Re: [Haskell-cafe] Shootout favoring imperative code

--- Sebastian Sylvan wrote:

My point here was that even though you _can_ generate this data in Haskell, there's no point in requiring (because the order doesn't matter for the benchmark itself).

We do need to agree on which 30 permutations should be used in the validation of the benchmark (just to make sure that the algorithms are producing correct output). Perhaps we could specify the 30 (or perhaps 'N') permutations as an input file, or perhaps require that they be hard-coded into the program? The problem with using an input file is that now we are involving file I/O in the benchmark, which introduces questions about where time is being spent (i.e., file access instead of pancake-flipping). -Brent