Re: XIRR Error State

[Morten Welinder <terra diku dk>]

The reason you cannot get it to work is, mostly, that it does not work...
There were two things wrong:

1. gnumeric_xirr needs to set sane xmin and maybe xmax.  I used -1 and
   +1000.  (The calculations involve (1+r)^t, so anything less than -1
   is problematic.)

2. Goal seek for the case without df was not good at picking an xstep.
   It used to be wrong in a different way, but I think I have all cases
   covered.

Sorry about the mess.

Morten