Jim Wowchuk <[log in to unmask]> wrote:
>
>Okay, we've just gone onto daylight savings time here in Oz this weekend and
>today's the first day I've been able to check the times... :(  Several days
>ago I raised a question about the hard & soft clocks, and while the issue of
>how to see it was cleared up the relationships wasn't.
>
>For example:
>----------------------------------------------------
>//FRED/SYS/LIB: showclks.pubxl.telesup
>  *************************************************************
>  ***                                                       ***
>  ***   Greenwich Mean Time : MON, OCT 30, 1995, 11:08 PM   ***
>  ***   GMT/MPE offset      : +10:00:00                     ***
>  ***   MPE System Time     : TUE, OCT 31, 1995,  8:08 AM   ***
>  ***                                                       ***
>  *************************************************************
>
>                 **** C Library Information ****
>
>        Current value of Time Zone(TZ) variable : EST-10EDT
>        CTIME function return :  Tue Oct 31 09:08:17 1995
>
>
>//FRED/SYS/LIB: showclock
>
>SYSTEM TIME: TUE, OCT 31, 1995,  8:09:02 AM
>CURRENT TIME CORRECTION:            0 SECONDS
>TIME ZONE:   10 HOURS  0 MINUTES EASTERN HEMISPHERE
>----------------------------------------------------
>The first question is: if the GMT/MPE offset is +10, then why is the system
>time at 8:08AM and not 9:08 AM?
 
The system time is whatever SHOWCLOCK says it is regardless of what the time
zone is or the SHOWCLKS utility says the hardware clock is. All times, once
the system is up and running, are based on the system time.
 
The SHOWCLKS display above could be the result of doing a SETCLOCK once the
system is running. A SETCLOCK with the DATE;&TIME; options will only affect
the system clock. A SETCLOCK with the ;TIMEZONE= affects both the time zone
and the system clock. Neither affect the Hardware clock directly. Much is said
about this in the Communicator.
 
>should make it 9:00am Tuesday.  I believed that the offset was a calculated
>figured derived from the difference of the hardware (GMT) and software
>(System) time.  Now I wonder if all three are stored discretely.  Can anyone
>offer an explanation (other than I am no longer mindful of addition :)
 
The system time is only calculated based on the hardware clock at system start.
At that time the hardware clock(which is set by the ISL utility CLKUTIL) and
the newly entered system time is used to calculate the time zone. After the
system is up and running the hardware clock is not referenced again. All
calculations are done based on the software clock. The SHOWCLKS utility in
TELESUP uses internal routines to access the hardware clock but that is not
SOP.
 
>I suspose another possibility is that SHOWCLKS.PUBXL.TELESUP is not
>reporting the GMT time correctly.  This will, I believe, require a re-boot
>to check, so I'm loathe.
 
My experience shows that SHOWCLKS does report the hardware, software, and
time zone correctly but they may not always calculate as you would expect.
You do not have control over the hardware clock except at ISL.
 
>BTW, the CTIME result is correctly adjusting the time one further hour (or
>maybe its not %), as the TZTAB file it uses is identical to our unix system
>which reports the time correctly.
 
The CTIME call is actually calculating the "correct" time only because your
system time and time zone are both wrong. Let me explain how I understand
CTIME to calculate the time. Again, the calculation starts with the actual
system time, 8:08 am in this case. It then adjusts for the time zone you see
with a SHOWCLOCK, +10:00 hours here. CTIME has calculated the UT to be 10:08
pm based on this calculation. Now it uses the TZ variable to recalculate the
correct time based on it's calculated UT and the info in TZTAB. In your case
it would adjust UT(10:00 pm) by 11:00 hours giving you the correct time of
9:08 am.
 
You can still correct your system time and time zone by doing another series
of SETCLOCK commands:
     SETCLOCK TIMEZONE=E11:00
     SETCLOCK CANCEL
     SETCLOCK DATE=11/??/95;TIME=current system time(not UT)
 
Depending on what direction the clock is being adjusted you may see a gradual
clock adjustment but all should be well when that is done. You will still
probably want to restart the system sometime and set the hardware clock to UT.
(I have heard that some processors will automatically sync the hardware clock
in 24 hours so I would check it with SHOWCLKS.MPEXL.TELESUP before taking the
system down.)
 
>Finally, it was interesting to see that when powering up my PC, Windows 95
>dutifully reported a time change for daylight savings.  Unfortunately, I was
>away from my desk after booting up (as is my want now that Win95 takes so
>long to boot up).  It _appears_ but I can't confirm that while waiting it
>stopped the clock.  Thus when I affirmed the change, the clock was in fact
>now near 30 minutes behind. Did anyone else share this experience?
 
I had left my Win95 system on overnight and had the message the next morning.
My clock was off ~30 minutes!?
 
Enjoy, :)
 
----------------------------------------------------------------
Virgil Hovar            Internet:[log in to unmask]
Technical Consultant    Voice:   tn/309-664-4055
Hewlett-Packard Co.     Fax:     tn/309-664-4100
================================================================