[Shootout-list] scoring problem
Brent Fulgham
bfulg@pacbell.net
Tue, 6 Jul 2004 16:49:55 -0700 (PDT)
--- Stephen Weeks <sweeks@sweeks.com> wrote:
>
> Implementations that have a fastest time (or memory
> usage) of zero are
> assigned a score of zero, rather than one. This
> happens on (at least)
> the except and strcat benchmarks.
>
Yes -- I alluded to this in my first e-mail. I need
to review the "minibench.pl" driver to see what it
does when a test aborts (due to timeout or incorrect
output). It should not be providing data for the test
at all, but I think it actually generates an entry
with zero values.
This was okay with the original scoreing algorithm,
since it tossed out zero entries. But this is a
problem with our current setup.
I'll see what I can do.
Thanks,
-Brent