Monday, 9 January 2012

Structure of a .NET Application


DLL Hell
    DLLs gave developers the ability to create function libraries and programs that could be shared with more than one application. Windows itself was based on DLLs. While the advantages of shared code modules expanded developer opportunities, it also introduced the problem of updates, revisions, and usage. If one program relied on a specific version of a DLL, and another program upgraded that same DLL, the first
program quite often stopped working.
    Microsoft added to the problem with upgrades of some system DLLs, like comctl.dll, the library used to get file, font, color and printing dialog boxes. If things weren't bad enough with version clashes, if you wanted to uninstall an application, you could easily delete a DLL that was still being used by another program.
    Recognizing the problem, Microsoft incorporated the ability to track usage of DLLs with the Registry starting formally with Windows 95, and allowed only one version of a DLL to run in memory at a time. Adding yet another complication, when a new application was installed that used an existing DLL, it would increment a usage counter. On uninstall, the counter would be decremented and if no application was
using the DLL, it could be deleted.
    That was, in theory. Over the history of Windows, the method of tracking of DLL usage was changed by Microsoft several times, as well as the problem of rogue installations that didn't play by the rules--the result was called "DLL HELL", and the user was the victim.
   Solving DLL hell is one thing that the .NET Framework and the CLR targeted. Under the .NET Framework, you can now have multiple versions of a DLL running concurrently. This allows developers to ship a version that works with their program and not worry about stepping on another program. The way .NET does this is to discontinue using the registry to tie DLLs to applications and by introducing the
concept of an assembly.
    On the .NET Platform, if you want to install an application in the clients place all you have to do is use XCopy which copies all the necessary program files to a directory on the client’s computer. And while uninstalling all you have to do is just delete the directory containing the application and your application is uninstalled.

Metadata
     An Assembly is a logical DLL and consists of one or more scripts, DLLs, or executables, and a manifest (a collection of metadata in XML format describing how assembly elements relate). Metadata stored within the Assembly, is Microsoft's solution to the registry problem. On the .NET Platform programs are compiled into .NET PE (Portable Executable) files. The header section of every .NET PE file contains a special new section for Metadata (This means Metadata for every PE files is contained within the PE file itself thus abolishing the need for any separate registry entries). Metadata is nothing but a description of every namespace, class, method, property etc. contained within the PE file. Through Metadata you can discover all the classes and their members contained within the PE file.
     Metadata describes every type and member defined in your code in a Multilanguage
form. Metadata stores the following information:
• Description of the assembly
  o Identity (name, version, culture, public key).
  o The types that are exported.
  o Other assemblies that this assembly depends on.
  o Security permissions needed to run
• Description of types
  o Name, visibility, base class, and interfaces implemented.
  o Members (methods, fields, properties, events, nested types)
• Attributes
  o Additional descriptive elements that modify types and members
Advantages of Metadata:
Now let us see the advantages of Metadata:
Self describing files:
   CLR modules and assemblies are self-describing. Module's metadata contains everything needed to interact with another module. Metadata automatically provides the functionality of Interface Definition Language (IDL) in COM, allowing you to use one file for both definition and implementation. Runtime modules and assemblies do not even require registration with the operating system. As a result, the descriptions used by the runtime always reflect the actual code in your compiled file, which increases application reliability.
Language Interoperability and easier component-based design:
Metadata provides all the information required about compiled code for you to inherit a class from a PE file written in a different language. You can create an instance of any class written in any managed language (any language that targets the Common Language Runtime) without worrying about explicit marshaling or using custom interoperability code.
Attributes:
The .NET Framework allows you to declare specific kinds of metadata, called attributes, in your compiled file. Attributes can be found throughout the .NET Framework and are used to control in more detail how your program behaves at run time. Additionally, you can emit your own custom metadata into .NET Framework
files through user-defined custom attributes.

Assembly
Assemblies are the building blocks of .NET Framework applications; they form the fundamental unit of deployment, version control, reuse, activation scoping, and security permissions. An assembly is a collection of types and resources that are built to work together and form a logical unit of functionality. An assembly provides the common language runtime with the information it needs to be aware of type implementations. To the runtime, a type does not exist outside the context of an assembly.
An assembly does the following functions:
• It contains the code that the runtime executes.
• It forms a security boundary. An assembly is the unit at which permissions are requested and granted.
• It forms a type boundary. Every type’s identity includes the name of the assembly at which it resides.
• It forms a reference scope boundary. The assembly's manifest contains assembly metadata that is used for resolving types and satisfying resource requests. It specifies the types and resources that are exposed outside the assembly.
• It forms a version boundary. The assembly is the smallest version able unit in the common language runtime; all types and resources in the same assembly are versioned as a unit.
• It forms a deployment unit. When an application starts, only the assemblies the application initially calls must be present. Other assemblies, such as localization resources or assemblies containing utility classes, can be retrieved on demand. This allows applications to be kept simple and thin when first downloaded.
• It is a unit where side-by-side execution is supported.
Contents of an Assembly
• Assembly Manifest
• Assembly Name
• Version Information
• Types
• Locale
• Cryptographic Hash
• Security Permissions
Assembly Manifest
Every assembly, whether static or dynamic, contains a collection of data that describes how the elements in the assembly relate to each other. The assembly manifest contains this assembly metadata. An assembly manifest contains the following details:
• Identity. An assembly's identity consists of three parts: a name, a version number, and an optional culture.
• File list. A manifest includes a list of all files that make up the assembly.
• Referenced assemblies. Dependencies between assemblies are stored in the calling assembly's manifest. The dependency information includes a version number, which is used at run time to ensure that the correct version of the dependency is loaded.
• Exported types and resources. The visibility options available to types and resources include "visible only within my assembly" and "visible to callers outside my assembly."
• Permission requests. The permission requests for an assembly are grouped into three sets: 1) those required for the assembly to run, 2) those that are desired but the assembly will still have some functionality even if they aren't granted, and 3) those that the author never wants the assembly to be granted.
   In general, if you have an application comprising of an assembly named Assem.exe and a module named Mod.dll. Then the assembly manifest stored within the PE Assem.exe will not only contain metadata about the classes, methods etc. contained within the Assem.exe file but it will also contain references to the classes, methods etc, exported in the Mod.dll file. While the module Mod.dll will only contain metadata
describing itself.
The following diagram shows the different ways the manifest can be stored:
Fig::
For an assembly with one associated file, the manifest is incorporated into the PE file to form a single-file assembly. You can create a multifile assembly with a standalone manifest file or with the manifest incorporated into one of the PE files in the assembly.
The Assembly Manifest performs the following functions:
• Enumerates the files that make up the assembly.
• Governs how references to the assembly's types and resources map to the files that contain their declarations and implementations.
• Enumerates other assemblies on which the assembly depends.
• Provides a level of indirection between consumers of the assembly and the assembly's implementation details.
• Renders the assembly self-describing.
For more information on Assemblies refer:
http://msdn.microsoft.com/library/en-us/cpguide/html/cpconassemblies.asp

Modules
   Modules are also PE files (always with the extension .netmodule) which contain Metadata but they do not contain the assembly manifest. And hence in order to use a module, you have to create a PE file with the necessary assembly manifest. In C#, you can create a module using the /t:module compiler switch.
   There are a few ways to incorporate a module into an Assembly. You can either use /addmodule switch to add module/s to your assembly, or you can directly use the /t:exe, /t:winexe and /t:library switches to convert the module into an assembly.
Difference between Module and Assembly
    A module is an .exe or .dll file. An assembly is a set of one or more modules thattogether make up an application. If the application is fully contained in an .exe file, fine—that's a one-module assembly. If the .exe is always deployed with two .dll files and one thinks of all three files as comprising an inseparable unit, then the three modules together form an assembly, but none of them does so by itself. If the product is a class library that exists in a .dll file, then that single .dll file is an assembly. To put it in Microsoft's terms, the assembly is the unit of deployment in .NET.
An assembly is more than just an abstract way to think about sets of modules. When an assembly is deployed, one (and only one) of the modules in the assembly must contain the assembly manifest, which contains information about the assembly as a whole, including the list of modules contained in the assembly, the version of the assembly, its culture, etc.

Microsoft Intermediate Language (MSIL)
   When compiling to managed code, the compiler translates your source code into Microsoft intermediate language (MSIL), which is a CPU-independent set of instructions that can be efficiently converted to native code. MSIL includes instructions for loading, storing, initializing, and calling methods on objects, as well
as instructions for arithmetic and logical operations, control flow, direct memory access, exception handling, and other operations. Before code can be executed, MSIL must be converted to CPU-specific code by a just in time (JIT) compiler. Because the runtime supplies one or more JIT compilers, for each computer architecture it supports, the same set of MSIL can be JIT-compiled and executed on any supported
architecture.
    When a compiler produces MSIL, it also produces metadata. The MSIL and metadata are contained in a portable executable (PE file) that is based on and extends the published Microsoft PE and Common Object File Format (COFF) used historically for executable content. This file format, which accommodates MSIL or native code as well as metadata, enables the operating system to recognize common language runtime images. The presence of metadata in the file along with the MSIL enables your code to describe itself, which means that there is no need for type libraries or Interface Definition Language (IDL). The runtime locates and extracts the metadata from the file as needed during