Is there any difference between Regression Testing and Sanity Testing?

Regression testing is applying the previously failed and all the related tests on the modified built.
Eg. Let us suppose there are 25 test cases for the functionality. During the initial testing, the test case numbers 11 and 12 have failed and we can't continue testing unless the 2 bugs are resolved. In such case we first report the bugs. Once we receive the modified build from the developers, we apply previously failed test cases (i.e. 11 & 12) and all related test cases according to the bug (say 6, 7, 8 and 9). After that we continue applying the remaining test cases (i.e. 13 to 25).

Sanity testing is a level 0 testing to test the testability of the s/w build, i.e. to test the s/w for its stability. It is tested to verify if it is Understandable, Operable, Controllable, Observable, Consistent, Simple, Maintainable and Automatable. These 8 stages are called Octagonal testing.

Hi,
sutnarcha

let me know what is meant by initial testing?
Do u mean initial testing and sanity testing are the same?

Bharathi

As Sanity testing is done in the initial stages of the s/w testing process and is done on the initial s/w build (that we get from the dev team for the first time), if it is termed as initial testing, then I would say its not incorrect.

what my doubt is that Do we start regression testing at level 0 testing phase when we get defects? and

during sanity testing we are not applying all testing methods. Here only we observe whether it is understandable (all mentioned u previously).

when we conduct regression testing and after which stage in testing process?

No there cannot be regression testing at level 0 testing phase. coz regression testing starts when we receive the modified build from the dev team.
Let us follow the chain below with all possibilities in a testing process.

1) We receive the very first s/w build from dev team.
2) We apply level 0 sanity testing to test its testability and understandability.
3) If we find it is not, then the sanity test is failed and we report the problems as to why it is not testable or understandable.
4) We get a fresh s/w build.
5) We apply level 0 sanity testing to the fresh build again. And let us say it is passed this time. Once a build is passed Sanity testing, it is called a "Stable build".
6) We now start our actual testing process by applying the written test cases (functional testing).
7) We found some of the test cases failed. We report the bugs to the dev team.
8) We receive modified build (after resolving the reported bugs).
9) We start regression testing on modified build.
10) Now go back to step 7.

Step 7, 8 and 9 are ongoing……

As you see above, the very first occurrence of the word ‘regression’ is only at 9th stage.

I hope this is now clear.
Please don't mind asking me if any more clarifications are required.

Now It is clear to me.

Thanks for giving detailed information about regression testing and sanity testing

You are welcome for any such clarifications. I will answer if i can. I may have to search the answer and reply. Thats one of the good and interesting way to learn things. Isn't it.

Please read through the thread "Difference between sanity and smoke test" for more differences on the two.

Hi
sutnarcha

there was no information about this thread.

In the search menu of GeekInterview shown above, just type 'Smoke' and choose 'show posts' option. Then click GO. There is a thread that has about 8 replys and is viewed for over 700 times. This has some info about diff between sanity and smoke tests.

Dear Friends,

As i know, Sanity testing is the testing in which we run the general test cases to test whether the application is working properly or not, means to say that it is sane enough to run exhauting testcases on application. During sanity testing we run only positive testcases.


Whereas Regression testing is needed whenever there is a change in build.It may be due to addition of new functionality or due to bug fixing.


With Regards,
Brijesh Jain.

Thanks for your instructions. I wonder if I understand correctly or not that we have SANITY TESTING only one time when we get the new s/w? Due to all the next modified builds, we call it REGRESSION TESTING. I mean for each s/w, we have only one period time we call SANITY TESTING. Please correct this to me if I am wrong. Thank you very much!

hi nhumai,

when we get the new software we do smoke testing, smoke determines whether it is possible to continue testing. for ex: if there form we check the responsivesness of the form.
if somke fails it is highly difficultto continue testing..

once if smoke testing passes we do sanity testing, sanity really determines the smallest subset of the functaionality, which determines whether the functionality is right or wrong..

Regression testing:
regression is done when there is a change in module which is previously developed we do regression to check whether the newly made change in the existing software has some side effects or not.

Thanks
Deepasree

Hi,

Thanks for your Sanity and Regression testing information.
what is meant by unit testing? Is there any relation with sanity and Regression testing?

Adv thanks,
Bhushan

Hi bhushan,

Unit testing is part of testing done by devolopers. In this testing they will check the code (logical part of the code). In this unit testing they mainly concentrate on Conditonal statements etc.... in coding part only. After this they will go for Integration testing.

I have read all answer and i am convinced but regarding to Sanity testing , Sanity testing is same to smoke testing ? pl explain difference with example in steps
thanks

hai hithesh,

There is no difference between sanity and smoke.

Thanks
Deepasree

Simply put, qa sanity testing (or smoke testing) occurs at the front end of testing (after the code drop, and before qa testing begins), while regression testing usually occurs at the tail end of the qa testing (at least the last regression pass). Sanity testing is to ensure the "testability" of the applications, regression testing occurs, after a mid-cycle code drop, to ensure nothing else is broken. Mid-cycle code drops occur because of a fix to a defect found during the course of testing. Regression will ensure that the defect fix did not break something else somewhere else. While regression testing is a simple effort on a one application per environment testing situation, in real life at an enterprise level, it is seldom that simple. When you have multiple projects making changes to same set of applications, and all projects and their applications sharing the same environment to ensure "production-like" test conditions, regression becomes a complicated, yet extremely critical exercise to success. Typically, in such real life scenarios, i would like my enterprise qa lead to plan for at least three regression passes, and the final pass to occur after the final code drop to the environment (regardless of which project is dropping code). The regression team will then run a regression suite, that includes all test cases for all applications that are living in the environment. This includes incremental test cases since the first version of the application had launched. Ideally, the final test pass would execute (and pass) 100% of the regression suite for 100% of the applications in the environment. However, when there is a time crunch to complete the "ideal scenario regression testing", the following could be acceptable - 100% of the regression suite for the applications that took a code drop prior to the start of final pass, end-to-end test scenarios for those applications, and 100% of the regression suite for any applications downstream to the application that took code drop. E.g. - crunch time scenario: there are 20 applications in the qa environment (shared testing). App1 took a code drop just prior to the start of final regression pass. App3 and app7 are downstream to app1 and depend on data passed from app1. minimum regression for crunch time scenario: 100% execution of regression suite for app1, app3 and app7; plus end to end test scenarios for app1.

Regression Testing: Regression Testing involves reuse of the sam etest, so you can retest after change. There are three kinds of regression tests
1. Bug Fix Regression: Carried out after reporting a bug and hearing later on that its fixed. The goal is to prove that the fix is no good.
2. Old Bugs regression: To prove that a change to the software has caused an old bug fix to become unfixed.
3. Side-effect regression/Stability Regression: Involves retesting of substantial parts of teh product.

Sanity Testing: Test things which you expect to work, and if they dont you will suspect that the program was built with the wrong file or that something basic is broken.

Thanks
Sushma