[Caml-list] Lazyness and exceptions

[Caml-list] Lazyness and exceptions

[Jean-vincent.Loddo]

Hello,

I have observed a strange behaviour of lazy values when an exception is 
raised. Here a minimal code that reproduces the behaviour:

# let action =
     let thunk () =
       Printf.printf "aaa\n";
       failwith "bbb"
     in
     lazy (thunk ());;

val action : unit lazy_t = <lazy>

# Lazy.force action ;;
aaa
Exception: Failure "bbb".

# Lazy.force action ;;
Exception: Failure "bbb".

The lazy value seems to be (correctly) not re-evaluated but the 
exception is (strangely) raised again. Why does this happen?
Jean-Vincent Loddo

Re: [Caml-list] Lazyness and exceptions

[Török Edwin]

On 11/22/2012 12:29 PM, Jean-vincent.Loddo@lipn.univ-paris13.fr wrote:
> Hello,
> 
> I have observed a strange behaviour of lazy values when an exception is raised. Here a minimal code that reproduces the behaviour:
> 
> # let action =
>     let thunk () =
>       Printf.printf "aaa\n";
>       failwith "bbb"
>     in
>     lazy (thunk ());;
> 
> val action : unit lazy_t = <lazy>
> 
> # Lazy.force action ;;
> aaa
> Exception: Failure "bbb".
> 
> # Lazy.force action ;;
> Exception: Failure "bbb".
> 
> The lazy value seems to be (correctly) not re-evaluated but the exception is (strangely) raised again. Why does this happen?

Its the documented behavior:
http://caml.inria.fr/pub/docs/manual-ocaml/libref/Lazy.html

Unfortunately the original backtrace will be lost.

--Edwin

Re: [Caml-list] Lazyness and exceptions

[David House]

In other words: what alternative semantics would you actually want?

On Thu, Nov 22, 2012 at 10:33 AM, Török Edwin <edwin+ml-ocaml@etorok.net> wrote:
> On 11/22/2012 12:29 PM, Jean-vincent.Loddo@lipn.univ-paris13.fr wrote:
>> Hello,
>>
>> I have observed a strange behaviour of lazy values when an exception is raised. Here a minimal code that reproduces the behaviour:
>>
>> # let action =
>>     let thunk () =
>>       Printf.printf "aaa\n";
>>       failwith "bbb"
>>     in
>>     lazy (thunk ());;
>>
>> val action : unit lazy_t = <lazy>
>>
>> # Lazy.force action ;;
>> aaa
>> Exception: Failure "bbb".
>>
>> # Lazy.force action ;;
>> Exception: Failure "bbb".
>>
>> The lazy value seems to be (correctly) not re-evaluated but the exception is (strangely) raised again. Why does this happen?
>
> Its the documented behavior:
> http://caml.inria.fr/pub/docs/manual-ocaml/libref/Lazy.html
>
> Unfortunately the original backtrace will be lost.
>
> --Edwin
>
> --
> Caml-list mailing list.  Subscription management and archives:
> https://sympa.inria.fr/sympa/arc/caml-list
> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> Bug reports: http://caml.inria.fr/bin/caml-bugs

Re: [Caml-list] Lazyness and exceptions

[Jean-vincent.Loddo]

Ok, I understand now why. In the case of my example this choice could 
be discussed because of the nature of the output (unit), because it's 
not clear if the exception belongs to the output domain of the function 
or not (it's a semantics question). However, in the general case we 
cannot provide a reasonable output without re-evaluating the expression.
Thanks.

On Thu, 22 Nov 2012 10:37:23 +0000, David House wrote:
> In other words: what alternative semantics would you actually want?
>
> On Thu, Nov 22, 2012 at 10:33 AM, Török Edwin
> <edwin+ml-ocaml@etorok.net> wrote:
>> On 11/22/2012 12:29 PM, Jean-vincent.Loddo@lipn.univ-paris13.fr 
>> wrote:
>>> Hello,
>>>
>>> I have observed a strange behaviour of lazy values when an 
>>> exception is raised. Here a minimal code that reproduces the 
>>> behaviour:
>>>
>>> # let action =
>>>     let thunk () =
>>>       Printf.printf "aaa\n";
>>>       failwith "bbb"
>>>     in
>>>     lazy (thunk ());;
>>>
>>> val action : unit lazy_t = <lazy>
>>>
>>> # Lazy.force action ;;
>>> aaa
>>> Exception: Failure "bbb".
>>>
>>> # Lazy.force action ;;
>>> Exception: Failure "bbb".
>>>
>>> The lazy value seems to be (correctly) not re-evaluated but the 
>>> exception is (strangely) raised again. Why does this happen?
>>
>> Its the documented behavior:
>> http://caml.inria.fr/pub/docs/manual-ocaml/libref/Lazy.html
>>
>> Unfortunately the original backtrace will be lost.
>>
>> --Edwin
>>
>> --
>> Caml-list mailing list.  Subscription management and archives:
>> https://sympa.inria.fr/sympa/arc/caml-list
>> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
>> Bug reports: http://caml.inria.fr/bin/caml-bugs