Summery of Combined Static and Dynamic Automated Test Generation for MS and PHD
Area: Software System Quality Assurance
By Sai Zhang, David Saff, Yingyi Bu
The motivation for automated test in today’s fast moving world, there is a challenge for any society to constantly defend and improve the character and efficiency of software system evolution In many software projects, software testing is ignored, there are many factors behind this like cost and time and so on that may result in deficiency of product quality and customer dissatisfaction and in the end to increase the overall software quality cost. Poor test approach, misjudge the effort of test case generation, delay in testing and following test maintenance are the main reason behind for cost mostly.
Mostly A unit test consists of a flow of approaches calls that create and modify objects, then use them as a parameter to a method under test. It is challenging task to automatically generate sequences that are original correct. This paper purposes a combined static and dynamic, automated test generation to address these problems for code without proper specification. Our first tactic uses dynamic analysis to suppose a call sequence from a sample program execution, then we use static analysis to recognize method craving relations based on the subjects they may say and compose. At the end, we combine the both dynamic model and the statically identified dependence information lead a random test generator to create a legal behaviorally test.
There are many several past research tools that follow an approach similar to us, but they neglect the two or three stages of our approach. Randoop, Palul and RecGen are different testing tools that are used in past. Paul presents the dynamic random approaches. RecGen uses a static does not have dynamic form and applies a static analysis to implement random test generation. Randoop is a pure random test generation tool.Compared to old approaches, Palus increases the structural coverage of generating test and improve their ability to detect errors. We applied it on half a dozen popular open source applications. The test results by Palus attained much higher structural coverage and found more unknown bugs than the other overtures.
Palus approach consists of four component names are, a load time instrumentation and dynamic model component, a static method analysis and a guided random test generation component. We present later a detailed introduction of all components
Related work for automated test generation techniques for OOP have been projected in the final decade. Like JCrasher creates test inputs by using a parameter graph to find method calls whose returns values can serve as input parameter. MSeqGen mines client code bases statically and extracts a frequent pattern as implicit programing rules that are practiced to support in generating tests. Another two alternatives approaches to create test input objects are with direct heap handling and using capture reply techniques.
For future work we are concerned with exploring two research directions. Mainly we plan to use different models like ADABU to guide an automated test generation.. Second, we are interested to train machine learning techniques to complement the dynamically inferred model we introduced here in the paper. Recently there is some work are done on machine learning techniques and proceed along.
0 comments:
Post a Comment