Research On Compatibility Issues Detection For Android Applications

With the development and maturity of Android ecosystem,Android devices and applications are rapidly gaining popularity.Due to the differences between devices,it is difficult for an Android application to ensure its consistency of functions among these devices,and compatibility issues arise from this.In this situation,Android developers use software testing to troubleshoot the compatibility issues in applications.Most of the existing studies focus on assisting developers in dynamic testing to reduce testing costs.However,how to identify,locate and fix the compatibility issues in Android applications at code level is still a challenge.Therefore,there is an urgent need for a new solution to the compatibility issues,which can automatically solve the compati-bility issues under the premise of ensuring efficiency.This dissertation has carried out the following research work around the solution to the compatibility issues in Android applications:1.This dissertation proposes a technology for identifying compatibility-issues-proneness applications based on multi-modal analysis.This research is dedi-cated to helping android developers quickly identify Android applications that may have compatibility issues,and then be able to test the targeted applications,and solve the cross-devices inconsistency on the applications.The challenge of this research is that the semantics of information in Android applications are com-plex,and the effectiveness of extracting information cannot be guaranteed.In order to solve this challenge,based on the insight into the causes of compatibility issues,this research carried out fine-grained and multi-modal extraction of ap-plication information,and manually constructed a data set that could reflect the degree of compatibility and finally used statistical testing methods to analysis of the validity of multi-modal information,and select useful information to iden-tify the target application.Experimental results show that this technology can be applied to the detection of the applications with compatibility issues.2.This dissertation proposes a technology for automatic identification of compat-ibility defect compatibility-related-API based on biased distribution.This re-search is dedicated to helping developers identify APIs in the code layer that can cause compatibility issues,and then be able to modify the application in the code layer to reduce the development cost of application compatibility issues.The challenge of this study is that the analysis cost caused by the magnitude of the API is huge;and the compatibility-related-API derived from different causes,and the existing automation solutions can only identify a specific type of target API.In order to solve this challenge,this study uses a combination of dynamic and static analysis to analyze the control dependencies of APIs in the application,and then identify the compatibility-related code segments,and build a candidate set of tar-get APIs to narrow the search space;then leverage the biased distribution of the compatibility-related-APIs among code segments to accurately identify the target API in the candidate set.Experimental results on real Android application data sets show that this method can effectively identify compatibility-related-APIs de-rived from different causes.3.This dissertation proposes an automatic code repair technology for compatibility issues in Android applications.The purpose of this research is to further reduce the developer's overhead in solving compatibility issues,and to automatically im-prove the code segments with compatibility defects for developers.The challenge of this research is how to judge whether compatibility defects arise and how to fix the problem code.In order to solve this challenge,this research first identifies the compatibility-related-APIs in the application,and uses backward analysis to determine whether the target API will cause compatibility problems;then capture the guard conditions of API,combined with the code instrumentation technology to construct the necessary conditional judgments,and then fixes compatibility issues.Experimental results prove that this technology can effectively repair po-tential compatibility problems in the code with acceptable time cost.