A reverse‐engineering approach to subsystem structure identification

Abstract

Reverse‐engineering is the process of extracting system abstractions and design information out of existing software systems. This process involves the identification of software artefacts in a particular subject system, the exploration of how these artefacts interact with one another, and their aggregation to form more abstract system representations that facilitate program understanding. This paper describes our approach to creating higher‐level abstract representations of a subject system, which involves the identification of related components and dependencies, the construction of layered subsystem structures, and the computation of exact interfaces among subsystems. We show how top‐down decompositions of a subject system can be (re)constructed via bottom‐up subsystem composition. This process involves identifying groups of building blocks (e.g., variables, procedures, modules, and subsystems) using composition operations based on software engineering principles such as low coupling and high cohesion. The result is an architecture of layered subsystem structures. The structures are manipulated and recorded using the Rigi system, which consists of a distributed graph editor and a parsing system with a central repository. The editor provides graph filters and clustering operations to build and explore subsystem hierarchies interactively. The paper concludes with a detailed, step‐by‐step analysis of a 30‐module software system using Rigi.

Using software evolution to focus architectural recovery

“…Rigi [25] also has a concept of omnipresent nodes, which is seemingly similar to the notion of Focus's domain classes. However, the two are defined and used in very different ways.…”

Applying Financial Time Series Analysis to the Dynamic Analysis of Software

“…But the key problem is to choose the relevant set of criteria (or similarity metrics) (Wiggerts, 1997) with which the "natural" boundaries of components can be found. In the reverse-engineering literature, the similarity metrics range from the interconnection strength of RIGI (Muller et al, 1993) to the sophisticated information-theory based measurement of (Andritsos, Tzerpos, 2003, 2005, the information retrieval technique such as Latent Semantic Indexing (Marcus, 2004) or the kind of variables accessed in formal concept analysis (Siff, Reps, 1999) (Tonella, 2001). Then, based on such a similarity metric, an algorithm decides what element should be part of the same cluster (Mitchell, 2003).…”

“…Therefore, many techniques have been proposed to recover the high level architecture from legacy code. Although the early approaches dealt with the static analysis of the source code, such as the well known fan in and fan out metrics of RIGI (Muller et al, 1993), dynamic analysis techniques have recently attracted much interest in the academic community. The central idea behind these techniques is to execute the legacy software following its use-cases and harvest the time-ordered list of methods or functions that have been executed: the execution trace (Andritsos, Tzerpos, 2003).…”

Fragmentation de graphes et applications au génie logiciel /

“…Les auteurs des systèmes logiciels d'ingénierie inverse Rigi [58] FlG. 5.14 -Deux partitions Pi et Pi ayant le même graphe quotient GQ.…”