C-Sharp current and future trends

Erik Borgenvik and Martin Saegebrecht 24 may 2006

Contents
0.1 0.2 0.3 Introduction . . . . . . . . . . . . The birth of C# . . . . . . . . . .NET Framework . . . . . . . . . 0.3.1 Typecategories . . . . . . 0.3.2 Class library . . . . . . . 0.3.3 MSIL and JIT . . . . . . 0.3.4 Garbage collection . . . . The rst version of C# . . . . . 0.4.1 Development . . . . . . . 0.4.2 Pointers and Unsafe code 0.4.3 Delegates . . . . . . . . . 0.4.4 Reection . . . . . . . . . The second version of C# . . . . 0.5.1 Generics . . . . . . . . . . 0.5.2 Partial Types . . . . . . . 0.5.3 Nullable Types . . . . . . 0.5.4 Anonymous Methods . . . 0.5.5 Iterator . . . . . . . . . . Design patterns in .NET . . . . . 0.6.1 Observer pattern . . . . . 0.6.2 Iterator pattern . . . . . . C# future . . . . . . . . . . . . . 0.7.1 Extension methods . . . . 0.7.2 Lambda Expressions . . . 0.7.3 Implicit variables . . . . . 0.7.4 Object initializers . . . . 0.7.5 Anonymous Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 4 5 5 6 6 6 7 8 9 9 10 10 11 13 14 14 16 17 17 18 19 20 20

0.4

0.5

0.6

0.7

Abstract What is a perfect programming language? There isnt a simple answer to that question, because all languages has there own dierent qualities. Some people say a perfect language is a language which is exible and easy to understand. A language with a rich class library and with a high abstract level makes it easier for the programmer to create exible programs. C# (C-Sharp) is such a language. It has a rich class library and takes care of a lot of things a programmer normally has to handle himself like memory deallocation etc. Are there any drawbacks then? Higher abstraction conceals a lot of underlying syntax from the programmer. Of course this makes it easier to create datatypes but there will also follow some complexity because the programmer will be more dependent of the built in types and looses a bit freedom to solve problems in his own way. Some parts of C# is really good and saves the programmer a lot of time while other parts just seem to increase the complexity and actually makes it harder to write code that is understandable. The report will cover features like delegates, anonymous types, nullable types and lambda expressions among with other interesting topics. To make you come up to your own opinion of C#, the report contains the most signicant parts of the language.

0.1

Introduction

This report will give you deeper knowledge about the object oriented programming language C#. It covers the most signicant programming parts of the language and gives a deeper knowledge about the underlying framework. We assume that you are familiar with the basic principles of object oriented programming and understand classes, abstract data types, structures etc. At the end you can also read about the future of C# with the new upcoming version 3.0.

0.2

The birth of C#

In the end of the 90s Microsoft started a big project meant to create a whole new infrastructure for software development and distributed systems. At that time Microsofts development tools were all working and executing independently from each other. The task was to develop one new model for objected oriented programming and another model for the execution of component based software. The project was called .NET (dotnet) which you can read more about in the following chapter. Another important milestone in the .NET project was to separate the type system (how variables are being created etc.) from the language so that all languages that support the type system could t into .NET. 1996 Anders Hejlsberg was recruited by Microsoft. As a chief Architect he and his group created C# as a part of the .NET project which were presented in the year 2000 and became an ISO standard in 2003. This language was created to support the same type system that .NET is using. C# is a modern objected oriented language which includes the most popular mechanisms from Java and C++ and has also got some inuences from Visual Basic. C# is simple to use, in comparison to C++, multiple inheritance from classes has been removed. Constants, variables and arrays must always belong to a class or a struct. Almost everything in C# are either classes and objects, even the classic primitive types like int, oat, etc can be seen as objects.

0.3

.NET Framework

The .NET Framework is a develop and execution environment that allows dierent programming languages and libraries to work together seamlessly to create Window-based applications which are easier to build, manage, deploy and integrate with other networked systems. Dotnet is based on the international standard called CLI (Common Language Infrastructure) which is central for libraries and languages to be able work together. Microsofts implementation of CLI is called CLR (Common Language Runtime) which provides that service. The CLI standard which languages like C# are based on contains the following principles Common Type System (CTS): Denes how types are declared, used and managed in the runtime environment. It also denes rules that languages must follow which ensure that objects written in dierent languages can interact with each other. According to CTS there are only three kinds of datatypes: classes, structs and interfaces. Common Language Specication (CLS): To fully interact with other objects regardless of the language they were implemented in there must be a set of simple types that all languages can understand and which they can include in there interfaces. Common Intermediate Language (CIL): A format for metainformation which after it has been JIT (Just In Time) - compiled becomes code for the CPU. Virtual Execution System (VES): Denes the environment of the execution, how classes should be loaded and automatic garbage collection.

Figure 1: CLI model.

0.3.1

Typecategories

When programming with CLI it is possible to use any language that supports the CTS/CLS. According to CLS the language are separated into two categories called consumers and extenders. Consumers only uses functions stored in the framework, but they dont create any new datatypes and dont have to support the object oriented mechanisms. Extenders are able to create datatypes according to the CLS and possible also according to the CTS. C# is an extender and supports both CLS and everything in CTS. If the framework are suppose to be independent from the choice of language there must be a set of simple types that all languages have to understand and not use any other types in there public interfaces. Therefore CLS types are as follows: CLS bool: can only be true or false char: 16 - bits character unsigned int8: 8 bits non - negative integer int16: 16 - bits integer int32: 32 - bits integer int64: 64 bits integer oat32: 32 bits oating number with precision of 7 numbers oat64: 64 bits oating number with a precision of 15 numbers One important thing to note here is that the CLS - types are simple types, not primitive types. Primitive types are variables not objects and they can only contain a value. Objects have both values and behaviors. As we mentioned, in CTS there are only three kinds of datatypes where class is the most central (the other two are structs and interfaces). The big dierence between a class and a struct is that a struct can be created on the stack which is not the case for class-objects. Interfaces are classes without implementation, only named methods.

0.3.2

Class library

The entire class library is a big deriving hierarchy where the base class is called Object. All datatypes inherits from Object. It is the root of the type hierarchy and supports all classes in the .NET Framework. The library is also divided into dierent namespaces where the root is called System. Namespaces are thought to give datatypes a division which corresponds to applicationareas. For example most things that has to do with threads and synchronization is placed in the System.Threading namespace.

Other example of namespaces is: System.Timers: allows you to trigger an event on a specic time interval. System.Security: Contains base classes for permission. System.Drawing: provides graphic functionality.

0.3.3

MSIL and JIT

When compiling code .NET translates the source code to a CPU - independent language called Microsoft Intermediate Language (MSIL) which can be converted to native code. MSIL includes instructions for loading, storing, initializing, control ow among other things. Before code can be run, MSIL must be converted to CPU - code usually by a Just - In - Time - compiler. This type of compiler has the benet of compiling code whenever it is needed instead of compiling the whole program at once. .NET supplies one or more JIT- compilers for each computer-architecture it supports. To describe the benets that MSIL is providing we rst have to explain the term metadata. Metadata is binary information that describes your program that is stored either in a common language runtime portable executable (PE) le or in memory. When you compile your code into a PE le, metadata is inserted into one portion of the le, while your code is converted to MSIL (Microsoft intermediate language) and inserted into another portion of the le. Every type and member dened and referenced in a module or assembly is described within metadata. When code is executed, the runtime loads metadata into memory and references it to discover information about your codes classes, members, inheritance, and so on. Metadata allows .NET languages to describe themselves automatically in a language-neutral manner.

0.3.4

Garbage collection

One interesting quality that VES is providing is the automatic garbage collection. Objects that dont have a reference will be erased automatically. Without garbage collection the programmer has to the job him self for example by calling the funcion delete in C++. t Example: MyClass obj = new MyClass(); Obj = null; When the reference is disappearing or being put to another value the object gets one less reference. When all references are gone the object is ready for garbage collection. To detect objects with no references is usually very hard because objects can be put together as a chain and it demands complicated algorithms to manage that with graph analyse. VES is supposed to handle those situations. 5

0.4
0.4.1

The rst version of C#

Developed as a part of Microsofts .NET package C# was set to be an object oriented simplication of C++, although when released it also contained many features from Delphi, Visual Basic and Java. This was much due to chief designer Anders Hejlsberg who previously had been working on the development of Delphi and Visual J++. An ongoing theme during the initial and continuous development of the language was to add simplicity over simplexity, Hejlsberg describes this in an interview as follows: No one ever argues that simplicity isnt good, but people dene simplicity in a variety of ways. Theres one kind of simplicity that I like to call simplexity. When you take something incredibly complex and try to wrap it in something simpler, you often just shroud the complexity. You dont actually design a truly simple system. In some ways you make it even more complex, because now the user has to understand what was omitted that they might sometimes need. Thats simplexity. So to me, simplicity has to be true, in the sense that the further down you go the simpler it gets. This thinking opened up new opportunities for C# whereas it now could also be used for RAD (Rapid Application Deployment) where usability and simplicity are highly valued. Many important design choices had to be made during the development of C#, one was not to use multiple-inheritance, the reason for this was to avoid complications with CLIs structure and at the same time make the user unable to struggle with multiple dependant classes when coding. Even though C# 1.0 lacks the use of multiple inheritance, a class can still implement an endless number of interfaces to it. One thing that was overlooked in the early version of C# was polymorphism, no support for templates as in C++ was made and generics was rst to come in a later version.

0.4.2

Pointers and Unsafe code

Since C# is much like Java and use safe references you might think it doesnt support the use of pointers but thats in fact not all true. Even though C# was made to be safe and simple, pointers can still be used although in a somewhat limited fashion. Code that uses a user-dened pointer will be marked as unsafe and only the proper application with the right permission will be able to execute the unsafe code. When operating in unsafe code it is much like writing C code but in a C# environment. A common misconception is that unsafe code is bad and shouldnt be used but dont let the name unsafe fool you. Ironically unsafe code is very safe compared to a C or C++ program where basically the whole program would be marked as unsafe. In C-Sharp you can strictly control the access by using the keyword unsafe and the execution engine also makes sure the code isnt executed in an untrusted environment. However unsafe code shouldnt be used unless it is really called 6

for. As mentioned earlier C# doesnt directly allow manipulation of pointers and direct memory management but this still needs to be done. When you read about VES in the previous chapter you were informed about the automatic garbage collection. When needed the garbage collection frees up memory and thus managed memory doesnt have to be freed explicitly by the user.

0.4.3

Delegates

Delegation is one of the more interesting features C# 1.0 has to oer. Using a delegate enables the user to pass-on the execution to another object, in C++ this could be solved by the use of function pointers. Delegates however has some distinct advantages over function pointers, they not only oer the full use of object orientation, but also encapsulates an object instance and method. To better understand delegates and how they work we will look at a rst example where a delegate is made to t the correct methods. delegate int D(int i, double d); class A { public static int M1(int a, double b) {...} } class B { public static int M1(int f, double g) {...} public static void M2(int k, double l) {...} public static int M3(int g) {...} } In order for a delegate to be instantiated and work properly it has to have the correct parameters and return value. The delegate D used here has a return type of int and takes two parameters of type int and double. This delegate can now be used on either M1 in class A or M1 in class B since both those methods fulll the type requirements. Methods M2 and M3 in class B however have either a wrong return type or parameter type and therefore cant be used when the delegate is instantiated. A delegate can consist of an instance method or static method but a delegate can also consist of another delegate. This allows the user to create a chain of delegates which can then be executed in order. The + or += operator is used to combine two or more delegate into a sequence which will then be executed when the original delegate is called. Heres an example of how it works:

delegate void D(int x); class A { public static void M1(int i) {...} public void M2(int i) {...} } class B { static void Main() { D cd1 = new D(A.M1); A a = new A(); D cd2 = new D(a.M2); D cd3 = cd1 + cd2; cd3 += cd1; } Here we create 3 new delegates cd1, cd2 and cd3. The rst delegate cd1 is created from the static method M1, thereafter cd2 is created through instance method M2 and nally we create a delegate cd3 from the two delegates we just made. Last we add the cd1 delegate to cd3 once again using the += operator. Note that the type rules still apply when using a set of delegates and thus methods with incorrect parameters or return type cant be used. Just as you can add delegates you can also remove them using the -= operator. In the case where more then one of the same delegate is existent in a set just like cd1 in the previous example, then the last added delegate of cd1 will be removed. If you were to use a return value in the methods of a set of delegates the output value would be of the last executing delegate.

0.4.4

Reection

Another interesting feature C-Sharp 1.0 oers is the use of reection. For people new to this, reection or Computational reection enables a program to observe and modify its behavior or structure at runtime. This can prove to be extremely useful in some situations such as le loading at runtime. This is the same principal you read about in the dotnet chapter about MSIL.

0.5
0.5.1

The second version of C#

Parameterized types, also known as generics is essentially the biggest change from C# 1.0 In short, generics is the ability to have dierent parameters on a type, much like C++ templates but without any complications. You might say, yes but this can already be made using the object type. This is true but using object does come with some drawbacks. Consider the following example of an object implemented stack. Stack stack = new Stack(); stack.Push(new Customer()); Customer c = (Customer)stack.Pop(); The Customer object in this case is pushed onto the stack and that is ne since any object can be used. Although later when its time to pop you explicitly have to cast it to a Customer since the stack itself doesnt know what spesic object was pushed on in the rst place. In the end it will lower performance becouse of the type checking made at runtime. The same applies when the input is a value of a value type, such as a number. The the number will automaticlly be boxed and later has to be unboxed using cast. This will lead to dynamic memory allocations being used as well as type checking at run-time. Lets now look at how generics work and the advantages it comes with. We start o by looking at an example of a Stack, implemented using generics. class Stack<T> { T[] items; int count; public void Push(T item) {...} public T Pop() {...} } In this example we use a type parameter called T witch act as a placeholder until a type is specied. Unlike using the object type we now arent forcing any type conversions, instead we accept the data type for what it is and store it. When we instanciate the stack as in the example below, T will simply be replaced with the int type.
>

Stack<int>stack = new Stack<int>(); stack.Push(3); int x = stack.Pop(); We now not only eliminate the problem with having to type cast back and forth, we also get a much greater storage ecency as we now use an int array 9

instead of an object array. Generics will also promote strong typing in the sence that trying to push an int onto a stack with Customers objects will result in a complite time error. Also there is no limitation to how meny type parameters that can be used. We only used T in the example above but some situations might call for the use of two or more parameters.

0.5.2

Partial Types

Many people agree on the fact that splitting up related code into two dierent sections or even place related code in two dierent les is an outrage and ultimately just a step in the wrong direction. Partial Types however was created and added into C# 2.0 for just this reason. It was mostly because of the extensive use of source code generators among programmers and designers. Partial types allow the user to separate the generated and user written code and thus simplifying and speeding up development. The designers are now free to generate new code without overriding the code added by programmers. Partial types can also be used to restructure very big classes into smaller sections for a better overview and possibly easier maintenance. The code below shows a simple example of what a partial type could look like partial class Door { private bool isOpen; public Door() {...}; } partial class Door { public void Open(int Key); {...} public void Lock(int key) {...}; } When compiled these partial classes will be merged in to one class.

0.5.3

Nullable Types

Using and handling none-value types and null values in a language has always been a bit troublesome. Dierent solutions exist but they all have aws. For example you can use predened null types but this would only work for existing types. Another solution is to use a certain value to represent null such as -1, but this only works as long as you dont use that special value. C-Sharps nullable types however solve this problem by using a ? operator. Heres an example of how the ? type modier works. int? x = null; int? y = 10; if(x.HasValue == true) return x; if(y.HasValue == true) return y;

10

First of all two nullable types x and y are declared using the ? modier. A nullable type consist of a boolean type plus the underlying value type,int in this example. The boolean value of x is then called by using the HasValue command which returns true if its a none-null type and false if its a null type. Since x is set to null HasValue will return false and thus x wont be returned. The opposite happens for y, HasValue will return true since y has a value of 10 and thus y will be returned. Nullable types can be implicitly converted from nullable type to a none-null type and the other way around. They can also be used with other operators just as normal types. The following examples will show how nullable, nonenullable types and operators work together. int? x = null; int? y = null; if(x == y) int? z = x + 1; In this second example the nullable type x has a value of 0 and y is still null. Then we try to compare the types using > and will end up getting a false expression. The same would happen if we were to use two null values. On the other hand z will now consist of x + 1 and not null since x now has a value of 0. Appart from the ? type modifer a new null coalescing operator ?? exist. The expression x ?? y should be interpeted as follows: if(x != null) return x; else return y; This operator can be specially convinient when dealing with nullable and non-nullable types as this nal example will show: int? int? int? int i x = GetNullableInt(); y = GetNullableInt(); z = x ?? y; = z ?? -1;

You can see the result of x ?? y is a nullable value z, but the result of z ?? -1 is a non-nullable value. In other words the ?? operator in the latter case will remove the ? from the type and at the same time supplies a default value to use in case z is null.

0.5.4

Anonymous Methods

An anonymous method is anonymous in the sence that it lacks a name, so the question arises how do you run such a method if you cant call it by its

11

name. The answer is though a delegate. An anonymous method might use a delegate to execute but seem to act far from it. Instead of delegating the work elsewhere the code gets exectuted in line with the rest of the code or so it might seem. In truth C# converts the code within the anonymous method into a unique method within a unique class. The delegate is then set to reference the new uniqly named method. In order for this to work you need to fulll the requirements of delegates ofcourse. Heres an example of a click event done using delegates but without an anonymous method: class A { ListBox listBox; TextBox textBox; Button button; public A() { listBox = new ListBox(...); textBox = new TextBox(...); button = new Button(...); button.Click += new EventHandler(AddClick); } void addClick(System.Object o, System.EventArgs e) { listBox.Items.Add(textBox.Text); } } Click in this case is a delegate which we reference to the method addClick, where we then add the current text in a textbox to a listbox. Now to do this without using the addClick method we can use an anonymous method. The example below will show how this is done. class A { ListBox listBox; TextBox textBox; Button button; public A() { listBox = new ListBox(...); textBox = new TextBox(...); button = new Button(...); button.Click += delegate(sender, e) { listBox.Items.Add(textBox.Text); }; } 12

What we do here is simply add brackets after the keyword delegate and then input the code we want to execute. As you can see we can still use and manipulate the variables just like if it was a normal method. Note that the parameters sender and e of type Object and EventArgs has to be passed even with an anonymous method this is because the click delegate is constructed of those two parameters and thus according to C# delegation rules we have to pass them along when we use the click delegate.

0.5.5

Iterator

A new form of iterator was introduced along with C# 2.0. The new iterator supports the foreach statement and also uses a new yield keyword. For the iterator to work properly you will need to implement a GetEnumerator method from System.Collections.IEnumerator. This is also where the new keyword yield comes to use. Lets look at an example before we start and discussion how the yield statement is used. publicclass List : System.Collections.IEnumerable { Int items[] = new int[5] 1, 2, 3, 4, 5 ; publicSystem.Collections.IEnumerator GetEnumerator() { for (int i = 0; i <items.Length; i++) { yield return items[i]; } } }

class MainClass { static void Main() { List list = new List(); foreach (int i in list) { System.Console.Write(i + ); } } } Here we rst create a list class which contains ve items, but the interesting part is of course the GetEnumerator method where we basically decide how to iterate our items. By implementing this method we can now decide how the 13

items should be iterated using yield return and yield break. In this example we simply choose to go through all the items in the list and return them one after another using yield return. We then use the foreach statement in the main class to output all the data on the screen. By implementing the GetEnumerator method accordingly, using the keyword yield and including the IEnumerable interface we get a fully Enumerable type. This means that we dont have to implement the MoveNext, Current or Reset methods since C# already has provided us with them and on top of that we now have the foreach statement at our disposal. The yield statement can be used more than once in the iterator as this second example of GetEnumerator will show. public System.Collections.IEnumerator GetEnumerator() { yield return Hello; yield return there; yield return !; } Using foreach or calling MoveNext manually will result in the rst item in the iterator body, Hello in this case. The next iteration will resume after the previous yield statement and then go on until it reaches a yield break statement or the end of the body is encountered. Something to remember when using the new iterator is that calling the Reset method will result in Current returning a none-dened value. To avoid an exception being raised we therefore have to call MoveNext before the rst value in a list can be used.

0.6

Design patterns in .NET

In this chapter we will talk about which design patterns are used in the BCL (Base Class Library) of .NET. Within the .NET framework the use of design patterns are so prevalent that it has become a built in part of the programming languages themselves instead of just being represented by classes. To describe this more briey we will give you two examples of how to use the observer- and iterator pattern which is supported by C# and other languages.

0.6.1

Observer pattern

A common scenario when classes have to interact with each other occurs when a class (observer) needs to know if something has changed in another class (subject). One way to x this problem is to let the observers implement an interface with a method called notify.

14

public interface observer { Void notify(); } All the subjects need a list of interested observers to call notify on when its state changes. Therefore the subject must derive from a subjectbaseclass to handle that task. public abstract class subjectbaseclass { private Arraylist observers = new ArrayList(); public void add(observer o) { observers.add(o); } public void remove(observer o) { observers.remove(o); } public void notify() { foreach(observer o in observers) { o.notify(); } } } This solution works but it is quite extensive to let all subjects in the system inherit from the subjectbaseclass and all observers implement the observer interface. Therefore a much simpler solution is provided in C# using delegates and events. In Windows, events arise when a button is pressed. If an observer needs to be notied about that event it is possible to couple one of its methods with a delegate. Principally this is the same as pointers to functions but on a higher abstract level.

15

public delegate void EventHandler(); publicclass Subject { public EventHandler Event; publicvoid raiseEvent() { EventHandler ev = Event; If(ev != null) ev(); } } public class observer1 { public observer1(Subject s) { s.Event += new EventHandler(handleEvent); } public void handleEvent() { Console.Writeline(An event has occurred!); } } With this solution, whenever an event occurs the observer will be aware of it through the method applied to the delegate (handleEvent).

0.6.2

Iterator pattern

When working with collections of objects, you often need to traverse and visit each element in dierent orders. Therefore C# uses the interface IEnumerable and the class IEnumerator. Int[] values = new int[] 1,9,4,3,2; IEnumerator e = ((IEnumerable)values).GetEnumerator(); while(e.MoveNext()) { Console.Write(e.Current.ToString() + ); } This iterator pattern lets you easily traverse a collection without expose the inner workings of the collection. It uses the interface IEnumerable which has one

16

single method GetEnumerator() that returns an object which implements IEnumerator. The IEnumerator class contains the code necessary to iterate through the collection. All of the collection classes in the System.Collections namespace, as well as arrays implement IEnumerable and can therefore be iterated all over. Other patterns used widely in .NET is Adapter-, Factory-, and Strategy pattern.

0.7

C# future

In October 2005 Microsoft proposed some new features that were meant to extend the current 2.0 version. The 3.0 version, also called C# Orcas is now under production and is suppose to contain those features. When reviewing the new promised features it seems as C# is taking a step towards functional style programming with the support for higher order functional style class libraries. The extended features are: Extension methods Lambda expressions Implicity typed local variables Object initializers Extended anonymous types Query expressions Expression trees This section will cover the rst ve.

0.7.1

Extension methods

You can say that extension methods are extended static methods. A static method is something you can call only from a class and not from an object. Example: Class X { static int val static void getVal() {return val;} } The method getVal here is only possible to invoke like this: int val2 = X.getVal(); Extension methods have the ability to be invoked from objects but they have to use the keyword this. Extension methods appear like additional methods on the types that are given by their rst parameter. Extension methods can only be created in static classes. Example: 17

Public static class Extensions { static void addOne(this int val) {return val++;} } int val2 = 1; int val3 = val2.addOne();

0.7.2

Lambda Expressions

As we mentioned in the beginning of the chapter, Microsoft wants to strive for more functional style syntax in their new upcoming version. Functional languages like Haskell, Lisp and ML are based on lambda calculus which can describe functions and their evolution. With lambda expression it will be easier to create and describe anonymous methods. A lambda expression has the form: (param) =>expr param is the parameter to the function and the expr is the body of the function. The parameters of a lambda expression can also be explicitly or implicitly typed. In an explicitly parameter list the types are explicitly stated which is not the case for implicitly once. Example: Implicit type x = >x + 1 Explicit type (int x) =>x + 1 The main reason why lambda expression is a better way to create anonymous methods is because you dont have to create large code blocks for the function body. To illustrate this we will show two examples rst with delegates and then with lambda expression. Lets say we have a list of ints and we want to create a new list with only even numbers. Example with anonymous methods using delegates: List<int>list = new List<int>(); list.Add(1); list.Add(2); list.Add(3); list.Add(4);

List<int>evenNumbers = list.FindAll(delegate(int i) {return (i % 2) == 0;}); In this case we have to create a codeblock for the delegate and the types must be explicitly stated, in our case, i have to be declared as an integer. Lambda expression can save us some work. Example with anonymous methods using 18

lambda expression: List<int>list = new List<int>(); list.Add(1); list.Add(2); list.Add(3); list.Add(4); List<int>evenNumbers = list.FindAll(i =>(i % 2) == 0); As you can see lambda expression is more eective. We dont have to capsulate syntax in a code block and the expression can be implicitly typed (we dont have to declare i as an integer).

0.7.3

Implicit variables

Implicitly typed local variables are declared using the type var. The type witch will actually be used is determined by the expression where the variable was declared. For this to work implicitly declared variables must always be initialized upon declaration. The following examples will show what an implicitly typed variable might look like and what it corresponds to: var var var var var i = 1; s = Hello; d = 1.0; numbers = new int[] 1, 2, 3; strings = new[] Hello, there, !;

These implicitly typed variables will be converted into the following when used: int i = 5; string s = Hello; double d = 1.0; int[] numbers = new int[] 1, 2, 3; An implicitly typed variable or a set of variables must always be initialized by an expression and not by an object or anything else. Thats why we have to use the new expression when declaring a set of variables. The last case where an array containing strings is declared but without any string type is actually a special case where both object initlializers and an anonymous type is used. These features and what they corresponds to will be explained more in detail later on. A unique occasion arises when there is a user dened type named var in existance, C# will then always use the user dened type over the actual implicit type. A nal example will show the wrong use of implicitly typed variables: var x; var y = 1, 2, 3; var z = null; In the rst case the declared variable x isnt initialized and therefore and error will occur. The Vaiable y however is initlialized but not with an expression and also generates an error. Last we try to set z to a null value and this will also result in an error since C# cant derive null to a spesic type. 19

0.7.4

Object initializers

This new feature oers an alternative to the use of a constructor. With object initializers you can initialize member variables or elements in a collection where the object is being created. A rst example will show how an initializer is used:

class Customer { public string name; string address; public string Adress { get { return adress; } set { adress = value; } } } This syntax example denes a initializer(Adress) for the class Customer. We can later use it when creating an object by adding bracketts and dening values like the example below. var customer = new Customer{name = John Doe, address = 123 23rd Street}; As you might notice name is also set at creation, this can be done becouse we have public access to it, this is not the case for the variable adress which justifys the use of an initializer. An object initializer can only consist of one member variable and thus another object initializer would be needed if name didnt have public access. Its important to remember that object initializers doesnt replace a constructor. A constructor can still be used as normal, if fact you can even use a constructor and object initializers at the same time. This is simply done by calling the constructor with its parameters and then adding bracketts after, like the prevoius example.

0.7.5

Anonymous Types

The use of object initializers enables a user to create and use anonymous types. An anonymous type is basiclly a temporary class without any name or methods. An example will show how to create an anonymous type: var a1 = new Name = John Doe, Age = 32 ; var a2 = new Name = Jane Doe, Age = 24 ; if(a1 == a2) a1 = a2; Here we actually dont have two anonymous types but two instances of the same anonymous type. If an anonymous type has the same number of parameters, with matching names and in the same order, an instance of the same type is created. This enables you to compare anoymous types or assagin new vaules to instances of the same type as shown in the example. Anonymous types can be confusing to read and hard to keep track of at times and they are therefore best put to use in temporary situations.

20

Bibliography
[1] Anders Forsberg. Programmering i C#. studentlitteratur, 2005 [2] codersource. http://www.codersource.net [3] wikipedia. http://en.wikipedia.org [4] microsoft. http://msdn.microsoft.com [5] microsoft. http://msdn2.microsoft.com [6] microsoft. http://blogs.msdn.com [7] gotdotnet. http://www.gotdotnet.com [8] genamics. http://genamics.com [9] csharp-station. http://www.csharp-station.com

21