Type inference: Understand that T is always implicitly convertible to T?

[masonwheeler]

I have a function with this signature:
public static T? Foo<T>(T? l, T? r) where T : struct, IEqualityOperators<T, T, bool>

Code that invokes it as Foo(26, 1953) gives an error:

CS0411: The type arguments for method 'Foo' cannot be inferred from the usage. Try specifying the type arguments explicitly.

I see no ambiguity here. This should obviously infer the types as int. But per discussion on Discord, the type inference engine doesn't understand that because there are no nullable values being provided, and it's not quite smart enough to reason that T is always implicitly convertible to T?.

Could this be improved in an upcoming C# version?

[CyrusNajmabadi]

@masonwheeler

Could this be improved in an upcoming C# version?

Yes. Best way would be to provide a potential spec revision blurb here (you can reference the spec directly, and use things like ```diff sections), stating what change to make to make this work.

[masonwheeler]

All right. Where can I find the spec?

[333fred]

Type inference is specified here: https://github.com/dotnet/csharpstandard/blob/draft-v8/standard/expressions.md#1263-type-inference.

[masonwheeler]

Thanks! I'll look it over when I have some time, probably this weekend.