IDENTIFICATION AND QUANTIFICATION OF VARIABILITY MEASURES AFFECTING CODE REUSABILITY IN OPEN SOURCE ENVIRONMENT

Open source software (OSS) is one of the emerging areas in software engineering, and
is gaining the interest of the software development community. OSS was started as a
movement, and for many years software developers contributed to it as their hobby
(non commercial purpose). Now, OSS components are being reused in CBSD
(commercial purpose). However, recently, the use of OSS in SPL is envisioned
recently by software engineering researchers, thus bringing it into a new arena. Being
an emerging research area, it demands exploratory study to explore the dimensions of
this phenomenon. Furthermore, there is a need to assess the reusability of OSS which
is the focal point of these disciplines (CBSE, SPL, and OSS). In this research, a mixed
method based approach is employed which is specifically 'partially mixed sequential
dominant study'. It involves both qualitative (interviews) and quantitative phases
(survey and experiment). During the qualitative phase seven respondents were
involved, sample size of survey was 396, and three experiments were conducted. The
main contribution of this study is results of exploration of the phenomenon 'reuse of
OSS in reuse intensive software development'. The findings include 7 categories and
39 dimensions. One of the dimension factors affecting reusability was carried to the
quantitative phase (survey and experiment). On basis of the findings, proposal for
reusability attribute model was presented at class and package level. Variability is one
of the newly identified attribute of reusability. A comprehensive theoretical analysis
of variability implementation mechanisms is conducted to propose metrics for its
assessment. The reusability attribute model is validated by statistical analysis of I 03
classes and 77 packages. An evolutionary reusability analysis of two open source
software was conducted, where different versions of software are analyzed for their
reusability. The results show a positive correlation between variability and reusability
at package level and validate the other identified attributes. The results would be
helpful to conduct further studies in this area.