|
HP OpenVMS Systemsask the wizard |
|
The Question is: This problem comes twice a year. After time change between daylight savings and standard, system logical SYS$TIMEZONE_DIFFERENTIAL needs to be set manually. Otherwise, C function localtime() and system time will not agree. However, C function gmtime() wil l return the proper UTC given the right system time. My understanding is gmtime() = system time - time offset using zone rules pointed to by logical SYS$TIMEZONE_RULE, and localtime() = gmtime() + TDF as defined by logical SYS$TIMEZONE_DIFFERENTIAL. The offset from zone rules is calculated based on current date/time, which will always be correct give proper system time. TDF requires manual adjustment. If TDF does not agree with offset, localtime() will obviously not the same as system time. If rules exist, why do we as system administrators need to maintain another TDF which is subject to human error? Can the next VMS release or C runtime use only rule-based offset for ALL time functions? There are applications (and we won't know which until problems occur) relying on proper localtime(). We have NTP / DTSS to set time. We have zone rules to tell the offset. Get TDF out of the way, please. I also notice the TDF logical exists starting VMS 6, but C runtime does not use it until VMS 7. So why is it there? Can it be ignored / deassigned on VMS 6.x? The Answer is : Timezone rules are available in OpenVMS, the current C RTL already uses the rules, an automatic switchover to daylight savings time is presently available via DECnet-Plus DECdtss, and an OpenVMS-integrated (and optionally enabled) switch-over and better documentation of the daylight savings time switch-over is planned for inclusion in a future OpenVMS release. C is not the only consumer of the TDF. An update to the OpenVMS FAQ is also planned, as this question crops up semi-annually.
|