Search for: All records

Distributed systems are seeing wider use as software becomes more complex and cloud systems increase in popularity. Preforming anomaly detection and other log analysis procedures on distributed systems have not seen much research. To this end, we propose a simple and generic method of clustering log statements from separate log files to perform future log analysis. We identify variable components of log statements and find matches of these variables between the sources. After scoring the variables, we select the one with the highest score to be the clustering basis. We performed a case study of our method on the two open-source projects, to which we found success in the results of our method and created an open-source project log-matcher. 

Code clones are fragments of code that are duplicated in the codebase of an application. They create problems with maintainability, duplicate buggy code, and increase the size of the repository. To combat these issues, there currently exists a multitude of programs to detect duplicated code segments. However, there are not many varieties of languages among the benchmarks for code clone detection tools. Without covering enough languages for modern software development, the development of code-clone detection tools remains stunted. This paper describes a novel tool that will take a seed of Python source code and generate Type 1, 2, and 3 code clones in Python. As one of the most used and rapidly-growing languages in modern software development, our testbed will provide the opportunity for Python code-clone detection tools to be developed and tested. 

Static analysis is one of the most important tools for developers in the modern software industry. However, due to limitations by current tools, many developers opt out of using static analysis in their development process. Some of these limitations include the lack of a concise, coherent overview, missing support for multiple repository applications and multiple languages and lastly a lack of standardized integration mechanisms for third-party frameworks. We propose an evaluation metric for static analysis tools and offer a comparison of many common static analysis tools. To demonstrate the goal of our metric we introduce the Fabric8-Analytics Quality Assurance Tool as a benchmark of a tool which successfully passes our evaluation metric. We demonstrate usage of this tool via a case study on the Fabric8-Analytics Framework, a framework for finding vulnerabilities in application dependencies. We issue a challenge to developers of modern static analysis tools to make their tools more usable and appealing to developers.