CSC/ECE 517 Fall 2009/wiki1b 1 ch: Difference between revisions
No edit summary |
|||
| Line 11: | Line 11: | ||
* Localizing fault | * Localizing fault | ||
* Correcting fault | * Correcting fault | ||
==Failure Detection== | ==Failure Detection== | ||
| Line 26: | Line 27: | ||
From the results printed in the console window, we can evaluate whether the code is doing what it supposed to do or not. However, when the complexity of the code is growing and you need to work with multiple classes, it will be more convenient to automate the testing process. Automation tools such as [http://en.wikipedia.org/wiki/JUnit JUnit] comes in handy and provides programmers with information on which class or function that was failing. The next step is to locate the exact position of the fault. | From the results printed in the console window, we can evaluate whether the code is doing what it supposed to do or not. However, when the complexity of the code is growing and you need to work with multiple classes, it will be more convenient to automate the testing process. Automation tools such as [http://en.wikipedia.org/wiki/JUnit JUnit] comes in handy and provides programmers with information on which class or function that was failing. The next step is to locate the exact position of the fault. | ||
==Fault Localization== | ==Fault Localization== | ||
| Line 38: | Line 42: | ||
* Failure-inducing changes (adding / updating / deleting variables, functions, classes, etc.) | * Failure-inducing changes (adding / updating / deleting variables, functions, classes, etc.) | ||
* Failure-inducing states (key presses, button presses, mouse clicks, etc.) | * Failure-inducing states (key presses, button presses, mouse clicks, etc.) | ||
===Whyline=== | ===Whyline=== | ||
Revision as of 03:29, 29 September 2009
Advances in Debugging
Introduction
Debugging has been an important part of software development life cycle ever since the first computer program being introduced. Since debugging process often performed at the very end of software development life cycle, the cost of fixing bugs are higher. Finding the cause of bugs within the code with complex architecture could be difficult, slow, tedious, and sometimes frustrating. Researchers have been finding ways to increase the efficiency of debugging process, thus reducing the time required for the process.
During the last decade, debugging techniques and tools have advanced immensely. Many of the most popular and widely used debugging tools are available for free and developed as open source projects. The most popular tools used by programmers, from beginners to professionals, are probably Eclipse and NetBeans IDE. These IDEs are not only used by individual developers, but also by developers working on commercial products.
Debugging process is divided into three steps of:
- Detecting failure
- Localizing fault
- Correcting fault
Failure Detection
Failure detection occurs during software testing. Failure is first detected when program fails to run or does not pass the unit tests. The process of finding the failures started as simple as implementing the print line command such as below:
int i = 5;
int j = 3;
int k = i+j;
System.out.println("i is " + i);
System.out.println("j is " + j);
System.out.println("i+j is " + k);
From the results printed in the console window, we can evaluate whether the code is doing what it supposed to do or not. However, when the complexity of the code is growing and you need to work with multiple classes, it will be more convenient to automate the testing process. Automation tools such as JUnit comes in handy and provides programmers with information on which class or function that was failing. The next step is to locate the exact position of the fault.
Fault Localization
Delta Debugging
Delta Debugging, developed by Professor Andreas Zeller's group at Saarland University in Germany, is an automatic and systematic technique that deals with isolation of the source of the failure -- often the developer's longest, most demotivating, and most repetitive activity when carried out manually.[1] The algorithm isolates the difference between the passing and failing test case hence simplifies the number of test cases needed.
Delta Debugging can be applied to isolate the cause of these different applications:
- Failure-inducing input
- Failure-inducing changes (adding / updating / deleting variables, functions, classes, etc.)
- Failure-inducing states (key presses, button presses, mouse clicks, etc.)
Whyline
Whyline is an interactive debugging tool which is designed to help programmers locating the fault by allowing them to ask questions along the source code based on their dependency.
Visual Interactive Debugging Aids (VIDA)
Visual Interactive Debugging Aids (VIDA) is a debugging tool which continuously recommends breakpoints for the programmer based on the analysis of execution information and the gathered feedback from the programmer.
Debugging and The Open-Source Community
There are a lot of open source debugger tools that have become popular among professional developers. They range from stand alone applications to Integrated Development Environment (IDE). They provide many choices for developers to choose whether they want tools for specific languages, operating systems, algorithms, optional plug-ins, etc. Two of the most popular IDEs that are widely used are Eclipse and NetBeans. Both tools are open source, easy to use, and provide cross-platform environment.
Eclipse
IBM developed Eclipse in the late 1990s and turned it over in 2001 to the nonprofit Eclipse Foundation. It became an open-source platform which enabled anyone to build applications using the technology. Eclipse is a popular choice, especially among Java developers, and often compared to NetBeans.
NetBeans
NetBeans is an open-source IDE developed in Java using the NetBeans Platform. NetBeans offer bundles for specific development needs, such as Java EE, Ruby, PHP, and C/C++.
Debugging in Eclipse and NetBeans
Eclipse and NetBeans come with debugging facilities, such as running and stopping the program, setting up breakpoints, keeping track of variables' value, watchpoint, hit count, etc. A nice tutorial for Java debugging with Eclipse can be found here.
Advanced Debugging Tools
Tarantula
Tarantula is a visualization system that displays the results of running suites of tests against software systems. By showing what portions of the code are executed by passed and failed tests, the system helps people identify faults in their programs. (http://pleuma.cc.gatech.edu/aristotle/Tools/tarantula/) Tarantula automates the time-consuming debugging process of stepping through the code and verifying the values at each step. It visualizes the suspicions of statements by different colors, ranging from red, yellow, to green. The statements that are suspicious will be marked with red color, whereas the innocent statements marked by green color.
(Insert example screenshot)