From: Guillaume Yziquel <guillaume.yziquel@citycable.ch>
To: Goswin von Brederlow <goswin-v-b@web.de>
Cc: Andreas Rossberg <rossberg@mpi-sws.org>, OCaml List <caml-list@inria.fr>
Subject: Re: [Caml-list] Recursive subtyping issue
Date: Sat, 27 Feb 2010 14:11:40 +0100 [thread overview]
Message-ID: <4B891A0C.604@citycable.ch> (raw)
In-Reply-To: <87k4tyoq3m.fsf@frosties.localdomain>
Goswin von Brederlow a écrit :
> Guillaume Yziquel <guillaume.yziquel@citycable.ch> writes:
>
>> My goal is to implement a type inference barrier.
>>
>> You can do
>>
>>> type 'a q = private w
>> and from the type inference point of view, int q and float q are two
>> distinct types, that you can subtype to a common type.
>>
>> What I want is also to have the reverse, i.e.
>>
>>> type w = private 'a q
>> But that doesn't work out this way because of the fact that 'a is unbound.
>
> But then int q :> w :> float q and float q :> w :> int q. That would
> make the whole thing somewhat pointless. Everyone could convert the type
> to anything. I guess it would protect from accidentally passing the
> wrong 'a q while allowing purposefully to pass any 'a q.
Exactly. It's the situation you have when you're trying to do OCaml
binding of libraries written in dynamic languages: You want to have type
inference on the side of semantics, hence a typing reflecting this. Bt
you also want to type lower-level details about objects. This last
sentence is not so true with Python, for instance, but with R, it is
(despite the argument I had with Simon Urbanek on the r-devel@ mailing
list).
You want to have an 'a t type with 'a reflecting the corresponding
typing in OCaml. And an 'underlying' type representing the low-level value.
Doing 'a t = private underlying allows you to create a type inference
barrier. However, you also want to be able to cast from underlying to 'a
t, when you get the result of a function in R or Python, for instance.
So that's exactly the use case you mentionned above.
> Why not supply conversion functions that do any additional checks to
> ensure the conversion is a valid one? Consider the following:
Because that's exactly what I try to avoid. R and Python are already
slow enough and dynamically type-checked at every corner. I'm not happy
to add another type-checking layer.
> module M : sig
> type w = Int of int | Float of float
Ugly. (IMHO)
> type 'a q = private w
> val add : 'a q -> 'a q -> 'a q
> val print : w -> unit
> val as_int : w -> int q
> val as_float : w -> float q
Ugly. (IMHO)
> end = struct
I've been looking all over at this issue, but simply cannot find a way
out. While experimenting on this, I've stumbled on a number of quirky
issues with the type system.
First one: http://ocaml.janestreet.com/?q=node/26
Second one:
> # type 'a q = <m : 'a>;;
> type 'a q = < m : 'a >
> # let f : 'a q -> 'a q = fun x -> x;;
> val f : 'a q -> 'a q = <fun>
> # let o = object method m : 'a. 'a -> 'a = fun x -> x end;;
> val o : < m : 'a. 'a -> 'a > = <obj>
> # f o;;
> Error: This expression has type < m : 'a. 'a -> 'a >
> but an expression was expected of type 'b q
> The universal variable 'a would escape its scope
> #
All these issues seem to be somehow related, in a way I'm not yet able
to articulate clearly.
--
Guillaume Yziquel
http://yziquel.homelinux.org/
next prev parent reply other threads:[~2010-02-27 13:11 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-02-27 1:52 Guillaume Yziquel
2010-02-27 6:38 ` [Caml-list] " Andreas Rossberg
2010-02-27 10:25 ` Guillaume Yziquel
2010-02-27 11:49 ` Goswin von Brederlow
2010-02-27 13:11 ` Guillaume Yziquel [this message]
2010-02-27 16:52 ` Andreas Rossberg
2010-02-27 18:10 ` Guillaume Yziquel
2010-02-27 19:52 ` Guillaume Yziquel
2010-02-27 20:32 ` Guillaume Yziquel
2010-03-01 10:55 ` Stéphane Glondu
2010-03-01 11:21 ` Guillaume Yziquel
2010-03-01 12:28 ` Stéphane Glondu
2010-03-01 12:49 ` David Allsopp
2010-03-01 13:06 ` Guillaume Yziquel
2010-03-01 12:49 ` David Allsopp
2010-03-01 13:28 ` Goswin von Brederlow
2010-03-01 20:12 ` David Allsopp
2010-03-02 10:22 ` Goswin von Brederlow
2010-03-01 13:33 ` Guillaume Yziquel
2010-03-01 20:18 ` David Allsopp
2010-02-28 9:54 ` Goswin von Brederlow
2010-02-28 11:08 ` Guillaume Yziquel
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=4B891A0C.604@citycable.ch \
--to=guillaume.yziquel@citycable.ch \
--cc=caml-list@inria.fr \
--cc=goswin-v-b@web.de \
--cc=rossberg@mpi-sws.org \
/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).