Regression testing strategy

　 Regression testingIs a test activity throughout all stages of the test. The purpose of this function is to check whether the detected defects have been correctly modified and whether new defects have been caused during the modification process. The software is being tested or OthersAfter the defects found during the activity are modified, the regression test is required. Different strategies can be used for regression testing.

Strategy (1) You can choose to completely repeat the test. All test cases are fully executed to confirm the correctness of the problem modification and whether the surrounding area is affected after modification.

The disadvantage is that the execution of all the use cases will increase the project cost and affect the project progress. Therefore, it is difficult to fully execute the test, so the regression testing strategy (2) is introduced.

Strategy (2) repeated tests can be selected. You can select a part for execution to confirm the correctness of the problem modification and whether the surrounding area is affected after the modification. So how do we select the use case? There are three methods: 1. overwrite the repair and modification method for the module where an error occurs. Select all the use cases of this module for testing. in this way, we can only verify whether there are still defects in this module. However, we cannot guarantee that the modules associated with this module will not cause defects due to this change. so we will introduce 2nd methods, that is, 2. peripheral impact method. in addition to executing the use case of the error module, we also execute the use cases of the modules associated with the module to ensure the quality of regression testing. of course, we can also analyze the quality of the module from a quantitative perspective. For example, after a series of regression tests above, we can check whether the defect rate left behind is within the permitted range, we can end this return test based on this criterion. that is, the third method I want to mention 3. indicator achievement method.

　　Regression testing process

1. Formulate regression testing strategies during the formulation of testing policies

2. Confirm the regression test version.

3. The regression test version is released. Perform the regression test according to the regression test policy.

4. The regression test is passed and the defect tracking ticket is closed.

5. The regression test fails, and the defect list is returned to the developer. Re-Modify and perform the regression test again.

Every time a new module is added as part of an integration test, the software changes. When a new data stream path is created, new I/O operations may also occur, and new control logic may be activated. These changes may make the originalWorkA normal function produces an error. In the environment of the integrated test policy, regression testing re-performs some subsets of some tests that have already been performed to ensure that the above changes do not spread unexpected side effects.

In a wider environment, successful test results (of any type) are error-prone, and errors are to be modified. When the software is modified, some aspects of software configuration (Program, document, or data) have also been modified, and regression testing is used to guarantee (for testing or other reasons) changes do not cause unexpected behaviors or other errors.

Regression testing can be performed manually by re-executing a subset of all test cases, or by using an automated capture and playback tool. The capture playback tool allows software engineers to capture test cases and then play back and compare them. Regression test set (the subset of the tests to be performed) includes three different types of test cases:

* Representative test cases that can test all functions of the software.

* Additional Testing of software functions that may be affected by modifications.

* Tests the modified software components.

Regression testing may become very large during integration testing. Therefore, regression testing should be designed to only test the main functions of modules with errors, it is impractical and inefficient to re-execute all tests for each program function.