Research On Testing Component-Based Software Based On Observation Model

Component-based software developing technology builds the new system using the existing components. Some characteristics of component however, such as source code invisibility, distributed execution, platform independence, cause the traditional software test theory and technology can not be completely applied to component-based software.In this paper, we analyze the features of component-based software, discuss limitations of traditional integration testing methods and advantages and disadvantages of the current component-based software integration testing methods. The existing component-based software integration test methods can not describe the whole behavior of component-based software, thereby they can not effectively support the implementation of test, so we propose a component-based software integration test method based on DFA observation model with independent relationships. We analyze the characteristic of algebra system which the software system maps, make the new description of event and state in component-based software combining with automata theory, use event and state sequences to describe software dynamic behavior, provide an unified research frame for both sequential and concurrent behavior. We study the characteristics of concurrent behavior combining with the definitions and properties relevant to independent relationships in the recognizable stable trace languages theory, build the observation model of component-based software. We research the sequential and concurrent behavior capturing method of component-base software combining with observation theory, present the test adequacy criteria of component-based software, and discuss its validity. The goal of test is achieved through the recognition of the observation model to actual behavior trails of software system. Finally, the validity of the component-based software integration test method based on the observation model is proved by an example.