[Shootout-list] Safety vs. speed

Bengt Kleberg bengt.kleberg@ericsson.com
Wed, 22 Sep 2004 08:24:33 +0200


Einar Karttunen wrote:
...deleted
> I think the tests *should* note implementors when they have to be
> wary of overflow. Languages should support larger numbers (even C 
> does with long long), but having bignum arithmetic slowing down
> other tests wouldn't be very nice.

the implementor gets the test data (values of N). we add a warning that 
test data may increase in the future. it is then up to the implementor 
to choose how to code the test solution.
one way of handling a potentially unsafe solution is to come back 
periodically and check if it has turned erronous.


...deleted
> ps. Is anyone working on "Ring of Message-Sending Processes" ? 
> I could add a Haskell solution tonight.

i have already written one, but not in haskell :-). however, i am 
lobbying for a change. i think the number of processes should be static, 
and N should change the number of times the message goes around the ring.
you see, i have also written a process instantination test. there the 
number of processes varies with N.
i think these are two different things (message passing and process 
creation), and therefore needs two different test.


bengt