.NET Patterns, Part I
- Naming Conventions
- Using Constructors
- Method Overloading
- Summary
Like this article? We recommend
Like this article? We recommend
Like this article? We recommend
Now that Visual Studio .NET (VS .NET) and the .NET Framework have been released, development using the new tools and technology can proceed in earnest. However, as developers begin to build new solutions and rework existing code to take advantage of XML Web Services, ADO.NET, and other features of .NET, it's good to revisit some fundamentals concepts that will make development in .NET more efficient. In particular, adhering to the design guidelines covered in these two articles will make it easier to maintain and extend your .NET applications by using common naming conventions, usage, and design patterns.
As you develop for .NET, you'll notice that the framework classes that you'll work with—for example, the class of the System.Xml, System.Data, and System.IO namespaces, adhere to standard naming conventions and programming patterns. These conventions and patterns can be applied in any .NET language and, if emulated in your class libraries, will allow other developers to use and extend your classes more efficiently. The basis for the guidelines discussed in these two articles can be found in the online documentation that ships with VS .NET. By opening the documentation and searching for "class library design guidelines" in the index, you'll be presented with a wealth of topics, the most important of which we'll explore in this article.
Naming Conventions
You'll probably notice right away when working with VS .NET that the capitalization conventions for classes, variables, data types, arguments, and so forth are slightly different than those you might have used in VB 6.0 or VBScript. In fact, the framework uses three different styles of capitalization: Pascal case, camel case, and upper case. Pascal case is the most prevalent; it refers to identifiers with no spaces and with each individual word capitalized. In .NET, Pascal case should be used to identify classes, enumerated types, events, public fields, interfaces, methods, namespaces, and properties. For example, the declaration of a class in VB .NET would use Pascal case as follows.
Public Class QuilogyException : Inherits ApplicationException
The second most prevalent convention is the use of camel case. Camel case refers to identifiers in which the first word in the identifier is left in lower case while subsequent words are capitalized. In the framework, camel case is used for parameters and protected fields—for example, a constructor of the QuilogyException class would be defined as follows:
Public Sub New(ByVal message as String,ByVal innerException as Exception)
Notice that the parameters use camel casing while the names of the method and the types use Pascal casing. Upper case is reserved only for identifiers that consist of two or fewer letters. As a result, you'll typically see upper case used for namespaces such as System.IO and System.Web.UI. Keep in mind as well that different languages handle capitalization differently, so you do not want to create code that requires case sensitivity. In other words, if you were writing a class in C#, you would not create both ReadXml and readXml methods: VB .NET is not case sensitive, so VB .NET clients would not be able to differentiate between the two.
In VB and VBScript, parameters and variables often are prefixed with a two- or three-letter prefix that denoted the data type, perhaps along with a prefix that denote the scope—for example, str for strings and int for integers, prefixed with m for the module's level or l for local. This Hungarian notation was informative but resulted in long variable and parameter names. As shown in the previous snippet, Hungarian notation is no longer used in .NET where the identifier should be in camel case and simply describes the meaning of the variable or parameter. The only prefix that you'll typically see is an underscore to denote that a field or method is private to a class. The reasons for abandoning Hungarian notation include the fact that VS .NET already provides copious type information through the object browser, ToolTips, and the Watch window, as well as for the previously mentioned reason that you don't want to create variable and parameter names that include any language-specific information.
As mentioned previously, namespaces should use Pascal casing. Namespaces come in handy when you're designing classes that will be reusable and want to group them into a hierarchical structure that makes them easier to work with. For example, you might want to organize all of the classes in your organization that represent and access data in a hierarchical namespace structure. The general convention is that the top-level namespace in the hierarchy should be your organization name, with as many nested namespaces as necessary to fully represent your hierarchy. For example, Quilogy might group all of its data access classes, interfaces, and enumerated types into a Quilogy.Data namespace that contains nested namespaces for each core area of the company, including Quilogy.Data.Consulting, Quilogy.Data.Education and Quilogy.Data.Hosting. In a structure such as this, the classes and interfaces in the lower-level namespaces should have a dependency on those in the next highest level; classes in the Quilogy.Data.Consulting namespace might inherit from classes and interfaces in the Quilogy.Data namespace.
Finally, you should use consistent naming rules when naming your classes and interfaces. In the previous example of the QuilogyException class, the class name is a noun or noun phrase, uses Pascal casing, and is not prefixed with a C, as you sometimes see particularly in previous versions of VB. In addition, class names should not use underscores and should avoid abbreviations where possible. In this case, the class name also contains a piece of its base class name to identify that QuilogyException is a type of Exception. Although this is not a hard-and-fast rule, it can be useful in helping developers using your class get an immediate feel for how your class might behave.
Interfaces follow many of the same rules, including the prohibition on underscores and abbreviations and the use of Pascal casing. However, interfaces should be prefixed with an I and can be named using an adjective phrase that describes the functionality exposed by the interface instead of using a descriptive noun or noun phrase. For example, the framework contains the ICloneable interface that is implemented by a whole host of classes such as String, Stack, Array, SortedList, and Delegate, among others. Here, cloneable is an adjective that describes the functionality that the class gains by implementing the interface. Class and interface names also should differ only by the I prefix when the class is the standard implementation of the interface. For example, the QuilogyDataAccess class might provide the standard implementation of the IQuilogyDataAccess interface, as shown here:
Interface IQuilogyDataAccess Function GetById(ByVal id As Integer) As DataSet Function Save(ByVal ds As DataSet) As Boolean End Interface Public Class QuilogyDataAccess : Inherits IQuilogyDataAccess Public Function GetByPk(ByVal id As Integer) As _ DataSet Implements IQuilogyDataAccess.GetById ' implementation End Function Public Function Save(ByVal ds As DataSet) As _ Boolean Implements IQuilogyDataAccess.Save ' implementation End Function End Class
One of the interesting differences between VB .NET and C# is in the syntax used to identify the methods that map to the interface that is implemented (inherited in C#). In the previous example, the VB .NET code uses the Implements keyword on each method to identify which member of the interface the method maps to. This allows the method name exposed to the outside world to be different from the interface method name. In other words, the GetByPk method in QuilogyDataAccess can be mapped to the GetById method of Interface. When referencing the class, I can call the method simply using GetByPk; when referencing the interface, I can use GetById. As a result, the following code snippet ends up calling the same method twice:
Dim o As New QuilogyDataAcess() Dim d As IQuilogyDataAccess d = o o.GetByPk(3) d.GetById(3)
However, C# does not allow you to implement the members of an interface in your class using alternate names. Still, you can explicitly place the name of the interface in the method name, as shown below:
public class QuilogyDataAccess : IQuilogyDataAccess
{
DataSet IQuilogyDataAccess.GetById(int id)
{
//implementation
}
public Boolean Save(DataSet ds)
{
//implementation
}
}
In this case, the GetById method is implemented using the name of the interface, whereas the Save method is not. Doing so exposes the GetById method only through the interface, whereas the Save method can be called both through the reference to the class and through the interface, as exemplified in the following snippet:
QuilogyDataAccess o = new QuilogyDataAccess(); o.Save(ds); //legal o.GetById(3); //illegal ((IQuilogyDataAccess)d).GetById(3); //legal ((IQuilogyDataAccess)d).Save(ds); //legal
This is particularly handy when your class implements two interfaces that expose a member of the same name. However, when you define the method using the explicit name from the interface, access modifiers such as public, protected, and internal are not allowed and the method cannot be marked as virtual.