Question

Code to illustrate :

    public struct MyStruct
    {
        public int SomeNumber;
    }

    public string DoSomethingWithMyStruct(MyStruct s)
    {
        if (s == null)
            return "this can't happen";
        else
            return "ok";
    }

    private string DoSomethingWithDateTime(DateTime s)
    {
        if (s == null)
            return "this can't happen";  // XX
        else
            return "ok";
    }

Now, “DoSomethingWithStruct” fails to compile with : “Operator ‘==’ cannot be applied to operands of type ‘MyStruct’ and ‘<null>‘”. This makes sense, since it doesn’t make sense to try a reference comparison with a struct, which is a value type.

OTOH, “DoSomethingWithDateTime” compiles, but with compiler warning : “Unreachable code detected” at line marked “XX”. Now, I’m assuming that there is no compiler error here, because the DateTime struct overloads the “==” operator. But how does the compiler know that the code is unreachable ? e.g. Does it look inside the code which overloads the “==” operator ? (This is using Visual Studio 2005 in case that makes a difference).

Note : I’m more curious than anything about the above. I don’t usually try to use “==” to compare structs and nulls.

EDIT : I’ll try to simplify my question – why does “DoSomethingWithDateTime” compile, when “DoSomethingWithMyStruct” does not. Both arguments are structs.

Answer 1

It knows that a struct is never null (Nullable<T> aside); that is enough to emit the warning.

There is a known compiler issue in this area, that arose between the C# 2.0 compiler and the C# 3.0 compiler (and remains in the C# 4.0 compiler at the moment) [I’m not sure why you are seeing it on VS2005, though]. The equality test does not raise an unreachable-code warning for custom structs with == / != operators. DateTime has these operators; your struct doesn’t – hence the difference.

This issue is logged on connect, and has been looked at fairly recently by the compiler team (who are keen to fix it when opportunity arises).