Rothermel, G. and Harrold, M. J. (1994a) `A framework for evaluating regression test selection techniques', in Proceedings International Conference on Software Maintenance, IEEE Computer Society Press, Los Alamitos CA, pp. 201--210.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... de ne a test to be a 3 tuple identi er, input, output , where identi er is the name of the test, input refers to the input values for the given test, and output is the expected output for the test given by S(input) The regression testing problem can be characterized in the following fashion [20]: Problem 1 (Regression Testing) Given a program P , its modi ed version P 0 , and a test set T that was used to previously test P , nd a way to utilize T to gain sucient con dence in the correctness of P 0 . 2 2.2 Selective Retest Techniques Selective retest techniques attempt to reduce ....

....it causes the two versions of the program to produce di erent outputs. Since there is no e ective procedure for precisely identifying the tests in T that are fault revealing or modi cation revealing, selective retest strategies have focused on the selection of tests that are modi cation traversing [20, 21]. A test is modi cation traversing for P and P 0 if and only if it executes new or modi ed code in P 0 while also exercising code that was formerly executed in P and deleted in P 0 [20] It has been shown that any modi cation traversing test is also modi cation revealing and fault revealing ....

[Article contains additional citation context not shown here]

G. Rothermel and M. J. Harrold. A framework for evaluating regression test selection techniques. IEEE Transactions on Software Engineering, 1994.

....techniques by Harrold, Gupta, and Soffa [HGS90] We expect to see a cost versus benefit tradeoff, and hope that an empirical comparison will lead to practical suggestions about when to use which technique. We also would like to compare our procedure within the framework of Rothermel and Harrold [RH94] 7 ACKNOWLEDGEMENTS We would like to thank Paul Ammann for helpful discussions, and particularly for pointing out that test cases that are executed early in the testing process have a weaker filter applied than test cases executed late in the process. ....

G. Rothermel and M. J. Harrold. A framework for evaluating regression test selection techniques. In Proceedings of the Sixteenth International Conference on Software Engineering, pages 201--210, Sorrento, Italy, May 1994. IEEE Computer Society Press.

....Select T 0 T , a set of tests to reexecute on P 0 . 3. Retest P 0 with T 0 , establishing P 0 s correctness with respect to T 0 . 4. If necessary to satisfy some adequacy criteria, create new tests for P 0 . 5. Create T 00 , a new test set history for P 0 . In previous work[21], we described a framework for evaluating regression testing techniques. Using this framework, we classify a regression testing technique as safe if it selects all tests from T that could possibly exhibit different output when run on P 0 . However, a safe approach may include many tests in T 0 ....

....as precise if it avoids choosing tests that will not cause P 0 to produce different output than P . Since we cannot find an algorithm to determine, for an arbitrary choice of P 0 , whether a test will exhibit different output in P 0 than in P , no technique can be both safe and precise[21]. When evaluating the efficiency of selective retest methods, it is important to recognize that regression testing is typically accomplished in two phases. In the initial phase of regression testing, while code is being modified, regression testing is not on the critical path; testers may be ....

[Article contains additional citation context not shown here]

G. Rothermel and M.J. Harrold. A framework for evaluating regression test selection. In Proceedings of the 16th International Conference on Software Engineering, pages 201--10, May 1994.

....on P 0 . 3. Retest P 0 with T 0 , establishing P 0 s correctness with respect to T 0 . 4. If necessary to satisfy some coverage criteria, create new tests for P 0 . 5. Create T 00 , a new test set for P 0 , and gather test execution information for T 00 . In previous work[21], we described a framework for evaluating regression testing techniques. Using this framework, we classify a regression testing technique as safe if it selects all tests from T that could possibly exhibit different output when run on P 0 . However, a safe approach may include many tests in T 0 ....

....dependence predecessor (cd predecessor) of B, and B is a control dependence 3 Since we cannot find an algorithm to determine, for an arbitrary choice of P 0 , whether a test will exhibit different output in P 0 than in P , no technique can be both safe and precise. For more discussion, see[21]. 4 Although there are several types of data dependence defined in the literature, we are only interested in flow dependence. 5 Traditional data flow analysis[2] or improved analysis that considers the effects of aliasing[13] can be used to gather the data flow information. 2 R1 R2 R5 R4 ....

[Article contains additional citation context not shown here]

G. Rothermel and M.J. Harrold. A framework for evaluating regression test selection. In Proceedings of the 16th International Conference on Software Engineering, pages 201--10, May 1994.

No context found.

Rothermel, G. and Harrold, M. J. (1994a) `A framework for evaluating regression test selection techniques', in Proceedings International Conference on Software Maintenance, IEEE Computer Society Press, Los Alamitos CA, pp. 201--210.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC