[ntp:hackers] Leap second support in versions prior to 4.2.6

Mike S mikes at flatsurface.com
Mon Jun 4 19:21:18 UTC 2012


On 6/4/2012 2:30 PM, Hal Murray wrote:
>
> mikes at flatsurface.com said:
>> I think you're complaining about systems which don't do it right, not  leap
>> seconds themselves.
>
> But POSIX pretends that leap seconds don't exist, so that's a lot of systems.

POSIX also pretends that the current definition of UTC existed in 1970. 
It obviously wasn't well designed, from a timekeeping perspective.

Since it's present on so many systems, that merely argues that code 
written to run on a POSIX system should be written to handle its 
timekeeping peculiarities. Putting this burden on application 
developers, instead of in the OS/utils, is another poor design decision.

> Are there any common OSes and/or APIs that do it correctly and cleanly?

If I understand things correctly, you can use the "right" zoneinfo TZs 
on Linux (and some other *nix) OSs to do so (making them non-POSIX in 
this regard). You just need to keep the zoneinfo files updated so they 
contain new leap second info.


More information about the hackers mailing list