On 2016-12-13 17:26, Tim Parenti wrote:
On 2016-12-13 19:12, Paul Eggert wrote:
Officially, in the 1918 and 1919 fallback transitions the clock for the long day kept going after 24:00, reaching 24:59 the previous day before falling back an hour to 00:00 the next day. As this behavior is not representable in this database, it is modeled instead as the clock reaching 00:59 the next day before falling back to 00:00. Er, remove "the previous day" for consistency about which day we're talking about. ;)
...and check the day you end up with is Sunday, unless explicitly documented otherwise, and even then, be very sceptical ;^> We encountered this in an earlier discussion, about war time Italy IIRC, where there was ambiguity in a similar "definitive" document containing non-explicit tabular data, resolvable by checking the weekdays on which time changes fell. For that and similar reasons, I now detest (standard?) date formats that do not include or allow day of week, as a quick reasonableness check, as well as UTC offset, for some certainty of the time meant. -- Take care. Thanks, Brian Inglis, Calgary, Alberta, Canada