Rich Bowen wrote: > leap seconds seem kinda arbitrary. Is that just a table that you look up > in?
Basically, yes. Just like leap years, but there is no rule (there is no exact calculations for it). Some (rarely) days will have one 0-60 seconds minute (instead of 0-59). See also: Date::Leapsecond - it's just a small table Dave Rolsky wrote: > We need to handle them. Flavio claims that using floating point seconds > makes this easier. Sorry, I was not talking about floating point seconds. I was talking about Date::Set. I think it could handle Leap seconds just like it handles timezones (that is, specify some rules, apply them). It will, of course, be slower than libtai64 just because libtai64 is 'C'. - Flávio S. Glock