[Haskell-cafe] Re: Progress on shootout entries
sebastian.sylvan at gmail.com
Wed Jan 4 09:25:44 EST 2006
On 1/4/06, Chris Kuklewicz <haskell at list.mightyreason.com> wrote:
> Sebastian Sylvan wrote:
> > On 1/4/06, Josh Goldfoot <j_goldfoot at yahoo.com> wrote:
> >>Keep in mind that the shootout requires that the first 30 permutations printed out by the Fannkuch benchmark to be exactly those given in the "example."
> > Well I'm one step closer to just not caring about the shootout anymore.
> > The spec says *nothing* about the order of permutation. So the fact
> > that they require them to be generated in a specific order (I'm sure
> > it's just coincidence that it's the order you get in thet typical
> > C-style permutation generator) is silly.
> > What's the point of a language benchmark if all it tests is your
> > language's ability to instruction-for-instruction implement a C
> > algorithm? It's certainly possible to implement the exact same
> > algorithm using Ptr Word8 etc, but what's the point? It's not
> > idiomatic Haskell anymore and as such has little or no interest to me.
> > This is silly!
> > /S
> It is silly. But real work almost always involves having to heed
> requirements that are annoying. And for a benchmark, it helps to keep
> everyone using a similar algorithm. That said, this is the code Bertram
> Felgenhauer posted to create the "right" permutation sequence:
In this case the benchmark was about indexing small sequences of
numbers, not about generating a specific sequence of permutations, so
it would've been better, IMO, to just let people use whatever
permutation algorithm they prefer.
More information about the Haskell-Cafe