Re: [office-formula] WEEKNUM reference

[Eike Rathke <erack@sun.com>]

Hi David,

On Thursday, 2007-02-22 10:47:00 -0500, David A. Wheeler wrote:

> Eike:
> > I just thought of the possibility that Excel might follow the idea of
> > enhancing the WEEKNUM functionality, but use a different numbering in
> > the parameters. Currently it generates a #NUM! error for values other
> > than 1 or 2 as the second argument, but in that case documents exported
> > to Excel would silently calculate different results. But maybe I'm too
> > precautious.
> 
> That's a valid concern.  Of course, that can be implemented as COM.MICROSOFT.WEEKNUM at that time, and eventually become a WEEKNUM_E or whatever.

My concern wasn't about how to write a new or incompatibly changed Excel
function to ODFF, but how to write the WEEKNUM as we defined it to Excel
instead. An app that wants to export the WEEKNUM function so far can
rest assured that if the calculation uses modes Excel doesn't know then
Excel will generate an error. If that changes, defining an additional
WEEKNUM_E later doesn't help.


> > > We can do much better, of course; we can implement a good WEEKNUM (with at least 0..7) and an ISOWEEKNUM so that the users can get whatever they need to get.
> > 
> > With the minimum-number-of-days parameter we could also drop ISOWEEKNUM,
> > since it could be written as WEEKNUM(d;2;4). Do we want that?
> 
> No, I think we want an ISOWEEKNUM as well.  There are already several implementations that _have_ an ISOWEEKNUM.  And in any case, it makes sense to have a function that makes it REALLY easy to access the ISO week number.

Ok, agreed.

  Eike

-- 
Automatic string conversions considered dangerous. They are the GOTO statements
of spreadsheets.  --Robert Weir on the OpenDocument formula subcommittee's list.