[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [office-formula] Who's doing what this week,Redux (including a request to Stefan Nikolas)
Hi David, On Wednesday, 2007-02-14 13:36:16 -0500, David A. Wheeler wrote: > David A. Wheeler wrote: > > > Robert Weir has volunteered to do NETWORKDAYS and WORKDAY, since they > > > have the same I18N issues. Basically, he's going to need to add an > > > additional optional parameter to handle internationalization > > Eike Rathke responded: > > If we add a parameter that no application is supporintg yet I suggest to > > define it the way that if the parameter is not given comply with the > > Ecma/Excel definition. Eases interoperability. > > Oh, absolutely, that was the reason I want it optional. Sorry I didn't make that clear. It's really not just Ecma/Excel; I think everyone does it the same way. Sorry, I didn't make that clear myself. I would not define it as something like "if omitted or 0 do like Excel", but instead "if omitted do like Excel, if present don't". So users are not tempted to have a calculated value that may act like Excel or may not, which is difficult to export when they expect that a calculated 0 should export fine to Excel. Of course with a flexible parameter there will always be the possibility to calculate it such that it acts identical. Just that we don't raise the expectation it could be interoperable with Excel. > > > - not EVERYONE's weekend is Saturday/Sunday. I see two major options: > > > (1) a parameter saying which day of the week begins a two-day weekend, > > > or > ... > > > (2) a parameter that's a boolean representation of the days of the > > > week that are to be considered non-workdays. Option #2 is more > > > flexible, but also is harder to implement and explain... does anyone > > > care which way he goes? > > > Eike adds a #3: > > It may also be necessary to include half days, as in "Saturday is half > > day off, Sunday one day". The most flexible IMHO would be to have an > > array with a factor in the range [0,1] for each weekday, if that's not > > considered overkill. > > Eek. Actually, you can even pile on the overkill further; many people in the U.S. work in 2-week cycles, so they work 5 days one week and 4 days in the alternate week. That would add yet another option: > > #4: The new optional parameter accepts an array; if 14 days instead of 7, it that describes a two-week cycle. Anyone coming up with a 4-week cycle? ;-) > I think #3 and #4 are more flexible but would be much harder to use. You could partly address that yet another option: > > #5: Merge #2, #3, #4. Basically, detect if the new parameter is a scalar number, and if so, interpret as #2. If the new param is array/multi-cell range, then interpret as #3 or #4 (depending on its length). That might make common cases easier to use, and still be very flexible - but it's more work to implement. > > I have to admit that #3, #4, and #5 seem awfully like overkill to me. Does anyone have an opinion on what is best for our purposes? Even if overkill, #5 actually sounds quite nice to me. If we don't want #3, #4 and #5, I would suggest #2. Users sometimes have a hard time though to figure out bit values.. Eike -- Automatic string conversions considered dangerous. They are the GOTO statements of spreadsheets. --Robert Weir on the OpenDocument formula subcommittee's list.
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]