[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Remaining critical bugs (IMHO)

bruce@pixar.com (Bruce Perens)  wrote on 29.04.97 in <[🔎] m0wMQLo-00IdTWC@golem.pixar.com>:

> I think POSIX was rather arrogant in intentionally codifying this
> mis-feature.

In my completely unhumble opinion, their choice was the only rational  
option available.

Leap seconds are just a curiosity for more than 99% of all possible  
computer applications. On the other hand, reliable future dates are  
important for a lot of these, and with any accurate leap second counters,  
you'll lose there - there's no way you could predict leap seconds.

Leap seconds are fine if you keep time as, say, year-month-day-hour-minute- 
second tuples. They are a serious problem, however, when you keep time  
with a second counter. And that is what Unix does.

When the only way to say "one day" is "86400 seconds", you don't want to  
have 86401 second days at unpredictable dates.

Setting the Unix clock to count leap seconds is, IMO, broken - and because  
of the large difference, I'd consider that a critical error.

By the way, this error might cause mirrors to refetch the distribution -  
mirror depends on dates, even depends on getting the time zone right.

MfG Kai

TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .

Reply to: