Towards proper differential analysis of static analysis engine changes

Horváth, Gábor and Kovács, Réka and Szécsi, Péter: Towards proper differential analysis of static analysis engine changes. Conference of PhD Students in Computer Science, (11). pp. 75-78. (2018)

Cikk, tanulmány, mű cscs_2018_088-091.pdf Download (128kB)

Abstract

The design and implementation of heuristics for static analysis engines require detailed knowledge about the code to be analyzed. Extensive testing is therefore required to validate whether a change to the analysis engine is beneficial for real-world software projects. The current practice of testing an analyzer on a fixed set of projects is not sufficient. Changes in the engine might affect language features that are utilized by only a fraction of the projects in the test suite. We explore a direction to ease the design of differential analysis experiments on a dynamic set of projects. This involves semi-automatic generation of the test set and evaluation of analysis results before and after applying changes to the engine. As the presented framework includes tools that aid the interpretation, reproduction, and sharing of analysis results, it might be valuable for a wide range of developers in the community.

Actions (login required)

View Item