An infrastructure for automated software repair based on code search: implementation of a Sourcerer instance

• Grant number: 13/04966-7
• Support Opportunities: Scholarships abroad - Research Internship - Scientific Initiation
• Start date: June 01, 2013
• End date: August 31, 2013
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Otávio Augusto Lazzarini Lemos
• Grantee: Felipe Capodifoglio Zanichelli
• Supervisor: Cristina Videira Lopes
• Host Institution: Instituto de Ciência e Tecnologia (ICT). Universidade Federal de São Paulo (UNIFESP). Campus São José dos Campos. São José dos Campos , SP, Brazil
• Institution abroad: University of California, Irvine (UC Irvine), United States

Abstract

The task of fault correction is an important obstacle in contemporary software development: studies point ou that about 90% of the total costs of a software project life-cycle accounts for maintenance. Code repair and other forms of software evolution are tasks that impact such costs. A way to save resources in this context is to automate the software repair activity using sistematic approaches. Weimer and others have proposed an automated code repair approach based on genetic programming. The basic idea is to use functional parts of the system itself to repair portions that may contain the fault. Failing test cases help out to indicate such code portions. Such approach can be improved by enlarging the search space, in the original proposal restricted only to the system itself. In fact, the Test-Driven Code Search (TDCS) approach, which is based on open source code repositories, could be adapted to support code repair. The goal of this BEPE project is to allow the colaboration between the candidate with Crista Lopes, a prestigious researcher from UCI, to implement part of the main IC project. The main goal is to allow the student to understand and implement a large instance of the Sourcerer infrastructure (developed in UCI), which is essential to the proposed code repair approach. (AU)