In computer programming, cohesion refers to the degree to which the elements inside a module belong together.[1] In one sense, it is a measure of the strength of relationship between the methods and data of a class and some unifying purpose or concept served by that class. In another sense, it is a measure of the strength of relationship between the class's methods and data.
Cohesion is an ordinal type of measurement and is usually described as “high cohesion” or “low cohesion”. Modules with high cohesion tend to be preferable, because high cohesion is associated with several desirable software traits including robustness, reliability, reusability, and understandability. In contrast, low cohesion is associated with undesirable traits such as being difficult to maintain, test, reuse, or understand.
Cohesion is often contrasted with coupling. High cohesion often correlates with loose coupling, and vice versa.[2] The software metrics of coupling and cohesion were invented by Larry Constantine in the late 1960s as part of Structured Design, based on characteristics of “good” programming practices that reduced maintenance and modification costs. Structured Design, cohesion and coupling were published in the article Stevens, Myers & Constantine (1974)[3] and the book Yourdon & Constantine (1979).[1] The latter two subsequently became standard terms in software engineering.
In object-oriented programming, a class is said to have high cohesion if the methods that serve the class are similar in many aspects.[4] In a highly cohesive system, code readability and reusability is increased, while complexity is kept manageable.
Cohesion is increased if:
Advantages of high cohesion (or "strong cohesion") are:
While in principle a module can have perfect cohesion by only consisting of a single, atomic element – having a single function, for example – in practice complex tasks are not expressible by a single, simple element. Thus a single-element module has an element that is either too complicated to accomplish a task, or too narrow and thus tightly coupled to other modules. Thus cohesion is balanced with both unit complexity and coupling.
Cohesion is a qualitative measure, meaning that the source code is examined using a rubric to determine a classification. Cohesion types, from the worst to the best, are as follows:
/*Groups: The function definitionsParts: The terms on each function*/Module A { /* Implementation of r(x) = 5x + 3 There is no particular reason to group functions in this way, so the module is said to have Coincidental Cohesion. */ r(x) = a(x) + b(x) a(x) = 2x + 1 b(x) = 3x + 2}
/*Groups: The function definitionsParts: The terms on each function*/Module A { /* Implementation of arithmetic operations This module is said to have functional cohesion because there is an intention to group simple arithmetic operations on it. */ a(x, y) = x + y b(x, y) = x * y}Module B { /* Module B: Implements r(x) = 5x + 3 This module can be said to have atomic cohesion. The whole system (with Modules A and B as parts) can also be said to have functional cohesion, because its parts both have specific separate purposes. */ r(x) = [Module A].a([Module A].b(5, x), 3)}
/*Groups: The function definitionsParts: The terms on each function*/Module A { /* Implementation of r(x) = 2x + 1 + 3x + 2 It's said to have perfect cohesion because it cannot be reduced any more than that. */ r(x) = 2x + 1 + 3x + 2}
Although cohesion is a ranking type of scale, the ranks do not indicate a steady progression of improved cohesion. Studies by Larry Constantine, Edward Yourdon, and Steve McConnell[5] indicate that the first two types of cohesion are inferior, communicational and sequential cohesion are very good, and functional cohesion is superior.