God i love haskell

thats really clever! thanks

Wasn't expecting this but wow

Great use of this property!

That's impressive, I often forget that there probably is some hidden pattern that I can leverage. Good one :).

I don't think ((>1) . length) is optimised away; that's just too much to ask. I did some testing once, though it may have been the testing itself that forced the O(n). Profiling is still beyond me.

It may not matter for the overall complexity in this case, but you're doing something in O(n) which could have been done in O(1) ( or maybe O(log n) ), which is generally a useful optimisation; those are always good to keep a sharp eye out for.

If @bkaest does it with not.null.tail, that's probably better. :P It may still be a micro-optimisation, that's certainly true.

Sure, but I don't know how many intermediate lists are left after optimization, so from my point of view it's adding 1 list (assuming it's not optimized away; IDK how advanced optimizations are) to an unknown number of lists keeping the asymptotic coplexity, so it can go either way.

Walking every group entirely ( length ) when all you're interested in is the second : constructor is not a Best Practice.

Composable streaming folds

Elegant.

Haha, very nice!