Tuple name inference fail to infer @this element name

[jcouv]

@SergeyTeplyakov commented on Sat Oct 28 2017

Version Used: master

Steps to Reproduce:

class Sample
{
    public void FooBar(int x)
    {
        // (Sample, int x)
        var tpl = (this, x);
    }
}

The inferred type is (Sample, int x), but could be (Sample @this, int x).

I work on perf critical application and to avoid closure computation I often pass a tuple into a method similar to ConcurrentDictionary.GetOrAdd:

class Sample
{
    private readonly ConcurrentDictionary<int, string> _cd = new ConcurrentDictionary<int, string>();

    public string GetValue(int value, bool extraState)
    {
        return _cd.GetOrAdd(
            value,
            (key, tpl) =>
            {
                return tpl.@this.ComputeValue(key, tpl.extraState);
            },
            (extraState, @this: this));
    }

    private string ComputeValue(int value, bool extraState) => value.ToString();
}

And in this case, I have to specify a name of the tuple element manually. I understand that my use case is not that common, but I'm curious is there any specific reason why such inference is not implemented? Or this case was just missed?

---

@jcouv commented on Sun Oct 29 2017

This is the same behavior as anonymous types: var x = new { this }; // not allowed vs var x = new { @this=this }; // ok.
I think that's currently by design, so I'll move this suggestion to csharplang repo for discussion.

[jcouv]

Here's the section of the spec that I think is relevant since inference of names follows the same rules for tuples as they did for anonymous objects:

A member declarator can be abbreviated to a simple name, a member access, a base access or a null-conditional member access. This is called a projection initializer and is shorthand for a declaration of and assignment to a property with the same name.