I am puzzled.During regression testing,select suitable test cases from the existed test suite to do the next test.How to choose? If consider test case's bug prediction ability,catch the most test cases that have high bug prediction ability.But,how to identify high or low bug prediction ability?
Regression testing is selective testing which is performed to make sure that any of the fixes and newly added features, changes in some existing features have not impacted the areas which were working fine previously. Hence those test cases should be executed which could get impacted due to any change. This comes with understanding of design. If a tester is involved from day 1 of project's life cycle and has contributed to all phase - requirements, design then should be able to understand and identify the features which might get impacted in such cases. Let's take an example -
Suppose an application has a feature that a user can enter and update comments to it. An issue was raised regarding the update functionality, it was getting added with a default user name say 'admin' every time and expected was that should be added with the correct user name. Then during regression, a tester should test the 'Enter' functionality and can also check both features with an 'Admin' user as well. But it is not needed to check the formatting, font size, style etc as those features would have been tested before during system testing round and would not get impacted due to the changes made to have Enter and Update feature working correctly.
This understanding also comes with experience. A tester can coordinate with the design team to understand where and how the changes were made and should be able to identify the test cases to be executed.
Get back to me in case of suggestions [img]/images/graemlins/smile.gif[/img]
[ QUOTE ]
How to choose? If consider test case's bug prediction ability,catch the most test cases that have high bug prediction ability.But,how to identify high or low bug prediction ability?
[/ QUOTE ]
You have to use your system testing experience. There must be certain areas of application where you might have seen problems during system testing. You can consider the test cases of those functionality.
If you didn't do system testing, take inputs from someone who did system testing.
I would re-visit some of the test cases of the functionality in which more bugs were found earlier.