Name of the participant: Sebastian Eder
Description of the IT-research project: The maintenance of software devours a large part of the total costs of a software system and is a task not to be underestimated. Especially the prioritization and planning of maintenance activities pose problems in practice. Static analyses already help here, for example, by enabling evaluations of the quality of software and thus identifying system parts that are difficult to maintain or identifying system parts that need urgent maintenance. However, there is information that cannot be taken into account by static analyses, such as the actual behavior of the actual users of the system.
Dynamic analyses are necessary to incorporate this information into the maintenance process. These are currently used very little in software maintenance, because often the collection of data about the actual use of a software system is already a major hurdle. This leads to the problem that there is hardly any knowledge about the actual use of industrial software systems. However, this information is missing in the targeted maintenance of software systems, since developers involved in maintenance can only estimate which system parts, for example, are frequently used and therefore require more attention.
The research project ANSE considers the analysis of usage data in the software evolution. This includes the pure analysis of this usage data and its comparison with maintenance efforts. First, however, the hurdle of collecting usage data must be overcome. Furthermore, there are program parts that cannot be executed due to the program structure. These program parts are also analyzed and related to maintenance work. Finally, usage data is not only collected in the productive environment of software systems, but also during testing. Here, usage data is used to collect coverage information from manual or automatic system tests in order to determine which system parts are affected by maintenance but have been badly tested.
The developed techniques should be firmly integrated into the context of software maintenance by examining their use in the continuous software evolution. For this purpose and for the evaluation of the proposed techniques tool support shall be implemented.
With the contribution of this research project, the usage analysis will be examined on a judgmental way (pure analyses) and on a constructive way (use in software evolution). Tools will be created that allow developers to better prioritize and assess maintenance efforts and to coordinate test efforts. Continuous evaluation ensures the quality of the developed results in an industrial environment.
Software Campus partners: TU München, Siemens AG
Implementation period:01.02.2013 – 31.01.2015