more specific method inference

[Dan Smith]

On Dec 19, 2013, at 5:09 PM, Rafkind, Jon <jon.rafkind at hp.com> wrote:

>> 'Integer' appears in incorporation: Q <: Number because Q extends Integer.
> 
> Is this because a type variable counts as a proper type?

Yes.  Both Q and Number are proper types, so "the constraint reduces to true if S is a subtype of T (4.10), and false otherwise."  Just vanilla subtyping.

> I should have
> been more clear about my confusion. 18.2.3 explicitly mentions T can be
> a type variable in (S <: T) but does not say S can be a type variable.
> The only case that seems to allow S to be a type variable is the first case
> 
>  If S and T are proper types, the constraint reduces to true if S is a
> subtype of T (4.10), and false otherwise.
> 
> If so, then it seems the first case would subsume some cases that are
> supposed to be handled by the case starting with "If T is a type
> variable..."

If T is a type variable, then in all subsequent cases, S is not a proper type.  S might be an inference variable, or a parameterized type Foo<alpha>, or an array type, or ...

If S is a type variable and T is not a proper type, then any of the rules involving "the supertypes of S" may apply.

—Dan