Time zones can be an issue for time intervals. For example, an airline flight from Houston, Texas to Atlanta, Georgia. The difference of departure and arrival times do not give the correct travel time. As you say, changes of Daylight Savings Time cause problems. I had the problem of generating reports on computer jobs being run in London and in Houston and reports had to be given to both areas on jobs run in both areas, but the reports had to be given in the local times of each area. I converted all local times by the various computer systems accounting to GMT, combined the data, then generated reports converting back to local time. Daylight Saving Time didn't cause problems going forward, but when it went back the accounting did not report 2:30AM as before the change to Daylight Savings or after. Jobs ended before they started. An IMS system we had solved the problem by requiring that it be shut down over that ambiguous hour. Otherwise its log file would be corrupted.
And to just make things a little more sticky, Great Briton's legislature determines when the changeover occurs. They wouldn't do that more than a year or so in advance. Or at least they didn't fifteen years ago. Sometimes it's tough when systems would only reboot every six months or so. Maybe we ought to go to syderial time and forget it. On Sun, Jan 4, 2009 at 6:56 PM, neville holmes <[email protected]>wrote: > Don Guinn wrote (in respect of my suggestions for primitive > representation of dates and time intervals): > > > What about time zone? > > Time intervals are independent of time zone. Dates are > of course dependent on time zone, and have been all along. > But differences of dates (time intervals) are independent > of time zone. > > Perhaps a more important geographical consideration is the > dependence of time intervals on idiosyncracies like switching > to and from daylight saving time. Arithmetic of time intervals > would ideally handle such matters. Here in Australia the > treatment of daylight saving differs from state to state and > is even done in two different ways in New South Wales. > > These complications would surely be better handled as > I suggest by the J interpreter rather than having them done > by functions that are complex to find and use and keep > up to date, and in any case are a trap for the unwary who > need protection. > > Incidentally, the term "time interval" seems to me to be > too complex itself. Why not just "interval" ? My Oxford > English Dictionary confirms that this is usual, and even > denigrates as "formerly" the phrase "interval of time". > Of course there are spatial intervals (and this was the > original meaning of "interval" in Latin) but these are far > less frequent in ordinary life and there the term "spacing" > is normally used anyway. > > Neville Holmes, P.O. Box 2412, Bakery Hill 3354, Victoria > > > Stay connected to the people that matter most with a smarter inbox. > Take a look http://au.docs.yahoo.com/mail/smarterinbox > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
