Citation Context

...ing a set of associated linear constraints and using the results to refine previous values. Genetic algorithms have been used to achieve the minimization of cost functions associated with branches in =-=[1, 15, 16, 17]-=-. Usage of a moving target to satisfy coverage criteria was proposed in [16], while proper fitness measures to estimate the distance from a given execution branch are studied in [1]. All these works a...

Citation Context

...n of infeasible paths (notably in the case of loops with a variable number of iterations), the treatment of aliases and the complexity of the formulae which are gradually built up. Dynamic approaches =-=[1, 5, 9]-=- avoid the problems of symbolic execution by dispensing with the path predicate and using general heuristic function minimisation techniques to modify the input data so that the test objective is cove...

Citation Context

...inite domains is NPcomplete in the worst case but heuristics have been developed which seem to give an acceptable performance for most test case generation problems [8][14]. Dynamic approaches [5][11]=-=[15]-=- avoid the problems of symbolic execution by running an instrumented version of the program under test on arbitrary data and then using function minimisation to try to modify the data so that a path c...

Citation Context

...equential programs coded in an imperative language. Our testdata generator prototype, also called PathCrawler, treats ANSI C programs. Like the dynamic approaches [FK96], [Kor96], [GN97], [GMS99] and =-=[MM98]-=- to test data generation, our method is based on dynamic analysis, but instead of heuristic function minimisation (except for [GMS99] which uses an iterative relaxation method), we use constraint logi...

Citation Context

...d as well as differential GA in their test data generator to achieve condition-decision coverage test adequacy criteria, which has higher coverage than the previous approaches [9] [10] [11] [12] [13] =-=[14]-=-. A test-data generator for stress testing was developed by Alander et al. to produce test cases in order to find problematic situations, e.g. processing time extremes. They use response time or memor...

Citation Context

...most often singlesfunctions) written in restricted programming languages. Thesresults thus obtained revealed that GA achieved maximumsbranch coverage among all other techniques (Michael et al.,s1997; =-=Michael and McGraw, 1998-=-).sIn 2003 Two GP-based procedures was introduced forsselection and evaluation of test data. These procedures weresrelated to two important questions: how to select test casessto reveal as many faults...

Citation Context

...pply an EA to SBST [108] to perform structural testing of a program. Most SBST research is focused on structural testing. These works can be organized into the sub-categories of branch-distance based =-=[70, 54, 55, 108, 67, 68]-=-, control based [78, 48], or a combination of both [93, 98, 99]. In general, a major issue for SBST structural testing is that the relationship between the components of a test case and the resulting ...