Zheng Dongjie "Software Engineering" learning notes
Thunder
Any engineering product (note is that any engineering product) can be tested using one of the following two methods.
Black box test: The functional design specifications of known products can be tested to prove whether each implemented function meets the requirements.
White box test: The internal work process of known products can be proved to prove whether each internal operation meets the design specifications, all internal components are inspected.
The software's black box test means that testing should be done at the interface of the software. This method is to treat test objects as a black box. Test personnel do not consider the logical structure and internal characteristics of the program, and only according to the requirements of the program, check whether the function is in line with its functional description. Therefore, the black box test is also called function test or data-driven test. The black box test is mainly to find the following types of errors:
1. Is there any incorrect or omission?
2, on the interface, can input it correctly? Can you output the correct result?
3. Is there a data structure error or external information (such as a data file) access error?
4, can performance meet the requirements?
5. Is there an initialization or termination error?
The software's white box test is a detailed inspection of the process of process for software. This method is to treat test objects as an open box, which allows testers to test all logical paths for programs using logical structures and related information, design or selection of test cases. By checking the program status at different points, it is determined whether the actual state is consistent with the expected state. Therefore, the white box test is also called a structural test or logical drive test. White box test mainly wants to check the program module:
1. Test at least one of the independent executive paths of the program module.
2, the two cases of "true" and "false" can be tested at least once.
3. Perform a cyclic body within the boundary of the loop and the boundary of the operation.
4. Test the effectiveness of the internal data structure, and so on.
The above facts have shown that software test has a fatal defect, namely the incomplete, uncomfortable. Since any procedure can only perform a limited test of a small amount (relative to the huge number of exhaustions), there is no error in the program when no errors are found.
