The life-cycle development or waterfall approach breaks the development cycle down into discrete phases, each with a rigid sequential beginning and end (see Figure 6.1). Each phase is fully completed before the next is started. Once a phase is completed, in theory during development, one never goes back to change it.

In Figure you can see that the first phase in the waterfall is user requirements. In this phase, the users are interviewed, their requirements are analyzed, and a document is produced detailing what the users' requirements are. Any reengineering or process redesign is incorporated into this phase.

In the next phase, entity relation diagrams, process decomposition diagrams, and data flow diagrams are created to allow the system to be broken down into manageable components from a data and functional point of view. The outputs from the logical design phase are used to develop the physical design of the system. During the physical and program unit design phases, various structured design techniques, such as database schemas, Yourdon structure charts, and Warmer– Orr diagrams, are used to produce a design specification that will be used in the next phase.

In the program unit design phase, programmers develop the system according to the physical design produced in the previous phase. Once complete, the system enters the coding phase, where it will be written in a programming language, unit or component tested, integration tested, system tested, and finally, user tested (often called acceptance testing).

Now the application is delivered to the users for the operation and maintenance phase (not shown in Figure 6.1). Defects introduced during the life-cycle phases are detected and corrected, and new enhancements are incorporated into the application.

Psychology of Life-Cycle Testing

In the waterfall development life cycle, there is typically a concerted effort to keep the testing and development departments separate. This testing organization is typically separate from the development organization, with a different reporting structure. The basis of this is that because requirements and design documents have been created at specific phases in the development life cycle, a separate quality assurance organization should be able to translate these documents into test plans, test cases, and test specifications. Underlying assumptions include the belief that (1) programmers should not test their own programs and (2) programming organizations should not test their own programs.

It is thought that software testing is a destructive process and that it would be very difficult for a programmer to suddenly change his perspective from developing a software product to trying to find defects, or breaking the software. It is believed that programmers cannot effectively test their own programs because they cannot bring themselves to attempt to expose errors.

Part of this argument is that there will be errors due to the programmer's misunderstanding of the requirements of the programs. Thus, a programmer testing his own code would have the same bias, and would not be as effective testing it as someone else.

It is not impossible for a programmer to test her own programs, but testing is more effective when performed by someone who does not have a stake in it, as a programmer does. Because the development deliverables have been documented, why not let another individual verify them?

It is thought that a programming organization is measured by its ability to produce a program or system on time and economically. As with an individual programmer, it is difficult for the programming organization to be objective. From the point of view of the programming organization, if a concerted effort were made to find as many defects as possible, the project would probably be late and not cost effective. Less quality is the result.

From a practical point of view, an independent organization should be responsible for the quality of the software products. Product test or quality assurance organizations were created to serve as independent parties.

0 comments