Paper Abstract

Many software systems have been developed as single products before Software Product Lines (SPLs) have emerged. Although some promising approaches have been proposed, extracting an SPL from existing software products is still expensive and time consuming. This paper presents an exploratory study that relies on a test-based SPL extraction from systems already developed. We aim to evaluate testing as the main mean to locate feature code and different sorts of existing artifacts to support the test-based location. We conduct two case studies starting from the derivation of the SPL feature model to the feature code location. Our preliminary results indicate (i) good rates of precision for feature seed location, where seed means a small portion of the feature code that allows the identification of the remaining portion, and (ii) good rates of recall for locating the whole feature code.

Introduction

This page contains the documentation used and the data collected on a exploratory study on SPL extraction by using a test-based technique.

The data is organized in three Sections.

One Section is dedicated to documentation files, which are shown in tabs as follows:

• JBook Tab: contains the JBook files and links .
• WebStore Tab: contains the WebStore files and links.

A second Section is dedicated to shadowing files, which are shown in tabs as follows:

• Manual Shadowing Tab: contains the feature reference code used in this study.
• Test Shadowing Tab: contains the shadowing collected with the execution of the test suites.
• Comparison Shadowing Tab: contains the shadowing resultant of the comparison between a "Manual shadowing" and a "Test Shadowing".

A third Section is dedicated to the metric collection files, which are shown in tabs as follows:

• JBook Tab: contains the JBook files.
• WebStore Tab: contains the WebStore files.

Finally, we enumerate the involved people and made some acknowledgments.

(top)

Documetation

Shadowings

• JBook
• WebStore