Re: [Haskell-cafe] benchmarking pure code

Ok, thank you for all your answers. I'm going to use NFData as advised by everyone. Paul On Wed, Mar 31, 2010 at 10:38:50AM -0700, Bryan O'Sullivan wrote:

On Wed, Mar 31, 2010 at 4:12 AM, Paul Brauner wrote:

...

Thank you, I will look at that. But it seems that criterion uses NFData no?

I do not know of anything wrong with NFData. What you're seeing is much more likely to be a bug in either the benchmarking library you're using, or in your use of it. Most of the benchmarking frameworks on Hackage are extremely dodgy, which was why I wrote criterion.