AN INVESTIGATION OF COMMON CODING ERRORS IN OPEN SOURCE GRAPHICAL USER INTERFACE CODE

Abstract

Introduction: This thesis investigates the occurrence of coding errors in the Open Source Software (OSS) Graphical User Interface (GUI) code. Characteristics of coding errors in the OSS GUI code are explored and analyzed so that guidelines are proposed to lower the influence of coding errors in OSS GUI software.
Background: This thesis recognizes the increased prominence of, as well as the increased total volume of OSS GUI code within modern software applications. This thesis seeks to identify whether specific types of errors manifest themselves more frequently in GUI code. The rationale behind this investigation is that: if specific errors are known to occur in specific locations then they can be more easily identified; if specific errors are due to specific causes then they can be more easily recognized in earlier stages of software development.
Methods: Common coding errors were selected and examined in example OSS code using an automatic code inspection. An analysis of results from this inspection identified the frequency and location (i.e. in GUI or non-GUI code) of the common coding error. An initial sample of OSS GUI projects was selected to be examined and a wider range of OSS GUI projects was randomly chosen to evaluate the results that were obtained from the initial sample.
Results: It was found that there are some differences in the type of errors within differing portions of source code. Certain types of coding errors were more frequently identified in GUI code than non-GUI code and corresponding typical GUI coding error-prone behaviors were summarized.
Discussion: Awareness of these differences helps predict errors during the earlier stages of software development. Comprehension of these GUI coding error-prone behaviours contributes to prevent typical GUI coding errors as much as possible during the whole lifecycle of OSS GUI projects.