On 25 January 2005 14:25, Simon Marlow wrote:

my plan was that the whole "time offset" issue would be a layer on top of this library, as others have suggested. You can already do simple calculations using CalendarTime.

One other thing I just remembered: IIRC there was a proposal to handle the whole timezone offset issue with a single function normalizeCalendarTime :: CalendarTime -> CalendarTime which would essentially just roll over the out of range fields in CalendarTime. This is essentially the same as the idea Marcin proposed in his message, I believe. Of course, normalizeCalendarTime needs to know about leap seconds (some minutes don't have 60 seconds!). Cheers, Simon