caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Gabriel Scherer <gabriel.scherer@gmail.com>
To: Gregory Malecha <gmalecha@gmail.com>
Cc: Jacques Garrigue <garrigue@math.nagoya-u.ac.jp>,
	Mailing List OCaml <caml-list@inria.fr>
Subject: Re: [Caml-list] "Type constructor b would escape its scope"
Date: Mon, 28 Mar 2016 10:07:54 +0200	[thread overview]
Message-ID: <CAPFanBG6sOSX7r5pavPzsMhTcB1fwSjq36huMXyMNHD5bT=yEA@mail.gmail.com> (raw)
In-Reply-To: <CAGraiHJVq6qMWRWoGNBe4ZPaFfBucoUMb1VuRykRvDbs4USceQ@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 3125 bytes --]

Locally abstract types scope over the definition body where they can
be used as type names or to refine GADTs by pattern-matching.

module type Ty = sig type t end

(* we need the scoping of the locally abstract type in this example;
   note that (type a) would work as well *)
let mkTy : type a . unit -> (module Ty with type t = a) =
  fun () ->
    let module T = struct type t = a end in
    (module T)

> val mkTy : unit -> (module Ty with type t = 'a) = <fun>

(* a type variable 'a only scopes over the type annotation *)
let mkTy : 'a . unit -> (module Ty with type t = 'a) =
  fun () ->
    let module T = struct type t = 'a end in
    (module T)

>  let module T = struct type t = 'a end in
>                                 ^^
> Error: Unbound type parameter 'a


type _ tag = Int : int tag | Prod : 'a tag * 'b tag -> ('a * 'b) tag

(* we need the ability to be refined by GADT matching here *)
let rec default : type a . a tag -> a =
  function
  | Int -> 0
  | Prod (ta, tb) -> (default ta, default tb)

> val default : 'a tag -> 'a = <fun>

(* a type variable just unifies with the first case and fails on the second
*)
let rec default : 'a . 'a tag -> 'a =
  function
  | Int -> 0
  | Prod (ta, tb) -> (default ta, default tb)
>   ^^^^^^^^^^^^^
> Error: This pattern matches values of type ($0 * $1) tag
>        but a pattern was expected which matches values of type int tag
>        Type $0 * $1 is not compatible with type int

(* the difference by "(type a) ..." and "type a . ..." is that the latter
   allows polymorphic recursion (just as "'a . ..."), so this example fails
   with just (type a) *)
let rec default (type a) : a tag -> a =
  function
  | Int -> 0
  | Prod (ta, tb) -> (default ta, default tb)
>                             ^^
> Error: This expression has type $0 tag but an expression
> was expected of type 'a
> The type constructor $0 would escape its scope

On Mon, Mar 28, 2016 at 6:20 AM, Gregory Malecha <gmalecha@gmail.com> wrote:
> Thanks. I'm not sure I understand the difference between the two though.
Is
> there an example showing when they produce different results?
>
> On Sun, Mar 27, 2016 at 6:12 PM, Jacques Garrigue
> <garrigue@math.nagoya-u.ac.jp> wrote:
>>
>> On 2016/03/28 03:52, Gregory Malecha wrote:
>> >
>> > Thanks, Leo --
>> >
>> > This is exactly what I needed to know. I thought that the (type a)
>> > annotation was forcing the function to be polymorphic. For future
reference,
>> > is there any way to write the annotation (that forces the function to
be
>> > polymorphic) without having to write explicit ‘fun's?
>>
>> You should either write
>>
>>   let generic_search_stream : ‘b. int option -> (internal_result, ‘b)
>> result_stream = …
>>
>> or
>>
>>   let generic_search_stream : type b. int option -> (internal_result, b)
>> result_stream = …
>>
>> where the latter one both requires polymorphism and defines a locally
>> abstract type,
>> i.e. probably the behavior you expect with (type b).
>>
>> Jacques
>>
>
>
>
> --
> gregory malecha

[-- Attachment #2: Type: text/html, Size: 3998 bytes --]

  reply	other threads:[~2016-03-28  8:08 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-03-27  6:19 Gregory Malecha
2016-03-27  7:04 ` Leo White
2016-03-27 18:52   ` Gregory Malecha
2016-03-28  1:12     ` Jacques Garrigue
2016-03-28  4:20       ` Gregory Malecha
2016-03-28  8:07         ` Gabriel Scherer [this message]
2016-03-28 15:35           ` Gregory Malecha
2016-03-28 19:20           ` Arnaud Spiwack
2016-03-28 20:51             ` Leo White
2016-03-28 21:00             ` Gerd Stolpmann

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAPFanBG6sOSX7r5pavPzsMhTcB1fwSjq36huMXyMNHD5bT=yEA@mail.gmail.com' \
    --to=gabriel.scherer@gmail.com \
    --cc=caml-list@inria.fr \
    --cc=garrigue@math.nagoya-u.ac.jp \
    --cc=gmalecha@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).