[bitc-dev] Haskell question: exceptions

[Jonathan S. Shapiro]

On Thu, 2005-08-11 at 16:42 -0700, Iavor Diatchki wrote:
> Hello,
> Hmm... I am not sure which rule you are referring to.  Could you post
> a link so that I can look at the details?
> -Iavor

Actually, I cannot. I was repeating a statement made by Swaroop. I have
just spent some time looking for such a requirement within the
definitions of SML and O'Caml, and I cannot find one. So: let me back
up.

Swaroop has noticed the following behavior in O'Caml that we cannot
explain:

        Objective Caml version 3.08.3
# exception Xn of int;;
exception Xn of int
# try raise (Xn 10) with Xn n -> function x -> x;;
- : '_a -> '_a = <fun>


The weird part is that the assigned type is not:

	- : 'a -> 'a = <fun>

Minor variations on this example *do* end up with 'a rather than '_a.

Swaroop: can you post the other cases to illustrate what is puzzling us?

So far as I can tell, there is no reason why ML-style exception handling
should have any effect on polymorphism. That is why we are puzzled.

Since my original question concerning Haskell exceptions, I *have* found
a conclusive statement that the only exceptions in Haskell are
restricted to the IO module, and that there is no generalized exception
mechanism within the language.

shap