[ Prev ][ Table of Contents ][ Front Page ][ Talkback ][ FAQ ][ Next ]
LINUX GAZETTE
...making Linux just a little more fun!
Programming Bits: Meeting C# and Mono
By Ariel Ortiz Ramirez

C# (pronounced C-sharp) is a new object-oriented programming language designed to take advantage of Microsoft's .NET development framework. It has many similarities with other popular object-oriented languages such as C++ and Java, yet it offers some new goodies.

Linux offers the opportunity to develop C# applications thanks to a project called Mono. Mono is an open source implementation of the .NET platform. In the following sections, I will describe the main elements of the current implementation of the Mono system.

The Mono Project

At this time, Mono implements two standards: the C# programming language (Standard ECMA-334) and the Common Language Infrastructure (Standard ECMA-335). Both of these specifications were developed by Microsoft and submitted to ECMA (a vendor consortium formerly known as the European Computer Manufacturers Association) on October 2000. They were formally approved on December 2001, and they will probably become ISO standards (thanks to a "fast-track" agreement that ISO has with ECMA) some time before the end of next year.

The Mono project is sponsored by Ximian, the same company that brought us the GNOME graphical desktop environment. Mexican hacker and Ximian CTO Miguel de Icaza currently leads the development of this project. In my opinion, the people involved with the development of Mono have done a remarkable job in quite a short amount of time. By the way, the word "Mono" means monkey in Spanish. These guys at Ximian really like monkeys.

Hello Mono World!

Lets follow a simple programming example in order to understand how C# and the different Mono components fit together. To see this in action, make sure you have a working Mono installation (see the resources section for information on downloading and installing Mono). 

The following figure summarizes the process we will follow in order to compile and run our C# program:

Compiling and running C# programs.

First, we will create a simple C# source program (the classical "Hello World!" couldn't be missing). Type the following program using your favorite text editor and save the file as hello.cs:

class Hello {
    public static void Main() {
        System.Console.WriteLine("Hello Mono World!");
    }
}

This program is composed of a class named Hello which contains a method called Main. This method establishes the program entry point, in the same way that the main function is the start of a C/C++ program. In this example, the Main method prints to the standard output the message "Hello Mono World".

We can now compile the program using the Mono C# compiler, called mcs. At the shell prompt type:

mcs hello.cs

We now should have a file called hello.exe in the current directory. But don't get baffled about the .exe file name extension. It is not a Windows executable file, at least not in the way we're used to. Contrary to what happens when we compile a program written in languages like C or C++, the C# compiler does not generate a machine-specific object file (for example a Linux ELF x86 object file), but instead generates a special binary file called an assembly, which is made up of metadata and intermediate language (IL) instructions. These two together represent a platform-agnostic translation of the program source code. This means, of course, that when we actually run the program contained in the assembly, its intermediate language code has to be translated to the native code of the computer where the program is being run. This last translation phase is carried out by a virtual machine, whose behavior is defined by the Common Language Infrastructure (CLI) specification. The CLI defines an object oriented runtime environment that supports a base class library, dynamic class loading and linking, multiple thread execution, just-in-time compilation, and automatic memory management. The Microsoft implementation of the CLI specification is usually referred as the Common Language Runtime (CLR). We say that the CLR is a superset of the CLI because the CLR contains some extensions that are not part of the CLI.

To execute our assembly, we must invoke the program called mono, which is the Mono virtual machine. Type at the shell prompt the following:

mono hello.exe

The output should be:

Hello Mono World!

Behind the Scenes

Lets see how to examine the contents of our assembly. The program monodis (Mono disassembler) reads the binary information of an assembly and produces a textual representation of its contents. Type at the shell prompt:

monodis hello.exe

The disassembler output should be something like the following:

.assembly extern mscorlib
{
  .ver 0:0:0:0
}
.assembly 'hello'
{
  .hash algorithm 0x00008004
  .ver  0:0:0:0
}
  .class private auto ansi beforefieldinit Hello
        extends [mscorlib]System.Object
  {

    // method line 1
    .method public hidebysig  specialname  rtspecialname
           instance default void .ctor()  cil managed
    {
        // Method begins at RVA 0x20ec
        // Code size 7 (0x7)
        .maxstack 8
        IL_0000: ldarg.0
        IL_0001: call instance void valuetype [corlib]System.Object::.ctor()
        IL_0006: ret
    } // end of method instance default void .ctor()

    // method line 2
    .method public static
           default void Main()  cil managed
    {
        // Method begins at RVA 0x20f4
        .entrypoint
        // Code size 11 (0xb)
        .maxstack 8
        IL_0000: ldstr "Hello Mono World!"
        IL_0005: call void class [corlib]System.Console::WriteLine(string)
        IL_000a: ret
    } // end of method default void Main()

  } // end of type Hello

The first part of this output corresponds to the metadata. It contains information about the current version of the assembly, any optional security constraints, locale information, and a list of all externally referenced assemblies that are required for proper execution. The rest of the output represents the IL code. We can spot two methods in this code: the default class constructor called .ctor, provided automatically by the compiler, and our Main method. As described before, when the virtual machine is asked to run this code, it uses a just-in-time (JIT) compiler to translate the IL into the native machine code of the hosting environment. The native code is not generated until it is actually needed (thus the name just-in-time). For our example, the following is the native x86 machine code (in AT&T assembly language syntax) that gets generated for the Main method:

push   %ebp
mov    %esp,%ebp
sub    $0x30,%esp
push   $0x80c9eb0
mov    0x805462c,%eax
push   %eax
cmpl   $0x0,(%eax)
mov    (%eax),%eax
call   *0x94(%eax)
add    $0x8,%esp
mov    0x805462c,%eax
push   %eax
cmpl   $0x0,(%eax)
mov    (%eax),%eax
call   *0xb4(%eax)
add    $0x4,%esp
leave
ret

Mono also comes with an interpreter called mint. If you use this program, the IL instructions are interpreted instead of being compiled to native code by the JIT. Actually, our simple program might be a little bit faster when run under mint because the JIT compiler will take some time to compile the code of our program and store it some where in memory. Of course, subsequent execution of the native code already in memory is definitely faster than interpretation. Currently the Mono JIT compiler is only available for x86 machines. The Mono interpreter must be used in any non-x86 machine. To see the interpreter running, type at the shell prompt:

mint hello.exe

If you're familiar with Java, you might be thinking that all this technology sounds pretty much like the way that the Java platform works. And this is indeed so. The CLI virtual machine is the key factor for platform independence. This means that I can write and compile a program in Linux using Mono, and then run it in a Windows computer with the .NET framework. There is no need to rewrite or recompile the source code. But in contrast to the Java Virtual Machine, which is tightly coupled to the Java programming language, the CLI specification not only allows platform independence, it also allows language independence. Windows has compilers that target the CLR from a number of languages. The most important ones are part of Microsoft's Visual Studio .NET development environment: Visual Basic .NET, JScript .NET, Managed C++ and C#. Other languages supported, from third party vendors, include APL, COBOL, Eiffel, Forth, Fortran, Haskell, Mercury, Mondrian, Oberon, Pascal, Perl, Python, RPG, Scheme and SmallScript. The Mono project only has a C# compiler at this time, but we will probably see in the near future other languages being supported.

Another important element of the CLI is the Common Type System (CTS). The CTS fully describes all the data types supported by the virtual machine, including how these data types interact with each other and how they are represented as metadata inside the assemblies. It is important to note that not all languages available for the CLI support all data types in the CTS. So there is a Common Language Specification (CLS), that defines a subset of common types that ensure that binary assemblies can be used across all languages that target the CLI. This means that if we build a CLI class that only exposes CLS compliant features, any CLI compatible language can use it. You could create a class in Eiffel, subclass it in C# and instantiate it in a Visual Basic.NET program. Now this is real language interoperability.

Some Advantages

Using a CLI compliant platform, such as Mono or the .NET framework, has some important advantages:

C#, as a programming language, has also some important features:

I will discuss these and other C# issues more thoroughly in later articles. 

Resources

http://www.go-mono.com/
The official Mono home page. The download and install instructions can be found here.
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/dndotnet/html/deicazainterview.asp
A very interesting interview with Miguel de Icaza about the Mono project and the use of ECMA standards.
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/cscon/html/vcoriCStartPage.asp
Information on the C# programming language at MSDN.
http://www.ecma.ch/ecma1/STAND/ecma-334.htm
The Standard ECMA-334 C# Language Specification.
http://www.ecma.ch/ecma1/STAND/ecma-335.htm
The Standard ECMA-335 Common Language Infrastructure.

Copyright © 2002, Ariel Ortiz Ramirez. Copying license http://www.linuxgazette.com/copying.html
Published in Issue 84 of Linux Gazette, November 2002

[ Prev ][ Table of Contents ][ Front Page ][ Talkback ][ FAQ ][ Next ]