泛型函数和带有Type参数的函数之间的区别

Question

I just wondered what is the difference between using a function that takes a parameter of type Type (and some more parameter) and a generic function that just wants the type in angle bracket ?

Idea behinde: I want to write a function that saves a type on ma filesystem as XML file. Since the XmlSerializer needs to have the type of object to serialize, I want to know: What is better?

private bool SerializeIt(object o, Type t, string filePath)
{
    bool result = false;
    try
    {
        if (File.Exists(filePath))
            File.Delete(filePath);

        XmlSerializer serializer = new XmlSerializer(t);
        using (FileStream fs = new FileStream(filePath, FileMode.CreateNew, FileAccess.Write))
        {
            serializer.Serialize(fs, o);
        }
        result = true;
    }
    catch (Exception ex)
    {
        result = false;
        Debug.WriteLine(ex);
    }

    return result;
}

private bool SerializeIt<T>(T o, string filePath)
{
    bool result = false;
    try
    {
        if (File.Exists(filePath))
            File.Delete(filePath);

        XmlSerializer serializer = new XmlSerializer(o.GetType());
        using (FileStream fs = new FileStream(filePath, FileMode.CreateNew, FileAccess.Write))
        {
            serializer.Serialize(fs, o);
        }
    }
    catch (Exception ex)
    {
        result = false;
        Debug.WriteLine(ex);
    }

    return result;
}

Answer 1

Your first method will only work with the type SomeData1 , since that is the type you specify. The other method works with any type passed through T .

You could solve the above issue with o.GetType() instead of typeof(SomeData1) . Still, the second method gives you the opportunity to use a base class as serialization base, which will clear out all properties in the deriving class.

Answer 2

Function private bool SerializeIt(object o, Type t, string filePath) is evaluated at runtime. Where function private bool SerializeIt<T>(T o, string filePath is evaluated at compile time. That is inserting the specified type in IL code.

Answer 3

There is a lot of differencies. Firstly you can create constraints for generic parameters, eg:

private bool SerializeIt<T>(T o, string filePath) where T: ISomething

Secondly - free type inference for generic types.

private bool SerializeIt<T>(T o, string filePath)

Something something = new Something();
o.SerializeIt(something, ""); // dont need to pass type (can infer from first argument).

Thirdly - you work on concrete strong type, eg:

private bool SerializeIt<T>(T o, string filePath)

Generally it makes interface simpler and cleaner. You dont have to use loosely typed objects.