Concurrent Software Testing Sequence Generation Method Based On State Space Pruning

Concurrent software plays an important role in the software industry,and reliable software quality requires comprehensive testing to ensure.The execution process of concurrent software is very complicated,and the number of states of the system is huge.However,most of the current software testing methods have low coverage,high redundancy,or low test efficiency.Therefore,it is very difficult to generate a test sequence that completely covers the test target without guaranteeing redundancy.In this thesis,the test behavior set is used to describe the test purpose,and the test coverage standard is defined as the full coverage of the execution path and the data-related behavior execution path.In order to improve the efficiency and efficiency of the test sequence,a test sequence generation method suitable for concurrent software based on the Colored Petri Net(Colored Petri Net,CPN)is proposed.The execution path unrelated to the test purpose by state node projection is removed,the state node pruning,arc pruning and the like in the state space to reduce the state space size.For the method of generating concurrent software test sequences based on state space pruning,the following four aspects are worked:(1)The test behavior set is used to describe the test purpose,and the test coverage standard is defined as the path coverage of the behavior to be tested and its data related behavior.The CPN model is extended to divide the changes in the software system model into four categories:(1)behavior to be tested,(2)related behaviors,(3)concurrent unrelated behaviors with concurrent relationships with the test and related behavior,(4)Non-concurrent unrelated behaviors that do not have a concurrent relationship with the test and related behavior.The basis for the proposed pruning for the two unrelated behaviors is laid down below.(2)A method such as state node projection is proposed.Based on the state space of the extended CPN model,state space pruning is performed on the execution path of non-concurrent unrelated behavior.A state space subgraph composed of a sequence to be tested,a related behavior,and a concurrent unrelated behavior execution sequence is obtained,and the purpose of reducing the state space in the first step is achieved.(3)Methods such as state node pruning and transition arc pruning are proposed to pruning concurrent unrelated behaviors.According to the positional relationship of concurrent unrelated behaviors in the model,it can be subdivided into four types: Prue,Before,Middle and After unrelated behaviors.The state node pruning operation can remove all execution paths of Prue,Before and After unrelated behaviors;the transition arc pruning operation can remove redundant execution paths of irrelevant behaviors.Both pruning operations further reduce the size of the state space,minimizing the range of test sequence generation.(4)A method such as full sequence connection is proposed,and a test sequence is generated based on the state space after pruning.The test sequence completely covers all execution paths of the tested and related behaviors.For concurrent unrelated behaviors and non-concurrent unrelated behaviors,only one shortest execution path is selected for coverage.This results in a collection of complete test sequences from the initial state of the system to the state of the end.And the test sequence is a test effect that the execution path of the test to be tested has full coverage and no redundancy.Finally,the full coverage and non-redundancy proof of the test sequence generation algorithm is completed.This method is used to complete the test sequence generation of the police management system.The complete coverage and non-redundancy of the algorithm are verified.The test results are compared with other methods to verify the efficiency of the algorithm.