spec: Interplay between type conversions and embedded structs is not clear.

[mattdee123]

The go 1.18 spec seems to disagree with the implementation regarding how type conversions and embedded structs interplay. Specifically:

type E struct {}

type A struct {
  E
}

type B struct {
  E E
}

func main() {
  _ = A(B{})
}

Results in

cannot convert B{} (value of type B) to type A

However, the spec seems to imply that this should work:

A non-constant value x can be converted to type T in any of these cases: [...] x's type and T are not type parameters but have identical underlying types.

The definition of type identity then says

Two struct types are identical if they have the same sequence of fields, and if corresponding fields have the same names, and identical types, and identical tags.

Finally, the section on struct type says of embedded structs

The unqualified type name acts as the field name.

So, in the case above, the corresponding fields do have the same names and types, which would imply that they are identical and can be converted. I think the definition of type identity should be updated to clarify that corresponding fields must have different "embedded-ness" as well.