On Fri, 31 Aug 2018 15:30:29 -0700, Charles Mills wrote:
>From The New York Times:
>
>E.U. Will Let Countries Decide Whether to Use Daylight Saving
>
>Some countries have lobbied to end the requirement that all 28 member states
>spring forward and fall back each year.
>
> https://www.nytimes.com/2018/08/31/world/europe/eu-daylight-saving.html
>
Is it likely that enough countries will repeal to start an avalanche? The plan
for local option rather than uniform repeal suggests that even the proponents
are doubtful.
And if even a significant minority retains the practice, IT is obliged to
continue
to support it.
Mainframers feel particularly intense about this topic. I believe that the
cause
is that z/VM and z/OS have such woefully primiitive support of timezones --
there are really more than the two that the TIME service supports.
A correct implementation does not update system control blocks such as CVTLDTO
or configuration files semiannually, but only when the convention changes; the
most recent such change in the US was in 2007.
TIME, STCKCONV, and CONVTOD should be enhanced to support multiple
timezones and to convert historic timestamps correctly.
Critical timestamps should be stored in GMT (UTC, whatever) and adjuted for
display by that enhanced STCKCONV. A conspicuous offender is ISPF which
(still, I believe!) keeps member timestamps in "local" time.
An interesting phenomenon, similar to Parkinsons Law of Triviality:
https://en.wikipedia.org/wiki/Law_of_triviality
... we adjust our time base by 24 hours every 4 years and few complain (ignoring
three major world religions. A bitterly divisive change was made around 1572.)
We adjust our time base by one hour semiannually and IT professionals complain
bitterly. We adjust our time base by one second every couple years and servers
crash.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
