
The "roles" that an actor plays within your system can be much easier (IMO) be modeled using generalization.
Actor <--- "extraspecial_user_role"
^
|--- "general_user" <-- "teller" <-- "supervisor" <-- "manager"
etc
BTW, I have always found that doing this and then "coloring" in the various roles' "heads" is a lot easier than shapeshifting (or whatever it's called)
-
[size=18]...[/size]I think the idea behind this behavior is that the stick man representation already is an indication of the stereotype (<<actor>>) so there is no more need to show the stereotype.
If this is the reasoning, it is obviously flawed.
Actually Geert, I think <<actor>> is a keyword, not a stereotype (Superstructure 16.3.1 Actor (from UseCases)).
Paolo,
That's exactly the reason why I said the reasoning is flawed. (I could have explained myself better though.) ;)
Geert
-
The "roles" that an actor plays within your system can be much easier (IMO) be modelled using generalization.
Actor <--- "extraspecial_user_role"
^
|--- "general_user" <-- "teller" <-- "supervisor" <-- "manager"
etc
Agreed! But that having been said, what you've done is describe types of actors. In my modelling I've also found a lot of use for specifying the nature of the actor. For this you need stereotypes.
Besides as Magnus says, it's a rendering issue. Especially, if you have the [ ] Strict UML option unmarked, Actors should behave consistently with other classifiers.
BTW, I have always found that doing this and then "colouring" in the various roles' "heads" is a lot easier than shapeshifting (or whatever it's called)
You can colour the heads via stereotyping (as you know) - with or without ShapeScripts.
Paolo