ocaml-developer Google Group

ocamlhackers.ning.com is open

martin_jam...@emailuser.net (Martin Jambon) — Thu, 26 Jun 2008 20:39:24 UT

I couldn't resist creating an OCaml social network at Ning:
[link]
It's free and easy. Allows you to have your OCaml blog and exclusively OCaml
friends. If you like this, join now :-)
Martin

[Newbie] Problems building ocamlnet2

phillip.old...@gmail.com (Phillip B Oldham) — Thu, 15 May 2008 19:40:08 UT

I'm having problems building the ocamlnet2 examples, and I'm not sure
if their either broken or I'm not doing it right.

I've downloaded and unzipped ocamlnet-2.2.9, cd'd to the examples/cgi/
netcgi2 directory, and because there's a Makefile there I'm just
trying `make`, but get the following:

Re: [ocaml-developer 120] Simple tutorial on linking to/using a C library?

oandr...@gmail.com (Olivier Andrieu) — Tue, 08 Apr 2008 10:06:50 UT

On Thu, Apr 3, 2008 at 7:49 AM, Phillip B Oldham
You can also check the chapter 12 of the O'Reilly book "Developing
Applications With Objective Caml"
[link]

Documenting arguments ?

fabrice.march...@orange.fr (Fabrice Marchant) — Sun, 06 Apr 2008 15:11:01 UT

Hi !
Maybe is it a good place to post such a question here ?
Not an OCaml specific question, however I'm trying to feed properly the 'doc' field for 'Arg.parse'.
Unix command :
man man
says :
bold text type exactly as shown.
italic text replace with appropriate argument.
[-abc] any or all arguments within [ ] are optional.

Re: [ocaml-developer 131] Re: Some opinion about OCAML

sylv...@le-gall.net (Sylvain Le Gall) — Fri, 04 Apr 2008 16:26:48 UT

I am not sure about that. Coming to mix different type are not that
common. In fact, most of the time, i am really annoyed by the
different type cast done by the compiler... I really do prefer to do
typecast explicitely. At least half of the time i don't get the expected
behavior and this is a huge source of bugs in C/C++ program. I like the

Re: [ocaml-developer 130] Re: Some opinion about OCAML

j...@ffconsultancy.com (Jon Harrop) — Fri, 04 Apr 2008 13:38:14 UT

Yes. However, that is not likely to catch many bugs.
This is a fundamental design flaw in OCaml that will almost certainly never be
fixed.
Even for scalars it is common to mix different numeric types: int8, int, int64
etc.
In a different thread, Sylvain wrote a pair of mutually recursive functions

Re: Some opinion about OCAML

daniel.c.buen...@gmail.com (daniel.c.buenzli) — Fri, 04 Apr 2008 13:17:28 UT

You cannot issue such a warning if you are using List.mem with an
abstract datatype. However I think
it can be done if the application of = is made on an abstract datatype
in a monomorphic context (witness the specialisation done by ocaml for
the comparison of ints).

The bugtracker.

Re: [ocaml-developer 128] Re: Some opinion about OCAML

j...@ffconsultancy.com (Jon Harrop) — Fri, 04 Apr 2008 09:50:47 UT

Alas, that is impossible with OCaml's current design because it can't do
monomorphization. All the OCaml compiler ever sees is a completely
polymorphic:
a = b
and it has no idea that this will later be erroneously applied to an abstract
type as well as its correct applications to other types.
Your best chance of a simple solution is to copy SML's equality types, which

Re: Some opinion about OCAML

daniel.c.buen...@gmail.com (daniel.c.buenzli) — Fri, 04 Apr 2008 07:04:47 UT

Well not in that case it seems. Frankly both approaches seem to have
problems. Best I think would be compile time warnings for caml when
abstract types are being compared with structural equality (a wish you
could have put in the bt by the way).

Not it is not, there's just a balancing act to make, but most go in a

Re: [ocaml-developer 126] Re: Some opinion about OCAML

j...@ffconsultancy.com (Jon Harrop) — Thu, 03 Apr 2008 22:09:39 UT

Sure. F# simply cures a practically important class of bugs that OCaml leaves
wide open. Other bugs still exist.
That is an awful policy. Not only is type inference an obvious counter-example
but the complexity required by OCaml to solve mundane problems is clearly
counter-productive. Look at the current thread on the caml-list about how

Re: Some opinion about OCAML

daniel.c.buen...@gmail.com (daniel.c.buenzli) — Thu, 03 Apr 2008 18:51:26 UT

Well you need to tell the compiler at least once otherwise he won't
know that e.g. you want to use physical equality on the strings for
this set instead of say structural equality. As a side effect this can
be the source of interesting and hard to catch bugs (forgot to coerce
to the right subtype) that you won't get with the more explicit

Re: [ocaml-developer 123] Re: Some opinion about OCAML

j...@ffconsultancy.com (Jon Harrop) — Thu, 03 Apr 2008 18:22:00 UT

The appropriate call is selected by the compiler after monomorphization. So
there is no ambiguity or need for type annotations etc.

Re: [ocaml-developer 120] Simple tutorial on linking to/using a C library?

hca...@gmail.com (Hezekiah Carty) — Thu, 03 Apr 2008 13:48:51 UT

On Thu, Apr 3, 2008 at 1:49 AM, Phillip B Oldham
This tutorial page is a very nice introduction to wrapping C libraries
for use in OCaml:
[link]
Chapter 18 in the manual provides more detail if you need it:
[link]

Re: Some opinion about OCAML

daniel.c.buen...@gmail.com (daniel.c.buenzli) — Thu, 03 Apr 2008 10:04:17 UT

So then I need inject/project functions for this new type or explicit
type anotations to make it use the right equality ?

Daniel

Re: [ocaml-developer 121] Re: Some opinion about OCAML

j...@ffconsultancy.com (Jon Harrop) — Thu, 03 Apr 2008 09:49:45 UT

Yes. You are just overriding the default comparison method provided for all
classes in .NET. To use a different comparison for an existing type, you just
create a subtype and override the comparison method.