Re: Bug in round function

[Morten Welinder <terra diku dk>]

(I notice that this email was sent December 19.  There must have
been a delay somewhere.)

Anyway, this has been "fixed" in cvs.  I say "fixed", because there
really is no good solution to problems like this that are due to
finite-precision math.  But current cvs code should act more like
most (non-comp-sci) users expect.  This was done by moving the
rounding break from (say) 1.555 to ever to slightly less.

Morten

   Date: Tue, 19 Dec 2000 21:38:11 +0100
   From: Pere Pujal Carabantes <ppujal hola fou net>

   Hello!
   I am trying gnumeric and see that:

   =round(1,5;0)  gives 2 --> correct
   =round(1,55;1)  gives 1,6 --> correct
   =round(1,555;2)  gives 1,55 --> Not correct
   =round(1,5555;3)  gives 1,556 --> correct
   =round(1,55555;4)  gives 1,5555 --> Not correct

[...]