[time-nuts] Leap Quirks

Poul-Henning Kamp phk at phk.freebsd.dk
Sun Jan 4 11:18:37 EST 2009

In message <4960CE3D.4080005 at erols.com>, Chuck Harris writes:

>> time_t is better defined as:
>> 	d * 86400 + min(rs, 86399)
>> where:
>> 	d = Number of complete days since 1970-01-01H00:00:00Z 
>> 	rs = number of seconds since UTC midnight.
>> Eliminating leapseconds would make it correct however.
>I believe your use of the phrase "make it correct" shows your bias
>towards removing the leap second corrections from UTC.  This is my
>bias as well.

Correct, I don't think leap seconds provide any benefit that
come close to the problems they cause, and I see the cost of
fixing POSIX and auditing all relevant apps after such a change
as totally pointless time & effort spent.

Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

More information about the time-nuts mailing list