caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: "Daniel Bünzli" <daniel.buenzli@erratique.ch>
To: "Milan Stanojević" <milanst@gmail.com>
Cc: Caml-list <caml-list@inria.fr>
Subject: Re: [Caml-list] GADT existential escape
Date: Thu, 19 Mar 2015 11:44:46 +0100	[thread overview]
Message-ID: <A9833AA6DB2C40058D2E66B6A381E599@erratique.ch> (raw)
In-Reply-To: <CAKR7PS_fFML9gvqF1Zr0qa+2uEMUg4FQfp2f9gXuhN1Ee7_n=A@mail.gmail.com>

Le jeudi, 19 mars 2015 à 02:27, Milan Stanojević a écrit :
> This is doable in OCaml.

Cool, a new trick. My understanding of this is basically to apply to types the same idea as Univ by representing them as a first class module.  

However are you sure your example compiles (tried to check, but Core fails on me in the toplevel) ? Trying the following [1] self-contained implementation of the idea [1] the compiler still complains about escaping types.  

> Note that Univ was possible in OCaml even before first-class
> modules, GADTs and extensible types (which are all used in Type_equal)

Yes, see http://alan.petitepomme.net/cwn/2010.02.09.html#1 if you are interested how.  

Thanks,

Daniel

[1]

module Key = struct
  type _ t = ..
end

module type W = sig
  type t
  type _ Key.t += Key : t Key.t
end

type 'a witness = (module W with type t = 'a)

let witness () (type s) =
 let module M = struct
   type t = s
   type _ Key.t += Key : t Key.t
 end
 in
 (module M : W with type t = s)

type ('a, 'b) eq = Eq : ('a, 'a) eq

let eq (type r) (type s) (r : r witness) (s : s witness)
  : (r, s) eq option
  =
  let module R = (val r : W with type t = r) in
  let module S = (val s : W with type t = s) in
  match R.Key with
  | S.Key -> Some Eq
  | _ -> None

type 'a key = 'a witness
type binding = B : 'a key * 'a -> binding
type dict = binding list

let lookup : dict -> 'a key -> 'a option = fun d k ->
  let rec find = function
  | [] -> None
  | B (k', v) :: bs ->
      match eq k k' with
      | None -> find bs
      | Some Eq -> Some v
  in
  find d




  reply	other threads:[~2015-03-19 10:45 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-03-19  0:29 Daniel Bünzli
2015-03-19  0:39 ` Milan Stanojević
2015-03-19  1:07   ` Daniel Bünzli
2015-03-19  1:27     ` Milan Stanojević
2015-03-19 10:44       ` Daniel Bünzli [this message]
2015-03-19 11:02         ` Jeremy Yallop
2015-03-19 11:34           ` Daniel Bünzli
2015-03-19 13:22             ` Daniel Bünzli
2015-03-19 16:46             ` Milan Stanojević
2015-03-19 17:14               ` Daniel Bünzli
2015-03-19 17:15                 ` Frédéric Bour
2015-03-19 17:22                   ` Milan Stanojević
2015-03-19 17:41               ` Jeremy Yallop
2015-03-19 18:29                 ` Alain Frisch
2015-03-19 16:31           ` Milan Stanojević
2015-03-19 11:05         ` Jeremie Dimino
2015-03-19 11:34           ` Daniel Bünzli
2015-03-19 11:59             ` Jeremie Dimino
2015-03-20 13:50               ` Yaron Minsky

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=A9833AA6DB2C40058D2E66B6A381E599@erratique.ch \
    --to=daniel.buenzli@erratique.ch \
    --cc=caml-list@inria.fr \
    --cc=milanst@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).