Using Hyper/J to implement Product-Lines: A Case Study

Aspect-Oriented Programming (AOP) is an emerging technology whose goal is to modularize concerns that may involve several classes. The purpose of this report is to describe how one of the main representatives of AOP, namely Hyper/J, was used to implement a simple yet illustrative product-line of graph algorithms. 1 The Graph Product Line (GPL) The Graph Product-Line (GPL) [5] is a family of classical graph applications that was inspired by work on software extensibility [4, 8]. GPL is typical of product-lines in that applications are distinguished by the set of features that they implement, where no two applications implement the same set.1 Also typical is that applications are modeled as sentences of a grammar. Figure 1a2 shows this grammar, where tokens are names of features. Figure 1b shows a GUI that implements this grammar and allows GPL products to be specified declaratively as a series of radio-button and check-box selections. The semantics of GPL features, and the domain itself, are straightforward. A graph is either Directed or Undirected. Edges can be Weighted with non-negative numbers or Unweighted. Every graph application requires at most one search algorithm: breadth-first search (BFS) or depth-first search (DFS); and one or more of the following algorithms [2]: • Vertex Numbering (Number): Assigns a unique number to each vertex as a result of a graph traversal. • Connected Components (Connected): Computes the connected components of an undirected graph, which are equivalence classes under the reachable-from relation. For every pair of vertices x and y in an equivalence class, there is a path from x to y. • Strongly Connected Components (StronglyConnected): Computes the strongly connected components of a directed graph, which are equivalence classes under the reachable-from relation. A vertex y is reachable form vertex x if there is a path from x to y. 1. A feature is a functionality or implementation characteristic that is important to clients [3]. 2. The grammar does not preclude the repetition of algorithms, whereas the GUI does.