A document-like software visualization method for effective cognition of c-based software systems

Abstract

It is clear that maintenance is a crucial and very costly process in a software life cycle. Nowadays there are a lot of software systems particularly legacy systems that are always maintained from time to time as new requirements arise. One important source to understand a software system before it is being maintained is through the documentation, particularly system documentation. Unfortunately, not all software systems developed or maintained are accompanied with their reliable and updated documents. In this case, source codes will be the only reliable source for programmers. A number of studies have been carried out in order to assist cognition based on source codes. One way is through tool automation via reverse engineering technique in which source codes will be parsed and the information extracted will be visualized using certain visualization methods. Most software visualization methods use graph as the main element to represent extracted software artifacts. Nevertheless, current methods tend to produce more complicated graphs and do not grant an explicit, document-like re-documentation environment. Hence, this thesis proposes a document-like software visualization method called DocLike Modularized Graph (DMG). The method is realized in a prototype tool named DocLike Viewer that targets on C-based software systems. The main contribution of the DMG method is to provide an explicit structural re-document mechanism in the software visualization tool. Besides, the DMG method provides more level of information abstractions via less complex graph that include inter-module dependencies, inter-program dependencies, procedural abstraction and also parameter passing. The DMG method was empirically evaluated based on the Goal/Question/Metric (GQM) paradigm and the findings depict that the method can improve productivity and quality in the aspect of cognition or program comprehension. A usability study was also conducted and DocLike Viewer had the most positive responses from the software practitioners.