What is the difference between Convert.ToInt32 and (int)?

Question

The following code throws an compile-time error like

Cannot convert type 'string' to 'int'

string name = Session["name1"].ToString(); int i = (int)name; 

whereas the code below compiles and executes successfully:

string name = Session["name1"].ToString(); int i = Convert.ToInt32(name); 

I would like to know:

1. Why does the the first code generate a compile-time error?

2. What's the difference between the 2 code snippets?

Answer 1

(int)foo is simply a cast to the Int32 (int in C#) type. This is built into the CLR and requires that foo be a numeric variable (e.g. float, long, etc.) In this sense, it is very similar to a cast in C.

Convert.ToInt32 is designed to be a general conversion function. It does a good deal more than casting; namely, it can convert from any primitive type to a int (most notably, parsing a string). You can see the full list of overloads for this method here on MSDN.

And as Stefan Steiger mentions in a comment:

Also, note that on a numerical level, (int) foo truncates foo (ifoo = Math.Floor(foo)), while Convert.ToInt32(foo) uses half to even rounding (rounds x.5 to the nearest EVEN integer, meaning ifoo = Math.Round(foo)). The result is thus not just implementation-wise, but also numerically not the same.

Answer 2

(this line relates to a question that was merged) You should never use (int)someString - that will never work (and the compiler won't let you).

However, int int.Parse(string) and bool int.TryParse(string, out int) (and their various overloads) are fair game.

Personally, I mainly only use Convert when I'm dealing with reflection, so for me the choice is Parse and TryParse. The first is when I expect the value to be a valid integer, and want it to throw an exception otherwise. The second is when I want to check if it is a valid integer - I can then decide what to do when it is/isn't.