Haskell Data.Sequence Implementation

[mroman42]

A faster and more concise Haskell implementation using immutable structures.

[mroman42]

I am concerned about two problems here:

• The resolution of the standard clock is maybe too small
• The lazy evaluation could prevent some benchmarks to be really evaluated

[mroman42]

I am trying to solve those problems using the Criterion library for benchmarking in Haskell.

[mroman42]

At the end, the Criterion library uses a similar timing method as we are using already. The source code can be read here: https://hackage.haskell.org/package/criterion-0.8.0.1/docs/src/Criterion-Measurement.html#time_

[mroman42]

There is a solution here for the problem: http://stackoverflow.com/questions/14163072/how-to-force-evaluation-in-haskell/14163112#14163112

[JJ]

So let's see if I got this correctly. It's being evaluated, right? Because deepseq is not used on the code.

[JJ]

I'm getting 0s in the bitflip and onemax... So either we need more resolution or it's not been evaluated.

[mroman42]

It's deepseq what forces them to be evaluated. I will use seq and deepseq in the code and update the pull request shortly.

[JJ]

Please remember to add your name to the paper, well deserved.

[JJ]

I don't have time to add it to the paper now. It'll stay there for the next version.