Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

A comparison of maintainability measures using the interactive combination of metrics

: Hernandez, Elena; Aldekoa, Gentzane; Knodel, Jens

Volltext urn:nbn:de:0011-n-692938 (899 KByte PDF)
MD5 Fingerprint: 4f93169ecc89c3c7178610b44d15ab31
Erstellt am: 12.2.2008

Kaiserslautern, 2007, VIII, 47 S. : Ill., Lit.
IESE-Report, 068.07/E
Reportnr.: 068.07/E
Bericht, Elektronische Publikation
Fraunhofer IESE ()
configuration management; goal question metric approach; maintainability; metric; quality assurance; ArQuE; software architecture

Maintainability is considered as one of the most crucial attributes of software quality due to the fact that maintenance in software systems consumes a high proportion of the total effort spent in the lifecycle of a system.
Measuring maintenance is to have base decisions on a sound foundation. This report presents a comparison of several state-of-the-art maintainability measures and evaluates their correlation for the example of the Apache Tomcat system.
To enable the comparison, the SAVE (Software Architecture Visualization and Evaluation) tool has been extended by two plug-ins. The first plug-ins allows the interactive combination of metrics: complex metrics can be derived and stored persistently. The second plug-in allows extracting several states from the configuration management systems. Both plug-ins support the process to select, calculate, combine and interpret maintainability metrics.
The selection of maintainability measures is based on the application of the GQM paradigm. The selected maintainability measures analyze the correlation between traditional metrics (e.g., lines of code, cyclomatic complexity, maintainability index, etc.) and object oriented metrics (coupling between classes, weighted methods count, etc.). The results for a subset of Apache Tomcat suggest that there no correlation between the maintainability index and object-oriented metrics. However, future work has to confirm these results with a larger data set.