Overload resolution should be smarter about unambiguous Methods.

[TonyValenti]

1. I understand that generic constraints are encoded as metadata and because of that, you can't have two methods in the same class that vary only by constraints.
2. I believe you should be able to have two extension methods in different classes that vary only by constraints and C# should be able to resolve the differences. The following code doesn't work but I believe it should:

public static class Extensions1 {
  public static T DoSomething(this T This) where T:class {
  ....
  }
}

public static class Extensions2 {
  public static T DoSomething(this T This) where T:Person {
  ....
  }
}

public static void Test(){
  var P = new Person()
   .DoSomething() //This causes the issue.
   ;
}

[HaloFour]

Should this not have been resolved by "bestest betterness" or did that not apply to extension methods?

[HaloFour]

Actually, "bestest betterness" only removes candidates that don't meet the constraints. It doesn't attempt to determine a "best fit" set of constraints.

[Joe4evr]

Actually, "bestest betterness" only removes candidates that don't meet the constraints. It doesn't attempt to determine a "best fit" set of constraints.

Correct. The new rules do allow overloading by generic constraint now, provided that the constraints don't overlap. Ergo, between where T : class and where T : struct. But since where T : Person implies where T : class, the constraints overlap and thus your call is not unambiguous.