On Sat, 4 Sep 2010 23:38:37 -0700, Ed Gould wrote: >http://www.pcworld.idg.com.au/article/358024/time_waits_no_one_leap_seconds_may_cut/ > Which says:
In the revised ITU plan, the divergence between UTC and UT will be allowed to grow over the next few hundred years, and could be reconciled by a single leap hour at some point. We have recognized the problem and squarely turned our backs to it. z/OS now idles user tasks down for a second at positive leap seconds. Is this an impact? I'm sure an hour would be unacceptable. z/OS has a half-hearted approach to leap seconds in idling user tasks. They implemented leap seconds but simply didn't have the courage to allow the TIME functions to return times such as 23:59:60 as specifed by the definition of UTC. UT1 is smoother than UTC; corrections from UTC to UT1 are issued daily with a granularity of 0.1 second. I'd propose an even smoother approximation, a chordwise version of UT1, with parameters published sufficiently in advance to support conversion of smoothed UT1<-->TAI for those rare cases where TAI is required. Of course, the mapping is undefined for more than a few months in the future. UNIX is worse than MVS. POSIX effectively prohibits any recognition of leap seconds in common time manipulation functions. NTP has a bizarre but probably satisfactory accomodation to leap seconds in shifting the epoch origin forward whenever a leap second is inserted. and cites: For instance, NTP itself can accommodate leap seconds by use of a parsable file of leap seconds that can be downloaded from the National Institute of Standards and Technology. Our site jumped aboard leap seconds when we got a Sysplex Timer. We abandoned them a year or two later beause of incompatibilities with ISV software. The problem was not with the disruption every year or two, but because vendor products were inconsistent in employing the correction in CVTLSO. Here I blame not the design of leap seconds, but vendor carelessness in their adaptation. Somewhere I saw a quotation to the effect that "It is unreasonable to expect that a time specified as a number of seconds since the epoch should reflect the actual number of seconds since that epoch." I've lost the source. I'd be delighted to recover it. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
