At 11:40 +0000 on 07/01/2015, Vernooij, CP (ITOPT1) - KLM wrote about
Re: Leap Second today!:
One of the particular appearances of the problem was, that the
software could not cope with the 60th second presented by time
servers in the last minute of the day.
These program are IMO BAD (Broken As Designed). It is not that hard
to write a Sanity Check. A leap second can only occur on June 30 or
December 31. When a program starts, it can check the date and set a
flag for the routine that will go bonkers if there is a 23:59:60
+0000 as opposed to 00:00:00 one second after 23:59:59 to warn it
that a leap second MIGHT occur that day.
A better solution would be to have the program read a record with the
date of the next leap second (once it is announced - there is
adequate warning to update the record. Until then it is the last leap
second) to set the "It is going to occur TODAY" flag.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
