Type system should treat function valued getters as methods?

[gbracha]

Given

interface I { int f();}

class C1 implements I {
   int f() => 1;
}

typedef int G();

is

class C2 implements I {
  G get f() {return () => 2;}
}

a legitimate implementation of I? Currently, this is disallowed. Yet some have requested such functionality. After all,

new C2().f();

works just as well as

new C1().f();

[DartBot]

This comment was originally written by @seaneagan

---

Starred!

It should also work in reverse, C1 should be a legitimate implementation of:

interface I { G get f();}

Note as well that getting the Function value works the same with either implementation:

new C1().f;
new C2().f;

I think the way to spec. it is to have methods desugar into Function valued getters, which yields closurization semantics for free, and avoids needing to specify Function getters separately from other getters, which would be strange.

Also see doc linked in issue #1616 (should be merged with this issue or vice versa) for more comments.

[anders-sandholm]

Added this to the M1 milestone.

[anders-sandholm]

Removed this from the M1 milestone.

[anders-sandholm]

Added this to the M1 milestone.

[gbracha]

Removed this from the M1 milestone.
Removed Type-Defect label.
Added Type-Enhancement label.

[gbracha]

Added this to the Later milestone.

[kasperl]

Removed this from the Later milestone.
Added Oldschool-Milestone-Later label.

[kasperl]

Removed Oldschool-Milestone-Later label.

[lrhn]

With the type system of Dart 2, we are going in the opposite direction from this.
I can guarantee that this will not happen short of another complete replacement of the type system.