RE: [office-formula] 6.9.13 NETWORKDAYS

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

On Tue, 2009-02-24 at 19:15 +0000, Eric Patterson wrote:
> Looking at NETWORKDAYS further, I see that the 4th parameter, Workdays is defined as a list of logical values for the days of week, starting on SUNDAY.
> 
> ISO8601 defines a calendar week to be a "time interval of seven calendar days starting with a Monday".
> 
> I propose that OpenFormula should use the same convention in this case.

I would agree with this. Is there currently an implementation that
implements the fourth argument? Excel, Gnumeric and Openoffice appear
not to.

Andreas

> 
> The amended test cases would be as follows.  Note, the results of the last 2 were incorrect in the current draft.
> 
> Expression                                                                                              Result
> =NETWORKDAYS(DATE(2007;1;1),DATE(2007;1;12))                                            10
> =NETWORKDAYS(DATE(2007;1;1),DATE(2007;1;12);DATE(2007;1;1))                             9
> =NETWORKDAYS(DATE(2007;1;1),DATE(2007;1;14);DATE(2007;1;6))                             10
> =NETWORKDAYS(DATE(2007;1;1),DATE(2007;1;14);;{0;0;0;0;1;1;0})                   10
> =NETWORKDAYS(DATE(2007;1;1),DATE(2007;1;14);DATE(2007;1;6);{0;0;0;0;1;1;0})     10
> 
> -Eric
> 
> -----Original Message-----
> From: robert_weir@us.ibm.com [mailto:robert_weir@us.ibm.com]
> Sent: Monday, February 23, 2009 7:04 PM
> To: office-formula@lists.oasis-open.org
> Subject: Re: [office-formula] 6.9.13 NETWORKDAYS
> 
> Eric Patterson <ericpa@exchange.microsoft.com> wrote on 02/23/2009
> 07:47:51 PM:
> 
> >
> > Reading the definition of the NETWORKDAYS() function in the current
> > draft of the specification, I see that the function definition
> > includes an optional 4th parameter.  Microsoft Excel's definition of
> > this function only includes 3 parameters.
> >
> > I'm trying to understand the implications of this.  If the intent of
> > OpenFormula is to create a portable syntax how should differences
> > like this be handled?  Implementers that only use the first 3
> > parameters could share documents, but what is expected if the 4th
> > parameter is included in a document?  I would expect it to return an
> > error value to provide a warning to the user.
> >
> > What effect would implementing only the first 3 parameters have on
> > application and document conformance?
> >
> 
> The level of granularity is the function.  Think of it like a .NET
> interface.  You either implement it or you don't.  The contract is
> IEnumerable and if you want to say you implement IEnumerable then you
> implement all of IEnumerable.  Same with NETWORKDAYS().  You either
> implement it or you don't.  There is no 75% conformance at the level of a
> function.
> 
> IMHO, user expectations are (in order):
> 
> 1) That the spreadsheet will calculate properly in Excel, to give the same
> results it did in the original application.
> 
> 2) That if Excel lacked a particular function or parameter in the
> function, that it would fail in an obvious (noticeable) way rather than
> silently give a wrong answer.
> 
> Of course, you are free to be conformant and meet user expectations at the
> same time.  Simply add support for that 4th parameter.  I know we've heard
> from users, especially in the Middle East, that the work-week assumptions
> of the 3-parameter Excel function did not work well for them, since their
> legally-defined weekend was Th/Fr or Fr/Sa and the 3-parameter function
> would yield incorrect results for them.
> 
> -Rob

-- 
Andreas J. Guelzow
Concordia University College of Alberta