Abstract:
This paper reports on an empirical study aiming at investigating the influence of employing the Object Constraint Language (OCL) upon how early learners approach system development. OCL is an object oriented specification notation that is applied to Unified Modeling Language (UML) models. It is recognized as a valuable means of articulating design details beyond what is offered by UML itself.
Twenty six undergraduate second year Software Engineering students and ten postgraduate advanced software engineering students took part in this study as early learners of OCL. They were assigned to a relatively simple specification task. Two main solution types in the form of two models were suggested to the subjects as possible approach to the specification. In stage1 the solutions were informal and based on two models representing different data structures. In stage 2 the two models were represented in OCL. These models and solution types are briefly presented in this paper. The subjects were asked for their preferences for one against the other solution type in the two stages. A major shift in the solution preference between the two stages is observed and reported. More subjects opted for simple solution type rather than the generic solution type in stage 2 after considering the OCL representation.
The statistical results suggest that despite the formal representation increasing the awareness of the characteristics of a given specification problem, the notation appears to be detrimental in the consideration of good quality generic solutions. The different solution types that were used in the study are presented. A comparison to the data for the Z formalism is made and the implications for the choice of simple solutions are discussed.
Twenty six undergraduate second year Software Engineering students and ten postgraduate advanced software engineering students took part in this study as early learners of OCL. They were assigned to a relatively simple specification task. Two main solution types in the form of two models were suggested to the subjects as possible approach to the specification. In stage1 the solutions were informal and based on two models representing different data structures. In stage 2 the two models were represented in OCL. These models and solution types are briefly presented in this paper. The subjects were asked for their preferences for one against the other solution type in the two stages. A major shift in the solution preference between the two stages is observed and reported. More subjects opted for simple solution type rather than the generic solution type in stage 2 after considering the OCL representation.
The statistical results suggest that despite the formal representation increasing the awareness of the characteristics of a given specification problem, the notation appears to be detrimental in the consideration of good quality generic solutions. The different solution types that were used in the study are presented. A comparison to the data for the Z formalism is made and the implications for the choice of simple solutions are discussed.