hello Andreas, thanks for your input,
hmmmm ...
not too easy to deal with excel, either i'm confused or they contradict themself.
support.microsoft.com - FLOOR function:
either this: I.) 'Rounds number down, **toward zero**, to the nearest multiple of significance.',
(think that's mimiced by gnumeric),
or that: II.) 'If the sign of number is negative, a value is rounded down and adjusted **away from zero**.'
(think that's mimiced by LO Calc),
or that: III.) '=FLOOR(-2.5,-2) Rounds -2.5 down to nearest multiple of -2. **-2**'
sample with behaviour like I.)
excel requires the input of a significance parameter, in gnumeric you can omit it, then '1' ( or partly -1 ) is taken as default,
elsewhere - odf export - gnumeric solves the problem by specifying a 'fake' parameter 'sign(x)'.
This way seems to make more sense to me since it:
1. preserves the function name,
2. gives the possibility to get back the original _expression_ when re-converting to gnumeric formats by omitting the added sign(x).
( Once the function name is changed there is no way to tell if the origin was a floor or rundown _expression_. )
Why I am looking for a better solution to this rather small problem? In the too many messages that the tests give out,
some small things get lost ( '
Fefe calls such a 'bug wave' ), or it is tedious and time consuming to check the effects of changes.
In this respect, any help is welcome even if my suggestions do not always fit, I'm just searching.
b.
