Name
Abstract | ||
|---|---|---|
Various record-replay techniques are developed to facilitate web application debugging. However, it is time-consuming to inspect all recorded events that reveal a failure. To reduce the cost of debugging, delta-debugging and program slicing are used to remove failure-irrelevant events. However, delta-debugging does not scale well for long traces, and program slicing fails to remove irrelevant events that the failure has program dependence on. In this paper, we propose an effective and efficient approach to remove failure-irrelevant events from the event trace. Our approach builds constraints among events and the failure (e.g., a variable can read any of its earlier type-compatible values), to search for a minimal event trace that satisfies these constraints. Our evaluation on 10 real-world web applications shows that our approach can further remove 70% of events in the reduced trace of dynamic slicing, and needs 80% less iterations and 86% less time than delta-debugging. |
Year | DOI | Venue |
|---|---|---|
2016 | 10.1145/2950290.2983964 | SIGSOFT FSE |
Keywords | Field | DocType |
JavaScript,failure,event trace reduction | Program slicing,Data mining,Computer science,Slicing,Real-time computing,Web application,JavaScript,Debugging | Conference |
Citations | PageRank | References |
2 | 0.37 | 17 |
Authors | ||
1 | ||