Hi, several people have noted that printing integers is slow. The attached patch implements a faster version of the jtos (read: positive integer to string) function in GHC's Num library. The algorithm is a divide and conquer algorithm, that converts numbers to base b by converting them to base b^2 first, and then splitting the digits. Notes: - I've tested the speed on my computer (which has an Athlon XP processor) and it seems to be of comparable speed as the original for small numbers and way faster for big numbers (read: a few hundred or thousand digits). - The changed conversion code is a much better list producer than the original one, which generates digits starting with the last. - There's similar code in Numeric.lhs, that deals with arbitrary bases. The current patch does not deal with this. (see questions below) - musabi on #haskell mentioned that there's interest in replacing GMP as the bignum implementation. This is independent of changing this string conversion, as far as I can see. Questions: - Is this really GHC specific? - Unfortunately I see no nice way to share code between the Numeric and Num modules. Does anyone have an idea how to do this without sacrificing performance for the common base 10 case? regards, Bertram