An Assessment of Maintainability of an Aspect-Oriented System

Table 1

Summary of maintainability metrics.

Study	Dependent variable	MetricstTested	Summary of result
Burrows et al. [9]	Fault-proneness	All Ceccato and Tonella metrics	In addition, they introduced a new metric base aspect coupling (BAC), which measures the coupling between base class and aspect. The study showed that the two metrics that displayed the strongest correlation to faults were CDA and BAC
Eaddy et al. [10]	Fault-proneness	DOSC, DOSM, CDC, CDO	Assessed the correlation between faults and crosscutting concerns. They found out that the more scattered a concern is the more faults in its implementation are. Concern metrics used to predict the scattering of a concern. These metrics are independent of the program size
Kumar et al. [11]	Changeability	WOM	Assessed the correlation between changeability and WOM metric. They found that the WOM can be used as an indicator of maintainability but it is a weak indicator. Change impact is less in AOP systems as compared to OO systems. Maintenance effort was measured in terms of the number of modules changed
Kulesza et al. [12]	Coupling, cohesion, separation of concerns	Sant’Anna metrics which includes LCOO, WOC, VS	VS and WOC cannot be used as predictors of maintainability as the increase in such metrics was always accompanied by less development effort. LCOO metric inconclusive for measuring maintainability
Shen et al. [13]	Changeability (coupling and maintenance tasks)	Ceccato and Tonella metrics for coupling (CFA, CMC, RFM, CAE, and CDA)	Coupling metrics correlated with maintainability.
Przybyek [14]	Modularity	CBO and LCOM	CBO and LCOM used to measure the modularity of a system. Aggregate coupling and cohesion should not be considered as coupling should be measured independent of the number of modules in the system.