Towards a classification of logical dependencies origins

Abstract

Logical dependencies are implicit relationships established between software artifacts that have evolved together. Software engineering researchers have investigated this kind of dependency to assess fault-proneness, detect design issues, infer code decay, and predict likely changes in code. Despite the acknowledged relation between logical dependencies and software quality, the nature of the logical dependencies is unknown in the literature.Most authors hypothesize about their origins, but no empirical study has been conducted to investigate the real nature of these dependencies. In this paper, we investigated the origins of logical dependencies by means of a case study involving a Java FLOSS project. We mined the project repository, filtered out irrelevant data based on statistical analyses, and performed a manual inspection of the logical dependencies to identify their origins using information from the revision comments, code diffs, and informal interviews held with the developers of the analyzed project. Preliminary results showed that logical dependencies involved files that changed together for a series of different reasons, which ranged from changing software license to refactoring classes that belonged to the same semantic class.

Uncovering the Hidden Co-evolution in the Work History of Software Projects

“…A solution addressing these requirements can partially build upon research in three main areas: i) work on Mining Software Repositories (MSR); ii) Process Mining (PM); iii) and software visualization. Only two labels for processes Zimmermann and Nagappan [26] Only functional dependencies Abate et al [1] b Only functional dependencies DAmbros et al [8] Oliva et al [15] Weicheng et al [23] Ruohonen et al [20] Lindberg et al [13] b Activity variations PM Kindler et al [12] Goncalves et al [9] Poncin et al [17] Beheshti et at. [4] b Mittal and Sureka [14] b Only bug resolution process Bala et al [2] b b Unlabelled Gantt chart Visualization Voinea and Telea.…”

“…Contributions in this area focus on the users and the artifacts, mining co-evolution or co-change of project parts [25,8] and network analysis of file dependency graph based on commit distance [26,1,23]. Hidden work dependencies are mentioned as logical dependencies [15]. Also techniques for trend analysis [20] and inter-dependencies between developers [13] are proposed.…”

“…After all the aggregated events ae have been explored, the algorithm moves on to computing the metrics (lines [13][14][15][16][17][18]. In this loop, the algorithm iterates through all the pairs of files.…”

Developers' perception of co-change patterns: An empirical study

“…Co-Change mining is used to predict changes [13], [32], to support program visualization [12], [33], to reveal logical dependencies [34], [35], to improve defect prediction techniques [36], and to detect bad smells [37]. Zimmermann et al propose an approach that uses association rules mining on version histories to suggest possible future changes (e.g., if class A usually co-changes with B, and a commit only changes A, a warning is raised recommending to check whether B should not be changed too) [13].…”

Recommending mentors to software project newcomers

“…System high level architecture is straightforward to extract the relationship among source code artifact and developer -'who changed what' -simply checking the history. It is also possible to infer logical dependencies among source code artifacts by checking the co-changes [13] -'which artifacts were changed and committed together'. Also using the source code repository it is possible to gather implicit relationships among developers using the method proposed by Cataldo et al [14] -'which developers cooperated by changing related artifacts'.…”