Improving evolutionary class testing in the presence of non-public methods

Created by W.Langdon from gp-bibliography.bib Revision:1.4333

  author =       "Stefan Wappler and Ina Schieferdecker",
  title =        "Improving evolutionary class testing in the presence
                 of non-public methods",
  booktitle =    "ASE '07: Proceedings of the twenty-second IEEE/ACM
                 international conference on Automated Software
  year =         "2007",
  pages =        "381--384",
  address =      "Atlanta, Georgia, USA",
  publisher_address = "New York, NY, USA",
  publisher =    "ACM",
  keywords =     "genetic algorithms, genetic programming, SBSE, unit
                 testing, object-orientation, encapsulation,
                 evolutionary testing, POSTER SESSION",
  isbn13 =       "978-1-59593-882-4",
  URL =          "",
  DOI =          "doi:10.1145/1321631.1321689",
  size =         "4 pages",
  abstract =     "Automating the generation of object-oriented unit
                 tests is a challenging task. This is mainly due to the
                 complexity and peculiarities that the principles of
                 object-orientation imply. One of these principles is
                 the encapsulation of class members which prevents
                 non-public methods and attributes of the class under
                 test from being freely accessed.

                 This paper suggests an improvement of our automated
                 search-based test generation approach which
                 particularly addresses the test of non-public methods.
                 We extend our objective functions by an additional
                 component that accounts for encapsulation.
                 Additionally, we propose a modification of the search
                 space which increases the efficiency of the approach.
                 The value of the improvement in terms of achieved code
                 coverage is demonstrated by a case study with 7
                 real-world test objects. In contrast to other
                 approaches which break encapsulation in order to test
                 non-public methods, the tests generated by our approach
                 inherently guarantee that class invariants are not
                 violated. At the same time, refactorings of the
                 encapsulated class members will not break the generated
  notes =        "Also known as \cite{1321689}",

Genetic Programming entries for Stefan Wappler Ina Schieferdecker