I am a junior tester and had a exp of 6 months.In a interview in a MNC recently I faced a Q that was tricky. Interviewer said that for a application there are 5000 TC's. Say for 100th TC there is a failure. So after fixing that as a part of regression testing how much TC's you will conduct?. He said suppose you don't know the TC's that are related or not .... also he added that continuing if the failure occurs at the 4000th TC, would you run all the Tests again as you don't know the related TC's....if you run only the failed onces and if error occur in others as say some other code gets disturbed..etc....how will you do the testing? If you do all then what abt the time that you spend? ....Actually I answered that I will chk the related functionality......Well I think I it is towards the right track but not sure.But pls anyone can guide me???
My answer is that I would stop testing until the depenancies are clearly known. How you can write 5000 test cases without any undserstanding of depenancies is beyond me and is an approach that is completely (on many levels) un-acceptable.
The only answer (based on the very stupid assumption that you don't know the dependancies) is that you will have to re-run every test again !
Its only with good understanding of the software design that you can optimise regression testing, but testing without knowing the dependancies..... if you worked for me and conducted yourself that way..... start looking for another job !