Boundaries as browser visible objects

[Paolo F Cantoni]

There has been discussion over the years about how Sparx has (for their own reasons) NOT placed all elements on the browser.

We are finding in the enterprise space that boundaries are proliferating throughout our repository.  We might have 10 copies of the same boundary element (often with the same description!)  In addition, we observe that the ArchiMate_Grouping element is defined (again, for Sparx's own reasons) as an element of type=Component, when clearly it is conceptually more like a boundary and looks like a package.

We have extended the ArchiMate_Grouping and would like to use the element type=Boundary since it offers some benefits to our users.  BUT, they can't find them in the browser!

Is it time to ask Sparx to revisit this?  Perhaps an option to allow the "hidden types to be visible in browser".

Should I put in a feature request?

Thoughts?

Paolo

[qwerty]

Yes. To extend the options zoo it would be fine to have configuration switches as to what is shown in the browser. Extending the Relationship window uniting with the browser would be an idea. Other tools show even the relations of elements in the browser (making them more to 1st class citizens). So with the appropriate option one could still leave the info level for the browser as it is or make it a full featured browser which will not need  subordinary browser-helpers.

q.

[Glassboy]

I'm in two minds about this one.  I think exactly what being a first class citizen really means needs to be clearly defined.  I can see it causing more problems than it solves.

[Paolo F Cantoni]

I'm in two minds about this one.  I think exactly what being a first class citizen really means needs to be clearly defined.  I can see it causing more problems than it solves.

Can you clarify?  Are you talking in general or just about Boundaries?

If Boundaries, I see them as precursors to the ArchiMate_Grouping element and therefore an aggregate object of some kind with an identity.

If in general, then that's a more open question.

Paolo

[Glassboy]

I'm talking about Boundaries but Archimate groupings would be the same (if the shape script wasn't dreadful).  They are mostly used as a decoration and are used in different ways on different viewpoints.

People tend to use them as a soft owner which doesn't really matter as it's a decoration.  Exposing them to the browser makes them suddenly a hard owner, where it would be better to actually model the reality; e.g. a "macro" component owning the rest of the components.

[Paolo F Cantoni]

I'm talking about Boundaries but Archimate groupings would be the same (if the shape script wasn't dreadful).  They are mostly used as a decoration and are used in different ways on different viewpoints.

People tend to use them as a soft owner which doesn't really matter as it's a decoration.  Exposing them to the browser makes them suddenly a hard owner, where it would be better to actually model the reality; e.g. a "macro" component owning the rest of the components.

Yes, that's what we've done... They are much more formal.  It's not about ownership, it's about relationships (between the items and the grouping and between the items in the grouping) in general.

Anyway, that's why we need to see them on the browser.  It's too confusing for users.

Paolo

[Glassboy]

Yes, that's what we've done... They are much more formal.  It's not about ownership, it's about relationships (between the items and the grouping and between the items in the grouping) in general.

Yes but what is that relationship exactly?

[KP]

Yes but what is that relationship exactly?

In Archimate 2, the grouping element just indicates that the grouped elements "have something in common", so you the architect need to answer that question.

There's a slightly different emphasis in Archimate 3. The spec shows an example of a relationship from a grouping to some other element which is taken to be the equivalent to the same relationship from each member of the group to the target element.

[Glassboy]

Yes but what is that relationship exactly?

In Archimate 2, the grouping element just indicates that the grouped elements "have something in common", so you the architect need to answer that question.

The answer to that question is generally they're all rectangles :-)

[Paolo F Cantoni]

Yes but what is that relationship exactly?

As KP says, the user decides...

Firstly we've combined the idea of the UML Generalization Set and the ArchiMate Grouping into a single element the Grouping Set.  This defines a set of elements which are related together under that grouping.  Say all the elements are related by the fact that they are all assignments from another element (the origin) - the grouping set is then categorised as an Assignment set and decorated accordingly.  The example in the ArchiMate 3 specification shows a Realization Set (in our parlents) and is a way of collecting all the elements that have a realization (or equivalent) relationship to the service.  If you removed the Grouping, within the model you should still have the relationship between the leaf elements and the service.  The aggregation relationships are actually derived from the underlying model.  We ensure that he Grouping Set is correctly manifesting the underlying relationships\ and semantics.

That's the nutshell view.  (bus stop coming up)...

Paolo

[Glassboy]

My experience of these sort of things that if the relationships aren't explicitly documented and clear, the next set of architects make incorrect assumptions about what they're looking at.

[Paolo F Cantoni]

My experience of these sort of things that if the relationships aren't explicitly documented and clear, the next set of architects make incorrect assumptions about what they're looking at.

Exactly! 

We believe our specification, implementation, and the use of the automatic diagrammer (to manage the linkages between the Grouping Set object and the related items) makes everything explicit and validatable.  Using our implementation, you can look at a diagram and tell whether you are looking at all the items in the set or some subset thereof, the nature of the relationship between the set items etc.  We have also allowed a relationship between an element and a Grouping Set to apply to all the items in the set.

The specification and implementation is currently being revised, so I can't be more explicit at this time.

Paolo