Exoneration-based fault localization for SQL predicates

•Faults in SQL clauses are automatically localized by new algorithms.•Faults are found with high precision, at the clause level.•Non-faulty clauses are exonerated by a novel algorithm.•The new algorithms are more effective and efficient than nine previous methods. Spectrum-based fault localization (...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:The Journal of systems and software Ročník 147; s. 230 - 245
Hlavní autoři: Guo, Yun, Li, Nan, Offutt, Jeff, Motro, Amihai
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Inc 01.01.2019
Témata:
Exoneration-based fault localization
Fault localization
Spectrum-based fault localization
SQL
WHERE clause
WHERE clause
Fault localization
Spectrum-based fault localization
Exoneration-based fault localization
SQL
ISSN:0164-1212, 1873-1228
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:•Faults in SQL clauses are automatically localized by new algorithms.•Faults are found with high precision, at the clause level.•Non-faulty clauses are exonerated by a novel algorithm.•The new algorithms are more effective and efficient than nine previous methods. Spectrum-based fault localization (SFL) techniques automatically localize faults in program entities (statements, predicates, SQL clauses, etc.) by analyzing information collected from test executions. One application of SFL techniques is to find faulty SQL statements in database applications. However, prior techniques treated each SQL statement as one program entity, thus they could not find faulty elements inside SQL statements. Since SQL statements can be complex, identifying the faulty element within a faulty SQL statement is still time-consuming. In our previous paper, we developed a novel fault localization method based on row-based dynamic slicing and delta debugging techniques that can localize faults in individual clauses within SQL predicates. We call this technique exoneration-based fault localization because it can exonerate “innocent” elements and precisely identify the faulty element, whereas previous SFL techniques simply ranked all the elements in an SQL statement based on suspiciousness. This paper improves the exoneration-based fault localization technique with a new algorithm that considerably reduces the execution time. We also conducted an empirical study that compared nine existing SFL techniques with the exoneration-based technique in localizing faulty clauses in SQL predicates. Results indicate that the new exoneration-based technique surpasses the other techniques both in terms of effectiveness and efficiency.
ISSN:0164-1212
1873-1228
DOI:10.1016/j.jss.2018.10.037