At 19:39 30/09/04 +0200, Tomasz Zielonka wrote:
What I like about GHC is that I can start from simple, high-level, sometimes slow solutions, but if there are efficiency problems, there is a big chance that I can solve them without switching the language.
That's a very good point, I think. One to hang on to.
But I did wonder if it wouldn't be possible to also abstract out the I/O element of your 'fileIterate', using instead something like: streamIterate :: [b] -> (a -> b -> a) -> a -> IO a
It seems to be a variant of foldl. You can eliminate IO from return type, or is there some reason for it?
Doh! (Why didn't I spot that?) All roads lead to Rome, or something like that? There seems to be a recurring tension between how much to specialize and how much to generalize. Maybe it should be something like: streamIterate :: (Monad m) => [b] -> (a -> b -> m a) -> a -> m a ? Er, but that's similarly a variation of foldM, right? Or maybe my earlier idea was closer: streamIterate :: (Monad m) => (c -> m (b,c)) -> c -> (a -> b -> m a) -> a -> m a ? Hmmm... I feel like a (intellectual) bull-in-a-china-shop here. I'm blundering about on the trail of a delicate and elegant idea that I'm sure others could dissect far more clearly. What I'm trying to capture (I think) is that there's some baggage to do with accessing the raw data and emitting the desired result that needs to be carried along (or interleaved) with the required computation on that data. Does this make any sense, or am I descending into farce here? #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact