Mailing list for all users of the OCaml language and system.
 help / color / mirror / Atom feed
From: Zheng Li <li@pps.jussieu.fr>
To: caml-list@inria.fr
Subject: Re: Compiler feature - useful or not?
Date: Wed, 14 Nov 2007 15:37:31 +0100	[thread overview]
Message-ID: <87mytg28r8.fsf@pps.jussieu.fr> (raw)
In-Reply-To: <473A363F.2080301@gmail.com>

Edgar Friendly <thelema314@gmail.com> writes:
> When one writes
>
> type row = int
> type col = int
>
> This allows one to use the type names "row" and "col" as synonyms of
> int.  But it doesn't prevent one from using a value of type row in the
> place of a value of type col.  OCaml allows us to enforce row as
> distinct from int two ways:
>
> 1) Variants:
> type row = Row of int
> type col = Col of int
>
> Downside: unnecessary boxing and tagging
> conversion from row -> int: (fun r -> match r with Row i -> i)
> conversion from int -> row: (fun i -> Row i)

I agree with Gerd, this is probably the least-effort solution. If I'm
not making mistake, even with the new feature introduced by Pierre (Btw,
really nice feature!), you still have to make the restriction at the
module level. I.e., only the outside world of current module will
benefit from this protection, and if you want to use this protection
right now, you still have to wrap type row/col as modules.

You may consider using variant with the syntax convention in Gerd's
post, or you can consider using record. E.g.

type row = {r:int} and col = {c:int}

then the "from" method are just record construction like {c=3} and "to"
method are just field access like x.r.

On efficiency, note that if your single variant (record) has multiple
fields, e.g. "type row = Row of int * int", this encoding doesn't
introduce more (un)boxing than "type row = int * int". The difference
happens when the single variant (record) has single field itself. There
was discussion on this topic before. I do agree that eliminating extra
(un)boxing on single variant type (and single immutable field record
type) would be nice.

> 2)  Functors:
> module type RowCol =
> sig
>   type row
>   val int_of_row : row -> int
>   val row_of_int : int -> row
>   type col
>   val int_of_col : col -> int
>   val col_of_int : int -> col
> end
>
> module Main = functor (RC: RowCol) -> struct
>  (* REST OF PROGRAM HERE *)
> end
>

If you want to make the protection effective in "current" module, you
anyway needs some module-level abstraction in current OCaml. The following
functor tries to do the simple wrapping for any type, and can be defined
once for all:

module type ANY = sig type t end
module type ABS = sig type t type nat val box: nat -> t val unbox: t -> nat end
module Abstr(T:ANY) : ABS with type nat = T.t = struct
  type t = T.t and nat = T.t
  let box x = x and unbox x = x
end

(* Test *)
# module Row = Abstr(struct type t = int end)
# module Col = Abstr(struct type t = int end)
# let x = Row.box 3 and y = Col.box 5;;
val x : Row.t = <abstr>
val y : Col.t = <abstr>
# let f x y = Row.unbox x + Col.unbox y;;
val f : Row.t -> Col.t -> int = <fun>

-- 
Zheng Li
http://www.pps.jussieu.fr/~li


      parent reply	other threads:[~2007-11-14 14:35 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-11-13 23:41 Edgar Friendly
2007-11-14  0:08 ` [Caml-list] " Yaron Minsky
2007-11-14  0:21   ` Martin Jambon
2007-11-14  7:58     ` Pierre Weis
2007-11-14 12:37       ` Alain Frisch
2007-11-14 13:56         ` Virgile Prevosto
2007-11-14 14:35         ` Pierre Weis
2007-11-14 16:38           ` Alain Frisch
2007-11-14 18:43             ` Pierre Weis
2007-11-14 19:19               ` Edgar Friendly
2007-11-15  6:29               ` Alain Frisch
2007-11-15 13:26                 ` Pierre Weis
2007-11-15 17:29                   ` Edgar Friendly
2007-11-15 20:28                     ` Fernando Alegre
2007-11-16  0:47                       ` Brian Hurt
2007-11-15 22:37                     ` Michaël Le Barbier
2007-11-15 22:24                   ` Michaël Le Barbier
2007-11-16  0:30                   ` Yaron Minsky
2007-11-16  1:51                     ` Martin Jambon
2007-11-16  9:23                       ` Alain Frisch
2007-11-16 14:17                         ` rossberg
2007-11-16 15:08                         ` Martin Jambon
2007-11-16 16:43                           ` Martin Jambon
2007-11-16 16:46                             ` Till Varoquaux
2007-11-16 17:27                             ` Edgar Friendly
2007-11-16 17:47                               ` Martin Jambon
2007-11-16 17:54                                 ` Edgar Friendly
2007-11-16 18:10                                   ` Fernando Alegre
2007-11-16 19:18                                     ` David Allsopp
2007-11-16 19:32                                       ` Fernando Alegre
2007-11-16 19:50                                         ` Gerd Stolpmann
2007-11-16 17:31                             ` Fernando Alegre
2007-11-16 17:43                               ` Edgar Friendly
2007-11-16  0:46                   ` Christophe TROESTLER
2007-11-16  8:23                     ` Andrej Bauer
2007-11-16  8:58                       ` Jean-Christophe Filliâtre
2007-11-16  9:13                         ` Andrej Bauer
2007-11-16  9:48                           ` Christophe TROESTLER
2007-11-14 16:57       ` Edgar Friendly
2007-11-14 21:04         ` Pierre Weis
2007-11-14 22:09           ` Edgar Friendly
2007-11-15  0:17         ` Jacques Garrigue
2007-11-15  6:23           ` Edgar Friendly
2007-11-15 10:53             ` Vincent Hanquez
2007-11-15 13:48               ` Jacques Carette
2007-11-15 14:43                 ` Jon Harrop
2007-11-15 16:54                   ` Martin Jambon
2007-11-14 16:09   ` Edgar Friendly
2007-11-14 16:20     ` Brian Hurt
2007-11-14 11:01 ` Gerd Stolpmann
2007-11-14 10:57   ` Jon Harrop
2007-11-14 14:37 ` Zheng Li [this message]

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=87mytg28r8.fsf@pps.jussieu.fr \
    --to=li@pps.jussieu.fr \
    --cc=caml-list@inria.fr \
    /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