Software Quality Model

From Wikipedia, the free encyclopedia

[edit] Software Quality Model

Software quality can be defined as 'conformance to requirements' and/or 'fitness of use'. Quality achievements start with a loud and clear definition of what "quality of source code" means to your organization or project. In simple terms all the stakeholders must be well informed of what is expected, what are the goals to be achieved, what is evaluation criteria and how they can contribute to achieve the goal.

The first step in this direction is to decide the goals and their definitions, for example the following are some of the common goals of any software development / maintenance project.

Testability: Ease of testing. Maintainability: The ease of change.

This solves the first level of ambiguity, but the job of defining software quality has just begun. The very next step is to determine what predicates a quality goal. There can be several factors which can influence a particular goal positively or negatively, we can call these factors as predicates of a goal and these predicates can be as simple as a naming convention to be followed or as complex as a desired "useful" comments percentage per class. In general these predicates are nothing but the "Best Practices" to be followed and Standard "Software Metrics" which are collected to determine different quality aspects of the software solution.

Goal Oriented Measurement: Let's take the goal "Testability" for instance. The software metric "Cyclomatic Complexity" has considerable influence on the testability of a function, module and eventually the software as a whole. In simple terms CC(Cyclomatic Complexity) increase as the number of conditional constructs in a function increases, which actually mean that more testing effort is needed. So, we can consider Cyclomatic Complexity as a predicate for the goal "Testability", and as CC increases Testability becomes tougher (decreases).

Predicate Contribution: There can be many factors which can influence the quality goal "Testability". Swat4J provides a normalized framework to include as many predicate as we see fit. Predicate contribution to the goal can vary; Swat4J allows tagging weight to each predicate to prioritize their contribution. Simply, by setting weights we can control which predicates contribute more to the goal of concern. Now, that might seem a lot of "do it yourself", to overcome this Swat4J comes with prepackaged configuration which are widely used in the industry, it can be customized to fit specific needs.