How to determine the end point of a software test

　　 How to determine the end point of a software test

　　 This problem is mentioned in every test book , It's not much fun just to list them , Just sum it up ：

　　1、 Forced exit at the organization level ：

　　- Project to suspend , Usually, there are serious problems in the project or force majeure factors

　　- The funds are exhausted , Usually, the project funds are not well controlled , A bankrupt , This kind of situation is relatively rare now

　　- Overdue , This is the most common , Especially in the traditional waterfall mode , Development has been delayed repeatedly , Resulting in insufficient testing time , Had to forcibly stop , It's up to God

​

　　2、 Reached the functional quality index . It's too difficult to list all the quality indicators , Put several common exit indicators ：

　　- Test case coverage

　　- Execution rate of test cases

　　- Test platform coverage , Like language , operating system , Hardware types, etc . For some special tests, such as configuration testing , localization / International testing is crucial

　　- Repair rate of serious defects

　　- Whether the unrepaired defects are recorded

　　- * The speed of opening new defects

　　- * The speed of repairing defects

　　- Whether the regression test is well performed

　　- Defect detection rate of regression test .( This is often overlooked , Introducing new defects from the repaired defect code is an important source of testing risk )

　　- * Change trend of the total number of unrepaired defects , It is usually considered a good time to end the test only when it converges quickly

　　- Document completeness , Problems that you decide not to fix must be noted in the release document

　　 The three curves formed by the three indicators marked with an asterisk are usually the most important reference basis for testing

　　3、 Reached the non functional index

　　- Performance indicators ( Expansion is too complicated )

　　- Availability metrics ( Although it is an indicator , But there are few hard indicators )

　　- Compatibility index , Especially for multi-component applications , For the compatibility between various components of different versions , Sometimes even developers are confused

　　- Safety index , Extremely important

　　 finally , I can't help but talk about agile development . Iterative development process will get different defect curves . It seems that there is no book or article to discuss this problem well . From personal experience, the above three key curves will reach a peak before the customer demonstration . And the exit time of each iteration is usually not long before the demonstration . Another important exit condition is that unit tests must reach a high coverage . Because in agile development, the quality of software largely depends on the quality of unit testing . From the whole project life cycle, the defect curve is wavy . Agile development has another special feature . In many cases of agile development , It is not mandatory that all requirements be submitted in the end . Some functions with low priority but high cost may be cancelled . So the final exit conditions may be more flexible and empirical .

　　 For more information on testing techniques, please pay attention to ： Shenzhen duosurveyor software and Technical Service Co., Ltd