Date of Award

2011

Degree Type

Thesis

Department

Electrical and Computer Engineering

First Advisor

Sridhar, Nigamanth

Subject Headings

Computer programs -- Verification, Relational databases, Program Analysis, Static Analysis, SQL, C++, Software Engineering, Software, Comprehension, Bugs, Source Code, XML, Relational, Relational Query, Relational Database

Abstract

Code comprehension is pivotal to reducing errors in software. Reading source code improves code comprehension and enables effective fixes but as a code base grows meta-data become increasingly important. Static Analysis techniques provide an avenue for software developers to learn more about their code through meta-data while also helping them safely detect potential errors in their source. Unfortunately, many Static Analysis tools have a steep learning curve and are limited in scope. This thesis seeks to make Static Analysis accessible and extensible by asking what ubiquitous tools like SQL and relational databases can offer and what they cannot. We begin to answer these questions by exploring the source code of three C++ projects (libodbc++, log4cxx, C++ Sockets Library) using a new Static Analysis tool called Trike. Initial results indicate Trike is a promising and accessible tool for analyzing the structure of a code base. With further improvements, Trike should equal more established Static Analysis tools in scope and surpass them in usability

Download
COinS