– | primitive strict primStrict :: (a -> b) -> a -> b
strict f x = x `GHC.seq` f x

– | primitive strict primStrict :: (a -> b) -> a -> b
strict f a = let b = f a in b `GHC.seq` b

Neil Mitchell says that “x `seq` x” is a mistake. See http://neilmitchell.blogspot.com/2008/05/bad-strictness.html

It seems to me that the definition of scanl’ in gofer was sort of bugged, in that it’s still too lazy to be of any use. No strictness is actually forced because it’s buried under a lazy cons.

> scanl’ (+) 0 [1..] !! 1000000
*** Exception: stack overflow

My understanding is that seq has to be used at the top level of the expression to have effect.
If I use something like this instead

scanl” f q (x:xs) = y `seq` (q : scanl” f y xs)
where y = f q x
scanl” f q [] = [q]

it works, although this seems a little too eager since we are forcing a value we may never use.

Gofer was also my gateway drug leading to Haskell addiction. And you examples bring up some good old memories. In school we had an assignment to write a hangman program in Gofer using the continuation-based IO. I remember it as being quite painful. I’m glad it didn’t scare me away though. Not long after I found a tutorial on monadic IO which was some very welcome news.

It’s funny how the Gofer prelude still has its place somewhere in the back of my head. Even to this day I sometimes look for functions like sum, product and copy in the Haskell Prelude only to find that they aren’t there.

Thanks again.