Private members reflection breaks encapsulation principle and thus exposing your code to the following :
There are so cases, when you depend on a third party or you need some api not exposed, you have to do some reflection. Some also use it to test some classes they own but that they don't want to change the interface to give access to the inner members just for tests.
To mitigate the easy to break issue, the best is to detect any potential break by testing in unit tests that would run in a continuous integration build or such. Of course, it means you always use the same assembly (which contains the private members). If you use a dynamic load and reflection, you like play with fire, but you can always catch the Exception that the call may produce.
In the recent versions of .Net Framework, CreateDelegate beat by a factor 50 the MethodInfo invoke:
// The following should be done once since this does some reflection
var method = this.GetType().GetMethod("Draw_" + itemType,
BindingFlags.NonPublic | BindingFlags.Instance);
// Here we create a Func that targets the instance of type which has the
// Draw_ItemType method
var draw = (Func<TInput, Output[]>)_method.CreateDelegate(
typeof(Func<TInput, TOutput[]>), this);
draw
calls will be around 50x faster than MethodInfo.Invoke
use draw
as a standard Func
like that:
var res = draw(methodParams);
Check this post of mine to see benchmark on different method invocations