An extended overview of the Mothra software testing environment

DeMillo, Richard A.; Guindi, D.S.; McCracken, Wendy; Offutt, A. Jefferson; King, K. N.

doi:10.1109/wst.1988.5369

Cited by 184 publications

(114 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…No mutants were created which change, for instance, which Exception is caught, and no test needs to be written to hit the catch block and therefore the catch block remains unexecuted. Research has indicated that killing every non-stubborn mutant will subsume line coverage given an exhaustive set of operators (DeMillo, et al, 1988). However, we find this subsumption only holds when no created mutants are stubborn and when the operator set is exhaustive of all language constructs.…”

Section: Unexecuted Statementsmentioning

confidence: 60%

“…When any existing test case detects the mutant, the mutant is said to be "killed." Mutants that have yet to be detected are referred to as "living" (DeMillo, et al, 1988). A mutation score, calculated by taking the number of killed mutants and dividing by the total number of mutants, summarizes the results of mutation testing.…”

Section: Mutation Testingmentioning

confidence: 99%

“…Mutation analysis is computationally expensive and inefficient (DeMillo, et al, 1988). If the process of killing all mutants is to be integrated with a software development methodology, information on how to produce the most test cases and achieve the highest line coverage from mutation analysis is of critical importance.…”

Section: Operator Analysismentioning

confidence: 99%

See 2 more Smart Citations

On guiding the augmentation of an automated test suite via mutation analysis

Smith

Williams

2008

Empir Software Eng

View full text Add to dashboard Cite

show abstract

Section: Unexecuted Statementsmentioning

confidence: 60%

Section: Mutation Testingmentioning

confidence: 99%

See 1 more Smart Citation

On guiding the augmentation of an automated test suite via mutation analysis

Smith

Williams

2008

Empir Software Eng

View full text Add to dashboard Cite

show abstract

“…Mothra [30] and Proteum [31] were developed for Fortran and C, respectively. Jester [32], Jumble [33], MuJava [8,34,35], JavaMut [36] and Response Injection (RI) [37], as well as the tool presented in this paper, are dedicated to the Java language.…”

Section: Related Workmentioning

confidence: 99%

Judy – a mutation testing tool for Java

Madeyski

Radyk

2010

IET Softw.

View full text Add to dashboard Cite

Popular code coverage measures, such as branch coverage, are indicators of the thoroughness rather than the fault detection capability of test suites. Mutation testing is a fault-based technique that measures the effectiveness of test suites for fault localization. Unfortunately, use of mutation testing in the software industry is rare because generating and running vast numbers of mutants against the test cases is time-consuming and difficult to do without an automated, fast and reliable tool. Our objective is to present an innovative approach to mutation testing that takes advantage of a novel aspect-oriented programming mechanism, called "pointcut and advice", to avoid multiple compilation of mutants and, therefore, to speed up mutation testing. An empirical comparison of the performance of the developed tool, called Judy, with the MuJava mutation testing tool on 24 open-source projects demonstrates the value of the presented approach. The results show that there is a statistically significant (t(23) = −12.28, p < .0005, effect size d = 3.43) difference in the number of mutants generated per second between MuJava (M = 4.15, SD = 1.42) and Judy (M = 52.05, SD = 19.69). Apart from being statistically significant, this effect is considered very large and, therefore, represents a substantive finding. This therefore allows us to estimate the fault detection effectiveness of test suites of much larger systems.

show abstract

“…There is a strong similarity to Blackburn's algorithms and the algorithms used by Offutt's test data generator for mutation [19,20], the key difference being that Blackburn's is specificationbased, whereas Offutt's constraints are code-based.…”

Section: State-based Approachesmentioning

confidence: 99%