Regression testing strategy for comprehensive quality assurance system

stay 《 Grading of test cases for a comprehensive quality assurance system 》 The main application of use case grading mentioned in is regression testing ; The main problems of use case hierarchical application , Many companies have established use case hierarchical management systems , But it is really used because different people have different understanding of the business , The use case strategy chosen is different . such as

The same is true for logging in IM Chat business impact , Some people think the impact is very big , All use cases of chat business need to be regression tested ; Some people feel that the impact is not very big , Just test the chat function BVT、 Two higher levels of use cases are sufficient . How to choose it after all ？ It is an important criterion to test whether our application of test case classification is successful .

After continuous practice , A set of effective regression test scheme is summarized , It is based on use case classification , Functional impact analysis is the main basis for selecting test cases ; Adjust the scope of use cases in combination with historical test conditions and online problem feedback ; Use case review and function coverage table are used to supplement test cases ; Assist in real-time adjustment of the test scope through dynamic monitoring of the test process .

Regression test strategy model

1. Test case classification

Test cases are graded in 《 Grading of test cases for a comprehensive quality assurance system 》 Have a comprehensive introduction , Interested students can click to read .

2. Functional impact analysis

During the evaluation of test scope , The most important thing is to conduct functional impact analysis , The results of functional impact analysis directly affect the test quality . Continuously sort out the functional impact in the work and establish the functional impact analysis table . As shown in the figure below , The header is divided into two parts , List the basic function nodes , The horizontal column indicates the functions that will be affected after the change of the functions shown in the column ; Four colors are used to identify Column function Yes Horizontal function The importance of , The degree of influence is divided into four levels , For very high 、 high 、 in 、 low .

 

Through the above function impact analysis , A simple functional impact analysis matrix can be established , As shown in the figure below , The functional importance is divided into four levels, which are very high 、 high 、 in 、 low , The abscissa identifies the importance of the column function in the above figure , The ordinate indicates the influence degree of the influence function （ As shown in the above figure, the corresponding color of the hook ）. Through the value corresponding to the function influence matrix, you can directly see the function and function influence closeness . As shown in the picture ：

The biggest impact of login function changes is on personal data , Contacts 、 View mood phrases ;

Login time IM Core functions , For the value of the 4, The value of the profile for is 4; Viewing mood phrases is an extension , The corresponding value is 1

In this way, in the functional impact analysis table , The values for are 16,16, and 4

Functional interaction analysis

The functional failure effects are described in the following figure , The vertical coordinate indicates the importance of the change function , The abscissa represents the result of the above analysis , That is, the closeness of function and function influence . The range of test cases can be selected through the value of the functional failure impact matrix .

Functional failure impact analysis tong

 

Continue with the example of login ： Login is a function represented by vertical coordinates , The importance of login is 4; Login has the highest impact on profile functions （ The calculation result in the above figure is 16）, Converted to this figure is 4. After analyzing the change of login function , The impact value on personal data is in the red area （12 perhaps 16）, In this way, the test cases for personal data regression can be selected according to the strategy. The use case level needs to be selected as BVT、 high 、 in 、 All use cases at the lower four levels .

If the value of a function in the failure impact analysis table is yellow, the use case can be very high 、 high 、 Three levels in . And so on .

3. Historical test results and online problem feedback

In early practice , Use only the functional impact analysis table to select use cases , But there will always be some inexplicable problems after use , such as

Why do some modules focus on testing or appear 5 Level quantization ？

Why some bug Repeated online ？

Bug Get together （ Several successive versions bug The module bug There are more ）？

Through continuous summarization, it is found that , These problems are often caused by the failure to track and analyze the test history and online feedback . Through the analysis of historical test conditions and online feedback, continuous tracking and optimization practices are as follows ：

Analysis of historical situation

Module defect rate ： Modules with high defect rate （ Greater than average ） Focus on when testing , and Appropriately improve the test coverage （ In principle, the use case coverage of the top three modules with defect rate is increased by one level ）.

Defect stability rate ： Statistics 6 Test defect stability of sub module version , Modules with poor stability Group discussion on whether to add use case coverage .

Version module trend chart ： Tracking module bug Whether the trend is within the normal prediction range , The interval value is exceeded , Focus on and timely adjust use case coverage （ combination bug Daily execution monitoring ）.

Online continuous feedback tracking optimization

Total level 5 faults 5 Pieces of ,A Business 3 Pieces of ,B Business 1 Pieces of ,C Business 1 Pieces of ; The main problem is not clear about the business logic .

Strategy ： All involve A Full regression of business module testing ,B、C Business focus .

D Business bug More , But the complaints are low , Strengthen testing when resources are available ,A Business... Business bug Less , But it leads to five levels of quantification , Focus on testing and monitoring .

Strategy ： For online production bug modular , In the future return , All included in the test scope .

4. Use case review and function coverage table confirmation

Use case reviews every company is doing , This is not an elaboration . Briefly, the function impact coverage table is signed for confirmation . The functional impact coverage table records the list of functions involved in this regression test , And the range of tests that require regression , Generally no more than one page . Confirm and sign before each test , Its purpose is not to sign , Instead, make sure coverage is in place .

5. Dynamic monitoring of test process

Monitoring defects dynamically during testing does , And the degree of verification of defects , Then dynamically adjust the test range . The specific practice is as follows .

bug Level ： Level five bug Report to the project team , Level Four bug On duty test group , There are project team and test team to discuss bug influence , Dynamically adjust the scope of test cases .

modular bug trend ： Refer to the historical version , The test team decides whether to adjust the scope of test cases .

The choice of regression test strategy is a very difficult thing , This article provides some simple practical experience for your reference . If you have different suggestions or ideas, you are welcome to communicate and leave messages .