This site hosts the data files used in Evaluating and Improving Fault Localization Techniques. All these files can be generated using scripts in our BitBucket repository, but only over the course of tens of thousands of CPU-hours; we put them here to save future researchers that expense.
For each fault that we analyzed, there are several files:
(If a killmap is not present, it means either that it didn't finish within its timeout, or it was unnecessary (because, for example, we didn't need a 168h-killmap since the 32h-killmap completed; or we didn't need an unoptimized killmap because the optimization was sound).
We also provide a
.tar.gz file containing all of the information for the artificial faults we generated from Defects4J.
These files are organized like so:
data/ artificial-faults.tar.gz Chart/ 1/ gzoltar-files.tar.gz 32h-killmap-files.tar.gz 32h-unoptimized-killmap-files.tar.gz 168h-killmap-files.tar.gz 2/ ... ... Closure/ ... Lang/ ... ...
You may find the following commands, or variants of them, useful:
wget --recursive --no-parent http://fault-localization.cs.washington.edu/data
wget --recursive --no-parent http://fault-localization.cs.washington.edu/data/Time/3
wget --recursive --no-parent --accept gzoltar-files.tar.gz http://fault-localization.cs.washington.edu/data
wget http://fault-localization.cs.washington.edu/data/artificial-faults.tar.gz && tar xf artificial-faults.tar.gz --directory /path/to/defects4j