Why does NUnit's Assert.That(..) have ref overloads?

Question

Maybe I'm just being dumb but when and why would you use:

NUnit.Framework.Assert.That<T>(ref T, NUnit.Framework.Constraints.IResolveConstraint, string, params object[])
NUnit.Framework.Assert.That<T>(ref T, NUnit.Framework.Constraints.IResolveConstraint, string)
NUnit.Framework.Assert.That<T>(ref T, NUnit.Framework.Constraints.IResolveConstraint)

in place of:

NUnit.Framework.Assert.That(object, NUnit.Framework.Constraints.IResolveConstraint, string, params object[])
NUnit.Framework.Assert.That(object, NUnit.Framework.Constraints.IResolveConstraint, string)
NUnit.Framework.Assert.That(object, NUnit.Framework.Constraints.IResolveConstraint)

What advantage does passing by ref bring to this these methods?

Answer 1

Digging into NUnit source code, I found this:

static public void That<T>(ref T actual, IResolveConstraint expression, string message, params object[] args)
{
    Constraint constraint = expression.Resolve();

    Assert.IncrementAssertCount();
    if (!constraint.Matches(ref actual))
    {
        MessageWriter writer = new TextMessageWriter(message, args);
        constraint.WriteMessageTo(writer);
        throw new AssertionException(writer.ToString());
    }
}

    public virtual bool Matches<T>(ref T actual)
    {
        return Matches(actual);
    }

versus:

    static public void That(object actual, IResolveConstraint expression, string message, params object[] args)
    {
        Constraint constraint = expression.Resolve();

        Assert.IncrementAssertCount();
        if (!constraint.Matches(actual))
        {
            MessageWriter writer = new TextMessageWriter(message, args);
            constraint.WriteMessageTo(writer);
            throw new AssertionException(writer.ToString());
        }
    }

As you can see, there is no difference in the implementation. The Ref T actual overload allows you to pass value types as reference as well, while reference types are already passed as reference.