On Wed, 22 Aug 2007 15:12:14 +0100, Perryman, Brian wrote: > >We have some issues with a program which was recently rescheduled to run just >after midnight (British Summer Time) - but the date is a day out.. I'm >wondering if it's because it's an old program getting the date in the wrong >way somehow (that we can fix), or if that's just the way Cobol works..?? > Of course, when the semiannual time change occurs, I just reset my watch; I never find it necessary to adjust my calendar.
I have encountered some extraordinarily naive code that didn't take the time zone into account when calculating the date (and a developer who insisted it was unnecessary to do so). Can you try a few probe jobs late at night to see if the date is generally reported for GMT while the time is reported for Summer Time? Even if that's "just the way Cobol works", it still ought to merit a PMR. It's even possible that your scheduler fails to take leap seconds in to account, causing a job the scheduler believes it's starting one second after midnight to start a few seconds before midnight (the previous day), UTC. Further, there's cause to suspect that the date and time are computed independently, leaving you yet exposed to the Midnight pitfall: If one of the date and time is obtained a microsecond before midnight and the other a microsecond after, the aggregate time will be nearly 24 hours in error. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
