|
Using Code Metrics in Visual Studio Team System 2008
|
|
Rating: 9 user(s) have rated this article
Posted by: Subodh Sohoni,
on 3/31/2008,
in category "VSTS"
Views: this article has been read 12739 times
Abstract: Developers need to have feedback about the quality of code immediately after the code is written. Team Edition for Software Developers which is part of Visual Studio Team System provides an integrated tool named Code Metrics to calculate how the code is written based upon some quantifiable criterion and shows the results in a separate window. Calculations are based upon criterions like Cyclomatic complexity, number of lines in code and inheritance depth etc. all resulting in maintainability index of the code. In this article we will take an overview of Code Metrics built into VSTS.
Using Code Metrics in Visual Studio Team System 2008
Developers need to have feedback about the quality of code immediately after the code is written. Team Edition for Software Developers which is part of Visual Studio Team System provides an integrated tool named Code Metrics to calculate how the code is written based upon some quantifiable criterion and shows the results in a separate window. Calculations are based upon criterions like Cyclomatic complexity, number of lines in code and inheritance depth etc. all resulting in maintainability index of the code. In this article we will take an overview of Code Metrics built into VSTS.
Code Metrics provides feedback to developers on following factors:
1. Maintainability Index: this index is a computed using a formula as illustrated in Visual Studio Code Analysis Team Blog: Maintainability Index = 171 - 5.2 * log2 (Halstead Volume) - 0.23 * (Cyclomatic Complexity) - 16.2 * log2 (Lines of Code). This formula is based upon the work done at Carnegie Melon University.
Range of values for maintainability index is from 0 to 100 where 0 is the least maintainable code and 100 is perfectly maintainable code. Naturally higher values i.e. the ones tending towards 100 are going to be welcome. To make the things simple, code metrics tool in its results provides icons for the part of the code (a class or a method) along with index values to show how much maintainable it is. Red icon represents maintainability index values less than 10 which mean that those parts of the code need immediate attention of the developer. Values between 10 and 20 are accompanied with yellow icon to bring to notice the parts of the code which are not well maintainable. All the values above 20 are adorned with green icon to show that the represented part of the code is maintainable.
2. Cyclomatic Complexity: This value is the count, for each part of the code, of various branches in the part of the code. Branches can be because of various statements which can make decisions like if and switch statements or they can be loop statements like for, while, do etc. Lower values are better as they provide more readable and maintainable code.
3. Class Coupling: This value is a count for each class on which that class depends. If the coupling count for a class is high which means that it depends upon many other classes then the chances that the code of this class will break due to some changes in the dependency classes is high. That is why as a good practice, this number should be as low as possible. Solution is to provide loose coupling or Service Oriented Architecture if possible. To identify common components and abstract them as a service will be a good practice. Class coupling number does not take into account the primitives like int32 and String etc.
4. Inheritance Depth: This number represents the number of types which are above the type being reported in the inheritance tree. This value is measured from Object class which is at level 0.
Results Window for Code Metrics
Results of the code metrics calculation are made available in a separate window which can be opened from either the View menu or from the Analyze menu. Once the window is opened we can calculate the code metrics by clicking on the analyze button as shown with a red circle in Figure 1. Results are shown in a grid which has a row for each module (e.g. class, method) for which code metrics can be computed. For each measurable criterion as are mentioned above, there is a column in the grid. Figure 1 shows the Results Window
Figure 1
We can filter the results of code metrics based upon values of any of the columns in the results window. To set the filter we can use the menu bar of the code metrics results window. In the menu bar we can select the column which will be the criterion and then set the minimum and maximum values which we want to set filter on. It is not possible to set the complex filter.
It is also possible to remove some of the columns or add them back again in the results window. It is also possible to change the sequence in which they appear from left to right.
We can export the results of code metrics either to the clipboard or to MS Excel in a table form. These results then can be used to do further offline analysis.
One of the best features for management is the ability to create workitems directly from the code metrics results window. We can use the context menu to create various types of workitems and assign them to developers for refactoring the code which does not have good maintainability.
Limitations and Constraints
Some of the things that code metrics feature of Team Edition for Software Developers cannot do are as follows:
1. Calculating code metrics for only part of the code in a module.
2. Providing custom columns in the code metrics results window.
3. Pinpointing the code lines due to which maintainability of the code of a method is reduced.
4. Exclusion of certain type of code like generated code from metrics calculations and results.
5. Calculation of code metrics cannot be set as check-in policy
Conclusion
Microsoft has provided an excellent tool to get an idea about the maintainability of the code, immediately after it is written, in the form of Code Metric Calculation. It provides these results based upon scientifically measured parameters and proven formulae. Results are shown in a separate window and they can be filtered as well as their view can be customized to suite the requirements.
I hope this article was useful and I thank you for viewing it.