Re: Haskell 98 draft report

[Lennart Augustsson]

> It looks odd to me too.   I think this is just taken from some other
> standard, so I don't propose to alter it.
Were you asleep during your numerical analysis classes? :-)

If you always round x.5 up you will get numbers that are a little to
big on the average since x.5 is exactly halfway inbetween.  To remedy
this it's been agreed that you should always round to even in those cases.

     -- Lennart