Why does the tho syntaxes for the same not generate the same code?

[tndata]

I would expect the two methods below to generate the same code:

public class Product
{
    public int Id;
    public string Name;

    public void MethodA(out int id,
                        out string name) =>
     
      (id, name) = (Id, Name);

    public void MethodB(out int id,
                        out string name)
    {
        id = Id;
        name = Name;
    }
}

But in Sharplab the ValueTuple assignment in MethodA becomes

    public void MethodA(out int id, out string name)
    {
        int id2 = Id;
        string name2 = Name;
        id = id2;
        name = name2;
    }

    public void MethodB(out int id, out string name)
    {
        id = Id;
        name = Name;
    }

Bug or by design? The ValueTuple version seems a bit more inefficient?

[BhaaLseN]

I wouldn't be surprised if this is optimized out anyways.

But the reason is probably that this also allows you to swap values:

string left = "two";
string right = "one";
(left, right) = (right, left);
Console.WriteLine("{0} <-> {1}", left, right);

[canton7]

I wouldn't be surprised if this is optimized out anyways.

Nope

[mburbea]

The optimization seems to apply if you are setting a readonly field or a getter only property in a constructor.
See sharplab

[ufcpp]

Evaluation order is not the same:

using System;

class X
{
    public int A { get { Console.WriteLine("X.A"); return 0; } }
    public int B { get { Console.WriteLine("X.B"); return 0; } }
}

class Y
{
#if tuple
    // Y.A → Y.B → X.A → X.B
    public Y(X other) => (A, B) = (other.A, other.B);
#else
    // Y.A → X.A → Y.B → X.B
    public Y(X other)
    {
        A = other.A;
        B = other.B;
    }
#endif

    private int _a;
    private int _b;
    public ref int A { get { Console.WriteLine("Y.A"); return ref _a; } }
    public ref int B { get { Console.WriteLine("Y.B"); return ref _b; } }
}

[canton7]

@ufcpp While that's true, it's not the same code that the OP posted, so I'm not sure that it's relevant?

[tndata]

Even the x86 assembly and IL code in Sharplab for the original code is different. So I was just curious why there's a difference in Sharplab.

[canton7]

So I was just curious why there's a difference in Sharplab.

It seems that when fields are accessed as part of a tuple, those fields are always copied to locals. It seems there are some cases that nobody's bothered to optimize - your example is one, this is another:

using System;
public class Product
{
    public int Id;
    public string Name;

    public void MethodA() 
    {
        var (a, b) = (Id, Name);
        Console.WriteLine($"{a} {b}");
    }
}

Compiles to two sets of locals:

public void MethodA()
{
    int num = Id;
    string text = nName;
    int num2 = num;
    string arg = text;
    Console.WriteLine(string.Format("{0} {1}", num2, arg));
}

I think the relevant code is around here.