Can anyone point me to C# code for an interpreted stack machine?

Question

Hi! I'm looking for a C# implementation of a stack machine, preferably one with accompaning unit tests or at least couple of examples. The code at http://en.wikipedia.org/wiki/P-code_machine appears to be the sort of thing that I'm looking for. Unfortunately, it's been more than a decade since I programmed in Pascal and ran into a bunch of problems porting it to C#. Also, there was no example(s) of using the code.

Anyway, any help in this regard would be greatly appreciated....

Answer 1

Interpreted stack machines are very similar in concept to Reverse Polish notation.

The expression

3 + (6 - 2)

is expressed in RPN as

3 6 2 - +

This gets evaluated as follows:

Input   Operation    Stack      Comment

3       Push value   3

6       Push value   6
                     3

2       Push value   2
                     6
                     3

-       Subtract     4          Pop two values (6, 2) and push result (4)
                     3

+       Add          7          Pop two values (3, 4) and push result (7)

From there, it should be easy to build a simple interpreted stack machine in C#. E.g.

var stack = new Stack<int>();

var program = new[] 
{ 
    OpCode.Ldc_3,
    OpCode.Ldc_6,
    OpCode.Ldc_2,
    OpCode.Sub,
    OpCode.Add,
};

for (int i = 0; i < program.Length; i++)
{
    int a, b;
    switch (program[i])
    {
    case OpCode.Add: b = stack.Pop(); a = stack.Pop(); stack.Push(a + b); break;
    case OpCode.Sub: b = stack.Pop(); a = stack.Pop(); stack.Push(a - b); break;
    case OpCode.Mul: b = stack.Pop(); a = stack.Pop(); stack.Push(a * b); break;
    case OpCode.Div: b = stack.Pop(); a = stack.Pop(); stack.Push(a / b); break;
    case OpCode.Ldc_0: stack.Push(0); break;
    case OpCode.Ldc_1: stack.Push(1); break;
    case OpCode.Ldc_2: stack.Push(2); break;
    case OpCode.Ldc_3: stack.Push(3); break;
    case OpCode.Ldc_4: stack.Push(4); break;
    case OpCode.Ldc_5: stack.Push(5); break;
    case OpCode.Ldc_6: stack.Push(6); break;
    case OpCode.Ldc_7: stack.Push(7); break;
    case OpCode.Ldc_8: stack.Push(8); break;
    }
}

var result = stack.Pop();

with

enum OpCode
{
    Nop,    // No operation is performed.
    Add,    // Adds two values and pushes the result onto the evaluation stack.
    Sub,    // Subtracts one value from another and pushes the result onto the
            // evaluation stack.
    Mul,    // Multiplies two values and pushes the result on the evaluation
            // stack.
    Div,    // Divides two values and pushes the result onto the evaluation
            // stack.
    Ldc_0,  // Pushes the integer value of 0 onto the evaluation stack.
    Ldc_1,  // Pushes the integer value of 1 onto the evaluation stack.
    Ldc_2,  // Pushes the integer value of 2 onto the evaluation stack.
    Ldc_3,  // Pushes the integer value of 3 onto the evaluation stack.
    Ldc_4,  // Pushes the integer value of 4 onto the evaluation stack.
    Ldc_5,  // Pushes the integer value of 5 onto the evaluation stack.
    Ldc_6,  // Pushes the integer value of 6 onto the evaluation stack.
    Ldc_7,  // Pushes the integer value of 7 onto the evaluation stack.
    Ldc_8,  // Pushes the integer value of 8 onto the evaluation stack.
}

For a real world example, have a look at the fields of the OpCodes class in the .NET Framework.

Answer 2

The original Magpie interpreter was written in C# and compiled down to stack-based bytecode. Take a look at Machine.cs for the core intepreter class. The compiler that translates source code to this bytecode is in Magpie.Compiler.