Lightweight Defect Localization for Java
Lehrstuhl für Softwaretechnik (Prof. Zeller)
Universität des Saarlandes – Informatik
Informatik Campus des Saarlandes
Campus E9 1 (CISPA)
E-mail: zeller @ cs.uni-saarland.de
Telefon: +49 681 302-70970
Lightweight Defect Localization for Java.
To appear at 19th European Conference
on Object-Oriented Programming, Glasgow, Scotland, July 2005.
In the AspectJ ranking, the faulty class is at position 10 out of 542 executed classes—meaning that the programmer, starting at the top, has to examine only 1.8% of the executed classes or 3.3% of the executed code (0.3% of all classes or 0.8% of the entire code) in order to find the defect. (In comparison, comparing coverage of passing and failing run yields no difference in the faulty class, making coverage difference worse than a random guess.) (How Call Sequences Indicate Defects)
Get the paper in PDF format (22 pages, 1MB).
AbstractA common method to localize defects is to compare the coverage of passing and failing program runs: A method executed only in failing runs, for instance, is likely to point to the defect. Some failures, though, come to be only through a specific sequence of method calls, such as multiple deallocation of the same resource. Such sequences can be collected from arbitrary Java programs at low cost; comparing object-specific sequences predicts defects better than simply comparing coverage. In a controlled experiment, our technique pinpointed the defective class in 36% of all test runs.
See Also...<email@example.com> · http://www.st.cs.uni-saarland.de//papers/dlz2004/ · Stand: 2018-04-05 13:41