Citation Context

...g code. The study consisted of refactoring four different applications so that the code responsible for handling exceptions was moved to aspects. Three of these applications were originally written in Java and one was implemented in AspectJ. This study differs from the Lippert and Lopes study for a number of reasons. First, the targets of the study are complete, deployable systems, not reusable infrastructures, such as a framework. Hence, the exception handling code also implements non-uniform, complex strategies, making it harder to move handlers to aspects. Second, we employ a metrics suite [11] to quantitatively assess attributes such as coupling, conciseness, cohesion, and separation of concerns in both the original and the refactored system. Third, we evaluate how exception handling aspects interact with aspects implementing other concerns. Fourth, we do not attempt to move error detection code to aspects. Error detection involves checking the state of a program against a certain predicate when its control flow graph reaches a certain node, at runtime [7]. Thus, in many cases, error detection code is very strongly coupled with the code that implements a system’s normal behavior. W...

Citation Context

...n terms of code locality, reusability, composability and pluggability. Garcia et al. have used these implementations as a basis of their quantitative evaluation of the benefits and scalability of AOP =-=[5, 15]-=-. Their studies show significant improvements in the case of 13 out of 23 design patterns with regards to metrics such as separation of concerns, coupling, cohesion and size. These studies mostly repr...

Citation Context

...la AspectJ for this task. 1 Introduction While many studies have explored the capabilities of aspect-oriented programming (AOP) [1] to improve the modularity, customization, and evolution of software =-=[2, 3, 4, 5, 6, 7, 8]-=-, little is known on how AOP has been used. As AspectJ 1 is the most widely used AOP language, we want to know which AspectJ mechanisms are used, to what extent, and for what kinds of crosscutting con...

Citation Context

...ew. Along the same lines, Mehner proposes a series of steps to validate AOP metrics and their application [33]. An extensive study on modularizing design patterns have been performed by Garcia et al. =-=[20]-=-. They use Hannemann’s implementation of GoF patterns to apply seven metrics that extend and adapt to AOP Chidamber and Kemerer’s metrics [16]. Their metrics are informally defined and their results a...

Citation Context

...al studies have pointed out that the use of AO techniques can: (i) increase the number of undesirable concern couplings [12, 15], and (ii) decrease the cohesion of modules realizing a certain concern =-=[14]-=-. These concern modularity flaws make the change and removal of involved concerns error prone [11, 12] and lead to the manifestation of ripple effects [15]. Concern-driven design anomalies cannot be s...

Citation Context

..., an integration of AOP mechanisms into FOP [5]. AML preserve the advantages of FOP and prevent the difficulties associated to AOP (see above). 6. RELATED WORK Hannemann et al. [14] and Garcia et al. =-=[10]-=- transformed OOP implementations into AOP implementations to discuss the advantages of AOP mechanisms with respect to OOP. We extend their transformation to FOP and furthermore showed the duality of A...

Citation Context

...sess how applicable and useful the constructs we have used here really are when applied to a broader set of real-world problems. With that in mind, it would be interesting to relate the findings from =-=[7]-=- to an implementation of the same patterns in PT. 6. ACKNOWLEDGEMENTS The work reported in this paper has been done within the context of the SWAT project (The Research Council of Norway, grant no. 16...