Charles Lane wrote:
> Peter Prymmer wrote:
> > Charles Lane wrote:
> >> BTW, I'm seeing failures of the vmsish test #17 that might be due to being
> >> east of greenwich. This'll be a good opportunity to debug 'em.
>
> > I am curious: Are you at CERN or DESY now? What is your
> > SYS$TIMEZONE_DIFFERENTIAL?
>
> Gran Sasso lab in central Italy. Debugging Perl from 1km underground :)
Grazie. Looking for cosmic or artificial neutrinos? Then your
system time setting is going to be rather important isn't it?
> SYS$TIMEZONE_DIFFERENTIAL = 7200
> SYS$TIMEZONE_DAYLIGHT_SAVINGS = 1
> SYS$TIMEZONE_RULE = "MET-1MET_DST-2,M3.5.0/2,M10.5.0/2"
> SYS$LOCALTIME = SYS$SYSROOT:[SYS$ZONEINFO.SYSTEM]CET.
> MULTINET_TIMEZONE = CET
> MULTINET_TIMEZONE_RULES =...BUNCH OF STUFF..
>
> I think multinet is setting the time via NTP, but the localtime stuff for
> the CRTL is messed up.
Watch out for DCE DTSS running. It'll muck up either UCX or Multinet
NTP and would have to be turned off.
> You do "print scalar(localtime)" [ or the C equiv] and it's okay.
> But the C function tzset says the TZ offset is 3600, and stat is getting
> times that are one hour off from what DIR says they are.
>
> I do a :
> perl -e "print 'time is: ',(time()%86400/86400*24),' hours';"
> and get 22.3... hours (it's 10:20 PM here). But it *should be 20.3... for
> greenwich time -2 hrs from local.
Hmmm... Here is what I just obtained:
$ perl -e "print 'time is: ',(time()%86400/86400*24),' hours';"
time is: 21.6494444444444 hours
$ sho log sys$timezone_differential
"SYS$TIMEZONE_DIFFERENTIAL" = "-25200" (LNM$SYSTEM_TABLE)
$ sho time
8-MAY-2000 14:39:29
> So I don't think it's a Perl issue...it's a sys configuration issue.
> But the pathology is complicated enough that I doubt it will be fixed
> anytime soon.
>
> Maybe we need to start a "what these test errors mean" document, listing
> the errors we've run across on VMS, fixes or workarounds, severity, etc.
>
> It might help for the next N times the "glob-basic" test error shows up.
That could help. My web page originally tried to do that, and now
Dan is carrying the torch. Also, the various README.$osname files
tend to collect such info.
Peter Prymmer
