This would be too J-centric but it would be nice, however unlikely, to see
the pendulum swing the other way.

More to the point, in this particular case, as in many others, J has
something _much_ more powerful than sort: "grade" returns a permutation
vector which opens up far broader possibilities than "sort".

For instance, removing duplicates from a list but leaving it otherwise in
the original order is trivial and efficient using a grade vector but could
be a real pain without it.

However, once you start dealing with interesting concepts, you've moved far
beyond the simplistic realm of benchmarking.

On 12/8/07, bill lam <[EMAIL PROTECTED]> wrote:
>
> It seems the benchmark was designed for C.  I'm skeptical about this
> benchmark
> because it shows fortran get rather poor performance in number crunching
> and
> that's counter-intuitive.
>
> If the benchmark is tailored for J then there should be programming tasks
> for
> each of J's primitive, eg. a problem on sorting an array and  /:~ is an
> accepted
> solution.  This benchmark is of course extremely unfair, however if you
> can live
> without those primitives than you do not need J. :-)
>
> I once read an email/article quoting APL was not allowed in a programming
> exam.
>
> --
> regards,
> bill
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>



-- 
Devon McCormick, CFA
^me^ at acm.
org is my
preferred e-mail
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to