次の方法で共有


Code Metrics – Class Coupling

SKU: Premium, Ultimate

Versions: 2008, 2010

Code: vstipTool0133

 

 

As we continue looking at code metrics in Visual Studio (see vstipTool0129, “Calculating Metrics”), it’s time to look at class coupling. Class coupling also goes by the name Coupling Between Objects (CBO) as originally defined by [CK94]. Basically, class coupling is a measure of how many classes a single class uses. A high number is bad and a low number is generally good with this metric. Class coupling has been shown to be an accurate predictor of software failure and recent studies have shown that an upper-limit value of 9 is the most efficient [S2010].

 

According to the MSDN documentation, class coupling “[m]easures the coupling to unique classes through parameters, local variables, return types, method calls, generic or template instantiations, base classes, interface implementations, fields defined on external types, and attribute decoration. Good software design dictates that types and methods should have high cohesion and low coupling. High coupling indicates a design that is difficult to reuse and maintain because of its many interdependencies on other types.”

 

The concepts of coupling and cohesion are clearly related. To keep this discussion on topic, I will not get into depth with cohesion other than to give a brief definition from [KKLS2000]:

“Module cohesion was introduced by Yourdon and Constantine as ‘how tightly bound or related the internal elements of a module are to one another’ [YC79]. A module has a strong cohesion if it represents exactly one task […], and all its elements contribute to this single task. They describe cohesion as an attribute of design, rather than code, and an attribute that can be used to predict reusability, maintainability, and changeability.”

 

 

Class Coupling Example

Let’s look at class coupling in action. First, create a new console application and create a new class called Person with some properties in it then immediately calculate the code metrics:

image

 

Notice the class coupling is 0 since this class doesn’t use any other classes. Now create another class called PersonStuff with a method that creates an instance of Person and sets the property values. Calculate the code metrics again:

image

 

See how the class coupling value goes up? Also notice that, no matter how many properties we set, the class coupling value just goes up by 1 and not by some other value. Class coupling measures each class only once for this metric no matter how much it is used. In addition, can you see that DoSomething() has a 1 but the constructor, PersonStuff() , has a 0 for its value? Currently there is no code in the constructor that is using another class.

 

What if we put code in the constructor that used another class? Here is what you get:

image

 

Now the constructor clearly has code that uses another class and the class coupling metric shows this fact. Again, you can see the overall class coupling for PersonStuff() is 1 and DoSomething() is also 1 to show that only one external class is being used no matter how much internal code we have that uses it.

 

Next, let’s create another new class. I’ll call my class Blah and create some properties inside it:

image

 

Now consume the class in our DoSomething() method within the PersonStuff class and calculate code metrics again:

image

 

As you can see, the class coupling for the PersonStuff class goes up to 2 and, if we drill into the class, we can see that the DoSomething() method has the most coupling in it but the constructor still only consumes 1 class.  Using these metrics we can see the overall max number for a given class and drill down into the details on a per-member basis.

 

 

The Magic Number

As I mentioned with cyclomatic complexity (see vstipTool0131, “Code Metrics - Cyclomatic Complexity”), there is no limit that fits all organizations. However, [S2010] does indicate that a limit of 9 is optimal:

“Therefore, we consider the threshold values […] as the most effective. These threshold values [for a single member] are CBO = 9[…].” (emphasis added)

 

 

Code Analysis

When using code analysis, the Extended Design Guideline rule set contains a maintainability area:

image

 

Inside the maintainability area is a rule for class coupling:

image

 

This rule issues a warning when the class coupling is excessive. You can learn more about the rule here:

https://msdn.microsoft.com/en-us/library/bb397994.aspx

 

Unfortunately, the documentation doesn’t talk about the thresholds set for this rule but I did find a description on the Code Analysis Team Blog (https://blogs.msdn.com/b/codeanalysis/archive/2007/11/15/code-metrics-as-check-in-policy.aspx): “Warning at above 80 for class and above 30 for a method”.  These seem abnormally high but at least provide an extreme upper limit. If you hit this warning then something is almost certainly wrong.

 

 

Citations

[CK94]

Chidamber, S. R. & Kemerer, C. F. (1994). A Metrics Suite for Object Oriented Design (IEEE Transactions on Software Engineering, Vol. 20, No. 6). Retrieved May 14, 2011, from the University of Pittsburgh web site: https://www.pitt.edu/~ckemerer/CK%20research%20papers/MetricForOOD_ChidamberKemerer94.pdf

 

[KKLS2000]

Kabaili, H., Keller, R., Lustman, F., and Saint-Denis, G. (2000). Class Cohesion Revisited: An Empirical Study on Industrial Systems (Proceedings of the Workshop on Quantitative Approaches in Object-Oriented Software Engineering). Retrieved May 20, 2011, from Université de Montréal web site https://www.iro.umontreal.ca/~sahraouh/qaoose/papers/Kabaili.pdf

 

[SK2003]

Subramanyam, R. & Krishnan, M. S. (2003). Empirical Analysis of CK Metrics for Object-Oriented Design Complexity: Implications for Software Defects (IEEE Transactions on Software Engineering, Vol. 29, No. 4). Retrieved May 14, 2011, from University of Massachusetts Dartmouth web site https://moosehead.cis.umassd.edu/cis580/readings/OO_Design_Complexity_Metrics.pdf

 

[S2010]

Shatnawi, R. (2010). A Quantitative Investigation of the Acceptable Risk Levels of Object-Oriented Metrics in Open-Source Systems (IEEE Transactions on Software Engineering, Vol. 36, No. 2).

 

[YC79]

Edward Yourdon and Larry L. Constantine. Structured Design. Prentice Hall, Englewood Cliffs, N.J., 1979.

Comments

  • Anonymous
    February 09, 2014
    Good one! =)

  • Anonymous
    February 21, 2014
    Are (1)import&export coupling metrics & (2)import&export code metrics are both same or not? #JustAsking for my survey prep! =)

  • Anonymous
    February 21, 2014
    Hey Aswini :) I'm not aware of an import/export for just class coupling in VS. You can export all the code metrics which includes class coupling, cyclomatic complexity, depth of inheritance, etc... Z

  • Anonymous
    July 31, 2014
    are this good for calculating " Coupling between object  CBO " metric ?