Re: [Caml-list] First Class modules -A bug in 4.00?

[Jacques Garrigue <garrigue@math.nagoya-u.ac.jp>]

On 2012/06/09, at 5:25, Milan Stanojević wrote:

> On Thu, Jun 7, 2012 at 11:22 PM, Jacques Garrigue
> <garrigue@math.nagoya-u.ac.jp> wrote:
>> On 2012/06/08, at 11:35, Hongbo Zhang wrote:
>> 
>>> Hi, List
>>> I am not sure this is a bug or not?
>>> 
>>> Below is a contrived example:
>>> ------------------
>>> module type S = sig
>>>   type t=int
>>>   module  X : sig type u   end
>>> end
>>> 
>>> let f ( module X : S)  (y:X.X.u) =
>>> 3
>>> --------------------
>>> Error: This pattern matches values of type X.X.u
>>>     but a pattern was expected which matches values of type X.X.u
>>>     The type constructor X.X.u would escape its scope
>>> -- Thanks
>> 
>> Definitely, this is not a bug.
>> Type X.X.u is abstract, and showing it outside (as by taking an argument
>> of that type) would be meaningless.
> 
> What do you mean by "abstract type" here?

The problem here is that X.X.u is both abstract, and introduced by the
pattern (module X : S), and as such is only valid in the scope of X.

> For example, I'd call X.t in the following example abstract and it is
> perfectly fine to use it as argument.
> module X : sig type t end = struct type t = int end
> let f (x : X.t) = 3

You can actually get the same error by using a reference:

let r = ref [];;
module X : sig type t end = struct type t = int end
let f (x : X.t) = 3  ;;
r := [f];;
Error: This expression has type X.t -> int
      but an expression was expected of type X.t -> int
      The type constructor X.t would escape its scope

In case of a usual module definition, references defined before it
are out of scope.

	Jacques Garrigue