I was too hasty to complain on a bug in ghc-0.2-linux and to send a letter yesterday with the Gauss method script. I am sorry. It may well occure that this is not a bug. It occured to me that this version of the method is very unstable. Hence the small diference between my divMod (positive remainder) and Haskell's quotRem may lead to enormous difference in the staircase matrices. The thing needs more verification. Sergey Mechveliani [EMAIL PROTECTED]
