Abstract
This study proposes a method of reducing the number of software modules to be tested when software is updated on version-ups. The purpose of this method is to detect degradation of software quality with lesser efforts when update is made. To achieve this purpose, variables of a program in C language are classified according to the extent of their influences. In this classification, there are nine influence levels. However, only variables that come from the three levels of the classification are needed to accomplish the purpose of the study. To prove the validity and effectiveness of this method, seventeen programs with intentional bugs are generated and they are tested with the method. In addition five actual application programs are selected and tested too. After the test is done, the ratios of the modules chosen by the method to all modules of the program are measured. Also, the selected modules are looked into to see whether they are the modules to be chosen, that is, modules with degradation. The result is that the method extracts sixty percent of all modules and ninety percent of all bugs due to degradation are included in the extracted modules.
