1. Proceso kintamybių modeliavimas Modelling process variabilities Donatas Čiukšys

2. 2005-02-23Modelling process variabilities2 Key concepts Variability represents a capability to change or adapt a system, i.e. it abstracts from the varying qualities of the considered systems to capture them in generalized artefacts, controls the derivation of the family members from that artefacts Variabilities and commonalities are captured with the help of feature diagrams

3. 2005-02-23Modelling process variabilities3 Feature diagram

4. 2005-02-23Modelling process variabilities4 Key concepts Variation point is an identifier of one or more locations in a software asset at which the variation will occur [Jacobsen] Realization mechanism: architectural design patterns, aggregation, inheritance, parameterization, overloading, macros, conditional compilation, and dynamically linked libraries...

5. 2005-02-23Modelling process variabilities5 Capturing variability

6. 2005-02-23Modelling process variabilities6 Early and delayed variability

7. 2005-02-23Modelling process variabilities7 The Connection Between Variabilities, Variation Points and Assets

8. 2005-02-23Modelling process variabilities8 Key concepts Dependencies are restrictions on the variant selection of one or more variation points are indicated as a primary concern in software product families [JaringBosh2004] originate from the application domain (e.g. customer requirements), target platform, implementation details, or restrictions on quality attributes

9. 2005-02-23Modelling process variabilities9 Simple dependencies

10. 2005-02-23Modelling process variabilities10 Formalisation of simple dependencies [Michel Jaring and Jan Bosch, “Variability Dependencies in Product Family Engineering”, 2004] The set of all variation points: VP = {vp a, vp b, vp c, …} The set of variants for vp x : vp x = {v x1, v x2, v x3, …} The power set (the set of subsets) of all variants: V = {{v a1, v a2, v a3, …}, {v b1, v b2, v b3, …}, …} A relationship between vp x and v xn, i.e., vpx binds vxn: (vp x, v xn )

11. 2005-02-23Modelling process variabilities11 Formalisation of simple dependencies if vp x is bound then vp y should be bound: if vp x then vp y if vp x is bound then vp y should bind v yn : if vp x then (vp y, v yn ) if vp x binds v xn then vp y should be bound: if (vp x, v xn ) then vp y if vp x binds v xn then vp y should bind v ym : if (vp x, v xn ) then (vp y, v ym )

12. 2005-02-23Modelling process variabilities12

13. 2005-02-23Modelling process variabilities13

14. 2005-02-23Modelling process variabilities14 More details – variation point State (open & closed, may change from one development phase to the next): An open variation point is a variation point to which new variants can be added. A closed variation point is a variation point to which it is not possible to add new variants. Realization relationship: Variation points that have no associated realization mechanism in the product family artefacts are realized by variation points on a lower level of abstraction

15. 2005-02-23Modelling process variabilities15 More details - dependencies Validity – statically and dynamically analyzable dependencies Association to variation point: Predictable associations represent variation points whose influence of the variant selection on the validity of the dependency is fully known. The impact of variant selection on the validity of the dependency can be determined before the actual binding of the selected variant(s). Directional associations represent variation points whose influence of the variant selection on the validity of the dependency is not fully known. Instead, the dependency only specifies whether a (re)selection of variants will either positively or negatively affect the validity of the dependency. Unknown associations represent variation points of which it is known that the variant selection influences the validity of the dependency. However, the dependency does not specify how a (re)selection of the variants influences the validity.

16. 2005-02-23Modelling process variabilities16 More details - dependencies Type (logical, numerical, nominal): Logical dependencies specify a function valid, which yields the validity of the dependency for each selection of variants of the associated variation points. Variation points are only predictably associated to logical dependencies. Therefore, logical dependencies can always be analyzed statically. Numerical dependencies define a numerical value N. The value of N depends on the variants selection of the associated variation points. The validity of the dependency is expressed by the specification of a valid range on the numerical value N. These dependencies can have all three types of associated variation points. The new value of N after a (re)selection of a predictable associated variation point can be calculated. The dependency specifies whether the (re)selection of a directional associated variant will increase or decrease the value of N. It does not specify the influence of unknown dependencies.

17. 2005-02-23Modelling process variabilities17 More details - dependencies Nominal dependencies specify a set of categories. The binding of all variation points associated to the dependency map to one of the categories. The validity of the dependency is expressed by the specification of the valid categories. There are only unknown associated variation points associated to nominal dependencies. Dependency interaction – occurs if the process of resolving one dependency may affect the validity of other dependencies Example: dependencies “memory use” and “performance”

18. 2005-02-23Modelling process variabilities18 Variation management process Make Feature Diagram For each variation point: Abstraction level Assess binding time When it’s open Select Realization Technique + variant management technique (e.g. manual or automatic) Add variants Bind

19. 2005-02-23Modelling process variabilities19 Variability modelling notation UML extensions using standard extension mechanisms: lightweight: UML profile: Stereotypes, tags and tag values heavyweight: MOF: Variation Point as instance of UML Classifier Proprietary notation

20. 2005-02-23Modelling process variabilities20 Example of UML Profile

21. 2005-02-23Modelling process variabilities21 Example of proprietary notation

22. 2005-02-23Modelling process variabilities22 Requirements for variability modelling notation R1. Uniform and first-class representation of variation points in all abstraction levels. R2. Hierarchical organization of variability representation

23. 2005-02-23Modelling process variabilities23 Requirements for variability modelling notation R3. Dependencies, including complex dependencies, should be treated as first class citizens in the modelling approach. R4. The interactions between dependencies should be represented explicitly For efficient product derivation, software engineers require an overview on the interactions between dependencies to decide which strategy to follow when solving the dependencies.

24. 2005-02-23Modelling process variabilities24