Thomas Neustupny wrote:
> Hi,
>
>
Hi,
our profiles are not 100% UML2 profiles, because (at least at that time)
we did not support UML2, only UML1.3,
so I implemented UML 1.3 profiles with some ideas from UML2.
> our concept of profiles is different from what the UML2 spec is talking
> about. In ArgoUML, any XMI represented model (so no diagrams) can be a
> profile, but in UML2 a profile is a different meta package instance with only
> stereotypes in it.
I don't think there's a constraint on the elements in a profile package,
if you see the definition of the profile package you'll notice that the
set of elements inherited from the package metaclass is not restricted
in any way...
In fact, the process we use to link a project to a profile is non
standard, maybe we can improve this .
> So we can't put e.g. datatypes in it. On the other hand, stereotype
> definitions can't be in a UML model, they can be only in UML2 profiles, so we
> must support UML2 profiles anyway.
>
>
Where's this constraint?
> How should we map our profiles to UML2 profiles?
>
> My suggestions:
> - let's distinguish our profiles from UML2 profiles
> - consider naming change: 'project profile' and 'UML profile'
> - our profile ('project profile') will no longer be a model instance
> - it will consist of two elements: a UML2 profile and a model
> - it is a new meta class instance with 2 attributes (profile+model)
>
> Is this feasible, or is there a simpler way?
>
> Is it correct that in such a project profile we can't stereotype elements,
> because this would make it depend on another profile? For example, we can't
> require to stereotype the model with <<profile>>, because the "root" profile
> can't apply a <<profile>> stereotype.
>
>
The "profile" stereotype is not a real stereotype, it is just syntactic
sugar for profile that is a subclass of package...
so there's no problem in using this syntax here.
*maas*
> Thomas
>
------------------------------------------------------
http://argouml.tigris.org/ds/viewMessage.do?dsForumId=450&dsMessageId=2357034
To unsubscribe from this discussion, e-mail:
[[email protected]].
To be allowed to post to the list contact the mailing list moderator, email:
[[email protected]]
