The knowledge of how to effectively spot smelly code and replace it with proven patterns will ultimately lead to a more stable, maintainable and elegant code base.

Often times, candidate code for refactoring is based upon a source file's smell, which its spotting can take time to acquire and is largely based upon subjective determinations. The proper use of code metrics, such as Cyclomatic Complexity, Fan-In, Fan-Out, and Depth of Inheritance can also facilitate the discovery of candidate code which is in need of refactoring.

For example, Cyclomatic Complexity is adept at spotting methods containing a high degree of conditional logic, which, consequently, can be replaced with polymorphism as elaborated in Martin Fowler's seminal work, Refactoring with the Replace Conditional with Polymorphism pattern. Additionally, excessively deep hierarchy trees create problematic testing targets, which can be broken out into separate objects with Fowler's Replace Inheritance with Delegation and Collapse Hierarchy patterns. Fan-In and Fan-Out are quite effective at pinpointing brittle code, which can be refactored into a more stable state with a plethora of patterns including Extract Hierarchy and Extract Class.

The open source DbUnit framework provides an elegant solution for controlling a database dependency within applications by allowing developers to manage the state of a database throughout a test. With DbUnit, a database can be seeded with a desired data set before a test; moreover, at the completion of the test, the database can be placed back into its pre-test state.

Most enterprise applications are constructed with a database in mind. Databases, however, create interesting dependency scenarios when it comes to effective white box testing. On one level, isolating application code with mock objects from a data store is an effective means of testing such code. Nevertheless, this level of code segregation is not always easy. Unit testing of EJB Entity beans is practically impossible without a container and an associated data store. Furthermore, applications using JDO-like frameworks also have slight difficulties in creating isolation levels without databases.

The open source DbUnit framework provides an elegant solution for controlling a database dependency within applications by allowing developers to manage the state of a database throughout a test. With DbUnit, a database can be seeded with a desired data set before a test; moreover, at the completion of the test, the database can be placed back into its pre-test state.

No one will argue that JUnit has positively affected the quality of thousands of Java applications around the world. JUnit’s simplicity and ease of use ushered in a whole new era of code quality; however, as many developers have found, its simplicity has also limited its use. TestNG was designed from the ground up to overcome some of JUnit’s limitations; moreover, TestNG’s features make it a great tool to complement your JUnit tests.

In this session we’ll learn about TestNG’s flexible fixture model, its test categorization capability, dependent methods, and TestNG’s powerful parametric testing facility. We’ll also see how both TestNG and JUnit can play together in a build process.