Static Call Graph Combination to Simulate Dynamic Call Graph Behavior

Call graphs are fundamental for many higher-level code analyses. The selection of the most appropriate call graph construction tool an analysis is not always straightforward and depends on purpose results’ further usage. choice has a great effect following tasks’ execution time, memory usage, result quality. This research compares resulting static dynamic Java to assist in tools. Static graphs, as their name suggests, constructed by analysis, based source or bytecode, without executing tests any parts. means that project can be analyzed its early stages with fewer resources, but there concern this will less accurate, noisier since behavior programs estimated algorithms. Inaccuracies greatly affect analyses graphs. On other hand, created during actual program. calls included edges exactly those were executed run, so you expect more accurate. However, requires resources via test cases which provide high coverage. In work, we investigated relationship between Is generated really better? Can approximate even complement sound results? order find answers these questions, compared results five one analyzer. They evaluated three projects different sizes We comparison merged ourselves combining analyzer outputs. Not only did compare results, also validated found could mislead user. show should considered good, although golden standard they contain phantom calls, present code. Such limited synthetic calls. analyzers applied consideration either, combination does tend similar no