On 4/8/2016 5:31 AM, Rob van der Heij wrote:
> I did not want to bother the rest of the IBMVM list with this, but I could
> use some votes on what we expect DELAY to do after a SET TIMEZONE was issued.
For an interval, I would certainly expect it to wait for the specified
duration, irrespective of time zone.
For a time of day, when the clock is turned back, I would expect it to
wait until the specified time in the new time zone. When the clock is
turned ahead, I would expect unpredictable results based on whether the
pipeline had done anything since the time change. I wouldn't expect
DELAY to wake up every second or even every day to check the time zone,
or to guess when the time change was scheduled (tzdata notwithstanding).
I also like the option of specifying a time zone and/or offset on the
time of day. I'd want some generic designation for the *current* active
time zone (I'm in EDT now, wake up at 3AM EDT regardless of whether we
change back to EST). To specify an offset from the default, active time
zone, we'd also need some generic designation for that. Say it was *
for the default and = for the current:
03:00 -04:00 --wake up at 3AM EDT
03:00 *-02:00 --wake up at 3AM two zones west of here
03:00 = --wake up at 3AM in the time zone I'm in *now*,
whether or not the clock changes before then
It would be nice to accept offsets in all formats valid to DEFINE
TIMEZONE or ISO 8601, with or without a blank before the time zone or
offset. East, West, and Z could conflict with valid CP time zones, though.
¬R
