Unspecified behavior of generics

[DartBot]

This issue was originally filed by ms...@unipro.ru

---

In the section 9.Generics the specification states that "A type parameter T may be suffixed with an extends clause that specifies the upper bound for T." and "It is a static type warning if a type variable is supertype of its upper bound." But it does not explicitly specify behavior in case a type is not assignable to its upper bound.

Note, that test case
class A<T extends String> {}
main() {
  var x = new A<List>();
}
compiles and runs without errors in DartVM in scripting mode, produces compile-time error in DartVM in checked mode, produces compile-time error in dartc compiler in both modes and compiles without errors in frog in both modes.

[dgrove]

Set owner to @gbracha.
Added Area-Language, Triaged labels.

[gbracha]

1. The declaration is perfectly valid. It can be instantiated as follows: A<String>.
2. The usage should produce a static warning, per 10.10:

It is a static type warning if any of the type arguments to a constructor of a generic type G invoked by a new expression or a constant object expression are not subtypes of the bounds of the corresponding formal type parameters of G.

Hence the behavior of dartc is buggy.

If executed in checked mode, the usage should produce an error, per 10.10.1:

t is a dynamic type error if, in checked mode, Vi is not a subtype of [V1, ..., Vm/T1, ..., Tm]Bi, 1 <= i <= m.

 The VM behavior is therefore correct, and frog's checked-mode behavior is not.

Will open new bugs against dartc and frog.

---

Added Invalid label.

[gbracha]

Just to clarify: "produces compile-time error in DartVM in checked mode," makes no sense, as checked mode has no effect on compilation. I assume it's a run time error.

Also: "compiles without errors in frog in both modes" is correct; the mode is irrelevant to compilation. If it runs without error in checked mode, there is a problem.