Separating features using colored snippet graphs

Abstract

In feature-oriented software product lines, two complementary approaches to implement features exist: compositional and annotative ones. To combine the respective advantages, a unified representation of the two on the file-system level is desirable, yet unavailable. We pick up this demand and propose a novel concept for separating features using colored snippet graphs instead of directories and files. In particular, our proposed ideas are based on snippets, ASTs and coloring, and enable the implementation of product lines with both compositional and annotative views on real documents. As a basis for future tools, we provide a conceptual overview of our approach, discuss our ideas for leveraging snippets, and demonstrate applicability using an example scenario.

Integrating approaches for feature implementation

“…Moreover, CIDE only supports adding feature details, not removing or replacing them [25] [5]. We build upon the ideas introduced with CIDE, but omit visual markers and support the overriding of feature details [7].…”

“…Thus, we came up with the ideas presented in this proposal. Moreover, we provided a first step towards an integration by means of a vision paper [7].…”

“…In detail, we define that a snippet represents either an organizational unit or in case of source code an AST element. To outline the illustrated concepts, we proceed from bottom to top [7].…”

“…1a). These mechanisms are valuable for (1) the avoidance of traversing wrappers in terms of a color mismatch; (2) the local detection of ordering issues; (3) the realization of homogeneous crosscutting concerns [7].…”

“…We have provided an initial step towards answering RQ1 in prior work [7]. There, we have introduced (1) the idea of a snippet-based approach to separate features, (2) the organizational units, and (3) the variability-aware source code representation.…”