This paper presents CAVM, a new search-based C language test data generation tool. CAVM was developed to improve the data tool, CodeScroll. CodeScroll uses static analysis and input partitioning techniques to generate test data. Unlike Austin, the current state-of-the-art automated test data generation tool for C, CAVM generates dynamic data structure data using purely search-based techniques. We present a study in which CAVM was compared with Austin and CodeScroll using 49 experimental functions, involving small-scale anti-pattern case studies, actual product-level open source code, and commercial code. CAVM was able to cover branches that CodeScroll and Austin could not cover, gaining ... [more] higher coverage.

Bibtex Entry

@inproceedings{You2018,
  author    = "You, Byeonghyeon and Kim, Junhwi and Kwon, Minhyuk and McMinn, Phil and Yoo, Shin",
  title     = "Evaluation of {CAVM}, {Austin}, and {CodeScroll} for Test Data Generation for {C}",
  booktitle = "Korea Conference on Software Engineering (KCSE 2018)",
  year      = "2018"
}