Better generic type inference based on type parameter constraints.

[orthoxerox]

Steps to Reproduce:

public static void Foo<T, U>(T value)
    where T: IEnumerable<U>
{
	Console.WriteLine(value.FirstOrDefault().ToString());
}

public static void Main(string[] args)
{
	IEnumerable<int> x = new [] { 1, 2 };
	
	Foo(x);
	
	Console.ReadKey();
}

Expected Behavior:

The program compiles and runs. The type of U is inferred to be int, because of T: IEnumerable<U>.

Actual Behavior:

The type arguments for method 'Scratchpad.Program.Foo<T,U>(T)' cannot be inferred from the usage. Try specifying the type arguments explicitly. (CS0411)

[HaloFour]

I want to say that I've seen this come up before but maybe it was in the comments of a different proposal. I'm a fan of better generic inference in cases like this, although I have some minor concerns about how recursive such inference might be. The above case is relatively straightforward

[TonyValenti]

I think you might be referencing this issue:
##5023

[orthoxerox]

Covered by dotnet/csharplang#110