On Wed, 17 Jun 2009 10:38:23 +0200, Yitzchak Gale <g...@sefer.org> wrote:
While there are indeed certain very rare situations in which
you want foldr or foldl for sum, they are both joltingly wrong
as the default for typical usage.
In practice, I find this to be a small annoyance that occurs
so often that it becomes a major one. Can we please fix it
already?
Let it be noted that this discussion also applies to product.
Thanks,
Yitz
I have done some research on functions in the base libraries, whether they
can handle large lists; I didn't check them all, because there are so many
of them. sum and product are certainly not the only ones having problems.
For example:
Hugs> reverse [1 .. 999999]
ERROR - C stack overflow
An improved reverse function:
reverse' = foldl' (flip (:)) []
There is no need for reverse to be lazy, so this one could replace the
original one.
reverse' is not too strict:
Data.List> let reverse' = foldl' (flip (:)) [] in head $ reverse'
[undefined, 1]
1
Some of the other functions that have problems with large lists are:
foldM
maximum
minimum
scanl
scanr
scanr1
iterate
take (in GHCi, not in Hugs)
drop (in GHCi, not in Hugs)
splitAt (in GHCi, not in Hugs)
inits
By the way, sum and product are implemented with foldl' in Hugs.
--
Met vriendelijke groet,
Henk-Jan van Tuyl
--
http://functor.bamikanarie.com
http://Van.Tuyl.eu/
--
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
