Is a generic method that works on int and uint not possible in dot net?

Question

In C# is a generic method that works on int and uint not possible because the generic type must be a class and not a primitive or is this only on constraints?

public decimal shift<T>(T input, byte b) where T : int, uint
{
   ///....
}

Is there anyway to make this work without the constraints?

public decimal shift<T>(T input, byte b)  
{
   ///....
}

When I do the latter, I get further but still error out on a bit shift operation on the int or uint. It's as though, it doesn't know what type it is at run time.

http://stackoverflow.com/questions/32664/c-sharp-generic-constraint-for-only-integers — Preston Guillot, Jul 02 '13 at 04:18
Generally the way this is done is with method overloading. In one of my projects I have a class that manipulates bits on native types, I just make 8 separate overloads and it works great. — Robert Rouhani, Jul 02 '13 at 04:33
Note that in particular case of `int, uint` constraint taking `long` as argument may be an acceptable approach. — Alexei Levenkov, Jul 02 '13 at 04:37
That's not possible, http://stackoverflow.com/questions/10833918/generic-method-multipleor-type-constraint — Alireza Maddah, Jul 02 '13 at 04:46

Robert Rouhani · Answer 1 · 2013-07-02T04:53:46.020

In C#, writing special cases for native types is typically done by method overloading. There are several places in the BCL where you can see this kind of design, one example being the BitConverter class.

If you want a generic version with a few special cases, you can add a generic overload and C#'s type inference will make it transparent to the end user.

public decimal shift(int input, byte b)
{
    //....
}

public decimal shift(uint input, byte b)
{
    //....
}

public decimal shift<T>(T input, byte b)
{
    //....
}

Usage:

shift(5, 1); //resolves to the 'int'overload
shift(5u, 1); //resolves to the 'uint' overload
shift(new Point(2, 2), 1) //resolves to the generic overload with T = Point

I'm guessing you're trying to do bit shifting with a generic method. There are only a handful of types that this makes sense for (byte, sbyte, short, ushort, int, uint, long, ulong), so you might as well just write out the 8 overloads. I have a class in one of my projects that includes a lot of bit hacks, and I just write out the necessary overloads and it works out pretty well.

Piotr Stapp · Answer 2 · 2013-07-02T04:40:05.343

0

You have to use following code to make it working:

    public static decimal shift<T>(T input, byte b) where T : struct
    {
        return 1.0M;
    }

It is because int and uint are structs not classes

edited Jul 02 '13 at 04:40

answered Jul 02 '13 at 04:24

Piotr Stapp

19,392
11
68
116

I tried, it doesn't seem to fool the compiler: – Matt Jul 02 '13 at 04:25
'int' is not a valid constraint. A type used as a constraint must be an interface, a non-sealed class or a type parameter. 'uint' is not a valid constraint. A type used as a constraint must be an interface, a non-sealed class or a type parameter. – Matt Jul 02 '13 at 04:28
It's in LinqPad (don't think it should matter) – Matt Jul 02 '13 at 04:29
I completly changed my answer. No it compiles without problems – Piotr Stapp Jul 02 '13 at 04:40
get an error when attempting to perform a shift (assuming this is actually what Matt wants to do) – Kevin Nacios Jul 02 '13 at 04:45

score 0 · Answer 3 · answered Jul 02 '13 at 04:39

I think in this case, you'll have to check the type it is at runtime, but it would be more clear if you just had to overloaded methods that took an int and uint

public decimal shift<T>(T input, byte b)
{ 
    if (input is int)
        return (Convert.ToInt32(input) << b);
    else if (input is uint)
        return (Convert.ToByte(input) << b);
}

vs

public decimal shift(int input, byte b)
{ 
    return input << b;
}

public decimal shift(uint input, byte b)
{
    return input << b;
}

Is a generic method that works on int and uint not possible in dot net?

3 Answers3