Re: [office-formula] Impact of daylight savings time changes

["David A. Wheeler" <dwheeler@dwheeler.com>]

Rob Weir said:
> The OpenFormula work -- these have plenty of time and date functions.  Is 
> there anything there where time zones are a concern?  In particular, 
> "4/2/2006" + TIME(1,1,0) results in 4/2/2006 2:01AM..

Nits: you have to stick in VALUE or DATEVALUE around the string for portability,
and the result would be 1:01AM, not 2:01AM (days normally begin at midnight).

> when really it would 
> be 3:01AM because of daylight savings time.  Do we want to be explicit 
> that date arithmetic does not take into account daylight savings time?  Of 
> course that has nothing to do with the law change. 

I thinking making an explicit statement would be a good idea.
It's certainly true in practice.

In spreadsheet formulas, date/time values are immediately converted
into a number representing "number of days (and fractional days) since the epoch".
Text representations can track timezones, but since they are immediately converted
to a single number (as a practical matter), tracking timezones is difficult.
I know of no spreadsheet program that explicitly tracks timezones. To my
knowledge, they all assume that "all dates and datetimes that are
used with each other are in the same timezone"
(which MIGHT be UTC, but not necessarily so).

I can imagine functions that provided uses with daylight savings time conversions,
and timezone conversions in general.  I would suggest that applications work
on developing them, and NOT dealing with that right now.
Once they've proven themselves, THEN add them to the spec.

--- David A. Wheeler