[Uml-devel] Correction for patch

Achim Spangler Achim.Spangler at mnet-online.de
Thu Mar 18 00:49:02 UTC 2004


Hi Oliver,
this means, that the object representation of actors and use-cases must be 
extended by information on the containing folder, or not?

As far as I interprete the meaning of packages, which are relating to the idea 
of namespaces and Java-Packages, the logical elements like use-cases 
shouldn't be nested in packages.
Thus we need a tag in the object description of logical elements, which can 
rule the automatic positioning of corresponding listviewitems - comparable to 
class-elements and packages.

I suggest, that my patch is integrated with a define and/or Doxygen-TODO tag. 
Thereby the standardised way can be changed back, as soon as the new 
folder-nest-attribute is defined for logical UML elements.

The current behaviour is a real bug, as I can insert use cases and actors in 
folders, but the load-from-file process can't reproduce the positioning. This 
is very unconvenient for big project with a lot of use cases, where folders 
are used to divide the items into ( functional ) smaller units.


Bye,
Achim


Am Mittwoch 17 März 2004 20:39 schrieb Oliver Kellogg:
> Hi,
>
> "Achim Spangler" <Achim.Spangler at mnet-online.de> wrote:
> > Hi,
> > I was too early with my assumption to have fixed the problem.
> > I detected too late, that the UseCase and Actor listitem entries are now
> > inserted at wanted folder. But I missed due to my big project, that there
> > is a double item outside the wanted folder.
> > [...]
>
> Please keep in mind that the aim is to make the <listview>
> tag in the XMI file optional since it is a non-standard tag.
> (This is for closing bugs.kde.org #56184.)
> Thus, reliance on UMLListView::loadFromXMI() should be
> _lessened_ not increased. Moreover, most of the info
> of the listview tag can be inferred from the UML objects
> in the XMI file.
>
> Unfortunately I don't have time for this currently -
> if nobody else picks it up it'll have to wait till
> later this year.
>
> Thanks,
>
>  Oliver




More information about the umbrello-devel mailing list