RE: [office-formula] Conflict between Part 1 and Part 2

["Andreas J. Guelzow" <andreas.guelzow@concordia.ab.ca>]

On Thu, 2010-02-04 at 16:41 -0700, Eric Patterson wrote:
> I think that puts us back where we started.  We use the null-date and
> acknowledge that dates between 1-Jan and 28-Feb may be different.  No
> further attributes are needed.  Anyone disagree to this strategy?

I do. 

I think an application should be able to legitimately discover whether a
file it opens was created under the assumption that there is a skip in
the number->date mapping before 1900/3/1 or not.

As far as I am concerned having the standard suggest that one falsifies
the null-date when saving is unacceptable. 

If for whatever reason people object to discoverability and want the
difference in serial numbers for dates prior to 1900/3/1, then the
null-date ought to be defined differently, by perhaps giving the date
corresponding to serial=100.

Andreas


-- 
Andreas J. Guelzow <andreas.guelzow@concordia.ab.ca>
Concordia University College of Alberta