LIPN: comment y aller
Abstract: UML 1 poses a lot of problems concerning its semantics and a lot of work on that can be found in the literature, the new UML 2.0 has solved some of those problems. Our recent work on the UML construct of the collaborations, added in the UML 2.0, however, brought up that also the syntax of the UML is quite problematic.
Indeed, it is quite difficult to answer questions about the legal use of UML constructs by looking at the OMG specification documents [1] (“the arcane details” in the words of [2]). More generally, it is extremely difficult to come out with an explicit metamodel related to a construct, due to the very nature of the OMG specification built over many levels of indirect references and specializations (you may easily count up to 30 levels of specialization!) and this poses a serious problem to a developer, as it is recognized by the UML inventors, who in [3] suggest to a developer building his/her own, usually simplified, conceptual model of the constructs of interest. Their suggestion is confirmed and reinforced by our own experience in projects and SE courses using the UML. Hence, we have decided to come out with a simplified metamodel dealing with the most desirable and useful features, cutting down the almost inextricable complication of too many indirect references.
This “conceptual metamodel” of the UML should be also the basis for a precise definition of the (formal) semantics of the UML, helping to concentrate on the real problems (and, yes, there are still problems on the semantics of the UML 2.0) without the rumors due to “the arcane details” of the official metamodel.
In the talk we will present the fragments of our new simplified metamodel for the UML related to the most used constructs with the associated formal semantics.
[1] UML Revision Task Force: OMG Unified Modeling Language (OMG UML), Superstructure, V2.1.2. (2007)
[2] Booch, G., Grady, J., Jacbson, I.: The Unified Modeling Language Reference Manual (Second Edition). The Addison-Wesley Ob ject Technology Series. Addison- Wesley (2004)
[3] Booch, G., Grady, J., Jacbson, I.: The Unified Modeling Language User Guide (Second Edition). The Addison-Wesley Ob ject Technology Series. Addison-Wesley (2005)
Abstract: The verification of High-Integrity Real-Time systems combines heterogeneous concerns: preserving timing constraints, ensuring behavioral invariants, or specific execution patterns. Furthermore, each concern requires speciÞc veriÞcation techniques; and combining all these techniques require automation to preserve semantics and consistency. Model-based approaches focus on the deÞnition of rep- resentation of a system, and its transformation to equivalent representation for further processing, including verification and are thus good candidates to support such automation. In this paper, we show there is a strong requirement to automatically map high-level models to abstractions that are dedicated to specific analysis techniques taking full advantage of tools. We discuss this requirement on a case study: validating some aspects of AADL models using both coloured and time Petri Nets.