Neo: Smart contract return type is irregular when return Boolean result

[ProDog]

Describe the bug
Smart contract return type is irregular when return Boolean result, sometimes it's 1, sometimes it's true.
like:

public class Example : SmartContract
{
    public static bool main(int i, int j)
    {
        return i != j && (i > 0 || j < 2);
    }

}

If invoked with i = 0 and j = 1, I get a Boolean result type. If I invoke with i = 1 and j = 1 I get an Integer, the results are correct, but the type is irregular.

Expected behavior
Always return Boolean.

Platform:

• OS: Windows 10 x64
• Version: neo3 master

[ShawnYun]

This should be due to the following reasons：
For example, in IL Code, about the instruction Ceq,

Compare two values. If the two values are equal, the integer value 1 (int32) is pushed onto the calculation stack; Otherwise, 0 (int32) is pushed onto the compute stack.

But in NeoVM，the corresponding opcode is NUMEQUAL,

                case OpCode.NUMEQUAL:
                    {
                        var x2 = Pop().GetInteger();
                        var x1 = Pop().GetInteger();
                        Push(x1 == x2);
                        break;
                    }

it will push a bool value (true or false).

So IL uses 1 and 0 for true and false, but NeoVM doesn't.
It causes the return value type to be irregular.

[shargon]

for the vm, an integer 1 or true it's, more or less, the same, that's why we have the Abi, we sould process the output

[gsmachado]

for the vm, an integer 1 or true it's, more or less, the same, that's why we have the Abi, we sould process the output

That's exactly the problem: the NeoVM returns more or less something, and the ABI might say something else. Inconsistency as it's best. 😃 🤣

I think the compiler should take care of that...

[shargon]

I think the compiler should take care of that...

of course, anything than can be fixed, must be fixed :)

[roman-khimov]

NeoVM returns more or less something, and the ABI might say something else

Well, it can also be viewed as an output companion of #1891 which is about inputs.