A nastier example

A nastier example

[Julien Verlaguet]

Here is a nastier case than the one in the previous msg.

exception E of int;;

let x=E(1);;

exception E of bool;;

x=E(true);;

answer : true

I received many mails (in private), explaining to me how the equality 
works in OCaml, how exceptions are represented etc ...
Which is not what I was asking, I will try to reformulate :

Is this the behaviour we want ?

Can two values of different types be equal ?

I agree that this wouldn't be a problem if it would only work in a 
toplevel ... But it compiles in structures as well !!

J

Re: [Caml-list] A nastier example

[Eric Cooper]

On Sat, May 14, 2005 at 04:51:07PM +0200, Julien Verlaguet wrote:
> Here is a nastier case than the one in the previous msg.
> 
> exception E of int;;
> 
> let x=E(1);;
> 
> exception E of bool;;
> 
> x=E(true);;
> 
> answer : true
> 
> I received many mails (in private), explaining to me how the equality 
> works in OCaml, how exceptions are represented etc ...
> Which is not what I was asking, I will try to reformulate :
> 
> Is this the behaviour we want ?
> 
> Can two values of different types be equal ?

But these aren't different types -- they're both type "exn".
For example:
    let test e = (e = Exit)
    val test : exn -> bool = <fun>
So test can be applied to any exception, no matter what kind of
structure it has.

-- 
Eric Cooper             e c c @ c m u . e d u

Re: [Caml-list] A nastier example

[Marcin 'Qrczak' Kowalczyk]

Julien Verlaguet <Julien.Verlaguet@pps.jussieu.fr> writes:

> I received many mails (in private), explaining to me how the equality
> works in OCaml, how exceptions are represented etc ...
> Which is not what I was asking, I will try to reformulate :
>
> Is this the behaviour we want ?
>
> Can two values of different types be equal ?

They don't have different types. They both have type exn.

-- 
   __("<         Marcin Kowalczyk
   \__/       qrczak@knm.org.pl
    ^^     http://qrnik.knm.org.pl/~qrczak/

Re: [Caml-list] A nastier example

[Julien Verlaguet]

> They don't have different types. They both have type exn.

I was talking about the type of the parameters ...

Sorry for the confusion,

J

Re: [Caml-list] A nastier example

[Jon Harrop]

On Saturday 14 May 2005 15:51, Julien Verlaguet wrote:
> Is this the behaviour we want ?

Ideally, no.

> Can two values of different types be equal ?

In theory, no.

> I agree that this wouldn't be a problem if it would only work in a
> toplevel ...

Does code ever work differently in the top-level than when compiled (with 
ocamlc)? Should it?

> But it compiles in structures as well !! 

I think everyone would agree that there are caveats here but this is a 
question of trade-offs. So the question should be "how can we improve upon 
the current implementation?". To which the answer, for me at least, is "I 
can't".

There are many pitfalls associated with equality in OCaml but the existence of 
polymorphic comparisons is easily justified by their usefulness, IMHO.

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
Objective CAML for Scientists
http://www.ffconsultancy.com/products/ocaml_for_scientists

Re: A nastier example

[Vincenzo Ciancia]

Il giorno sab, 14/05/2005 alle 17.36 +0200, Marcin 'Qrczak' Kowalczyk ha
scritto:
> 
> > Can two values of different types be equal ?
> 
> They don't have different types. They both have type exn.

Yes but one usually does not expect structural equality to "equalize"
things with different structure, while the example posted does something
really strange and implementation dependent - I don't know what other
discussions have been around about this so sorry if this is an old
flame :) but exception equality implemented that way makes physical data
structures implementation observable (or am I fooled by something?)

Vincenzo

-- 
Per rispondermi in privato per favore usate vincenzo_ml at yahoo dot it
To send me a private reply please use vincenzo_ml at yahoo dot it

Re: [Caml-list] A nastier example

[skaller]

On Sun, 2005-05-15 at 05:09, Jon Harrop wrote:

> I think everyone would agree that there are caveats here but this is a 
> question of trade-offs. So the question should be "how can we improve upon 
> the current implementation?". 

> There are many pitfalls associated with equality in OCaml but the existence of 
> polymorphic comparisons is easily justified by their usefulness, IMHO.

Require the user to define the comparison, then provide a way
to generate them 'memberwise' for products and sums.

-- 
John Skaller, mailto:skaller@users.sf.net
voice: 061-2-9660-0850, 
snail: PO BOX 401 Glebe NSW 2037 Australia
Checkout the Felix programming language http://felix.sf.net

Re: [Caml-list] A nastier example

[Mark Shinwell]

On Sat, May 14, 2005 at 08:09:28PM +0100, Jon Harrop wrote:
> Does code ever work differently in the top-level than when compiled (with 
> ocamlc)?

Yes.  See for example "Reply 2" of:

http://pauillac.inria.fr/bin/caml-bugs/not%20a%20bug?id=3591;page=64;user=guest

Mark
-- 
Dr Mark Shinwell -- email: Mark.Shinwell@cl.cam.ac.uk
Theory and Semantics Group, University of Cambridge Computer Laboratory

Re: [Caml-list] A nastier example

[Jon Harrop]

On Sunday 15 May 2005 18:24, Mark Shinwell wrote:
> On Sat, May 14, 2005 at 08:09:28PM +0100, Jon Harrop wrote:
> > Does code ever work differently in the top-level than when compiled (with
> > ocamlc)?
>
> Yes.  See for example "Reply 2" of:
>
> http://pauillac.inria.fr/bin/caml-bugs/not%20a%20bug?id=3591;page=64;user=g
>uest

Very interesting. In summary, is it fair to say that Marshall has problems 
with functions when used in the top-level but not with ocamlc- and ocamlopt- 
compiled code?

Thanks for the link.

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
Objective CAML for Scientists
http://www.ffconsultancy.com/products/ocaml_for_scientists

Re: [Caml-list] A nastier example

[Damien Doligez]

On May 16, 2005, at 02:15, Jon Harrop wrote:

> Very interesting. In summary, is it fair to say that Marshall has 
> problems
> with functions when used in the top-level but not with ocamlc- and 
> ocamlopt-
> compiled code?

Yes, although it is a very small problem: it would pretty hard for the
average programmer to guarantee that the same function is compiled at
the same address in the top-level that unmarshals the closure.

So the usefulness of marshaling closures from the top-level would be
rather small.

-- Damien