Ausgeschriebene Abschlussarbeiten

Context

A software product line (SPL) comprises all software systems that can be created based on a common infrastructure including tools and software-related artifacts. Thus, an SPL supports the economic production of highly specialized software systems.

Similarities and differences of software systems of a SPL are described by so-called features. A feature is an important characteristic for a stakeholder. The systems of a SPL have common features and differ by variable features.

There is a high demand to find the parts of a software artifact, e.g. source code, that correspond to a certain feature. This problem is called feature location.

Recent research by Eisenecker & Müller introduces a formal approach to feature isolation as a necessary prerequisite for feature location (https://doi.org/10.1016/j.jss.2023.111938). The following topics for bachelor or master theses continue research based on this new approach.

Topics

a) Set operations for flat strings

A string is of course a sequence of characters. Therefore, one could consider the individual characters as ELements of a set. However, to calculate the intersection of two flat strings, the result of a diff operation (see en.wikipedia.org/wiki/Diff for more information) can also be used. In this paper, it shall be investigated which possibilities there are to compute the intersection of two strings using diff and related approaches. Furthermore, it will be investigated whether the analyzed approaches can also be used to define a union of flat strings and perform a set difference on the results of such processing.

b) Feature location for flat strings vs. AST

Assuming that the aforementioned work yields promising results, a comparison between the implementation of a feature location based on flat strings and a feature location based on an abstract syntax tree shall be performed. For this purpose, an SPL in the Java programming language could be used. The results presented in Müller, R., & Eisenecker, U. (2019): A Graph-Based Feature Location Approach Using Set Theory, In 23rd Systems and Software Product Line Conference SPLC '19 (pp. 161-165), Paris, France: ACM, Doi:10.1145/3336294.3342358, are based on the analysis of Java source code. Therefore, this could be a viable basis for performing such a comparison.

c) Feature Location with unsuitable number or unsuitable categories of features.

If the previous two theses provide technically sound results, they should serve as a basis for performing feature location with incorrect assumptions about categories and number of features. This will allow to evaluate the robustness of feature isolation with respect to feature location in practice.

d) Evaluation of (Fuzzy) parsers for C++ (including preprocessor directives)

In this work, (fuzzy) parsers that can process C++ source code, ideally including preprocessor directives, and that are available under a FOSS license are to be investigated. The parsers found are to be evaluated with a sample of specific C++ source files. The accuracy of the parsing results and the possibility of further processing the results are important criteria.

Prerequisites

Work on all topics requires a high level of formal thinking and a willingness to deeply explore the topic of feature location. Implementations should ideally be in C++, unless Java is specifically desired.

Contact

If you have questions or are interested in any of the topics, please contact Prof. Eisenecker