Thanks, excelent tip!

Agree that == True should be left of, but i like the style otherwise

True, I don't remember why I didn't do that

There have already been an (&) function in Data.Function that do the same thing like (|>), so we could just write

infixl 0 |>
(|>) = (&)

I'll be honest -- mostly I try to somewhat golf these just for the fun and challenge of it. Performance is only a secondary consideration, but good observation about nub... although I guess for unrestricted strings, someone could throw in some UTF-8 and then you have a slightly larger N, but still the same time complexity of course :D

λ> nub "áéíóá你好你好们你好你好"            
"\225\233\237\243\20320\22909\20204"

I find DestyNova's algorithm very elegant. It's not O(n^2) because we are dealing with a string (a list of chars). There are only 256 diferent chars, so the list that nub s produces is never longer than 256. This makes the algorithm O(n).

I do believe the nub function itself is not that efficient, possibly O(n^2). So as long as we use that standard library function, optimization is pointless anyway :p

you mean len(odds) == 1