From: Jacques Le Normand <rathereasy@gmail.com>
To: caml-list caml-list <caml-list@yquem.inria.fr>
Subject: GADT constructor syntax
Date: Sat, 4 Dec 2010 14:25:01 -0500 [thread overview]
Message-ID: <AANLkTimVTv7J_b6AY1gKKT9hdQx83a+4FBM7pqvHpAHM@mail.gmail.com> (raw)
Dear caml-list,
I would like to start a constructive discussion on the syntax of GADT
constructors of the ocaml gadt branch, which can be found at:
https://sites.google.com/site/ocamlgadt/
There are two separate issues:
1) general constructor form
option a)
type _ t =
TrueLit : bool t
| IntLit of int : int lit
option b)
type _ t =
TrueLit : bool t
| IntLit : int -> int lit
I'm open to other options. The branch has used option b) from the
start, but I've just switched to option a) to see what it's like
Personal opinion:
I slightly prefer option b), because it makes it clear that it's a
gadt constructor right from the start. This is useful because the type
variables in gadt constructors are independent of the type parameters
of the type, consider:
type 'a t = Foo of 'a : 'b t
this, counter intuitively, creates a constructor Foo of type forall 'd
'e. 'd t -> 'e t.
2) explicit quantification of existential variables
option a)
leave existential variables implicitly quantified. For example:
type _ u = Bar of 'a t : u
or
type _ u = Bar : 'a t -> u
option b)
specifically quantify existential variables. For example:
type _ u = Bar of 'a. 'a t : u
or
type _ u = Bar : 'a. 'a t -> u
Currently, the branch uses option a).
Personal opinion: I prefer option b). This is for four reasons:
I) the scope of the explicitly quantified variable is not clear. For
example, how do you interpret:
type _ u = Bar of 'a. 'a : 'a t
or
type _ u = Bar : 'a. 'a -> 'a t
In one interpretation bar has type forall 'a 'b. 'a -> 'b t and in
another interpretation it has type forall 'a. 'a -> 'a t. My
inclination would be to flag it as an error.
II)
In the example of option b), the 'a variable is quantified as a
universal variable but, in patterns, it is used as an existential
variable. This is something I found very confusing in Haskell where
they actually use the 'forall' keyword.
III) option a) is the current Haskell GADT syntax and I've never heard
anyone complain about it
IIII) I don't see how option b) improves either readability or bug prevention
I look forward to hearing your opinions.
--Jacques Le Normand
next reply other threads:[~2010-12-04 19:25 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-12-04 19:25 Jacques Le Normand [this message]
2010-12-04 19:36 ` [Caml-list] " gasche
2010-12-04 19:39 ` Basile Starynkevitch
2010-12-04 19:41 ` Lukasz Stafiniak
2010-12-04 20:14 ` Jacques Le Normand
2010-12-04 20:22 ` Lukasz Stafiniak
2010-12-04 20:54 ` Jacques Le Normand
2010-12-04 21:00 ` Lukasz Stafiniak
2010-12-04 21:06 ` Jacques Le Normand
2010-12-05 8:10 ` Lukasz Stafiniak
2010-12-05 8:16 ` Lukasz Stafiniak
2010-12-05 8:25 ` Lukasz Stafiniak
2010-12-05 8:35 ` bluestorm
2010-12-06 0:21 ` Jacques Garrigue
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=AANLkTimVTv7J_b6AY1gKKT9hdQx83a+4FBM7pqvHpAHM@mail.gmail.com \
--to=rathereasy@gmail.com \
--cc=caml-list@yquem.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;
as well as URLs for NNTP newsgroup(s).