R – the best way to use assembly versioning attributes

assembliesnetversioning

The AssemblyVersion and AssemblyFileVersion attributes are the built-in way of handling version numbers for .NET assemblies. While the framework provides the ability to have the least significant parts of a version number (build and revision, in Microsoft terms) automatically determined, I find the method for this pretty weak, and no doubt have many others.

So I'd like to ask, what ways have been determined to do the best job of having version numbers that better reflect the actual version of a project? Do you have a pre-build script that sets part of the version to the date and time, or repository version for your working copy of a project? Do you just use the automatic generation provided by the framework? Or something else? What's the best way to manage assembly/file versioning?

Best Answer

On my current project, we use the Subversion revision number as the least significant (build) part of the version number, and we use a Nant script to create the project AssemblyInfo file. We use the same version number for both the AssemblyVersion and AssemblyFileVersion attributes. (The other three parts are major.minor.point, where major.minor will be incremented every time there's a database schema change, and point is incremented for each release.)

We started out with the build number being simply incremented, but that required that the version file be checked in for every build, and caused conflicts when merging. When that proved unworkable, we started using CruiseControl.NET to generate the build number, but that made it difficult to reproduce specific builds manually. Eventually we went to the current (Subversion-revision) scheme.

Note: Unfortunately with .NET, it is not possible to perfectly recreate a build from a past revision, because the .NET compilers encode the current timestamp into the object file when compiling. Every time you compile the same code, you get a different object file.

This is the style that Microsoft tends to use in their examples.

It appears that the guidance in this area may have changed, as StyleCop now enforces the use of the C# specific aliases.

C# – the difference between const and readonly in C#

Apart from the apparent difference of

having to declare the value at the time of a definition for a const VS readonly values can be computed dynamically but need to be assigned before the constructor exits.. after that it is frozen.
const's are implicitly static. You use a ClassName.ConstantName notation to access them.

There is a subtle difference. Consider a class defined in AssemblyA.

public class Const_V_Readonly
{
  public const int I_CONST_VALUE = 2;
  public readonly int I_RO_VALUE;
  public Const_V_Readonly()
  {
     I_RO_VALUE = 3;
  }
}

AssemblyB references AssemblyA and uses these values in code. When this is compiled:

in the case of the const value, it is like a find-replace. The value 2 is 'baked into' the AssemblyB's IL. This means that if tomorrow I update I_CONST_VALUE to 20, AssemblyB would still have 2 till I recompile it.
in the case of the readonly value, it is like a ref to a memory location. The value is not baked into AssemblyB's IL. This means that if the memory location is updated, AssemblyB gets the new value without recompilation. So if I_RO_VALUE is updated to 30, you only need to build AssemblyA and all clients do not need to be recompiled.

So if you are confident that the value of the constant won't change, use a const.

public const int CM_IN_A_METER = 100;

But if you have a constant that may change (e.g. w.r.t. precision).. or when in doubt, use a readonly.

public readonly float PI = 3.14;

Update: Aku needs to get a mention because he pointed this out first. Also I need to plug where I learned this: Effective C# - Bill Wagner

Best Answer

Related Solutions

C# – the difference between String and string in C#

This is the style that Microsoft tends to use in their examples.

C# – the difference between const and readonly in C#

Related Topic