BOLD: an ontology-based log debugger for C programs

摘要

Program debugging is often an ad hoc activity, with a combination of manual and semi-automated processing. A challenge posed by debugging is the lack of standardized procedures for instrumenting, representing, and analyzing the execution trace. Further, debugging is often low level, getting into the nitty-gritty details of variables and their semantics—rather than at a high-level. The presence of libraries only exacerbates these issues. We propose BOLD, an Ontology-based Log Debugger, to unify various activities involved in the debugging of sequential C programs. The syntactical information of programs can be represented as Resource Description Framework (RDF) triples. BOLD automatically instruments programs by querying these triples. It represents the execution trace of the program also as RDF triples called trace triples. BOLD’s novel high-level reasoning abstracts these triples as spans. A span gives a way of examining the values of a particular variable over certain portions of the program execution. The properties of the spans are defined formally as a Web Ontology Language ontology. A developer can debug a given buggy program by querying the trace triples and reasoning with the spans. To empirically assess BOLD, we debugged the programs in standard Software-artifact Infrastructure Repository. Experiments related to debugging through automated reasoning show improvement in conciseness of the developers’ specifications and run time performance of BOLD compared to gdb-Python.