Extension of Actual_Function_Binding

[khoroshilov]

We find usable to have Virtual Bus and Bus in applies to clause of Actual_Function_Binding definition.

[reteprelief]

I understand the virtual bus. We also want to map virtual processors. What is an example where you want to map a bus functionally onto something else?

[khoroshilov]

The root cause of that is more or less 3.0 topic.

The problem is that difference between BUS and VIRTUAL BUS is quite illusive. Once you model a concrete system usually you have quite good understanding what is physical component (BUS) and what is not. Keep in mind that BUS often represents not only physical part (wire) but some protocol on top of it.

It is not an unusual situation that a year later you get a design of NextGeneration system, where BUS implemented in different way, for example, partially in SW including code working in ARINC653 partitions. As a result you have several options:

• to update old models and analysis tools to support new understanding of decomposition between SW/HW;
• to have incompatible versions of old/new models and analysis tools;
• to stop to distinguish BUS and VIRTUAL BUS components.

So, the main question here what is difference between BUS and VIRTUAL BUS and which recommendations should be provided by AADL specification so we became able to develop stable AADL libraries of reusable model components with long life cycle.

As for Bus in applies to clause of Actual_Function_Binding, we do not insist on it.