Re: [office-formula] DATEVALUE and TIMEVALUE with full DateTimesupport

["Andreas J. Guelzow" <aguelzow@math.concordia.ab.ca>]

On Wed, 2007-21-03 at 21:58 +0100, Eike Rathke wrote:
> On Wednesday, 2007-03-21 14:48:32 -0500, David A. Wheeler wrote:
> 
> > 100% agree, on all counts.
> > 
> > However, I think there needs to be a way to access the value without the chopping-off behavior.  I believe that everyone's VALUE() implementations already do this; is that correct?
> 
> At least it SHOULD. Excel and OOo do, but Gnumeric 1.4 and Kspread 1.4
> don't. So far we didn't require that. Should we make that a SHOULD or
> a SHALL?

If Excel and Gnumeric indeed differ in some basic function behaviour
(where Excel is not mathematically wrong) there is no problem with
respect to Gnumeric to go with the Excel behaviour. If we want to have a
different behaviour we will use a different function name and change our
function to correspond to Excel's behaviour. Essentially Excel and
Gnumeric will differ for one of two reasons:

1) Excel is simply mathematically wrong. (In that case we may want to
retain Gnumeric's behaviour, but we have been known to implement the
wrong version plus a corrected Gnumeric version)

2) Our testing has not shown the difference. If (1) doesn't also apply
we are very willing to change Gnumeric's behaviour.

Of course I really can only speak for myself but I am quite sure that
there is general agreement on this among Gnumeric developers.

Andreas 

-- 
Andreas J. Guelzow, Professor
Dept. of Mathematical & Computing Sciences
Concordia University College of Alberta