Roslyn: nameof requires a non-static member in a sub-expression

[alrz]

Version Used: https://github.com/dotnet/roslyn/commit/5fbdd5c567b854e9b5f6f1ea64944373a40efb26

Steps to Reproduce:

class C
{
    public string Instance;
    public static string Static;

    void M()
    {
         _ = nameof(C.Instance);             // OK         
         _ = nameof(C.Instance.Length);      // ERROR
         _ = nameof(C.Static);               // OK  
         _ = nameof(C.Static.Length);        // OK
    }
}

class C<T>
{
    public T Instance;
    public static T Static;

    void M()
    {
         _ = nameof(C<string>.Instance);    // OK
     _ = nameof(C<string>.Instance.Length); // ERROR
         _ = nameof(C<string>.Static);      // OK
         _ = nameof(C<string>.Static.Length);   // OK
    }
}

Expected Behavior: No error.

Actual Behavior:

CS0120: An object reference is required for the non-static field, method, or property 'C.Instance'

[alrz]

I suspect this is not per spec.

[sharwell]

I'm marking as a bug even though I'm not sure whether or not this is the correct behavior (it's either by design or a bug, so I figured this was the right label).

[gafter]

According to https://github.com/dotnet/csharplang/blob/master/spec/expressions.md#nameof-expressions this should be legal. It is therefore a compiler bug.

[gafter]

According to https://github.com/dotnet/vblang/blob/master/meetings/2014/LDM-2014-10-23.md (Yes, C# design notes in the VB repo because that's where the meeting was held)... the compiler was correct until @MadsTorgersen wrote a spec that didn't exactly agree.

[gafter]

Having said that, I like what @MadsTorgersen wrote into the spec and I'm happy to make the compiler do that.

[gafter]

Fixed in https://github.com/dotnet/roslyn/pull/24761 for dev15.7.x.