Jan Brosius <[EMAIL PROTECTED]> wrote > compared with Ocaml and SML , Haskell is a more sophisticated and more > elegant language. Does anybody know how efficient the Haskell compiler is > with regards to impure functional languages. > will Haskell compiled programs be faster by using more strictness > annotations; I especially think of Clean where strictness annotations are > "abundantly" used? I have programmed a lot of computer algebra in Haskell and found that there exist Haskell implementations which are, generally, on average, good in performance. Though, i never compared the performance on the same algorithm. For example, i expect that programming the simplest `merge' sorting in Haskell and in SML, Scheme, would not make much difference in performance. On the other hand, the programmer is sometimes bitten by the so-called un-needed laziness. For example, programming sum :: Num a => [a] -> a, minimum :: Ord a => [a] -> a, and such, in different styles, it is hard to predict which of the programs will take the run-time space proportional to the given list and which take the constant space. Iteration often turns into recursion when evaluated lazily. Strictness annotations in the data declarations (`!' - fields) often increase (a little) the efficiency, only need to be set in the right places, there is a simple consideration where to set them. While setting the strictness in the *functions* leads to that you would, probably, have a strict copy almost for each function in you program system (this is my impression, i am not very sure). Then, is not it better to write the whole thing in SML ? To my mind, the only (?) purpose of the laziness is to make the program look simpler, more algebraic. Setting the strictness in the functions destroys this effect. ------------------ Sergey Mechveliani [EMAIL PROTECTED]
