Random Testing: Theoretical Results and Practical Implications

Goal-oriented dynamic test generation

“…A number of automated test input generation techniques have been extensively explored, developed, and applied in academia, research labs, and industry. These techniques include random testing [3], symbolic execution [112], dynamic symbolic execution [29], search-based testing [87], and the chaining approach [52]. Noticeably, recent years have the development of techniques that have been applied to test industrial-scale software systems and which have uncovered many serious subtle bugs and security vulnerability defects in the Windows [66] and Linux [93] operating systems, saving millions of dollars [15].…”

“…These techniques include random testing [3], symbolic execution [112], dynamic symbolic execution [26], [62], [124], search-based testing [87], and the chaining approach [52]. The theoretical foundation of dynamic symbolic execution and the chaining approach is separately presented in chapters 3 and 4, respectively.…”

“…Random testing is a simple but effective technique for automated test input generation [3]. In random testing, the process of generating test inputs is performed randomly, based on sampling the interface of the software program under test.…”

“…Testing automation can be enhanced by automating the process of generating test inputs. Over the last three decades, techniques have been proposed to achieve this goal, ranging from random testing [3], symbolic execution [112], dynamic symbolic execution [26], [62], [124], search-based testing [87], and the chaining approach [52].…”

“…For each predicate σ i chosen in this depth-first order, σ i is negated to form a constraint system to be solved for a test input t (lines [3][4]. With this test input t, P is executed to obtain another path condition φ .…”